C++ for Computer

Science and Engineering

(4th Edition)

Vic Broquard

Broquard eBooks
103 Timberlane
East Peoria, IL 61611
[email protected]

ISBN: 0-9705697-2-6
C++ for Computer Science and Engineering
Vic Broquard
Copyright 2000, 2002, 2003, 2006 by Vic Broquard

All rights reserved. No part of this book may be reproduced or transmitted in any form without
written permission of Vic Broquard.

Fourth Edition

ISBN: 0-9705697-2-6
Brief Table of Contents

1 Introduction to Programming
2 Numerical Processing
3 Additional Processing Details
4 Decisions
5 Files and Loops
6 Writing Your Own Functions
7 More on Functions
8 Character Processing and Do Case
9 Arrays
10 Using Arrays
11 Strings
12 Multidimensional Arrays
13 Structures

Appendix A: How to Use Microsoft’s Visual C++ .NET 2005 Compiler

Appendix B: How to Use Microsoft’s Visual C++ .NET 2002 Compiler
Appendix C: How to Use Microsoft’s Visual C++ Version 6.0 Compiler
Index
To all of my dedicated, persevering students,
and to L. Ron Hubbard, who taught me to “Simplify”
Preface
This book assumes you have no previous programming background. It uses a step-by-step
building block approach to gradiently learn how to solve computer science and engineering
problems in the C++ language.

Each chapter has three sections. Section A presents the basic theory and principles of the
current topic. Section B illustrates these basic principles by using applications that are often
found in computer science. Section C illustrates these basic principles by using applications that
may be found in the various engineering disciplines. You should study the basic theory Section A
and then study the appropriate application section. Of course, anyone can benefit by also
reviewing the other application area, since they are frequently interrelated.

The book comes with a self-extracting zip file containing all of the sample programs in
the book along with all of the test data required for the programming assignments.

At the end of each chapter are Design Exercises, Stop Exercises and Programming
Problems. Before you tackle any programming assignments, you should do both the Design and
Stop exercises. The Design Exercises are paper and pencil activities that assist in solidifying the
basic design principles covered in the chapter. The Stop Exercises cover the new syntax of the
language, illustrating many of the more common errors beginners make in coding the language. If
you dutifully do these two sets of exercises before you start in on your programming
assignments, you will have a much better chance of success with drastically lower frustration
level.

If you find any errors or have any suggestions or comments, please email me at
[email protected]

v
 Contents

Chapter 1 — Introduction to Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
What is a Computer? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Designing Solutions — the Cycle of Data Processing . . . . . . . . . . . . . . . . . . 10
Building a Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
The Steps Needed to Create a Program — or — . . . . . . . . . . . . . . . . . . . . . 13
How to Solve a Problem on the Computer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
The Early Retirement Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
The Mechanical Robot Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
The Mechanical Mouse Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Basic Computer Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
The C++ Language and the Hello World Program . . . . . . . . . . . . . . . . . . . . 23
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Chapter 2 — Numerical Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Variables and Constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Integer Versus Floating Point (Real) Numbers . . . . . . . . . . . . . . . . . . . . . . . . 40
Which Type of Data Do You Use for Which Variable? . . . . . . . . . . . 41
Definition of Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
The Issue of the Case of a Variable Name . . . . . . . . . . . . . . . . . . . . . . 44
Defining More Than One Variable in the Same Statement . . . . . . . . 44
Where Are Variable Definitions Placed in a Program? . . . . . . . . . . . 46
Initializing Variables and the Assignment Operator . . . . . . . . . . . . . . . . . . . 46
Multiple Assignments — Chaining the Assignment Operator . . . . . . . . . . . 48
Input of Data Values into Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Chaining Extraction Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Always Prompt the User Before Inputting the Data . . . . . . . . . . . . . . 50
Output of a Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
The setw() Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Insertion of Floating Point Numbers into an Output Stream - setprecision
and fixed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Labeling Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Math Operators — Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Precedence or Priority of Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

vi
 Constant Data Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Math Library Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
The Most Nearly Accurate Value of PI . . . . . . . . . . . . . . . . . . . . . . . . 62
Other Math Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Some Additional Insertion Operator Details . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Breaking a Complex Calculation Down into Smaller Portions . . . . . . . . . . . 63
Section B: Computer Science Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Cs02a — Ticket Prices for a Concert . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Section C: Engineering Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Engr02a — Pressure Drop in a Fluid Flowing Through a Pipe
(Civil Engineering) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

Chapter 3 — Additional Processing Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
The Complete Integer Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Which Type of Data Do I Use in My Program? . . . . . . . . . . . . . . . . . 85
How Integer Data Is Stored in Memory . . . . . . . . . . . . . . . . . . . . . . . . 86
Integer Variable Overflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
The Complete Floating Point Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Principles of Data Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Assigning Smaller Sized Integers to Larger Sized Integers . . . . . . . . 90
Assigning Larger Sized Integers to Smaller Sized Integer Variables
(The Typecast) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Calculations Involving Multiple Floating Point Data Types . . . . . . . 93
Mixed Mode Math . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Constants and Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Additional Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
The Increment and Decrement Operators . . . . . . . . . . . . . . . . . . . . . . 98
The Compound Assignment Operators . . . . . . . . . . . . . . . . . . . . . . . . 99
Section B: Computer Science Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
CS03a — Vote Tally Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Section C: An Engineering Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Engr03a—Calculating the Power Supplied to a Load
(Electrical Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

Chapter 4 — Decisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

vii
 Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
The Components of an If-Then-Else Decision Structure . . . . . . . . . . . . . . . 113
The If-Then-Else Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
The Test Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Nested Decisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Compound Test Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
The Logical Not Operator — ! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Data Type and Value of Relational Expressions — The bool Data Type . . 124
The bool Data Type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
The Most Common Test Condition Blunder Explained . . . . . . . . . . . . . . . . 126
The Conditional Expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
The Precedence of Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Testing of Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Section B: Computer Science Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Cs04a — Compute the Total Bill By Finding the Sales Tax Rate . . . . . . . . 131
Section C: An Engineering Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Engr04a — Quadratic Root Solver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

Chapter 5 — Files and Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Input Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
I/O Stream States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Testing for Goodness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Testing for Bad Data Entry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
The End of File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Closing a File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
The Iterative Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Loops That Are to Be Executed a Known Number of Times . . . . . . . . . . . 158
Loops to Input All Data in a File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Sentinel Controlled Input Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Keyboard Data Entry Sentinel Controlled Loops . . . . . . . . . . . . . . . 162
Menus as Sentinel Controlled Loops . . . . . . . . . . . . . . . . . . . . . . . . . 162
Primed Input Loops that Detect End of File . . . . . . . . . . . . . . . . . . . 163
A More Compact Loop That Detects End of File . . . . . . . . . . . . . . . 165
Applications of Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Application: The Summation of a Series . . . . . . . . . . . . . . . . . . . . . . 166
Counters and Totals — Grand Totals . . . . . . . . . . . . . . . . . . . . . . . . 167
Finding the Maximum and Minimum Values . . . . . . . . . . . . . . . . . . 170

viii
 Bulletproofing Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Creating Output Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
The Do Until Instruction — An Alternative to the Do While . . . . . . . . . . . 178
The Do Loop or for Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Efficient Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Nesting of Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
An Example of Nested Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Section B: Computer Science Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
Cs05a — Acme Ticket Sales Summary Program . . . . . . . . . . . . . . . . . . . . . 186
Cs05b — Calculating N! (N factorial) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Section C: Engineering Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Engr05a — Summation of Infinite Polynomials . . . . . . . . . . . . . . . . . . . . . . 194
Engr05b — Artillery Shell Trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
New Syntax Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212

Chapter 6 — Writing Your Own Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Principles of Top-Down Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Writing your own functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Step A. Define the Function’s Prototype . . . . . . . . . . . . . . . . . . . . . . 221
Step B. Define the Function Header . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Step C. Code the Function’s Body . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Step D. Invoke or Call the Function . . . . . . . . . . . . . . . . . . . . . . . . . . 226
A Second Example, calcTax() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
How Parameters Are Passed to Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
The Types, Scope and Storage Classes of Variables . . . . . . . . . . . . . . . . . . . 232
Registers and the Stack — a Bit of Computer Architecture . . . . . . . . . . . . 235
How a Function Returns a Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
More on the bool Data Type and Functions that Return a bool . . . . . . . . . 239
The Shipping Cost Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
Functions that Return No Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
Where Should Error Messages Be Displayed? . . . . . . . . . . . . . . . . . . . . . . . 241
Controlling Leading 0's on Output — the setfill() Function . . . . . . . . . . . . 242
Inputting Integers that have Leading Zeros — The dec Manipulator
Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
Section B: Computer Science Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Cs06-1 — Employee Payroll Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Section C: An Engineering Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Introduction to Numerical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249

ix
 Numerical Analysis: Root Solving, the Bisection Method . . . . . . . . . . . . . . 251
Engr06a — Root Solving, the Bisection Method . . . . . . . . . . . . . . . . . . . . . . 254
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266

Chapter 7 — More on Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Reference Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
The Need for Reference Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
The Reference Variable Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
The Static Storage Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
The Global/External Storage Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
Using Global Variables in Other Cpp Files — the extern Keyword . . . . . . 287
Where are Global and Static Variables Actually Stored? . . . . . . . . . . . . . . 288
Philosophy on the Use of Global Variables . . . . . . . . . . . . . . . . . . . . . . . . . . 289
How to Pass iostreams to Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
Section B: Computer Science Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
Cs07c — Acme Ticket Sales Report — a Multi-page Report
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
Cs07a — Multiple Level Control Break Processing . . . . . . . . . . . . . . . . . . . 299
Cs07b — Summary Reports Based upon Control Break Processing . . . . . 307
Section C: Engineering Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
Bisection Revisited — Writing a Generic Bisection Function . . . . . . . . . . . 310
Engr07a — Using a Generic bisect() Function . . . . . . . . . . . . . . . . . . . . . . . 312
Engr07b — Molar Volume of Non-Ideal Gases . . . . . . . . . . . . . . . . . . . . . . 315
Faster Alternative Root Solving Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
The Regula Falsi Root Solving Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
Engr07c — Molar Volume of Non-Ideal Gases — Using Regula Falsi Method
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
Newton’s Method of Root Solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
Engr07d — Molar Volume of Non-Ideal Gases — Using Newton’s Method
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
The Secant Method of Root Solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Engr07e — Molar Volume of Non-Ideal Gases — Using the Secant Method
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
Summary of Root Solving Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344

Chapter 8 — Character Processing and Do Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351

x
 Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
The Processing of Character Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Defining Variables to Hold a Character of Data . . . . . . . . . . . . . . . . 351
Inputting Character Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
Using the Extraction Operator to Input a Character . . . . . . . . . . . . 352
Hexadecimal Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Using the get() Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
Output of Character Data — the put() Function . . . . . . . . . . . . . . . 355
How Are Character Data Stored? . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
The Escape Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
Numbers and Letters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
The Character Processing Functions . . . . . . . . . . . . . . . . . . . . . . . . . 361
Basic08a — A Word Counter Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
The Do Case Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
More on the break Statement and the continue Statement . . . . . . . 370
Enumerated Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
Section B: Computer Science Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
Cs08a — Inventory on Hand Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
Cs08b — Inventory on Hand Program — Using a Generic processFile()
Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
Section C: Engineering Examples — Numerical Integration . . . . . . . . . . . . . . . . . 391
The Trapezoid Method of Numerical Integration . . . . . . . . . . . . . . . . . . . . 391
Engr08a — Numerical Integration with the Trapezoid Rule . . . . . . . . . . . 394
Integration Using Simpson’s Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
Engr08b — Numerical Integration with Simpson’s Rule . . . . . . . . . . . . . . 397
Engr08c — Using Menus to Control Program Operation . . . . . . . . . . . . . . 399
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409

Chapter 9 — Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Definitions and Need for Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Defining Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Accessing Array Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
Methods of Inputting Data into an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
Method A: Inputting a Known Number of Elements . . . . . . . . . . . . 417
Method B: Inputting the Number of Array Elements To Be Input
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
Method C: Inputting an Unknown Number of Elements Until EOF Is
Reached . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
Working with Arrays — The Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . 420

xi
 Working with arrays: the Output Process . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
Initializing an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
Passing Arrays to Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
Section B: Computer Science Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
Cs09a — Sales Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
Section C: Engineering Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
Engr09a — Vector Coordinate Conversions . . . . . . . . . . . . . . . . . . . . . . . . 437
Engr09b — Plotting Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453

Chapter 10 — Using Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
Using an Array for Direct Lookup Operations . . . . . . . . . . . . . . . . . . . . . . . 458
Parallel Arrays and Sequential Searches — Inquiry Programs . . . . . . . . . 459
Inserting Another Element into an Unsorted Array . . . . . . . . . . . . . . . . . . . 461
Ordered (Sorted) Lists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
Inserting New Data into a Sorted List . . . . . . . . . . . . . . . . . . . . . . . . 464
Sorting an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
Section B: A Computer Science Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
Cs10A — Merging Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476
Section C: An Engineering Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
Engr10a — Statistical Computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
Least Squares Curve Fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498

Chapter 11 — Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
Defining Character Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
Inputting Character Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
Method A — All Strings Have the Same Length . . . . . . . . . . . . . . . 509
Method B – String Contains Only the Needed Characters, But Is the
Last Field on a Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
Method C — All strings Are Delimited . . . . . . . . . . . . . . . . . . . . . . . 512
Outputting Character Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
Passing a String to a Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514
Working with Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514
The String Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
How Could String Functions Be Implemented? . . . . . . . . . . . . . . . . . . . . . . 522

xii
 Section B: A Computer Science Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
Cs11a — Character String Manipulation — Customer Names . . . . . . . . . 523
Section C: An Engineering Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
Engr11a — Weather Statistics Revisited . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 538
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 540

Chapter 12 — Multidimensional Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
Defining Multidimensional Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
Physical Memory Layout Versus Logical Layout . . . . . . . . . . . . . . . . . . . . . 547
Initialization of Multidimensional Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
Passing Multidimensional Arrays to Functions . . . . . . . . . . . . . . . . . . . . . . 549
Loading a Multidimensional Array from an Input File . . . . . . . . . . . . . . . . 549
Working with Multidimensional Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
Section B: A Computer Science Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 557
Cs12a — Arrays of Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 557
Section C: Engineering Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
Matrix Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
Matrix Math Operations Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 565
Mathematical Theorems of Determinants . . . . . . . . . . . . . . . . . . . . . 567
The Gauss Method for Solving a System of Linear Equations . . . . . . . . . . 568
Gauss-Jordan Method of Solving Simultaneous Linear Equations . . . . . . 570
Engr12a — Aligning the Mirrors of a Telescope (Astronomy) . . . . . . . . . . 575
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581

Chapter 13 — Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588

Section A: Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
Defining Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
Creating Instances of a Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 590
How are Structures Initialized? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
How are Structure Members Accessed? . . . . . . . . . . . . . . . . . . . . . . 592
Rules of Use for Structure Variables . . . . . . . . . . . . . . . . . . . . . . . . . 593
User-Written Header Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
Binary Files and Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
Mechanics of Binary Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
Section B: Computer Science Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 600

xiii
 Cs13-1 — Credit Card Application with Sorting . . . . . . . . . . . . . . . . . . . . . 600
Cs13-2 — Writing a Binary File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
Cs13-3 — Reading a Binary File — . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612

Section C: An Engineering Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 616

Engr13a — Weather Statistics Revisited . . . . . . . . . . . . . . . . . . . . . . . . . . . . 616
Design Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 621
Stop! Do These Exercises Before Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 622
Programming Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624

Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler . . . . . . . . . . . . 631
C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
Making a New Programming Solution —
I Am Building a New Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
Continue to Work on an Existing Program — Starting Visual Studio . . . . . . . . . . 637
Bringing Files From Home to School . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638
Building a New Project in Which the Cpp Files Already Exist . . . . . . . . . . . . . . . . 638
Compiling and Running Your Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 639
Executing a DOS Console Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 641
Getting Source File Printouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 641
Getting a Printed Copy of the Program Execution Output . . . . . . . . . . . . . . . . . . . 642
Case 1: The Entire Output Fits on One Screen Without Scrolling . . . . . . . 642
Case 2: Using cout and There Are Too Many Lines To Capture With a
Screen Shot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 642
Case 3: Using an Output File Stream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 643
Visual Studio Operational Tips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 644
Debug Versus Release Builds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645
A Primer on Using the Debugger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646

Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
Using Microsoft’s VC 7 (.NET) Compiler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
C++ DOS Console Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
Making a New Programming Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
I Am Building a New Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
Continue to Work on an Existing Program — Starting Visual Studio . . . . . . . . . . 658
Bringing Files From Home to School . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
Building a New Project in Which the Cpp Files Already Exist . . . . . . . . . . . . . . . . 659
Compiling and Running Your Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 660
Executing a DOS Console Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662
Getting Source File Printouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662
Getting a Printed Copy of the Program Execution Output . . . . . . . . . . . . . . . . . . . 663
Case 1: The Entire Output Fits on One Screen Without Scrolling . . . . . . . 663

xiv
 Case 2: Using cout and There Are Too Many Lines To Capture With a
Screen Shot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663
Case 3: Using an Output File Stream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664
Visual Studio Operational Tips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665
Debug Versus Release Builds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 668
A Primer on Using the Debugger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669

Appendix C — How to Use Microsoft’s Visual C++ 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 676

Step 0. Get Organized . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 676
Step 1: Building the Program Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 678
Step 2. Transporting Programs to and from School Computers . . . . . . . . . . . . . . . 682
Step 3. Opening an Existing Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683
Step 4. Compiling the Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683
Step 5. Handling Compile Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684
Step 6. Where Is the Executable File (*.exe) Located? . . . . . . . . . . . . . . . . . . . . . . . 684
Step 7. Running The Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685
Step 8. Program Debugging and Execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
Step 9. The Help System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689
Step 10. Some VC6 Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 690
Step 11. Getting the hardcopy documentation for programs to hand in . . . . . . . . 691

xv
Introduction to Programming 1

Chapter 1 — Introduction to Programming

Section A: Basic Theory

Introduction
There are few areas of study that have more specialized terms and abbreviations to know than in
the computer field. As you study the text, key words and abbreviations are given good action
definitions as they occur. If a section of a chapter is blank in your mind, stop and look back
earlier and see if you can find a word or abbreviation that is not fully understood. Once the word
is fully understood, reread that blank section and it should now make sense.

At the end of each chapter are two practice sections designed to solidify the theory just
studied. The “Design Exercises” enhance your problem solving skills. The “Stop! Do These
Exercises Before Programming” exercises illustrate many of the common errors that a
programmer can make. Thus, if you work these exercises before you begin the actual
programming problems, you should make far fewer goofs, should have a much more enjoyable
time doing the programming, should greatly reduce the amount of time it takes to do your
assignments and should definitely lower the frustration level.

What is a Computer?
A definition of a computer is an electronic device that can input data, process data and output
data, accurately and at great speed. Data are any kind of information that can be codified in some
manner and input into the computer. Normally, we think of data as facts and numbers such as a
person’s name and address or the quantity or cost of an item purchased. However, data can also
be graphical images, sound files, movies and more.

A computer is capable of inputting information such as the quantity ordered and the cost
of that item. Processing data means to do something with it. Often we think of processing as
performing some kind of calculations. If the quantity and cost have been input, then the obvious
calculation would be to multiply cost times quantity to produce the total cost. However,
processing data can mean more than just calculations. Perhaps you have entered the series of
friends and their phone numbers. Processing the data can also mean sorting the friends’
information into alphabetical order by last names. Finally, to be useful, the computer needs to be
able to output information, the results, to the user in an accurate and timely manner. The user is
anyone that is making use of the results that the computer is producing.
Introduction to Programming 2

However, an abacus can input, process and output data. There must be more in this
definition. It is the qualifier, accurately and at great speed, that makes computers so powerful.
Let’s look at each of these in turn.

A computer is accurate and reliable; they do not make mistakes. But it did not used to be
this way. Back in the first generation of computers in the early 1950's, computers were built from
some 18,000 vacuum tubes. And tubes frequently burned out forcing their replacement.
Statistically, when one has 18,000 of these tubes in one machine, one expects one tube failure
every fifteen seconds! This is where the idea that computers are not reliable has its genus. There
was no reliability in those days. However, with modern computers now built from silicon and
germanium integrated circuits or chips (a device consisting of a number of connected electronic
circuit elements such as transistors fabricated on a single chip of silicon crystal), the failure rate
is about one chip failure ever thirty-three million hours of operation. Of course, if you drop a
computer or run it during an electrical storm, you can significantly shorten its lifetime. Thus,
modern computers are reliable. However, the software that runs on them is not necessarily error
proof.

The other qualifier is at great speed. Just how fast is a computer? Let’s compare the time
that it takes various computers to add two integer whole numbers. The unit of time measurement
is the nanosecond which is 10-9 of a second, or 1/1,000,000,000 of a second. Electricity travels
approximately 11.4 inches down a copper wire in a nanosecond. The following chart is an
approximation of how long it takes some computers to add two numbers. (MHz is short for
megahertz or a million cycles per second, GHz is gigahertz (1024 MHz), and ns is
nanoseconds.)
IBM-PC 4.77 MHz 600 ns
386 33 MHz 60 ns
486 100 MHz 10 ns
Pentium 200 MHz 5 ns
P-3 500 MHz 2 ns
P-4 2 GHz .5 ns

In other words, if you have one of the newer Pentium-3 500 MHz machines, in one
second the computer could perform many billions of additions. (Note that the addition instruction
is one of the fastest instructions the computer has. Many other instructions take substantially
longer to perform.)

Thus, it is the ability of the modern computer to perform reliably and to perform at great
speed that has made it so powerful.
Introduction to Programming 3

Computers have a fixed set of instructions that they can perform for us. The specific
instruction set depends upon the make and model of a computer. However, these instructions can
be broadly grouped into four basic categories:
Math instructions
Comparison instructions
Movement of data instructions
Input and output instructions

When one thinks of math instructions, the add, subtract, multiply and divide operations
immediately come to mind. However, for a mathematician, there are more complex math
operations as well, such as finding the trigonometric sine of an angle or the square root of a
number. Comparison instructions permit the computer to tell if one number is greater than, less
than or equal to another number. The computer can move data from one location in its memory
to another area. And of course, the computer can input and output data.

And that is all that a computer knows how to do. I sometimes joke that a computer is
basically a “moronic idiot.” That is, it is an “idiot” because of its limited instruction set, in other
words, what it knows how to do. The “moronic” adjective comes from the fact that the computer
always attempts to do precisely what you tell it to do. Say, for example, you tell the computer to
divide ten by zero, it tries to do so and fails at once. If you tell the computer to calculate a
person’s wages by multiplying their hours worked by their hours worked, say, forty hours this
week, the computer accurately and at great speed does the multiply instruction, and outputs their
pay as $1600!

Thus, we have this rule: If you tell the computer to do something stupid, the computer
accurately and at great speed does that stupid action! Your idea of a computer either
malfunctioning or making a mistake is likely coming from this aspect.

What is a program? A computer program is a series of instructions that tell the computer
every step to take in the proper sequence in order to solve a problem for a user. A programmer
is one who writes the computer program. When the computer produces a wrong or silly result, it
can be traced to an improper sequence of instructions or incorrect data being input to the
program. That is, the responsibility or blame lies on either the original programmer who wrote
out the instructions for the computer to follow or the user who has entered incorrect data.

For example, the latest Mars explorer satellite, after flawlessly traveling all the way to
Mars, disintegrated on attempting to go into orbit around the planet. The reason NASA
discovered is that the computer program controlling the satellite expected measurements to be in
English units and someone supplied those measurements in the metric system.

Thus, I have a new term for programs that have one or more errors in them — “mostly
working software.” When a program has an error in it, that error is often called a “bug.” And the
process of getting all of the errors out of a program is called debugging. The term originates in
Introduction to Programming 4

the first generation of computers when someone removed a fly that had gotten into the computer
circuitry and shorted it out - they were “debugging” the computer. In fact, mostly working
software is a pet peeve of mine. Mostly working software — a program with one or more errors
in it — is indicative of a programmer who has not done their job thoroughly for whatever
reason. What would you think about having an operation done by a mostly working surgeon?

Designing Solutions — the Cycle of Data Processing

Perhaps the single most important aspect of solving a problem on the computer is the initial
design phase in which one lays out with paper and pencil the precise steps the computer must
take. Nearly every significant program follows the same fundamental design and it is called the
Cycle of Data Processing, Figure 1.1.

The Cycle of Data Processing is Input, Process, Output. First the computer must input a
set of data on which to work. Once the data has been input into the computer, it can then process
that data, often performing some calculations on that information. When the calculations are
finished, the computer outputs that set of data and the results.

For example, suppose that we wanted to write a program that would calculate someone’s
wages. First, the computer must be instructed to input the person’s hours worked and their pay
rate. Next, the computer uses the values it has just input to calculate the wages. Now that the
wages are known, the computer can output the answer.

The Cycle of Data Processing is called IPO for short. IPO is the most basic design of a
program. Thus, when you are confronting a computer problem to solve, IPO is the starting point!
Input a set of information first. Then do the requisite processing steps using that information.
Last, output the results.

Also notice that in general, once that set of data and results have been output, the program
would repeat the entire process on the next set of data until there are no more sets of data to be
processed. It will be several chapters before we can implement all these steps.
Introduction to Programming 5

Any deviation from the IPO sequence generally yields silly results. Suppose that someone
tried to write a program to calculate a person’s wages by doing OPI instead? That is, have the
program output the answer before it knew what the hours that were worked or who was the
person for whom the wages were to be found! Nonsense. Have the program calculate the pay
before it has input the hours worked? How can it? You see, worked in reverse, it just makes no
sense at all.

Occasionally, by accident someone writes an IP program. That is, it inputs the data and
does the calculations, but fails to output the result. For example, you want a soda, so you input
your quarters into the pop machine, Input. You press the button and the internal machinery makes
lots of noise as it processes your request. But no can of soda ever appears, no Output! Or take a
PO program. You walk by the soda machine and all of a sudden you hear it making noises and a
can of soda appears! Or you walk by a piano and it starts playing — spooky when there is no
input! Occasionally, someone writes an O program by accident. Suppose the program needed to
print some headings at the top of a page and suppose the programmer made a booboo and
continually printed headings over and over and over. You would have an O program. Or take a P
program, the program just calculates, calculates, calculates, endlessly. This is sometimes called
an infinite processing loop.

Whenever you are trying to design a program, remember that it usually must follow the
IPO cycle. Now there are some exceptions, but they are rare. A possible exception might be
producing a mathematical table of trigonometric function values. For example, suppose that you
wanted to produce a table of the values of the sine and cosine of all angles from zero to ninety
degrees. In such a case, there would be no input, just a process-output series as the program
calculated each set of results and displayed them.

Building a Program
The computer internally operates on the binary number system. In the binary number system,
the only valid digits are 0 and 1. For example, if you add in binary 1 plus 1, you get 10, just as if
you added 9 + 1 => 10 in the decimal or base 10 system.

Why does the computer use binary? Electronic circuits can either have some electricity in
them or not. If a circuit element, such as a transistor, has electricity, it can be said to contain a 1;
if none, then a 0. This is the basis for computer operations. The actual instructions that make up a
program are all in binary, long strings of binary digits. But no one wants to try to write out these
long tedious series of 1's and 0's to try to direct the computer to solve a problem. Rather a high-
level language is used. In this case, we use the C++ language.

In a high-level language, we use various symbols and mathematical notations to create the
program which is called the source program. A source program is the precise series of high-
Introduction to Programming 6

level language statements in the proper order for the computer to follow to solve the problem.
For us, that source file has the file extension of .cpp.

Another piece of software called the compiler inputs our source program and converts it
into the machine language, binary equivalent. If you make some typos, these show up as syntax
errors when the compiler tries to convert the source program. A syntax error just means that you
have coded the C++ instruction incorrectly. When you first compile a program and suddenly see
a large number of compile errors, don’t panic. Often it is just one small syntax error that cascades
into many errors. Fix the original error and the others are automatically fixed. The output from a
successful compile run is called an object file with the .obj file extension. The obj file contains
the binary machine language instructions to control the computer in solving your problem.
However, it is not the final program; object files are missing something.

Although we know nothing about the C++ programming language at this point, we can
still understand what is missing in the object files. Suppose that as part of your program you
needed to input some value, then compute the square root of that value and lastly print out that
original number and its square root. Ignoring for the moment the input and output situation, how
can you calculate the square root of any given number? If you have a strong math background,
you probably are beginning to think of a method for doing just this. However, the C++ language
has already provided that coding necessary to calculate the square root of any number for us.
Why reinvent the wheel? We should use the solution provided by the compiler manufacturer.
These short solutions to common needs such as finding the square root of a number are called
functions.

A function is a subprogram, a collection of instructions that does a very precise action.

Our program invokes or calls the function. When we do so, the computer temporarily halts
execution of our instructions and goes to the instructions of the function and carries out the
function’s instructions. When the function has completed its task, it issues a return instruction.
The return instruction instructs the computer to go back to from where the function was called
and resume execution there. So in short, our program calls the square root function which then
does its thing and returns back to us with the answer for our program to use as we wish.
Functions are a vital part of any programming language.

So in the above example of inputting a number, finding its square root and then printing
it, the object file as created by the compiler does not have in it the provided functions that are
built into the language for our use. Specifically in this case, it is lacking the input, output and the
square root functions. These are located in the compiler’s Lib folder. In order to make the final
executable program, the .exe file, another piece of software called the Linker, must be run. The
Linker program inputs our object files and finds all the needed system functions, such as the
square root function, and builds the actual .exe file for us.

Finally, in order to make the entire process easy for us, from the initial editing or typing
of the cpp file, through compilation and linking phases, most compiler manufacturers provide an
Introduction to Programming 7

integrated development platform or environment known as an IDE. An IDE is simply a software

application that provides a convenient common environment to create, compile, link and test
execute our programs.

However, the price that the IDEs command for all this convenience is a project file. A
project file (also called a solution in .NET) is a compiler manufacturer specific file(s) that tell
the IDE everything it needs to know in order for it to build the final exe file. For example, it
needs to know the name and location of our source file(s), where to place the exe final program,
where the system libraries are located that contain all the system functions such as the square root
function, and so on.

The Steps Needed to Create a Program — or —

How to Solve a Problem on the Computer
The following steps represent an optimum procedure to follow to solve any problem on the
computer. Every time you begin to tackle another programming assignment, this IS the procedure
you should follow slavishly. In fact, I am letting you in on an inside programmer’s secret. This
series of steps, if followed precisely and honestly, results in the finished program in perfect
working order with the least amount of your time spent on it and with the least frustration on
your part. The reverse is true as well. If you want to spend vast amounts of time trying to get a
programming assignment completed with maximal frustrations on your part, simply completely
ignore these steps.

Here is the tale of one of my former students. She actually believed me about these steps
and followed them slavishly. In her programming class, whenever a new assignment was handed
out, she was known as the last person to ever get the problem coded into the computer, to get the
cpp source file created. She did get teased about this, but only briefly. She was always the very
first person to have the assignment completed and ready to turn in! Soon, everyone in the class
was turning to her for “help.” She was looked upon as a programming goddess.

Now that I have your attention, what are the steps to developing a program?

Step 1. Fully understand the problem to be solved. Begin by looking over the output, what
the program is supposed to be producing, what are the results? Then look over the input that the
program will be receiving. Finally, determine what general processing steps are going to be
needed to turn that input into the required output. If something about the problem is not clear,
usually your instructor can assist you in understanding what is to be done. It is pointless to try to
go on to the subsequent steps, if you are not 100% certain what must be done.

Part of this step of understanding the problem involves determining the algorithm to be
used. An algorithm is a finite series of steps for solving a logical or mathematical problem. In
Introduction to Programming 8

computer programming, there are a large number of algorithms or methods that have been
designed to assist us. Many of the Computer Science examples are illustrating common
algorithms often needed in such programming. Likewise many of the Engineering applications
are concerned with numerical analysis algorithms. Some are used to find statistical averages,
others to find roots of equations (where the graph crosses the x-axis), some for numerical
integration, and so on. Part of learning how to program problems on the computer is learning
about algorithms or methods to use.

Step 2. Design a solution using paper and pencil. This process involves two distinct
activities.

The first action is to design what function(s) would best aid in the solution. Note these are
functions that you must write, not those like the square root that are provided by the compiler
manufacturer. This process is greatly aided by a design technology called Top-down Design
which is covered in Chapter 6 where you first learn how to write your own functions. Until then,
no additional functions of our own design are needed and this action can be skipped until then.

The second action is crucial. Write out on paper the precise steps needed to solve the
problem in the precise sequence. This is often called pseudocode. It is done by using English and
perhaps some C++ like statements. You are trying at this point to say in English the correct
sequence of steps that must be followed to produce the result.

Even though we know nothing about C++ at this point, given only the Cycle of Data
Processing, we can still solve problems by writing out the pseudocode for them. Let’s do so now.
Suppose that the problem is to ask the user to input a number and then display the square root of
that number. Here is some beginning pseudocode to solve this simple problem.
Display on the screen: “Enter a number: “
Input the user’s number and store it in Number
Let Answer equal the square root of Number
Display on the screen Answer

Notice one crucial aspect of the solution above — in bold print. I have indicated where
in the computer’s memory to place the user’s inputted number; it is going to be placed into a
memory area known as Number. I have also shown that the result is going to be placed in a
memory area known as Answer. Both Number and Answer are known as program variables. A
variable is a memory location in which to store something. It is vital that the variable names are
100% consistent from line to line in your solution.

One common problem all programmers face is slightly different names. For example,
suppose I had sloppily coded this solution as follows.
Display on the screen: “Enter a number: “
Input the user’s number and store it in Number
Let Answer equal the square root of Num
Introduction to Programming 9

Display on the screen Ansr

Remember that the computer is an idiot. It is going to try to do precisely what you tell it to do. In
the above coding, the user’s data in input and stored in a variable called Number. However, the
variable used in the next line is not Number but Num. To us, it is obviously referring to
Number, but to the compiler and the computer, Number and Num are two completely different
things! Ditto on the result variable. Answer contains the square root return value, but I try to
display the contents of the variable Ansr — a completely different name! Both yield instant
compiler errors or produce erroneous garbage results. This then brings us to the most important
step in this entire process!

Step 3. Thoroughly desk check the solution. Desk check means to play computer and
follow slavishly and precisely the steps written down in the solution. You are looking for errors
at this point. When you desk check, you must learn to play the role of a moronic idiot. That is,
you do precisely what is written down, not what should be there, not what was intended, not what
ought to be there — just what is there, as it is. To desk check, one really needs to draw a picture
of the memory of the computer and place the variables as boxes in it so you can write in them.
Let’s see how the above incorrect solution could be desk checked. First we construct a picture of
memory with all the variable names found in the solution and place a ??? in each box as shown
in Figure 1.2.

Figure 1.2 Main Storage —Initial Setup

Then, as you step through each line of the solution, make needed changes in the boxes.
Assume the user enters 100. The square root is 10. But what does my erroneous coding do?
Following the precise steps, we get the following results as shown in Figure 1.3.

Figure 1.3 Main Storage —Wrong Results

Obviously, the solution is wrong. Here is how the correct version would be desk checked.
Again the starting point is to draw boxes to represent all the variables in the solution and give
them their initial values of ??? or unknown as shown in Figure 1.4.
Introduction to Programming 10

Figure 1.4 Main Storage —Correct Initial Setup

And when one has gone through the series of pseudocode steps, the following results as
shown in Figure 1.5.

Figure 1.5 Main Storage —Correct - Final Results

The benefits of desk checking cannot be undervalued! The whole purpose of desk
checking is to find all errors in the solution. Do not go on to Step 4 until you have thoroughly
tested the solution. The key word is thoroughly. This is the point that so many veterans fail to
do. Thoroughly means 100% completely under all conditions, all possibilities and so on. If you
mostly desk check your program, then you have a mostly working program!

Step 4. Code the solution into the programming language, C++ in our case. With the
pseudo coding and memory drawings at hand, it becomes a fairly simple matter to convert the
solution into a C++ source program. Your biggest challenge at this point is to get the syntax
correct.

Step 5. Compile the program. If there are any errors found by the compiler, these are
called syntax errors. Again a syntax error is just incorrect coding. Just fix up the mistyping and
recompile. Once you have a clean compile and built the program (and have an executable file),
go on to the next step.

Step 6. Test the program with one set of data. Try inputting one set of test data only.
Examine the output and verify it is correct. If you have done a good job with Step 3, Desk
Checking, there are no surprises; the results are correct. If they are not correct, this is a more
serious problem. An error here is called a runtime logic error. If the results are not correct, then
you have missed something. It is back to Step 1 or 2 to figure out what was missed. After you
discover what was missed, you then need to fix up the solution and re-desk check, then re-code,
then recompile and try the single set of test data again. Obviously, you cannot go on to the next
step until you have the program producing the right results for one set of test data.
Introduction to Programming 11

Step 7. Thoroughly test the program. At this point, one tests the program thoroughly and
completely. Often the problems in this text have some supplied test data sets you must use. These
are designed to thoroughly test your program. If in testing, you discover another error, it is again
a logic error. And it is once more back to Step 1 and 2. Then you have to redo all the steps in
order to get back here to this step. Now, if you have done a thorough job of desk checking your
pseudo coding, there are no more surprises — the program works perfectly no matter what tests
you do. Once more, you cannot go on to the next step until the program is working perfectly.

Step 8. Put the program into production. In the real world, this means that the program is
given to the users who now run it to solve their problems. In the case of student programs, they
are handed in to be graded by the instructor who plays the role of the user. I guarantee you that
users will indeed thoroughly test that program; users are known for doing all sorts of unexpected
things with programs! What happens if the user finds an error? It is once again all the way back
to Steps 1 and 2 once more. But if you have done a thorough job of desk checking and testing the
program itself, the users will find nothing wrong.

In the industry, dollar costs have been calculated in this bug finding process. If it costs the
company $1 to locate and find an error during Step 3 Desk Checking, then if that bug is found in
Step 6, it costs $10. If that same error is found during Step 7, the thorough testing phase, it costs
the company $100. However, if the program goes into production and the users find the error,
then it costs the company $1,000! Hence, there is a major incentive to find all the program’s
errors early in the development cycle.

The Early Retirement Program

Let’s apply the how to solve a problem logic to a simple problem. The Acme company wants to
have a listing of all their employees that might consider a new early retirement plan. The input
comes from their employee file which consists of one line per employee which contains the
following information: the employee’s id, their age and the years they have been employed at
Acme. To be considered a candidate, the employee must have worked for Acme for ten years and
be at least 55 years old. The report should display the id number, age and years employed. The
last line should contain the total number of possible candidates.

Looking over the input lines, three variables or fields are needed to store the incoming
data, id, age and years_employed. The output consists of these three fields. However, the last
line is a count which we can call total_number. Each time we discover that an employee is
qualified for early retirement, we need to display their information and add one to the
total_number. Our Main Storage diagram contains the three input fields and the total_number
and is shown in Figure 1.6.
Introduction to Programming 12

Figure 1.6 Main Storage —Early Retirement

Here is the English solution.

set total_number to 0
input an id, age and years
as long as there is a set of data, do the following
if the age is greater than or equal to 55 and
the years is greater than or equal to 10, then do the following
display the id, age and years
add one to total_number
end of the then clause
try to input another id, age and years
end of the do the following loop
display “The total number of possible early retirement candidates is ”
display total_number

We can test the program with the following information.

123 60 21
234 44 10
266 55 10
275 55 9
284 56 9
345 25 5
344 34 12
And the output of the program is
123 60 21
266 55 10
The total number of possible early retirement candidates is 2
Introduction to Programming 13

The Mechanical Robot Problem

To illustrate these design principles and to help you to get the feel for what is needed to be able
to write programs, consider the Mechanical Robot Problem. Your company has been given a
multimillion dollar robot in the shape of a person. For a demo test, you are to write out the
solution of the following problem. The robot is initially seated an unknown distance from the
wall. It has sensors in its fingers so that if its arms are raised, it fingers can tell if it is touching
any obstruction, such as a wall. You are to instruct the robot to stand up and walk forward until it
finds the wall, turn around and retrace its steps until it reaches the chair, at which point it should
turn around and sit down.

Sounds simple when said in normal English. But the problem is that the robot does not
understand English. Rather, as a computer, it understands a basic set of instructions. Here are the
only commands the robot understands.
Stand up
Sit down
Raise arms
Lower arms
Turn around
Are your fingers touching anything? It replies yes or no.
Take one step (all steps are a uniform distance)
Set an internal counter to 0
Add one to the internal counter
Subtract one from the internal counter
Is the internal counter 0? It replies yes or no.
And these are the only commands it knows how to do. If you give it a command other than these
precise ones, it stands there and does nothing.

Your job is to use only these commands and write out a solution that will work with all
possible distances the robot might be from the wall. For simplicity, assume that the robot is
always an integral number of steps from the wall. That is, the robot distance from the wall could
be 0, 1, 2, 3, 4 or more steps. All steps are uniform in size. Thoroughly desk check your solution.
Be sure it works if the robot is 0 steps from the wall as well as 5 steps. Note that if the robot is 0
steps from the wall, it still has room to raise its arms, at which point its raised arms would be
touching the wall, if asked.

Be prepared to share your solution with others.

Introduction to Programming 14

The Mechanical Mouse Problem

A mechanical mouse must run through a maze. The maze has only four “cells.” Two
outside walls of the maze are fixed as shown in Figure 1.7

Figure 1.7 Mouse in a Maze Cell

Baffle walls may be erected on any of the dotted lines, but a maze is valid only if it meets
these conditions:
1. One (only one) entry point on the entry side of the maze.
2. One (only one) exit point on the exit side of the maze.
3. An open passage from the entry point to the exit point.
4. Two of the four sides are open; two are closed on each cell that must be traversed.
Figure 1.8 shows three valid mazes.

Figure 1.8 Three Valid Mouse Mazes

At the beginning, an operator will place the mouse on the entry side of the maze,
in front of the entry point, facing the maze. The instruction, “Move to the Next Cell,” causes the
mouse to move into the middle of the entrance cell.

After that, the job is to move from cell to cell until the mouse emerges on the exit
side. If the mouse is instructed to “Move to the Next Cell” when there is a wall in front of
Introduction to Programming 15

it, it hits the wall. In this case, there will be a sharp explosion, and both the mouse and
maze will disappear in a cloud of blue smoke (and the game is lost). Obviously,
the mouse must be instructed to test if it is “Facing a Wall?” before any “Move.”

Your assignment: Write out a sequence of these permissible instructions which navigates
the mouse through any valid maze. The only permissible instructions are the following.

The Mechanical Mouse’s Instruction Set

A. Physical Movement
Move to the Next Cell (the mouse will move in the direction it is facing).
2. Turn Right
3. Turn Left (all turns are made in place, without moving to another cell).
4. Turn Around

B. Logic
1. Facing a Wall? (through this test the mouse determines whether there is
a wall immediately in front of it; that is, on the border of the cell it is
occupying, and in the direction it is facing).
2. Outside the Maze?

If the mouse is outside the maze, it can also make the following decisions:
3. On the Entry Side? (If so, it gets frustrated and detonates in an explosion as well.)
4. On the Exit Side?

When your solution works on the above three mazes, test it on this last maze, Figure 1.9.

Figure 1.9 Test Maze

Introduction to Programming 16

Basic Computer Architecture

In order to effectively write programs on the computer, some basic knowledge of computer
architecture is required. The computer can be viewed as having two major components, the
Central Processing Unit or CPU and main storage or memory. The CPU handles all of the
mathematical operations, comparisons and input and/or output actions. I/O is often used to mean
input and/or output operations. That portion of the CPU that carries out the mathematical
operations and the comparisons is called the ALU, arithmetic and logic unit.

Main storage or memory is a vast collection of storage units called a byte. A byte is
capable of storing one character of information. A byte is composed of eight connected bits. A
bit is the tiniest storage element and consists of a circuit element that can be either on or off,
representing a one or zero. A bit could be a tiny transistor located on a computer chip, for
example. A single bit cannot be accessed directly; rather memory is accessed in terms of one or
more bytes at a time. The term 1K or kilobyte represents 1024 bytes. It is 210 bytes which is why
it is not an even 1,000 bytes. The term 1M or megabyte represents 1024K. Personal computers
now typically have between 64M and 256M of main memory.

The computer can read and write the contents of a byte. But in order to do so, it must
specify which byte is to be referenced. Bytes are located by their memory addresses. The first
byte in memory is given the address 0. The next sequential byte is at address 1, and so on, rather
like post office box numbers. However, no two bytes can ever have the same address. Each is at
a distinct location. See Figure 1-9a below.

When data is to be input into the computer, it must be placed into some location in its
memory. When data is to be displayed on the screen, for example, that data normally comes from
some memory location. From the point of view of the high-level languages, such as C++, these
memory locations are known as variables. Some variables might occupy only a single byte, for
instance the letter grade you receive for the course. Other variables occupy many consecutive
bytes, such as a person’s name or address. Some kinds of variables always occupy the same
number of bytes; the numerical types of data are a prime example.

When the power to the computer is turned off, the contents of the computer memory bytes
are lost permanently. When power is subsequently turned back on, all of the main memory bytes
are reset to zero. This kind of computer memory is known as RAM, random access memory;
RAM is normal memory which can be both read and written — that is we can store something
in memory and then later retrieve it back. Some computers have a small amount of ROM, read-
only memory. This specialized type of memory has some permanent information stored or burned
into it so that when power is reapplied, the contents reappear. ROM is used to store parts of the
computer’s operating system code in some PCs. The key point is that data stored in memory is
gone once the program finishes its execution.
Introduction to Programming 17

Attached to the computer are many I/O devices. The keyboard is an input device while
the display screen, the CRT (cathode ray tube), is normally an output device. Floppy disks and
hard disk drives are called auxiliary storage devices because they can store vast amounts of data
on a semipermanent basis. Typically, programs read files of data stored on disk and can write
files of results back to disk.

Figure 1-9a Parts of a Computer

The C++ Language and the Hello World Program

The C programming language was developed at Bell Labs in 1972 by Dennis Ritchie. C as a
language makes extensive use of functions. The concepts of structured programming were
pioneered during this era. Structured programming defines specific ways that computer
instructions ought to be organized. Instead of coding instructions in any manner that suited the
programmer, structured programming dictates that all the instructions are organized into one of
three main groups: the sequence structure, the decision structure and the iterative structure.

The sequence structure represents one or more instructions that are to be executed one
after the other in the sequence they are written. The short program to calculate the square root of
a number was one sequence structure with four instructions in it.
Introduction to Programming 18

The decision structure allows one to ask a question that can be answered yes/no or
true/false. If some question is true, then the program can execute a series of instructions that are
only done when the question is true. If it is false, the computer can optionally execute a different
series of instructions.

The iterative structure performs a series of instructions over and over, a loop in other
words, while some condition is true. These are shown in Figure 1.10.

Figure 1.10 The Three Structured Programming Sequences

It has been mathematically proven that any problem that can be solved on the computer
can be solved using only these three organizations of instructions.

Over the years, it was realized that, while exceedingly powerful, C had some aspects that
made it hard to learn and error prone. Further, while complex problems were successfully broken
down into smaller functional units in C, the idea of treating the data and the functions that
operate on that data as an object or entity led to the development of C++.

Bjarne Stroustrup, also at Bell Labs, in 1985 developed the C++ language as an extension
of the C language. C++ encompasses most of the C language and adds new features to bypass the
more error prone coding sequences of C and added support for working with objects, or Object
Oriented Programming, OOP for short.

The adage you must learn to crawl before you can walk and run holds true. Before you
can dive into the OOP portion of the language, you must master the basics, that is, the C portion.

In 1998, the C++ language was formally standardized by ISO (International Standards
Organization). This means that now your C++ program can be written and compiled on any
computer platform (PCs, minicomputers, mainframe computers) that has a standard C++
compiler.
Introduction to Programming 19

In this chapter, we are going to examine the basic format of C++ so that we can write a
simple program to display the message “Hello World” on the screen.

Rule 1. C++ is a case-sensitive language. Each of these “cost” identifiers is considered

totally different from each other. Always be alert for the case of the various identifiers that are
used.
cost COST Cost cosT cOSt

Rule 2. All C++ programs must have a main() function. When DOS launches your
program, some compiler supplied coding known as the C++ startup code is what actually begins
executing first. The startup code prepares the C++ environment for your program. Once it has
everything set up, it then calls a function known as main(). Remember a function is like a
subprogram, it does its required processing steps and then returns back to the calling point. Take
the square root function, for example. When it is invoked, it finds the desired root and returns
that answer back to the calling program which can then use that answer as it chooses.

Notice that it is a lowercase main(). Notice also that there are () after it. Between the ()
one would pass to that function any values that function needed to do its work. In the case of the
square root function, we would have to pass it the number of which we wanted to find the root.
While the main() function is indeed passed some parameters from the C++ startup coding, we
must wait until a later chapter to be able to understand and utilize those parameters. When an
empty set of () are used, it means either that we are ignoring all the parameters or that the
function really does not have anything passed to it. With the main() function, we are ignoring
them for now.

Rule 3. A block of coding is surrounded by { } braces. The { brace indicates the start
of a block of instructions. The } brace indicates where it ends. All of the instructions that we
wish to have in our main() function must be surrounded by the pair { }.

Rule 4. The main() function does indeed return a value back to the C startup
program. That return value is a completion code which is in turn given back to DOS, and it
indicates whether or not the program ran successfully. A value of zero is interpreted by DOS to
mean that the program ran successfully. Any non-zero value indicates the program did not
complete successfully. Normally, DOS ignores that return code. These return codes are integers
or whole numbers — a number with no decimal point. And the kind of data the function is to
return is coded preceding the name of the function. Thus, we have the basic shell as
int main () {
... our instructions go here
}
Here the int is short for integer. The first line says that this is the main() function, that it is
accepting no parameters and that it returns an integer back to the caller which is the C++ startup
program. This first line is also called the function header, for it marks the beginning of the
function.
int main () { <- the function header
Introduction to Programming 20

... our instructions <- the function body

}
All three lines above are called the main() function. The first line is the function header. All
lines between the { and } braces are known as the function body.

Definition: White space, abbreviated ws, is a consecutive series of blanks, tabs, carriage
returns, line feeds, printer page ejects and vertical tabs (found only on main frame computer
terminals, not on PCs).

Rule 5. White space is the delimiter in C++. That is, white space is used to separate
things. Notice the function header for main just above. White space separates the return type of
data (int) from the name of the function (main). When you press the enter key while editing a
program, it generates a carriage return and a line feed. (A carriage return, as in a typewriter, goes
back to column one, while the linefeed advances to the next line.) Since white space is used to
separate things in C++ coding, you can use as much white space as you desire.

Rule 6. When coding a block of instructions, you need to use a consistent style of
indentation. In C++, we have an inside joke: C++ is a write once, never read language. That is, a
C++ program can be rather hard to read to see what it is doing. Thus, anything you can do to
make your program more readable, the better it is. There are two major coding styles in common
use. Here are the two.

Style A:
int main () {
... our instructions go here
... our instructions go here
}

Style B:
int main ()
{
... our instructions go here
... our instructions go here
}

Notice in each of these, our instructions are uniformly indented some constant amount.
That way, one can tell at a glance the “block structure” of your program or what instructions are
in what block of coding. Yes, soon our programs will have many begin — end braces as the logic
becomes more complex.

How much do you indent? It is a matter of style. I prefer Style A with a uniform
indentation of one space or blank. Through much experience, I have found that if one
accidentally has one too few or one too many } braces, with Style A, it is much easier to find and
Introduction to Programming 21

fix than it is in Style B. I also indent one space because I prefer to see as much of the line of
coding without horizontal scrolling as possible. Since I often put in lots of comments on lines of
code to make them more understandable, my lines tend to be long.

One caution. Many editors, such as the Microsoft Visual C++ editor, insert tab codes to
assist in maintaining the consistent indentation. Sometimes by accident one enters some blanks
or spaces by pressing the space bar to force things to line up. However, blanks and tab codes are
two different things. Tabs are often expanded by different amounts between a screen and a
printer. If you get tab codes and spaces (blanks) intermingled, while your source program may
look perfect displayed on the screen, when you print it, jagged edges in the indentation may
appear. The first action I always take when installing a new C++ compiler is to find the setting
that replaces all tabs with a fixed amount of actual blanks (1 blank in my case). Realize that none
of this affects the actual operation of the program. It only impacts its visual appearance in a
program editor.

Rule 7. Since the main() function is supposed to return back to the C++ startup code an
integer indicating a successful execution, we must code a return instruction and give it the
integer to return, a zero, to indicate that all is ok.
return 0;

Rule 8. All C/C++ statements end in a ; (semicolon). If it does not end in a semicolon,
it is not a C statement. The function header for main() — int main () { — is not a C statement.
Rather, it is a function header. Shortly we will see another example that is not a C statement.

Where do you place the return 0; instruction? It should be the last line of the main()
function because when it is executed, the computer passes control back to the C++ startup
program to terminate the program. Putting this together, we have thus far:
int main () {
... our instructions go here
return 0;
}

Rule 9. C++ supports literal constants. In the above return instruction, the 0 is an
integer constant or literal value. Some other literal numerical values are: 42, –10, 3.1415926 for
example. If you want to have a single character literal value, enclose that letter within a single set
of quote marks. ‘A’ might be a literal that represents the desired grade in this course. An ‘F’
might denote the sex of a customer. A literal string of characters must be enclosed within a set of
double quote marks (“series of characters”).

If we want to write a program to display Hello World on the screen, then this is precisely
what we need, a literal character string. We can code the message we want to display on the
screen as
"Hello World"
Introduction to Programming 22

Literal constants are covered more completely in the next chapter. All we now need is the output
instruction that displays the message on the screen.

C++ views the consecutive series of characters being displayed on the screen as a
program executes an output stream, rather similar to the water stream coming from a garden
hose. Instead of water, characters appear on the screen, one after the other, in sequence. Treating
the output to the screen as an object to be manipulated is actually an OOP (object oriented
program) construct. An object in C++ consists of all the data and functions to operate on that
data - all taken together as an entity. The blueprint for the compiler to follow to make an actual
instance of an object is called a class in C++. In simple terms, a class is just the model that
defines how a real object is to be constructed when one is needed.

For example, if one were to create a Car class in C++, one would want various
informational items or data to be a part of the object, including such things as the make, model,
color, VIN number, size of the gas tank, current amount of gas and miles per gallon. Also, the
class or model defines functions to operate on those items, such as FillGasTank(). Given the
class definition, then one can actually make a car object, say a Ford Bronco and fill it with gas
and so on. The Bronco is then an instance of the Car class.

In a similar manner, C++ defines the class ostream to represent output to the screen. The
class has various informational items and most importantly a way to output data to the screen.
The specific instance of that ostream class that we use to display information on the screen is
called cout. The function we use to output data is called the insertion operator or <<. The line
of code to output our literal string message is then
cout << "Hello World";
The insertion operator displays exactly the characters as we have them in the string. Visualize the
insertion operator as a directional arrow that is sending or flowing the data to its right to the
destination on its left, cout, which is the screen.

However, remember that the computer does precisely what you tell it to do. If we ask it to
display the above message, this is what appears on the screen. When the instruction is done,
notice where the cursor is located.
Hello World_
The cursor is positioned after the ‘d’. Normally, like a typewriter, when we have finished
displaying all the characters on a line, we want the cursor to be in column one of the next line. In
C++, we have to tell the ostream to go to a new line. This can be done in one of two ways.

The first way to generate a carriage return and line feed is to display the new line code in
our string. The new line code is \n. Thus, a better version of our message would be
cout << "Hello World\n";
Wherever the new line code is found, there is a new line at that point. What would be the output
from the following instruction?
cout << "Hello\n\n World\n";
Remember that it displays exactly what you tell it to display. The output is
Introduction to Programming 23

Hello

World
_
Two \n codes in a row cause double spacing. The ‘W’ character does not line up with the ‘H’
character because there is a blank or space before the ‘W’ character after the second new line
code.

The second way to generate the new line is to insert endl, the end line, value into the
output stream. This method is not as convenient in this example, but is coded this way.
cout << "Hello World" << endl;

The endl is called an output manipulator because it is manipulating the output stream in
some way, here adding a new line code at the point that it appears. If we wanted to have each
word on a separate line, code
cout << "Hello" << endl << "World" << endl;

The above line of coding is an example of chaining several separate pieces together into
a single output instruction. It could also be done with separate lines as follows
cout << "Hello";
cout << endl;
cout << "World";
cout << endl;
There is no limit to how many lines of output are displayed in a single cout instruction. To chain,
just code another insertion operator followed by the next piece of information to be displayed.
Here is how our first program appears thus far, though we are not yet finished.
int main () {
cout << "Hello World\n";
return 0;
}

How does the compiler know what cout is or that endl is an output manipulator? It
doesn’t unless we provide the compiler with the blueprints to follow. As it stands, if we were to
compile this program, we would get a bunch of error messages saying basically that the compiler
does not know what these two things are.

It is our job to include in our programs the needed blueprints for the compiler to use.
These blueprints are the class definitions and function prototypes. A function prototype is a
blueprint or model for the compiler to follow when it wants to call a function. A function
prototype looks very similar to the function header line. It gives the name of the function, its
parameters (if any) and what kind of information the function will be returning (if any). Until
Chapter 6, we use functions provided by the compiler manufacturer, the standard functions of the
C++ language. However, in Chapter 6, we will learn to write our own functions; function
prototypes are explored there in depth.
Introduction to Programming 24

In our beginning program, we need to tell the compiler to include the definitions of the
ostream class and the manipulators. This is done by issuing an order to the compiler to copy the
contents of some files into our program — the #include directive. Its syntax is
#include <filename>
The #include tells the compiler to copy a file into our program at this place in the program. The
<> tells the compiler to look for the file in its own \INCLUDE folders. Each compiler has one or
more include folders in which the various class definitions and standard C++ function prototypes
are located. Included files usually have the .h file extension, h for header file. However, many of
the newer C++ headers have no file extension. In our first program, we must code the following
#include <iostream>
#include <iomanip>
using namespace std;
The header file iostream contains the definition of the output stream class; the iomanip file
contains the definitions of the manipulators (for endl in this case). Notice that these are compiler
directives and not C++ statements and therefore do not end in a semicolon.

With all the possible identifiers in C++, a way to manage their use was added to the C++
language recently. A namespace is a collection of identifiers, class definitions and functions
grouped together for a program’s use. The C++ language provides the namespace std that refers
to all the normal C++ classes and function prototypes. When a program uses the standard
namespace, the header file includes take on an abbreviated form. The using statement notifies
the compiler that a particular namespace is to be used in this program. It is coded as follows.
using namespace std;

Where are header file includes placed in programs? The answer is simple.

Rule 10. Header file includes must be physically before the first usage of what they
define or contain. Thus, nearly always, the includes are the very first thing in the source
program. Here is our complete first program that displays Hello World on the screen.
#include <iostream>
#include <iomanip>
using namespace std;

int main () {

cout << "Hello World\n";

return 0;
}

Notice one small detail. I have added blank lines to separate key lines of coding to make
it more readable. Please use blank lines in your coding to assist the readers of your program. A
blank line in a source program acts the same way a blank line does in a book, marking the end of
paragraphs. In a book, the reader knows that they may take a breather when they reach the end of
a paragraph. It is the same way when reading a program. One can safely pause when a blank line
Introduction to Programming 25

is encountered. In programming, we use blank lines to group related instructions to make the
reading of the program easier.

Rule 11. Comments in C++ take two forms. The short form is //. When the // is not part
of a literal character string, everything after the // to the end of that line is considered a comment
by the compiler. The long form or C style comment is everything between a /* and a */ is
considered a comment, when not inside a string literal constant. For example
cout << endl; // display a blank line here
/* this is a comment */
/* this is a
longer
comment */

The following are not comments because they are imbedded in a character string literal
value.
cout << "This is a // strange message\n";
cout << "This is also strange /* not a comment */ \n";

To help document a program, I use block comments that look like this.
/*********************************************************************/
/* */
/* First Program in C++: display Hello World on the screen */
/* */
/*********************************************************************/
Why? Imagine someone hands you a C++ program that has no comments in it. How can you
figure out what that program does — what its purpose is? You would have to read through the
program coding to attempt to find out.

Now suppose someone handed you the following program. Notice you can tell at a glance
what it does without having to read a line of C++ coding.
/*********************************************************************/
/* */
/* First Program in C++: display Hello World on the screen */
/* */
/*********************************************************************/
#include <iostream>
#include <iomanip>
using namespace std;

int main () {

cout << "Hello World\n";

return 0;
}

Rule 12. Always document your program. Include some form of comment at the very
beginning outlining in twenty-five words or less what the purpose of the program is. Also include
additional comments where they are needed to help someone follow the logic and operation of
Introduction to Programming 26

your program. Throughout the text, you will see numerous examples of what I think a well
documented program ought to look like. Some of you may not care for the impressive visual
impact that my block comments make; in that case use a more gentle style. Style is not at issue,
but the comments are.

It has been said many times in this industry that a complex program with no internal
comments at all is practically worthless because it is nearly impossible for someone other than
the author to maintain. Please develop good habits by documenting your programs as you write
them.

What should you do next? Get your compiler installed and see if you can get this Hello
World program entered and to execute successfully.

Note that this chapter does not have the Computer Science or Engineering Examples
sections.

Design Exercises
1. How would you solve this problem? What is the answer? A bug wishes to climb to the top of a
12-foot tall telephone pole. During the day, it climbs 3 feet. However, while it sleeps at night, the
bug slides back down 2 feet. How many days does it take the bug to reach its objective, the top of
the pole?

2. Sketch a solution in pseudocode or English to solve this problem. A math teacher wishes to
have a program that displays the multiplication tables for her fourth graders. She wants the
program to accept any whole number (integer) from 1 to 9. The program then displays the
multiplication tables from 1 to that number. A sample run might be as follows. Note she enters
the underlined number 4.
Enter a number from 1 to 9: 4
1 x 1 = 1 x 1 = 1
1 x 2 = 2 x 1 = 2
1 x 3 = 3 x 1 = 3
1 x 4 = 4 x 1 = 4
2 x 2 = 2 x 2 = 4
2 x 3 = 3 x 2 = 6
2 x 4 = 4 x 2 = 8
3 x 3 = 3 x 3 = 9
3 x 4 = 4 x 3 = 12
4 x 4 = 4 x 4 = 16
Introduction to Programming 27

3. Sketch a solution in pseudocode or English to solve this problem. A manager of some carpet
store wishes a program that calculates the square footage of carpet a customer requires and
determines his cost for installation based on the square footage. The program first asks him to
enter the length and width of the room. It then displays the square footage. His installation cost is
found by multiplying the square footage by 7.5%. A test run might be:
Enter the length and width of the carpet: 10 20
The square footage is 200 and the service charge is $15.00

Stop! Do These Exercises Before Programming

Correct the errors in the following programs. If you are having trouble determining what is
wrong, you can always make a test program, enter this coding and see what the compiler
indicates is wrong.

1. Why does this program not compile? Show what must be done to fix it?
int main () {
cout << "Hi there!\n";
return 0;
}
#include <iostream>
#include <iomanip>

2. Why does this program not compile? Show what must be done to fix it?
#include <iostream>
#include <iomanip>
Int Main () {
Cout << "Great day outside!!\n";
return 0;
}

3. Why does this program not compile? Show what must be done to fix it?
#include <iostream>
#include <iomanip>
using namespace std;
int main () {
cout << Hi there! << endl;
return 0;
}
Introduction to Programming 28

4. Why does this program not produce any output? Show what must be done to fix it.
#include <iostream>
#include <iomanip>
Using Namespace Std;
int main () {
return 0;
cout << "John Jones successfully made this" << endl;
}

5. Why does this program not compile? Show what must be done to fix it?
#include <iostream>
#include <iomanip>
using namespace std
int main ()
cout << "Something is very wrong here" << endl;
return 0;
}

6. Why does this program not compile? Show what must be done to fix it?
#include <iostream>
#include <iomanip>
using namespace std;
int main (){
c out >> "Something is still wrong here" << endl;
Return zero;
}

7. Why does this program not compile? Show what must be done to fix it?
#include <iostream>
#include <manip>
using namespace std;
int main (){
cout << 'I cannot get this program to work!' << << endl;
return 0;
}
Introduction to Programming 29

8. This program compiles, but what is stylistically wrong? Show what must be done to fix it?
#include <iostream>
#include <iomanip>
using namespace std;
int main () {
cout << "Something is not quite right here" << endl;
return 0;}

9. A programmer has written the following solution to calculate wages for their company’s
weekly payroll. The programmer is convinced it will work just fine and has submitted it to you
for verification and approval. This means you must thoroughly desk check their solution. How
many errors can you find in it? Show how could they be repaired so that it would work.

Each line of input contains the employee’s number, the hours they have worked and their
pay rate. Any hours more than 40 are to be paid at time and a half.
Set total_payroll to 0
input employee_id and hours and payrate
as long as we got a set of data, do the following
multiply hours by payrate and store it in pay
if the hours is greater than 40 then do this
Pay = (hours - 40) times rate times 1.5
end the if
add the pay to the TotalPayroll
display the id number and the pay
try to input another employee_id and hours and payrate
end the do series here
display the total_payroll

Test the program with the following input lines of data

123455 40 5.00
245346 20 7.50
535323 60 6.00

Hint: draw a picture of what main storage or memory should be. Pay attention to the
names of the variables used in the solution.
Introduction to Programming 30

What is the precise output from these output instructions? Fill in the boxes to indicate the
results. Each box represents one column on the screen and each row represents one line on the
screen.

10.
cout << "One";
cout << "Two";
cout << "Three";

11.
cout << "One ";
cout << "Two ";
cout << "Three";

12.
cout << "One" << endl;
cout << "Two"<< endl;
cout << "Three" << endl;
Introduction to Programming 31

13.
cout << "One\n";
cout << "\nTwo\n";
cout << "Three";

14.
cout << "One\nTwo\nThree\n";

Programming Problems
Although we do not know enough about C++ to actually write programs yet, we have all the
skills needed to write out English step by step solutions and to desk check them. Developing the
skill to write out the sequential steps needed to solve a problem on the computer is always the
first step to solving a programming problem.

Problem Cs01-1 — Make a Sandwich

Write the steps that someone can follow to make a peanut butter–jelly sandwich. The bread,
peanut butter and jelly are located in the refrigerator. Utensils are located in a cabinet drawer. Be
as specific as you can. Imagine a 6-year-old trying to follow your instructions. That is, the person
is going to follow your instructions to the letter. Assume that the person has no prior knowledge
about how it should be done. As a guide, you probably do not need to use more than fifty
different words in the solution. (Note this is not supposed to be written as a computer program.)
Introduction to Programming 32

Problem Cs01-2 — Directions

A friend from out of town is staying with you and needs directions to the nearest bus station.
Write out the series of steps necessary to direct them from your house to the station. While they
are not likely to walk all the way, assume that they are navigating themselves, either on foot or by
car (a bit unrealistic). The object is to give directions to enable someone to get from your house
to the station. If you do not have a bus station, use the nearest train station or airport. (Note this is
not supposed to be written as a computer program.)

Problem Cs01-3 — Cost of Goods Sold

Using English statements, write out a solution to the Cost of Goods Sold problem. Acme
company has a file of sales data. Each line in the file contains the quantity sold and the cost of
that item. The report should look like the following
Qty Cost Total Cost
42 10.00 420.00
10 4.99 49.90

Grand Total Cost of Goods is 469.90

Use the Cycle of Data Processing as your guide. The basic steps are Input, Process,
Output and then back to Input until there is no more data to be processed. (Note this is not
supposed to be written as a computer program.)

Problem Cs01-4 — Your First Actual Real Computer Program

Write a program that outputs your personal information in a card-like format. Your card
information should be a series of lines containing the title line, your name, address, city, state, zip
and phone number. Format it similar to mine:

Program Cs01-4 by Vic Broquard

Name: Vic Broquard

Address: 10305 Ridge Line Road

East Some City, IL
61611

Phone: (309) 699-9999

Note that this program has no “variables” and consists of one or more cout lines. You can
output the information using six cout instructions or jam it all into a single cout.

Create the cpp source program, compile and execute the program.
Introduction to Programming 33

Problem Engr01-1 — Converting Degrees to Radians

Write out the steps to solve the following problem. Use English or mathematical equations as
needed. Follow the Cycle of Data Processing, Input, Process, Output and repeat until the user is
done. Do not actually write the C++ program.

The user enters an angle in degrees; the program converts it into radians and displays the
results. The program should convert as many angles as desired until the user enters a -999 for the
angle at which point the program stops. Here is a sample run of how the program should work.
Enter an angle in degrees: 0
0 degrees is 0 radians
Enter an angle in degrees: 180
180 degrees is 3.14158 radians
Enter an angle in degrees: -999
Thanks for using the Converter Program
Pay particular attention to what is constant literal character string information and what is
variable information on the input and output processes. (Note this is not supposed to be written
as a computer program.)
Numerical Processing 34

Chapter 2 — Numerical Processing

Section A: Basic Theory

Introduction
When one begins to write a program, the first action should be to define the variables that the
program needs; variables are places in memory in which to store data items. Next, one writes the
instructions to input values from the user (say, via the keyboard) and store those values in the
variables. Now that the variables have values, one can write the needed calculation instructions
and finally the instructions to display the results. This chapter concentrates on numeric type of
variables and constants along with their basic input and output instructions.

Variables and Constants

In C++, a data object is a region in memory in which to store data. A data value is the contents
of that object. A place to hold the quantity purchased, say 42, could be called qty but the actual
contents of qty, its data value, would be 42 in this case. A variable in C++ is a modifiable data
object, that is, it is a place to store data and the value stored there can change as we desire.

The opposite of a variable is a constant data object. With a constant data object, once
the initial value is defined, it can never be changed in any way. If we created a variable to hold
the number of months in a year, it should be a constant data object because its value, 12, ought
not ever be changed. Similarly, if we defined a variable to hold the value of PI, it should be a
constant as well, 3.14159.

Integer Versus Floating Point (Real) Numbers

Thinking about the number of months in a year and the value of PI illustrates that there are
inherently two types of numerical data, integer (whole numbers) and real (floating point)
numbers. Integer numbers are discrete values on the number system; they never contain any
fractional part. Some examples of integer numbers are 10, 42, 84, 99, and –88. Real numbers or
floating point numbers contain a possible fractional amount. Any number with a decimal point is
Numerical Processing 35

a floating point number. Some examples include 3.14159, 123.55, 10. and –.00123. Notice that
the addition of the decimal point on the 10. has changed it into a real number from an integer.

When entering some real or floating point constant numbers, often they are very large or
very small, with many leading or trailing 0's. In such cases, use the scientific notational form. For
example, 123.45 could be rewritten as .12345x103 and entered as a constant as .12345E3, where
the E stands for “exponent” and the 3 digit is the power of ten. Here are some other examples.
123000000.0 could also be coded as .123e9 or 1.23e8 or 12.3e7 or 123.e6
.000123 could also be coded as .123e–3 or 1.23e–4

Scientific notation closely resembles how the computer will store floating point numbers.
Specifically, if we had 1.23e+002 representing 123.00, then the 1.23 is called the mantissa; the e
stands for exponent power of 10; and the +002 is the exponent, meaning 102. When the computer
stores a floating point number, it stores both the exponent and the mantissa.

In summary, 123.00 represents the fixed point notation while 1.23e+002 represents that
same number in scientific notation. They are the same number. Which way the computer displays
them to us is controlled by the program. See the Insertion of Floating Point Numbers into an
Output Stream section below.

Which Type of Data Do You Use for Which Variable?

Now which data type you use when defining a variable is significant. The computer’s integer
math instructions are some of the fastest executing instructions. On the other hand, floating point
math instructions are some of the slowest. Furthermore, some variables represent discrete
integral values. How could one have 5.5249 months in a year? Similarly, a count of the number
of employees in a company would be an integer (how could you have 10.3487 employees)? The
quantity of cans of soup purchased would be an integer. However, the cost of one can of soup
would need to be stored in a floating point variable ($0.49).

One cannot just make all the numerical data fields in a program arbitrarily a floating point
number any more than you can make them arbitrarily all integers. The deciding factor is “will
this field or variable ever possibly have a fractional part?” If so, it must be defined as a floating
point number. If not, it should be an integer type.
Numerical Processing 36

Definition of Variables
The syntax to define a variable is actually quite simple.
datatype variable_name;

Let’s begin with the two most commonly used data types, the int and the double. The
int is a short form of the word integer. (Yes, you could spell it out, but most programmers
simply use int.) An int variable can only contain an integer or whole number. A double
specifies a floating point number, a number with a decimal point and possible fractional
portion. Thus, in the definition of a variable, the data type can be either an int or a double. In
the ensuing chapters, we explore other possibilities.

After the data type comes the name you wish to call this variable. What are the rules
for names in C+ + ?
1. Names can be from one character long to as many characters as desired; however,
only the first thirty-one characters are used by the compiler.

2. The name must begin with a letter of the alphabet or the _ character. (However,
names beginning with an _ generally have a special purpose meaning and should be
avoided.)

3. Each name must be unique.

4. Remember also that C+ + is case sensitive.

5. The _ character can be used to separate compound names.

6. A blank cannot be used in a variable name because a blank is a form of white space
and white space is the delimiter between language elements.

7. Numerical digits can be a part of the name after the first character of the name.

The following are valid names in C+ + : cost, Cost, COST, qty, quantityOnHand,
PI, department5, invoiceNumber, firstName, R2D2, resistance, ohms and grade.

These are invalid names in C+ + : invoice number and 3-d. Why? A blank cannot be
part of the name in the first one, and a name can neither begin with a digit nor contain a minus
sign (which is interpreted by the compiler to mean a subtraction operation).

8. There are some reserved words in C+ + and these words are the language “verbs”
and components. A variable cannot be a reserved word.
Numerical Processing 37

For example a variable name could not be int or double because these are the reserved
words for the two data types. While I could compile a list of all known reserved words, such
is not needed for two important reasons. Both are discussed under the next rule.

9. All variable names in a program must be meaningful names.

This rule is true in every programming language. The names you choose to call the
variables of your program absolutely must be representative of their meaning. Someone, you
included, must be able to read the program’s instructions and make sense of them. If your
program needed to have a variable to represent the quantity ordered and the item’s cost, then
there is nothing to prevent you from calling them Fred and Wilma, respectively, or x and y, or v1
and v2. However, doing so makes the program nearly impossible for another to decipher! In this
example, what would be good meaningful names? It is a matter of taste and clarity. If I were
naming them in my program I would likely call them qty and cost because I tend to favor shorter
names. However, if you are having trouble following coding sequences, longer more descriptive
names can aid you; try calling them quantityOnHand and costOfGoods, for example.

When solving Engineering problems, often a mathematical equation is involved. Suppose

we need to calculate the force on an object given its mass and acceleration. The equation is f = m
a. In such cases, it is acceptable to name the variables f, m and a. However, it is also acceptable
to call them force, mass and acceleration. It would not be meaningful to call them x, y and z. In
programming, when a programmer sees variables x, y and z, they often visualize a three-
dimensional coordinate system!

The key point is that no language element that is a reserved word would be a meaningful
variable name in most programming applications.

The second reason that it is not important to have a large list of reserved words is that
most compilers today provide an editor in which to type in a program. The editors have chroma-
color syntax highlighting systems. This system operates as you type in characters. Suppose that
you type in the following two letters: in. They appear in a normal font. However, the second that
you type the letter t as the next character, the editor turns the three letters into some form of
color-highlighting: int. It is trying to show you that this is a key C++ identifier. If you then type a
fourth letter, say you intend to spell out interest, as soon as you enter the letter e next, the color-
highlighting reverts back to plain text as it is now not that key identifier: inte. Once you have
typed in a couple programs, you will instinctively make use of that color-highlighting system.
Since the editor displays reserved words in a fancy manner, there is no need to have a lengthy list
of names to avoid.
Numerical Processing 38

The Issue of the Case of a Variable Name

What about case and compound variable names such as a variable to represent a person’s first
name? One could use any of the following: firstname, FirstName first_name, firstName. Of
course, there are many more possibilities than these four. However, these four are sufficient to
define the issue at hand. Take the first version, firstname. When a variable name is actually a
compound name formed from two words, using all lowercase letters all run together makes the
name harder to read. Programmers usually avoid that if possible.

A better solution is to use the _ character: first_name. However, using a form of

capitalization is often seen as an even better way to make compound variable names more
readable: FirstName and firstName. However, it is my personal suggestion that you avoid
capitalizing the first word of a compound variable name or the first letter of any variable name.
That is, don’t use FirstName or Quantity, rather capitalize the second and subsequent words in
the compound name: firstName and quantity. Why?

I am assuming that you are all going to master the beginning C++ programming and move
on into the advanced C++ courses and perhaps even into Windows programming as well. In
almost all of Windows programming and in much of advanced C++ work, the key identifiers and
function names often must be capitalized. Thus, if you capitalize your variable names, eventually
you will not easily be able to tell at a glance whether a given identifier is your variable or a C++
or Windows identifier. It is called “name space pollution” in the industry. The word name space
here means the universe of identifier names of the program. Throughout this text, when I use a
variable name that is formed from two or more words, the second and subsequent words are
capitalized. No normal variable name of mine ever begins with a capital letter unless there is no
better alternative.

Similarly, for two reasons, do not code all of your variable names in all uppercase letters,
such as QUANTITY. First, when you use all uppercase letters in a variable name, others tend to
perceive it as “shouting.” THIS SENTENCE IS VIEWED BY MOST READERS AS
SHOUTING. Second, an identifier in all uppercase letters tends to stick out readily when viewed
in a sea of lowercase letters. However, in our programs, we are soon going to have some special
identifiers used to represent very special items. By convention amongst all programmers, these
special identifiers are always fully upper-cased so that they do stick out as special items.

Defining More Than One Variable in the Same Statement

One can also define multiple variables of the same data type with a single statement. This
alternate syntax is:
data type name1, name2, name3, ... name_n;
If the problem required three integers, one for quantity, product id and year, one could code:
int quantity, productId, year;
Numerical Processing 39

or
int quantity,
productId,
year;
or
int quantity;
int productId;
int year;

Suppose that one needed some double variables to hold the charge, voltage and resistance
of a circuit element. One could code either of these to define the variables.
double charge, voltage, resistance;
or
double charge,
voltage,
resistance;
or
double charge;
double voltage;
double resistance;

Which style is preferred? Again, questions of style are best answered by each
individual programmer. Here are common arguments for and against each. Statistically, a
programmer has fewer errors in coding when they can see all of the coding on the screen at
one time. Defining multiple variables on one line reduces the total lines of a program thereby
making more of the program visible before you have to scroll. The disadvantage is twofold.

When multiple variables are defined on one line, it is very hard to find any specific
variable in the list. Consider for a minute the following set of lines that define a series of
variables.
int qty, hours, idNumber, minutes, prodId, seconds, count;
double cost, hoursWorked, payRate, yearToDatePay, pay;
How fast can one spot the variable for product id number or the person’s rate of pay? You have
to read through each line searching for them.

The other disadvantage is that one cannot insert comments to better document the
meaning of that variable. Frequently in the coding examples in this text you will see comments
beside the variable definitions providing further information about its meaning. For example,
int quantity; // quantity purchased under warranty
int productId; // manufacturer's id number of these
int year; // purchase year for start of warranty

Adding comments to variable definitions can provide a means to identify the units of the
values the variable holds. Consider these variable definitions.
double distanceTraveled; // in miles
Numerical Processing 40

double frequency; // in kilohertz

int time; // in seconds
Can you see the extra clarity the comments can provide? Also notice a subtle formatting action I
used. I aligned all the variable names in the same column because it greatly aids readability. For
aesthetics and readability, I aligned the comments as well. Finally, what happens if you need to
change the data type of a variable which has been defined along with four others on one line?

Where Are Variable Definitions Placed in a Program?

Where are the variable definitions placed within a program? The answer is that a variable must
be defined before its first use in an instruction. Normally, the variables to be used for input, those
for calculation results and those for output are defined at the beginning of the main() function.

For example, suppose that our program needed to define variables for a cost, a quantity
and a total. The program up to this point is as follows.
#include <iostream>
#include <iomanip>
using namespace std;

int main () {
int qty; // quantity ordered
double cost; // cost of one item
double total; // total cost of the order
...

Initializing Variables and the Assignment Operator

When a variable is defined, it has no starting or initial value. We refer to it as containing “core
garbage” or just random data. For some variables, this is perfectly fine. Suppose that the first use
of a variable is to store the data entered by the user in response to an input instruction. Storing
new data into the variable wipes out whatever was previously contained in that memory location.

It serves no purpose to initialize variables that are going to be used for input. Any such
initial value is going to be replaced by the incoming data. In fact, many compilers issue a
warning message notifying you that no use is being made of that initial value. Likewise, any
variable that is going to hold the result of a calculation need not be initialized.

However, any variable that is used as a counter or a total value must be initialized to zero.
That is, if your first use of a counter is to add one to it, that counter must have a value initially.

A variable whose value is not to be inputted from the keyboard can be given an initial
value in one of two ways. Both methods use the assignment operator, which is an = sign.
Numerical Processing 41

The assignment operator copies the data value on the right side of the operator to the
variable on the left side of it.

One method to assign values is to code an assignment instruction. For example, to place a
42 into quantity and 10.99 into cost, we code
quantity = 42;
cost = 10.99;
then the first line removes any previous value in quantity and places the integer value of 42 in it.
Likewise, the double value of 10.99 is placed into the variable cost.

Please note that the assignment operator should not be confused with the mathematical
equals sign that is also an equal sign. Assignment means to copy a data value on the right into the
variable on the left.

The second method that a variable can be initialized to a starting value is at the time the
variable is being defined. The syntax is:
datatype name = value;
Thus, we could have coded:
int quantity = 42; // quantity ordered
double cost = 10.99; // cost of one item

Several variables can be initialized as they are defined. Consider this line.
double cost = 10.99, total = 0, tax, grandTotal = 0;
Here when the compiler sets up memory for cost, it gives it the starting value of 10.99. Then it
sets up memory for total and gives it a value of zero. Next it sets up memory for tax and gives it
no starting value (presumably it is going to have to be calculated somehow). Finally, it sets up
memory for grandTotal, giving it a value of zero. Figure 2.1 shows what computer memory
looks like after the above variable definitions are completed.

Figure 2.1 Main Storage after Data Definitions

In programming, you will see variables being initialized or given a value both ways.
By initializing variables that must have a starting value when you define them, you eliminate
one extra line (the assignment instruction). This means that you can see more lines of the
program on the screen at one time which lessens the potential for errors.

Rule: When assigning data with the assignment operator, the data value on the right
must be compatible with the data type of the variable on its left side.
Numerical Processing 42

Consider this incorrect assignment.

double cost = "Hello World";
Coding the above like causes a compile time error. How can the character string “Hello World”
be converted into a numerical value? It cannot and such attempts generate error messages. One
can code
double cost = 0;
Here cost is a double but the number being assigned to cost is an integer — no decimal point.
However, the compiler can easily convert an integer into a double for us. This is known as data
conversion, a process in which data of one type is converted into another data type. Data
conversion and its rules are discussed in the next chapter.

In the above example, if we now assign tax a value, by coding

tax = 8.84;
then memory appears as shown in Figure 2.2.

Figure 2.2 Main Storage after Assignments

Multiple Assignments — Chaining the Assignment Operator

Just as we can chain insertion operators into a single output operation, so can the assignment
operator be chained.

Suppose that we have three total fields defined to hold the purchase price of tickets.
Children and senior citizens get a price reduction. Their ticket costs are defined as follows.
double costAdultTickets;
double costChildTickets;
double costSeniorTickets;
Suppose further that the program needed to set these for cost fields to 0. One could do so with
four assignment statements, such as this one.
costAdultTickets = 0;
Or one could initialize these to zero as they are defined as shown here.
double costAdultTickets = 0;

However, one can also chain and make a multiple assignment of 0 to each of these as
follows.
costAdultTickets = costChildTickets = costSeniorTickets = 0;
This moves a 0 into costSeniorTickets first and then copies that 0 into costChildTickets and
finally copies that 0 into costAdultTickets. While not commonly used, multiple assignments
occur mostly when setting variables to 0.
Numerical Processing 43

Input of Data Values into Variables

C++ defines an input stream that retrieves data from the keyboard as a consecutive series of
characters. The class or blueprint is called an istream and the specific instance that we use to
obtain keyboard input is called cin. This definition of cin, as an instance of the istream, is
parallel to our use of cost as an instance of the intrinsic double data type. It is the same with cout
being an instance of the ostream class of data. The istream class and the definition of cin are
contained in the header file iostream.h. The operator that causes transfer of data from the
keyboard into our variable is called the extraction operator which is >>. The extraction operator
can be thought of as extracting the next data item from the input stream of characters as they are
entered on the keyboard.

The extraction operator’s syntax is similar to the insertion operator that is used to send
data to the screen.
cin >> variable;

Assume that we have defined both qty and cost as an integer and a double as above. If
we code
cin >> qty;
then the input stream waits until the user has entered their data and pressed the enter key. Once
the enter key is pressed, the extraction operator attempts to extract the data requested which is
given by the data type of the variable. If it is successful, then that data is placed into the variable,
erasing what was previously contained in the variable. Thus, if the user enters 10 and presses the
enter key, the extraction operator above places 10 into our qty field in memory. If we next code
cin >> cost;
and the user types in 42.99 and presses the enter key, then the extraction operator attempts to
input a double value, since cost is a double. If successful, cost now contains 42.99. What
happens when incorrect data is entered is discussed in a later chapter. Until then, assume that no
user ever makes a mistake typing in any data. (I know, this is highly unlikely, but we need more
C++ language elements to sort it all out.)

Chaining Extraction Operators

Just as the insertion operator can be chained with other insertion operators to output more than
one item at a time, so can the extraction operator. For example, one can code
cin >> qty >> cost;
There is no limit on the number of data items that can be input on a single cin input operation,
just as there is none on cout.

However, just how must the user enter the two values? The answer is simple, white space
must be used to separate the two numerical values. Recall that white space is any consecutive
series of blanks, tabs, carriage returns and line feeds for example. Pressing the enter key
Numerical Processing 44

generates both a carriage return and a line feed. For most purposes, the two numerical values
would be separated by a blank. Thus, the user could enter both values as
10 42.99<CR>
where <CR> means press the enter key. In the following, b means a blank or space and t means a
tab key was pressed. The user could also enter these two values in any of the following ways.
10<CR>
bbttbbtbtbtbtbt<CR>
bttt42.99bttbbt<CR>
or
10<CR>
42<CR>
Only when that last <CR> is pressed does the extraction operator finally get to input the two
values. Realistically, no one is going to be that silly when typing in numerical data. Just a single
blank is commonly typed.

Always Prompt the User Before Inputting the Data

This is an important concept. Typically, our programs must have the user’s desired input before
calculations and the output of results can occur. But you cannot just start the program with
cin >> qty >> cost;
Why? Imagine what the user sees. They launch the program which immediately displays a
blinking cursor and sits there waiting for them to enter the input data. What are they supposed to
enter?

When inputting data from the keyboard, a program must always prompt the user notifying
them what data is to be entered at this point. A prompt is nothing more than a simple cout line.
For example, one could code
cout << "Enter the quantity: ";
cin >> qty;
cout << "Enter the cost: ";
cin >> cost;
On the screen the user sees the messages and responds by entering the keystrokes shown in
boldface below.
Enter the quantity: 10<CR>
Enter the cost: 42.99<CR>
Notice one nice touch. I put sufficient blanks in the “enter cost” prompt message so that the first
character of the user’s entry (the 4) aligned with the first character of the quantity (the 1). This
makes a more aesthetic appearance.

What would the prompt look like if the user was supposed to enter both values at the
same time? That depends on the “computer savvy” of your program’s users. The following might
be sufficient.
cout << "Enter quantity and cost, separated by a blank\n";
cin >> qty >> cost;
Numerical Processing 45

Here I am implicitly assuming that the users know that, after they have entered the two desired
values, they need to press the enter key. If they might not know that, then you should add even
more directions in your prompt message. In this example the user sees and responds as follows.
Enter quantity and cost, separated by a blank
10 42.99<CR>
In both of these input operations, the final result is our variables qty and cost now contain the 10
and the 42.99.

Now let’s examine some common errors a programmer can make. Can you spot the
errors?
Cin >> qty;
cin qty;
cin >> >> qty;
cin > qty;
cin >> 10 >> 42.99;
In the first line, cin was capitalized which makes it an unknown item. In the second line, there is
no extraction operator coded. In the third line, there are two extraction operators in a row. In the
fourth line, there is no extraction operator; the single > is the greater than sign that is used in
comparison operations. In the fifth line, there are no variables to store the inputted values; the
programmer is confusing the variables qty and cost with the user entered data values.

Now we are ready to continue with our program. Before we tackle the calculations, let’s
examine the output operation in more depth. That way, when we do get to the calculations, we
can display the answers.

Output of a Variable
Let’s assume that these three variables are defined and initialized as follows.
int qty = 10;
double cost = 1.99;
double total = 19.99;

To display output on the screen, cout is used similar to the Hello World Program of
Chapter 1. The insertion operator (<<) is used to send one or more variables to the output stream
onto the screen. To display the contents of the cost variable, code
cout << cost;

Further, chaining can be used just as it was in the Hello World Program. To display the
person’s order represented by the qty, cost and total variables, one could code
cout << qty << cost << total << endl;
Numerical Processing 46

However, there are some other aspects that must be considered because the above line
does not produce usable results.

Specifically, unless instructed otherwise, the insertion of a number into the stream
includes all significant digits of the value. Given the above values of these three variables, the
output from the above cout line is
101.9919.99

Oops. Notice that it does display all the significant digits in the values. However, there is
no spacing between each value! One advantage of the output stream is that it gives us total
control over the formatting aspects.

The first thing we need to do is to put some spaces between the numbers. There are two
ways to do this. The first method is quick and dirty; output a string with some blanks in it
between the numbers; here I used two blanks.
cout << qty << " " << cost << " " << total << endl;
This now yields the following.
10 1.99 19.99
For simple results, this can be perfectly fine.

However, soon the program may need to display a several sets of these values. Suppose
these represent a person’s order. What if the program processed several sets of orders displaying
the results for each using the above cout instruction? Here is an example of four orders being
displayed with this method.
10 1.99 19.99
1 .25 .25
100 .49 49.00
20 2.00 40.00
Can you read the columns of numbers?

Obviously the data should be aligned. With numbers, columns of data are aligned on the
right. If decimals are involved, usually the decimal points are aligned on successive rows. The
desired results for this sample are
10 1.99 19.99
1 .25 .25
100 .49 49.00
20 2.00 40.00
xxxxyyyyyyzzzzzzz
Now the results are readable. Notice that in this example, the maximum width of the quantity
column shown by the series of x’s is four. The width of the cost column shown by the y’s is six;
the total column shown by the z’s is seven.
Numerical Processing 47

The setw() Function

The setw() or set width function can be used to set the total width for the next item being
displayed. It applies solely and only to the next item to be displayed. The function takes one
parameter, the total width of the next item. For numeric types, the values are right justified within
the total specified width.

A more optimum way to display the line is as follows.

cout << setw (4) << qty
<< setw (6) << cost
<< setw (7) << total << endl;
This is in fact exactly what was used in the above proper columnar alignment example.

If you misjudge a needed width and make it too small, then the output stream ignores
your width and displays all of the significant digits. Suppose that I used a width of four columns
for the total variable,
cout << setw (4) << qty
<< setw (6) << cost
<< setw (4) << total << endl;
then, the output would have been
10 1.9919.99
1 .25 .25
100 .4949.00
20 2.0040.00

Also note that the setw() only applies to the very next item to be output! This would also
fail.
cout << setw (7) << qty << cost << total << endl;
Here the width of seven columns applies only to the qty variable; the others display only the
actual digits in them with no blanks between them.

In general, always use the setw() function to control the maximum width of numbers.

Insertion of Floating Point Numbers into an Output Stream - setprecision and

fixed

We have seen that the setw() function is used to define the total width of a field on output.
Included in the width is any negative sign and all significant digits. However, on output, floating
point numbers pose an additional problem. The output problem is two fold. First, since a double
has 15 possible decimal digits in it, how many of those do we desire to display? Do we show 4.0
or 4.0000000000000 or something in between? Second, when the value is large or the value is a
tiny decimal fraction, the ostream default is to show the result in scientific notation. Thus, we
are likely to see $4e+006 instead of $4000000. Or we see 4e–6 instead of .000004.
Numerical Processing 48

The first problem of the number of decimal digits to show is easily solved using the
setprecision() function. When not displaying numbers in scientific format, this function sets the
number of decimal digits to the right of the decimal point that we desire to see in the result.
Unlike setw(), the precision once set, applies to all floating point numbers that are output until a
new setprecision() function call occurs if any. Thus, if we desire that all the numbers in our
output contain two decimal digits, such as found in financial calculations, we need only to make
one call to setprecision() at the start of the program as shown below.
cout << setprecision (2);

An added benefit of using the setprecision() function is that it rounds the number as it is
displayed to that precision. Suppose that variable result contained 12.456789. Here are several
runs with different precisions; notice the effect of rounding. Notice with a precision of zero
digits, the fractional .4 is not .5 or greater so that the unit’s digit remains at 2.
double a = 12.456789;
cout << setprecision (5) << a << endl;
outputs: 12.45679
cout << setprecision (4) << a << endl;
outputs: 12.4568
cout << setprecision (3) << a << endl;
outputs: 12.457
cout << setprecision (2) << a << endl;
outputs: 12.46
cout << setprecision (1) << a << endl;
outputs: 12.5
cout << setprecision (0) << a << endl;
outputs: 12.

If the precision is not set, the output stream has a default of six digits. The setprecision()
function is a manipulator and requires the inclusion of the iomanip header file.

The second problem requires more work. We must tell the output stream that we do not
want to see floating point numbers in the scientific notational format, but rather in the fixed point
format that we are more used to seeing. The output stream contains a number of flags or switches
that indicate which format it is to use. The setf() or set flags function can be used to directly alter
those flags. The coding we need is
cout.setf (ios::fixed, ios::floatfield);
cout.setf (ios::showpoint);
These two work on any manufacturer’s compiler. However, Microsoft and others provide an
even more convenient way using the fixed manipulator function.
cout << fixed;
cout.setf (ios::showpoint);
The first function call to setf() or to fixed tells the cout stream to display all float fields in the
fixed point format, not in the scientific format. The second call tells the cout stream to show the
decimal point in all floating point numbers being displayed whenever the precision is set to 0
Numerical Processing 49

digits. Notice that the compiler must show a decimal point as long as the precision is greater than
0. However, when it is 0, do you or don’t you want to see the decimal point. showpoint forces
the decimal point to be displayed when the precision is 0. The identifiers prefixed with ios:: are
key identifiers found in the input/output stream class.

These two lines need to be coded before the first output of a floating point number to the
cout stream. Commonly, they are coded just after the variables are defined.

If one needed to return to the scientific notation for floating point numbers, one can code
cout.setf (ios::scientific, ios::floatfield);

There is yet another way to control the selection of fixed and scientific formatting of
floating point numbers. C++ standard namespace has a pair of manipulators to do this.
cout << fixed;
and
cout << scientific;
Once used, all floating point numbers are output in the fixed point format in the first case or the
scientific format in the second case. However, there is one small difference. If the variable a
from the above example that contained 12.456789 was subsequently output this way
cout << fixed << setprecision (0) << a << endl;
then the output is just 12 with no decimal point.

A final note. If the setprecision() function is used without using either the ios::fixed or
ios::scientific flags set by either of the two methods, then the precision passed to setprecision()
specifies the total number of digits to be displayed including those to the left of the decimal
point. If the setprecision() function call comes after specifically setting either fixed or scientific
formats, then the precision passed refers only to the number of digits to the right of the decimal
point.

Labeling Output

Commonly, output is labeled to make the meaning of the results clear to the reader. Let’s display
a single customer order given by the following variable definitions:
int qty = 10
double cost = 1.99
double total = 19.99;
When a program is to display a single set of results, it can be done as follows.
cout << "The quantity is: " << qty << endl
<< "The cost of each is: $ " << cost << endl
<< "The total of the order is: $" << total << endl;
And the result that this produces is this.
The quantity is: 10
The cost of each is: $ 1.99
Numerical Processing 50

The total of the order is: $19.99

Notice that I used sufficient blanks in the character string messages to make the numbers align in
a column and look good.

When displaying variables, it is important to also display the units of the resultant
numbers, if any. Suppose that the result of the calculations yielded the resistance of a circuit. The
output might be done as follows.
cout << "The resistance needed is "
<< resistance << " ohms.\n";
If the value currently in the variable resistance is 125, then the screen displays
The resistance needed is 125 ohms.

Similarly if a distance measurement resulted from calculations, one must output also the
units of that distance, otherwise it is a meaningless number.
cout << "The distance is " << distance << " km\n";
If the variable distance contains 125, this yields
The distance is 125 km

When displaying floating point numbers, there are some other additional details involved.
See the section below called Insertion of Floating Point Numbers into an Output Stream.

Math Operators — Calculations

The math operators in C++ consist of the following.
operator name Example Yields
+ addition 10 + 3 13
- subtraction 10 - 3 7
* multiplication 10 * 3 30
/ division 10 / 3 3
% remainder or 10 % 3 1
mod operator - applies only for integer types
Most of these operators are self-evident, except the integer divide and remainder operators.

When doing an integer division, the divide operator gives only the integer quotient. Thus,
10/3 gives 3 not 3.33333 because 3.33333 is a floating point or real number. If one did a floating
point division, as 10./3., then the result would be 3.33333.

The remainder operator gives the remainder of the integer division. When one codes
10 % 3, then quotient of 3 is discarded and the remainder of 1 is returned. From a mathematical
point of view, the remainder operator is sometimes called the mod operator. The remainder
operator cannot be applied to real or floating point numbers because there is no remainder
because the decimal fractional part is part of the result already.
Numerical Processing 51

Thus, if one wanted to calculate the totalCost, one could code

totalCost = cost * quantity;
This assumes that the variables cost and quantity have previously been given a value.

Assuming that the needed variables have been both defined and either given their initial
values or inputted from the user, many formulae can be computed. Here are some common ones.
force = mass * acceleration;
areaTriangle = .5 * base * height;
perimeterTriangle = side1 + side2 + side3;
salesCommission = sales * commissionRate;
average = sum / count;

Notice that in each of these lines, I surround every operator with one blank. It greatly aids
readability of the formula. Also note that all of the variables found on the right side of these
equations must have been previously given a value. The variables on the left side of the
assignment operator are given their values by the calculation results from the right side.

Precedence or Priority of Operators

Normally, the compiler performs the calculations from left to right as it works out the final value
of an equation. In the preceding example of calculating the perimeter of a triangle, the contents of
side1 are first added to the contents of side2 and then that result is added to the contents of side3
and the final result placed into perimeterTriangle.

However, multiply, divide and the remainder operators have a higher precedence or
priority than do the add and subtract operators. What is the order that the compiler follows in this
calculation?
e = a + b * c - d;
Here, the multiplication of b and c are done first. Next, the content of a is added to that result and
then the content of d is subtracted. Finally, the result is stored in e.

Use () to override the precedence of operators when it is needed. In the previous example,
if the result of a + b was supposed to be multiplied by the results of c - d, then parentheses are
required.
e = (a + b) * (c - d);

Here are some other examples that require parentheses.

sum = n * (n - 1) / 2;
x = a * ( b + c);
result = x * y / (z + 1);
Do not get parentheses happy and use parentheses where they are not needed. Why? A
common syntax error is one too many begin parentheses or one too few end parentheses.
Consider the following messy calculation.
Numerical Processing 52

result = (((a + b) + c) / (d - (e - (f * g)));

Can you spot the missing parentheses? The equation should have been written as follows.
result = (a + b + c ) / (d - e + f * g);
Now the intention of the programmer is quite clear.

Constant Data Objects

Frequently a program needs the use of a variable whose value is inherently a constant value, such
as the number of months in a year. In geometry problems, the value of PI is frequently needed; it
is a constant. In financial calculations, there are 100 pennies per dollar. Certainly one can simply
use a literal value, such as 12, for the number of months wherever that value is needed in a
program. Likewise, one can simply code 100 as needed in money problems. However, from a
program maintenance point of view, it is superior to have an actual constant identifier associated
with that value. It avoids confusion, aids program readability and facilitates changes when they
are required. Let’s see how.

Suppose a program that calculates monetary results also calculated some percentages. For
example, one might have the following.
percentLoss = Loss * 100 / Gross;
dollars = pennies / 100;
Now suppose that inflation strikes and the treasury decided that a dollar now requires 200
pennies. When changing the program, one might opt to globally change all 100 values into 200.
Yet, if that is done, suddenly all the percentage calculations are quite incorrect!

To alleviate the problem, one could define a variable penniesPerDollar as

int penniesPerDollar = 100;
and then the monetary calculations could make use of that variable like this.
percentLoss = Loss * 100 / Gross;
dollars = pennies /penniesPerDollar;
However, since penniesPerDollar is a variable, there is nothing to prevent the programmer from
accidentally coding
penniesPerDollar = 42;
Now the program produces correct results up to this point where the accidental assignment
occurred. What is needed is a way to denote that some variable is holding a constant value that
cannot be changed. This is known as a constant data object.

The syntax is quite simple — place the keyword const before the data type when defining
the variable. However, when making a variable a constant data object, one must at that time give
that variable the constant value that it is to contain. We could define the constant data object,
PenniesPerDollar and use it as follows.
Numerical Processing 53

const int PenniesPerDollar = 100;

percentLoss = Loss * 100 / Gross;

dollars = pennies /PenniesPerDollar;

There are two small points to be understood about using constant data objects. The first is
the case of the item’s name. By convention, programmers desire these constant objects to have a
clearly distinguishable name. Sometimes, all uppercase letters are used. However, in the case
above, the compound name would be unreadable, PENNIESPERDOLLAR. However,
capitalizing all of the words in a compound name works well.

The second point concerns the placement of the definition itself. Obviously, the definition
must occur before the first usage of that constant data object. Since these constant objects are
special, place them at the beginning of the main() function before all other variable definitions.
This location of a constant object will be changed later on in Chapter 6, when we learn to write
our own functions. Since the constants might well be used within these additional functions, they
have to be moved to the global namespace area above the function header for the main()
function.

Both of the following are correct.

#include <iostream>
#include <iomanip>
using namespace std;

const int PenniesPerDollar = 100;

int main () {
double pennies;
double dollars;

cout << "Enter the number of pennies: ";

cin >> pennies;

dollars = pennies / PenniesPerDollar;

cout << pennies << " equals $" << dollars << endl;
return 0;
}
and
#include <iostream>
#include <iomanip>
using namespace std;

int main () {
const int PenniesPerDollar = 100;
Numerical Processing 54

double pennies;
double dollars;

cout << "Enter the number of pennies: ";

cin >> pennies;

dollars = pennies / PenniesPerDollar;

cout << pennies << " equals $" << dollars << endl;
return 0;
}

However, the following is incorrect since the variables of main() are now defined outside
of main().
#include <iostream>
#include <iomanip>
using namespace std;

const int PenniesPerDollar = 100;

double pennies; // error defining outside of main
double dollars; // error defining outside of main

int main () {
cout << "Enter the number of pennies: ";
cin >> pennies;

dollars = pennies / PenniesPerDollar;

cout << pennies << "equals $" << dollars << endl;
return 0;
}
The above actually produces the correct results but is very bad program design. This is covered in
Chapter 7 at length. For now, making the variables of a program defined before the main()
function makes those variables known to the entire program everywhere. These two variables are
intended to be used only within main()’s calculations and nowhere else.

Math Library Functions

C++ has a large number of mathematical functions available for our use. To use any of the math
functions, be sure to use the #include <cmath> for the math header file.

For example, suppose that we needed to find the square root of a number. The built-in
square root function is called sqrt().
root = sqrt (number);
The prototype that is contained in the cmath header file defines the sqrt() function as taking one
Numerical Processing 55

double type parameter returning the square root of that parameter as a double type.

Have you noticed that there is no exponentiation operator in the C++ language? Suppose
that you needed to calculate xy or x raised to the yth power. The pow() (power) function is used to
handle exponentiation. The syntax of the pow() function is
double result = pow (base number, exponent number);
To calculate xy, we code
double answer = pow (x, y);

The pow() function has many uses, especially with scientific and Engineering formulae.
Suppose that we needed to find the 4th root of a number. We can code the following
double root = pow (number, .25);

Many trigonometric functions are provided; among these are sin(), cos(), tan(), asin(),
acos() and atan(). The sin(), cos() and tan() functions take an angle (of double type) in radians
and return a double result. To convert an angle in the more familiar degree units into radians,
remember that there are two PI radians in 360 degrees.

Suppose that one needed a program that would input an angle in degrees and display the
cosine of that angle. One could code the following; notice the comments help readability.
#include <iostream>
#include <iomanip>
#include <cmath>
using namespace std;

int main () {
const double PI = 3.14159;

double angle; // in degrees

double radians; // angle converted to radians
double result;

cout << "Enter the angle in degrees: ";

cin >> angle;

radians = angle * PI / 180.; // 2PI radians in 360 degrees

result = cos (radians);

cout << result << endl;

return 0;
}

Often it is clearer to solve the problem one step at a time. However, one could have coded
this problem with far fewer lines.
#include <iostream>
#include <iomanip>
Numerical Processing 56

#include <cmath>
using namespace std;
int main () {
const double PI = 3.14159;
double angle; // in degrees
cout << "Enter the angle in degrees: ";
cin >> angle;
cout << cos (angle * PI / 180.) << endl;
return 0;
}
Here the return value from the cosine function is immediately sent to the screen. No blank lines
are used to separate the distinct groups of thoughts either, running it all together. It is shorter but
much harder to read and write.

The Most Nearly Accurate Value of PI

Sometimes, one needs the most nearly accurate value of PI in a double, that is, the value of PI is
needed to 15 digits. Since there are 2 PI radians in 360 degrees, or PI radians in 180 degrees and
since the cosine of 180 degrees is –1., we can get PI to 15 digits by coding
PI = acos (-1.);
Thus, one of the best ways to define the most nearly accurate PI is
const double PI = acos (-1.);

Other Math Functions

Some other math functions include the following

Name Meaning Usage

abs absolute value of an integer int j = -3;

int x = abs(j);// x = 3

fabs floating point absolute value

exp expotential function ex double x = 1;

double y = exp (x);

log natural log double x = 9000.;

y = log (x); // 9.105

log10 log base 10 Y = log10(x);// 3.954

Numerical Processing 57

Some Additional Insertion Operator Details

The insertion operator << can be used to output either a constant, the contents of a variable or an
expression. Some examples are as follows.
// constants
cout << 42; // displays the integer 42
cout << "Hello World\n"; // displays a string "Hello World"

// variables
const double PI = 3.14159;
cout << PI; // displays 3.14159
int x = -123;
cout << x; // displays -123

// expressions
double angle = 60.;
cout << cos (angle * PI / 180); // displays .5

It is much easier to debug a program that does not output expressions but rather calculates
the result and outputs the result variable. Jamming everything into one line, like the above cosine
expression, is an “all or nothing” proposition. Either it comes out right or it doesn’t. If it doesn’t,
finding the error can be harder if there are no intermediate results to manually check.

Breaking a Complex Calculation Down into Smaller Portions

To illustrate the idea of breaking a complex equation down into more manageable
portions or sub-expressions, let’s consider an example from astronomy. Find the period of a
satellite in an orbit 100 kilometers above the earth. Here is the formula for determining the time
for one revolution or period.

In the equation Re is the earth’s radius of 6.378E6 meters, g is the force of gravity of 9.8
m/sec/sec and h is the height of the satellite in meters. In this case if a satellite is in orbit 100 km
above the earth, its period is about 1.4 hours which means that it passes by a ground-based
observer about 16 times a day. Note, it is not important how to derive this formula — that is the
arena of astronomy and orbital physics. Often, programmers are given a formula for some
problem and asked to write a computer program to solve it.

The best way to solve this complex equation is to first calculate various expressions and
then put the pieces together. First, define the constants in the problem, those that cannot vary.
This one has three
Numerical Processing 58

const double PI = acos (-1.);

const double Re = 6.378e6; // radius of earth
const double g = 9.8; // gravitational acceleration

Now define the variables and answer areas. The height of the satellite should be
considered a variable since we could easily change it.
double h = 100 * 1000; // convert using 1000 m per km
double period; // objective: find the period

Next, break the lengthy calculation into smaller pieces, defining a variable to hold each
piece.
double a = 2 * PI;
double b = sqrt (Re / g);
double c = 1. + h / Re;
double d = pow (c, 1.5);

Finally, put the smaller pieces together to form the result

period = a * b * d;
cout << period;

If anything goes wrong, you can add in debugging steps so that you can manually
determine which ones are incorrect.
cout << "a = " << a << endl
<< "b = " << b << endl
<< "c = " << c << endl
<< "d = " << d << endl;

However, one could also code a much shorter version as follows.

cout << 2 * PI * sqrt (Re / g) * pow(1.+h/Re, 1.5) << endl;
Here, one hopes that nothing goes wrong! Thus, unless there are some other overriding concerns,
always break complex calculations down into smaller more manageable units.

Here is the complete period of an orbiting satellite program. Notice the placement of the
various instructions.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic02a - Calculate the Period of a Satellite *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Basic02a Calculate the period of a satellite in orbit */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
Numerical Processing 59
* 11 *
* 12 int main () { *
* 13 const double PI = acos (-1.); *
* 14 const double Re = 6.378e6; // radius of earth *
* 15 const double g = 9.8; // gravitational acceleration *
* 16 double h; // convert using 1000 m per km *
* 17 double period; // objective: find the period *
* 18 *
* 19 // setup floating point format for output *
* 20 cout << fixed << setprecision (2); *
* 21 *
* 22 cout << "Enter the height of the satellite in Km: "; *
* 23 cin >> h; *
* 24 *
* 25 h = h * 1000; // convert to meters *
* 26 *
* 27 // compute subterms of the equation *
* 28 double a = 2 * PI; *
* 29 double b = sqrt (Re / g); *
* 30 double c = 1. + h / Re; *
* 31 double d = pow (c, 1.5); *
* 32 *
* 33 // compute final answer in terms of the subtrerms *
* 34 period = a * b * d; // period in seconds *
* 35 *
* 36 // output the results in km and in hours *
* 37 cout << "\nA satellite orbiting at a height of " *
* 38 << h / 1000 << " kilometers\nhas a period of " *
* 39 << period / 3600. << " hours\n"; *
* 40 *
* 41 return 0; *
* 42 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
And here is the output from a test execution.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic02a - Calculate the Period of a Satellite *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Enter the height of the satellite in Km: 100 *
* 2 *
* 3 A satellite orbiting at a height of 100.00 kilometers *
* 4 has a period of 1.44 hour *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Numerical Processing 60

Section B: Computer Science Example

Cs02a — Ticket Prices for a Concert
Acme Box Office requires a program to calculate the total cost of a customer’s tickets. Ticket
prices vary. Children less than 12 are charged 1/4 of the normal rate. Senior citizens pay 1/2 the
normal rate. Anyone person buying tickets could purchase a variable number of tickets from the
three categories. The program should display the number of tickets for each category along with
the cost for those tickets. A final line should contain the total cost for all of the tickets. Test the
program with a normal rate of $10.00 with two children, two adult and two senior citizen tickets
purchased.

Always design the solution first before attempting to program it. The starting point is to
determine what variables are going to be needed and draw the main storage or memory box. The
problem specifies that there is a basic ticket price; let’s call it basicPrice. However, since there
are two discount rates involved, let’s make those constant data objects, RateChild and
RateSenior. What other variables are needed? Three variables must hold the number of tickets
purchased in the categories. While the two rates and the basic price must be doubles, no one can
purchase .5967 of a ticket. The three number of purchased tickets should be integers. Let’s call
them numAdultTickets, numChildTickets and numSeniorTickets. Next, the problem
indicates we must display the total price of tickets purchased in each of these three categories.
We will need a double variable for each of these, say costAdult, costChild and costSenior.
Finally, the grand total cost of all tickets purchased can be called grandTotal. Figure 2.3 shows
what main storage for the program should be.

Now that we have drawn the picture and solidified the variable names, the program must follow
the Cycle of Data Processing. Thus, the next step is to input the data required. The following
accomplish this.

Figure 2.3 Main Storage for Tickets Sold Program

prompt “enter basic price of a ticket”

input basicPrice of a ticket
Numerical Processing 61

prompt “enter the number of adult tickets purchased”

input the numAdultTickets
prompt “enter the number of children’s tickets purchased”
input the numChildTickets
prompt “enter the number of senior citizen tickets purchased”
input the numSeniorTickets

With the input instructions finished, now work out the calculations that are required. In
this case, there are four simple ones. One can use English statements or pseudo-C++ lines. I’ll
use the latter to get the following:
costAdult = numAdultTickets * basicPrice;
costChild = numChildTickets * basicPrice * RateChild;
costSenior = numSeniorTickets * basicPrice * RateSenior;
grandTotal = costAdult + costChild + costSenior;

With the calculations complete, write a series of output instructions to display the title
and the results. Something like this should suffice.
print a nice heading
print identifier for adult tickets, numAdultTickets and costAdult
print identifier for child tickets, numChildTickets and costChild
print identifier for senior tickets, numSeniorTickets and costSenior
print identifier for grand total and grandTotal

Figure 2.4 Main Storage after the Program Is Desk Checked

When the design is complete, we must desk check the program for accuracy. Use the
main storage drawing as a sketch pad during desk checking. First, place the constant data value
of .25 in the RateChild box; .5 in the RateSenior box. Pretend you are running the program and
follow the series of prompt and input steps to input $10 in the basicPrice box and a 2 in each of
the three number of tickets variables. As you execute each line in the above calculations, carry
out that operation using the contents of the various boxes referred to in the calculation, placing
the result in the indicated box. For example, into the costAdult box place 2 * 10.00 or 20.00.
Into the costChild box goes 2 * 10.00 * .25 or 5.00. Into the costSenior box goes 2 * 10.00 * .5
or 10.00. In the grandTotal box, place the results of adding up the three cost boxes, 35.00.
Finally, carry out the print instructions. Look over the results displayed. Are they correct? If so,
thoroughly desk check the solution. Experiment with other initial starting values. Only when the
Numerical Processing 62

solution is 100% correct, do we then convert it into a C++ program. Figure 2.4 shows what the
main storage diagram should contain when this initial set of test data has been processed by the
program.

Here are the completed program and the output from the test run.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs02a — Calculate Ticket Prices *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs02a: calculate concert ticket prices */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 using namespace std; *
* 10 int main () { *
* 11 const double RateChild = .25; // children's rate *
* 12 const double RateSenior = .5; // senior citizen's rate *
* 13 *
* 14 double basicPrice; // basic price of a ticket *
* 15 int numAdultTickets; // number of adult tickets purchased *
* 16 int numChildTickets; // number of child tickets purchased *
* 17 int numSeniorTickets; // number of senior citizen tickets *
* 18 *
* 19 double costAdult; // the cost of all adult tickets purchased *
* 20 double costChild; // the cost of all child tickets purchased *
* 21 double costSenior; // the cost of all senior tickets purchased*
* 22 double grandTotal; // the grand total cost of all tickets *
* 23 *
* 24 // setup floating point format for output of dollars *
* 25 cout << fixed << setprecision (2); *
* 28 *
* 29 // output a title *
* 30 cout << "Acme Box Office Ticket Sales\n\n"; *
* 31 *
* 32 // prompt and input the user's data *
* 33 cout << "Enter the basic price of a ticket: "; *
* 34 cin >> basicPrice; *
* 35 cout << "Enter the number of adult tickets purchased: "; *
* 36 cin >> numAdultTickets; *
* 37 cout << "Enter the number of child tickets purchased: "; *
* 38 cin >> numChildTickets; *
* 39 cout << "Enter the number of senior tickets purchased: "; *
* 40 cin >> numSeniorTickets; *
* 41 *
* 42 // compute ticket costs *
* 43 costAdult = numAdultTickets * basicPrice; *
* 44 costChild = numChildTickets * basicPrice * RateChild; *
* 45 costSenior = numSeniorTickets * basicPrice * RateSenior; *
Numerical Processing 63
* 46 *
* 47 // compute grand total cost *
* 48 grandTotal = costAdult + costChild + costSenior; *
* 49 *
* 50 // output the results *
* 51 cout << endl; *
* 52 cout << "Number of adult tickets: " *
* 53 << setw(4) << numAdultTickets << " = $" *
* 54 << setw (6) << costAdult << endl; *
* 55 cout << "Number of child tickets: " *
* 56 << setw(4) << numChildTickets << " = $" *
* 57 << setw (6) << costChild << endl; *
* 58 cout << "Number of senior citizen tickets: " *
* 59 << setw(4) << numSeniorTickets << " = $" *
* 60 << setw (6) << costSenior << endl; *
* 61 cout << "Total purchase price: $" *
* 62 << setw (6) << grandTotal << endl; *
* 63 *
* 64 return 0; *
* 65 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from a Sample Run of Cs02a — Calculate Ticket Prices *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Acme Box Office Ticket Sales *
* 2 *
* 3 Enter the basic price of a ticket: 10.00 *
* 4 Enter the number of adult tickets purchased: 2 *
* 5 Enter the number of child tickets purchased: 2 *
* 6 Enter the number of senior tickets purchased: 2 *
* 7 *
* 8 Number of adult tickets: 2 = $ 20.00 *
* 9 Number of child tickets: 2 = $ 5.00 *
* 10 Number of senior citizen tickets: 2 = $ 10.00 *
* 11 Total purchase price: $ 35.00 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

There are a number of things about the completed program to notice. First, the extensive
use of comments greatly aids readability along with the use of descriptive names. Second, line
breaks separate each logical group of actions, such as variable definitions from calculations from
output operations. Third, lines 25-27 set up the cout output stream for proper floating point
output of dollars. Finally, lines 52-62 carefully control spacing so that all of the output results
form consistent columns making the report easy to read.
Numerical Processing 64

Section C: Engineering Example

Engr02a — Pressure Drop in a Fluid Flowing Through a Pipe
(Civil Engineering)
Consider the problem of an incompressible fluid being pumped through a pipe at a steady rate.
The drop in pressure from point one to point two in the pipe is given by
pressureDrop = P1 - P2 = d ( g h + Eloss)
where d is the fluid density, g the gravitational constant, h the difference in height between points
one and two, and Eloss is the energy loss per kilogram from internal friction with the walls of the
pipe. The energy loss expression is

where f is the friction factor, v is the velocity of flow, L is the length of the pipe and D is the
pipe’s diameter. The velocity of fluid flow is given by

where is the volume flow rate. Finally, for smooth pipes, the friction factor f depends only on the
Reynold’s number R given by

and u is the viscosity of the fluid. If R is less than or equal to 2,000, then the friction factor is
f=8/R for laminar flow (non-turbulent).

Calculate the pressure drop of ethyl alcohol whose density is 789.4 kg/m3, whose
viscosity u is 0.0012 kg/m-sec through a pipe that is .01 meters in diameter and 100 meters long
with a height difference of 10 meters at a volume flow rate Q of 0.00002 m3/sec.

The starting point is to design a solution on paper. In this problem, there are a large
number of constant initial values. They can be either constants or actual input values. Since the
problem did not specifically state that they must be input, we can store them as constant data
objects. Let’s identify those and their values first. The constant Density is 789.4; the constant
Viscosity is 1.2E–3; the constant Height is 10.; the constant Diameter is .01; the constant
Length is 100.; the constant Q is 2.0E–5; the gravitational constant g is 9.8; and finally PI. Draw
a series of memory boxes for each of these and place these constant values in them. Next identify
the variables needed for the calculations. Here I have called them velocity, reynolds, friction,
eloss and pressureDrop. Make up another five boxes and label them with the chosen names.
Figure 2.5 shows the completed Main Storage box.
Numerical Processing 65

Figure 2.5 Main Storage for Pressure Drop Problem

Next, write out in English or pseudo C++ the calculations that are required in the order to
find the resultant pressure drop.
velocity = 4. * Q / (PI * Diameter * Diameter)
reynolds = Density * velocity * Diameter / Viscosity
friction = 8. / reynolds
eloss = 4. * friction * velocity * velocity * Length / Diameter
pressureDrop = Density * (g * Height + eloss)

Finally, design how the results are to be displayed and code those instructions. It is an
excellent idea to echo these starting values or constants before displaying the final answer, the
pressure drop. Also, for debugging purposes, let’s also display the results of the four intermediate
calculations. So we should sketch the following:
print a title, the Density, Viscosity, Diameter, Length, Height and
the flow rate Q all appropriately labeled
print the velocity, reynolds, friction and eloss results, also labeled
print the final answer pressureDrop nicely labeled

Now desk check the solution. Using the numbers placed in the constant object boxes, step
through the program line by line, doing each calculation and placing the results in the
corresponding box, beginning with velocity. Use your handy-dandy pocket calculators as needed.
When you have verified the solution works, then convert it into a program. In our case, the only
difficulties are in the formatting of the printed results.

Here are the final program and the output from the test run.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr02a - Calculate the Pressure Drop of Ethyl Alcohol in a Pipe *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr02a: Calc pressure drop of Ethyl Alcohol in a pipe */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
Numerical Processing 66
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 int main () { *
* 12 const double Density = 789.4; // fluid density in kg/m/m/m *
* 13 const double Viscosity = 1.2E-3;// fluid viscosity in kg/m-sec *
* 14 const double Height = 10.; // height between P1 and P2 in m*
* 15 const double Diameter = .01; // pipe diameter in m *
* 16 const double Length = 100.; // pipe length in m *
* 17 const double Q = 2.0E-5; // volume flow rate- cubic m/sec*
* 18 const double g = 9.8; // gravity acceleration constant*
* 19 const double PI = acos (-1.); *
* 20 *
* 21 double velocity; // velocity in the pipe *
* 22 double reynolds; // Reynold's number *
* 23 double friction; // friction factor *
* 24 double eloss; // energy loss *
* 25 double pressureDrop; // the pressure drop between P1 and P2 *
* 26 *
* 27 // setup floating point format for output - set for 4 dec digits*
* 28 cout <<fixed << setprecision (4); *
* 31 *
* 32 // perform the calculations *
* 33 velocity = 4. * Q / (PI * Diameter * Diameter); *
* 34 reynolds = Density * velocity * Diameter / Viscosity; *
* 35 friction = 8. / reynolds; *
* 36 eloss = 4. * friction * velocity * velocity * Length / Diameter;*
* 37 pressureDrop = Density * (g * Height + eloss); *
* 38 *
* 39 // display the initial specifications *
* 40 cout << "Ethyl Alcohol Pressure Drop in a Pipe\n"; *
* 41 cout << " of density = " << setw (9) << Density *
* 42 << " kg/cubic meter\n"; *
* 43 cout << " viscosity = " << setw(9)<<Viscosity<<" kg/m-sec\n";*
* 44 cout << " pipe specs\n"; *
* 45 cout << " diameter = " << setw (9) << Diameter << " m\n"; *
* 46 cout << " length = " << setw (9) << Length << " m\n"; *
* 47 cout << " from height = " << setw (9) << Height << " m\n"; *
* 48 cout << setprecision (5); *
* 49 cout << " flow rate = " <<setw(9)<<Q<<" cubic meter/sec\n\n";*
* 50 *
* 51 // display the intermediate results *
* 52 cout << "Velocity: " << setw (12) << velocity << endl; *
* 53 cout << "Reynolds: " << setw (12) << reynolds << endl; *
* 54 cout << "Friction: " << setw (12) << friction << endl; *
* 55 cout << "Energy Loss: " << setw (12) << eloss << endl; *
* 56 *
* 57 // display the desired final pressure drop *
* 58 cout << "Pressure Drop: " << setw (12) << pressureDrop << endl; *
* 59 *
* 60 return 0; *
* 61 *
Numerical Processing 67
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Engr02a - Calculate the Pressure Drop *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Ethyl Alcohol Pressure Drop in a Pipe *
* 2 of density = 789.4000 kg/cubic meter *
* 3 viscosity = 0.0012 kg/m-sec *
* 4 pipe specs *
* 5 diameter = 0.0100 m *
* 6 length = 100.0000 m *
* 7 from height = 10.0000 m *
* 8 flow rate = 0.00002 cubic meter/sec *
* 9 *
* 10 Velocity: 0.25465 *
* 11 Reynolds: 1675.15883 *
* 12 Friction: 0.00478 *
* 13 Energy Loss: 12.38723 *
* 14 Pressure Drop: 87139.67970 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
All the constant data objects begin with a capital letter except g for gravity. One could
have uppercased all of these. However, g is the universal symbol for gravity; it is far better to use
universal symbolic names when they are available as legal C++ variable names.

Notice that the comments greatly aid the readability of the program and that the line
breaks tend to group the different logical actions such as defining constants, variables,
calculations and outputting results.

Observe how the uniform spacing of the output fields and text was achieved. By placing
each output line on a separate line, one can align the literal text strings. Also, use a uniform field
width whenever possible. Finally, notice lines 30 and 48. Initially, the precision was set to four
decimal digits. This was sufficient for all the constant initial values except the flow rate, Q,
which needed five digits. By having five digits in the result values, the friction result is well
displayed, but the Reynolds number and the final pressure drop certainly do not need so many
digits to the right of the decimal point. It was done this way so that the column of result numbers
aligned on their decimal points.

New Syntax Summary

Keyboard Input Stream: cin
extraction operator: >>
1. Skips over any whitespace (blanks, tabs, CR’s, LF’s) to first
non-whitespace character

2. Extracts characters that are appropriate for the type of data to be

inputted
Numerical Processing 68

3. Stops when it encounters whitespace, an inappropriate character,

or the end of the stream of characters (end of the file)

chaining of extraction operator: multiple values are separated by whitespace

cin >> qty >> cost;

CRT Output Stream: cout

insertion operator: <<
default: displays only the significant digits in the value

controlling output — the setw (n) function

1. The set width function applies only to the very next item output

2. The integer value of n instructs the insertion operator to make

this item occupy n columns whenever possible

3. By default, the value output is right aligned within the n columns

4. If there are more significant digits in the value than n, all the
significant digits are shown; the requested width is ignored

5. For numbers, the width n includes any minus sign and decimal
point (if the value is floating point)

Floating Point Output: scientific format .12300e3 versus fixed point format 123.00
Default is scientific format

Switching to fixed point format can be done two ways — with

either method, once set, it remains set, so this is often only done
one time as the program starts.
1. Use the fixed manipulator function
cout << fixed;
2. Use the set flags function with the ios::fixed flag
cout.setf (ios::fixed, ios::floatfield);

Use the setprecision manipulator function to set the

precision, which is the number of digits to the right of the decimal
point to be shown. Note that the output is always rounded at this
location. Once a precision is set, it applies to all subsequent
floating point output until another setprecision call changes it.
cout << setprecision (2);
This requests two digits, typically for a dollar amount field.
Numerical Processing 69

Exception: when the precision is set to 0, no digits to the right of

the decimal point, the default is to not show any decimal point. If a
decimal point is required, then use the set flags function passing
the ios::showpoint value.
cout.setf (ios::showpoint);

Commonly Used Math Library Functions Found (uses the header file <cmath>)
sqrt prototype: double sqrt (double number);
returns the square root of the number

pow prototype: double pow (double base, double power);

returns the base raised to given power

sin prototype: double sin (double angleInRadians);

returns the trigonometric sine of the angle passed

cos prototype: double cos (double angleInRadians);

returns the trigonometric cosine of the angle passed

tan prototype: double tan (double angleInRadians);

returns the trigonometric tangent of the angle passed

asin prototype: double asin (double sinevalue);

returns the angle in radians for the passed sine value

acos prototype: double acos (double cosinevalue);

returns the angle in radians for the passed cosine value

atan prototype: double atan (double tanvalue);

returns the angle in radians for the passed tangent value

abs prototype: long abs (long value);

returns the absolute value of the passed long integer value

fabs prototype: double fabs (double value);

returns the absolute value of the passed double value

exp prototype: double exp (double power);

returns e raised to the passed power — ex
Numerical Processing 70

Design Exercises
1. Mysterious “crop circles” sometimes appear in a farmer’s corn field. A crop circle is an area in
the middle of his corn field in which all of the corn stalks have been trampled flat, yielding some
design visible only from the air. Farmer Jones discovered just such a circle in his field. Since the
plants were smashed, Farmer Jones suffers a crop yield loss. His crop insurance covers some of
his lost income by paying him a rate of $2.25 per bushel of corn lost. His yield on the remainder
of that field is 125 bushels per acre. He measured the crop circle and found it was 50 feet in
diameter. How much money does he get from the crop insurance? Hint, the area of a circle is
given by PI times the square of the radius and an acre is 4840 square yards.

2. It’s party time. You are planning a party and have $40 with which to buy as many refreshments
as possible. But not every guest prefers the same refreshments. Six guests prefer pizza while
eight prefer to eat a hot dog. Four guests like Pepsi, eight prefer Coca-Cola, and two want Dr.
Pepper. A pizza comes with eight large slices and costs $9.00. Hot dogs cost $1.25 each. A six-
pack of any kind of soda pop costs $3.50. The rules you must follow are:
All guests must have something they like to eat and drink.
Soda pop can only be purchased in six-packs.
Pizza can only be bought as a whole pizza with eight slices.
What is the minimum that you must buy to satisfy these criteria? Do you have enough money to
pay for it? (Ignore sales taxes.)
Numerical Processing 71

Stop! Do These Exercises Before Programming

1. The programmer was having a bad day writing this program. Correct the errors so that there
are no compile time errors. Make whatever assumptions you can about data types that need fixing
up.
#includ <iostrea>
Const int TOTAL 100;
int main () {
Double costOfItem;
quantity = 42;
double total cost;
cupon_discount int;
const double AmountPaid;
cost of item = 4.99
AmountPaid = 9.99;
...

2. Circle the variable names that are invalid C++ variable names. Do not circle ones that are
legal, even though they might not represent the best naming convention.
CostOfGoodsSold
total Price
C3P0
3D-Movie
distance Traveled
sin
fAbs
Log
qty_sold
qty sold
qtySold

3. Convert each of these formulas into a proper C++ statement.

a. F = m a (Force = mass times acceleration)

b. A = PI R2 (area of a circle)

c.
Numerical Processing 72

d. x = sin (2 PI y);

e.

f.

g.

h.

4. Correct the errors in these C++ calculations.

a. cost = qty unitPrice; // cost is qty times unitPrice

b. sum = sum + + count; // add count to sum
c. count + 1 = count // add one to count
d. root = sqrt x * x + y * y;
// x is the square root of x squared + y squared
e. xy = Pow (x, y); // calculate x raised to the yth power
f. count + 1; // increment count

5. The equation to be solved is this

Assuming all variables are doubles, which of the following correctly calculates the
percentage? Next, assuming all variables are integers, which of the following correctly calculates
the percentage? Indicate which work for doubles and which work for integers by placing an I or a
D before each letter.

a. percent1 = salesTotal1 / salesTotal1 + salesTotal2 * 100;

b. percent1 = salesTotal1 / (salesTotal1 + salesTotal2 * 100);

Numerical Processing 73

c. percent1 = salesTotal1 / (salesTotal1 + salesTotal2) * 100;

d. percent1 = ((salesTotal1) / (salesTotal1 + salesTotal2)) * 100;

e. percent1 = salesTotal1 * 100 / salesTotal1 + salesTotal2;

f. percent1 = salesTotal1 * 100 / (salesTotal1 + salesTotal2);

6. Show the precise output from the following series of cout instructions. Assume these are the
initial values of the variables. Assume the ios::fixed has been set along with ios::showpoint.
The precision has not been set to any value initially.
int x = 123;
double z = 42.35353;

a. cout << setw (5) << x << x;

b. cout << x << setw (5) << x;

c. cout << setprecision (2) << z

<< setw (7) << setprecision (3) << z;

d. cout << setprecision (4) << setw (8) << z;

7. For each of these short calculations, show the result that is displayed. Assume that ios::fixed
and ios::showpoint have been set on cout and that the precision is set to two decimal digits
unless overridden.

a.
int x = 10;
int y = 4;
cout << x / y;
Numerical Processing 74

b.
int pennies = 123;
const int QUARTERS = 25;
int quarters;
quarters = pennies / QUARTERS;
pennies = pennies % QUARTERS;
cout << quarters << " " << pennies;

c.
double number = 100;
cout << sqrt (number);

d.
double num = 10;
double bignum;
bignum = pow (num, 2);
cout << setprecision (0) << bignum;
Numerical Processing 75

Programming Problems
Problem Cs02-1 — Conversion of a Fahrenheit Temperature to Celsius

When dealing with temperatures, one common problem is the conversion of a

temperature in Fahrenheit degrees into Celsius degrees. The formula is

Write a program that converts a constant Fahrenheit temperature into the corresponding
Celsius temperature. Prompt the user to enter a Fahrenheit temperature. Then, calculate the
Celsius equivalent and display the results using the format shown below. You should produce
precisely these results; observe the formatting. Make three test runs of the program entering the
indicated Fahrenheit temperatures.
100.0 F = 37.8 C
32.0 F = 0.0 C
212.0 F = 100.0 C

Problem Cs02-2 — Format Control

Write a program that inputs two integers and outputs their difference and their product
using this precise format.
123 123
- -10 X -10
-------....-------
133 -1230
Assume that both integers, their difference and their product do not exceed five digits. Either or
both may be negative. Test the program by entering the two integer values shown above. Then
test the program with these two additional sets.
–12345 and 2
1234 and –6
Numerical Processing 76

Problem Cs02-3 — Monthly Mortgage Payment Calculator

How much will my monthly payments be? This is a common question new home buyers
frequently need answered. The following formula calculates the monthly payment

where P is the monthly payment, A is the loan amount, r is the monthly interest rate and n is the
number of monthly payments to be made.

Unfortunately, the loan statistics are not often in these units. Prompt the user to enter the
values for the loan amount, the annual interest rate percentage and the loan length in years.
Then compute the monthly payment and output the results as shown below. Note that you must
convert the initial data as entered by the user into the proper quantities needed in the formula.
The monthly interest rate is 1/12th of the annual rate and is not in percentage format. All
variables should be doubles.

Make four test runs of the program entering the indicated values shown below. The
output should look like this
----------Input-------------- Results
Loan Annual Length Monthly
Amount Rate in Years Payment
$ 50000 11.50% 25.0 $ 508.23

2nd run
----------Input-------------- Results
Loan Annual Length Monthly
Amount Rate in Years Payment
$ 24800 7.80% 25.0 $ 188.14

3rd run
----------Input-------------- Results
Loan Annual Length Monthly
Amount Rate in Years Payment
$1000000 5.00% 30.0 $5368.22

4th run
----------Input-------------- Results
Loan Annual Length Monthly
Amount Rate in Years Payment
$ 9500 14.75% 5.0 $ 224.76
Numerical Processing 77

Problem Engr02-1 Falling Objects

The equation that describes the height of a falling object as a function of time is

where
y0 is the initial height of the object
v0 is the initial velocity of the falling object
y is the final height of the object
t is the time
g is the gravitational acceleration: 9.8 m/s/s

Write a program that, prompts the user to input a set of values for the initial height, initial
velocity and the final height. Then calculate and display the number of seconds until the object
reaches the final height. All data are in the metric system. Assume that the final height is less
than the initial height. All program variables should be doubles. Use the quadratic equation to
solve for t. However, you must determine whether to use the + or – root.

For the first run, use these values for the initial height, initial velocity and final height:
100., 0., 0. Then when that is producing the correct results, rerun the program and enter these
three: 1000., 100., 100.

With the program now verified as operational, use it to solve this problem. An
astronomer has detected an asteroid that is on a collision course toward the earth. When it is
detected, the asteroid is located 80,000,000 m away or about 50,000 miles moving directly
toward the earth at a speed of 20 m/s or 45 mph. How much time do we have to take preventive
measures, such as launching a nuclear strike to break the asteroid into small fragments? (Assume
the final height is zero for a collision; ignore orbital considerations as well as air friction effects.)

Turn in the output of the two initial test runs as well as the asteroid run.
Additional Processing Details 78

Chapter 3 — Additional Processing Details

Section A: Basic Theory

Introduction
This chapter introduces more of the numerical data types. How these data are really stored in
memory is discussed. This leads to a discussion of the effects of data types upon calculations or,
more precisely, data conversion. Finally, some additional C++ math operators are presented.

The Complete Integer Data Types

There are actually eight different integer data types. Integers fall into two categories, signed and
unsigned. A signed integer may have a sign — either + or –; if no sign is present, it is assumed
to be positive. Unsigned integers cannot ever be negative; they are assumed to be positive.

Some examples of a signed integer are: +10, –32, +42, and 11 where this last one is
assumed to be +11. Some examples of an unsigned integer are: 11, 42, 88, 1.

Each type of integer comes in four sizes. The following Table 3.1 shows these data types,
the number of bytes they occupy and the range of values that each can hold. This list will be
expanded when the new 64-bit computers arrive.

Note that the int and unsigned int are both platform dependent. Under old DOS, they are
2 bytes, but under Win32 console applications and similar 32-bit platforms like main frame
computers and Unix, they are 4 bytes in size.
Additional Processing Details 79

Table 3.1 The Integer Data Types

Data Type Number Range of values
of Bytes

signed:
char 1 +127 to -128
short 2 +32,767 to -32,768
int* 2 or 4 as a short or a long
long 4 +2,147,483,647 to -2,147,483,648

unsigned:
unsigned char 1 0 to 255
unsigned short 2 0 to 65,535
unsigned int* 2 or 4 as an unsigned short or unsigned long
unsigned long 4 0 to 4,294,967,295

* int and unsigned int: platform dependent

Why does C++ have all these different sizes? Would not one size, the long, serve all of
them? The major reason for the different sizes, besides backwards compatibility with existing
applications and data bases, is to reduce total memory requirements of a program and of files on
disk. While our programs are extremely simple at this point, it does not take much imagination to
envision a program storing vast quantities of similar data to be processed.

Suppose that no one could order more than 127 of any given item. Then, defining the
quantity as a char would make sense. If there were 1,000,000 customer orders in the data base,
the amount of memory saved by making the quantity be a char instead of a long would be
(4 – 1) * 1,000,000 or three million bytes!

Some variables are inherently small in range. Take for example the x and y screen
coordinates of a colored pixel dot on the CRT. In high resolution mode, there are 1024 dots
horizontally and 768 vertically. If your program were plotting points, it would make sense to
store those (x,y) coordinates as shorts not longs.

Here is another example; companies often give their departments a number to identify
them. Typically, the department numbers are stored as a char because few companies have more
than 127 departments.

Which Type of Data Do I Use in My Program?

Well, that all depends on the maximum value that each integer variable is to hold. Sometimes the
size is given in the problem specifications. If it is not, then it is the programmer’s task to
Additional Processing Details 80

determine the best choice to use. There is no escaping the fact that the programmer must know
what the range of values for any given variable is expected to contain.

However, if you do not know which to use, try using an int. In fact, the int is the most
commonly used integer data type, even though it is platform dependent. Why is the int so
commonly used? The reason is that C++ language specifications allow the compiler makers to
create the fastest possible executing integer math instructions for the int type of data. Thus,
program developers who use an int data type are guaranteed the best performance on any
platform. For example, under old DOS, performing math operations on long variables is
significantly slower than if int variables are used. However, if the program is compiled and run
on a Win32 console platform, then the longs and ints are entirely equivalent in terms of speed.

The decisions are destined to become even more complex. The newer computers just now
coming out have these high speed work register circuit elements 64-bits or 8 bytes in size. Look
for full support of 8-byte integers in the next release of the compilers. Following that in the not
so distant future are the 128-bit or 16-byte computers.

How Integer Data Is Stored in Memory

While a detailed knowledge of exactly how an integer is stored in memory circuits is not needed
for most programming applications, it greatly aids one’s understanding of the details of integer
math. Recall that the computer is really a binary machine, that is, everything is either on or off,
electricity or no electricity, 1 or 0. Suppose that one defined a char variable called x and stored a
1 in it.
char x = 1;
Since a char occupies one byte of memory and since a byte consists of eight connected bits or
circuit elements capable of storing a 0 or a 1, the decimal number 1 is stored in x as follows
0000 0001
where each binary digit represents a power of two. The 1 is in the 20 position and means 1 times
20 — anything to the 0th power is 1. So we have 1 as the decimal number.

The number 2 would be stored as

0000 0010
where the 1 means 1 times 21 or 2 in decimal.

The number 5 is stored as

0000 0101
which is 1x 22 plus 0x 21 plus 1x 20 which is 4 + 0 + 1 or 5.

The sign is always the first bit of the entire field, here the left-most bit. A 0 means the
number is positive and a 1 means it is negative. What is the largest number that can be stored in a
char? The left-most bit must remain a 0 so that the number is positive, but the rest of the bits are
1's.
Additional Processing Details 81

0111 1111
which is 1x26 + 1x25 +1x24 +1x23 +1x22 +1x21 +1x20 = 64 + 32 + 16 + 8 + 4 + 2 + 1 which is
+127.

Curiously, what happens if you add one to a char number that currently has +127 in it?
Let’s perform that binary addition problem and see.
0111 1111
+0000 0001
----------
1000 0000
In binary, 1 + 1 is 10, or a 0 and carry your 1. A carry and 1 is 10, or 0 and another carry. It works
just like elementary addition in the decimal system. But look what happened to the sign bit! It is
now a 1 bit which means the whole number is negative! In fact, this is actually how a –128 is
stored! What has happened is that the result has overflowed the contents of a char sized
variable.

On the other hand, if we define y to be an unsigned char, now the leftmost bit is part of
the number. A 1 bit here means 1x27 or 128 added into the total. So the maximum value that can
be stored in an unsigned char is 255
1111 1111
which is 1x27 + 1x26 + 1x25 +1x24 +1x23 +1x22 +1x21 +1x20
or 128 + 64 + 32 + 16 + 8 + 4 + 2 + 1 or 255.

Ok. What happens if we add one to an unsigned char that holds 255 currently? We have
the following
1111 1111
+0000 0001
----------
0000 0000
The leftmost carry is going into a nonexistent 9th bit and is simply pitched by the circuitry. Again
this is called an overflow.

Integer Variable Overflow

When integer math is done, the programmer must be alert for the possibility of overflow. Failure
to do so can result in silly results. Let’s see how this comes about. Suppose that the programmer
has defined the following variables and does the indicated calculation
short quantity = 10;
short cost = 10000;
short total;
total = cost * quantity; // potential error is here
Clearly, a short is an excellent choice for the quantity ordered and the cost of the item (say a
used car for example) is also fine as is. Both numbers, 10 and $10,000, fit nicely in a short. But
Additional Processing Details 82

what happens when the multiply instruction is executed? The numerical result is 100,000.
However, a short can only hold a maximum of 32,767. What occurs during this multiply
operation is overflow — namely all bits beyond the 16 bits that a 2 byte short can hold are
discarded. It is even worse, because in this case, the result has a 1 in the left-most bit position of
the short result, indicating the result is some negative number when it was supposed to mean
1x215 added into the result. In fact if you displayed the result, it is –31072.

How can it be fixed? The crudest way to fix the overflow problem is to make all variables
a long. But in the real world this is not often possible. In a later chapter our programs input
master files from disk where the data is stored in a short and must be input that way (See the
Binary File section in Chapter 13.). Thus, we must find a way to make this work short of making
all variables a long. After examining the floating point data types, we will examine this
conversion problem fully.

The Complete Floating Point Data Types

Just as there are more than one size of integer data types, so also there are additional floating
point data types. There are actually three types of floating point numbers as shown in Table 3.2.

Table 3.2 The Floating Point Data Types

Data Type Number Range Decimal Powers
of Bytes of values Digits of Ten

float 4 ±3.4x10-38 6 3/4 10+38 - 10-38

+38
±3.4x10
double 8 ±1.7x10-308 15 10+308 - 10-308
+308
±1.7x10
long double 10 ±3.4x10-4932 19 10+4932 - 10-4932
±1.7x10+4932
These values are for the Microsoft Visual C++ compiler, but most implementations are similar.
Note: in VC6, to get the long double as 10 bytes, you need to link to LIB/FP10.obj and include it
before LIBC.LIB, LIBCMT.LIB, MSVCRT.LIB on the linker command line.

As you look over these data types, pay careful attention to the number of decimal digits a
variable of that type can hold. The basic float data type is seldom used because it only offers six
decimal digits of accuracy. Suppose that you made the grand total monthly sales variable be a
float type. Only six digits can be accurately stored, $9,999.99 would be the largest total
accurately represented. In financial calculations, this would be considered a disaster!

Thus, the most frequently used floating point data type is the double. A double gives 15
digits of accuracy, which is usually totally sufficient. Notice that the exponent or power of ten is
seldom a problem with any of these. 1038 is a very large number; a one with 38 zeros after it.
Additional Processing Details 83

Also, math operations with the long double are the slowest math instructions on any computer
and are exceedingly rarely used.

Principles of Data Conversion

When math is performed on two variables or constants, the C++ rules specify that both must be
of the same data type. Thus, C++ can add two ints, two shorts, two chars, two unsigned longs,
for example. But it cannot do math on unlike items. Consider what happens when the compiler
encounters this calculation.
short quantity;
long cost;
long total;
total = cost * quantity;
The compiler is forced to do some data conversion, because it cannot do the long * short
multiplication. (Note: the C++ data conversion rules are a bit different than the older C
conversion rules.)

Data Conversion Rule 1. All types of char and short (signed and unsigned) are
automatically promoted to an int in calculations. (On an old DOS platform where an int is only 2
bytes, then an unsigned short is converted to an unsigned int instead because an unsigned
short can contain a larger value than can be held in a 2 byte int.)

This makes calculations involving a mixture of the smaller sized integer types very
convenient. For example, assume the following definitions and calculation.
char a;
short b;
int c;
c = a + b;
Here the compiler automatically promotes both a and b to an int by temporarily creating a pair of
int variables and converting a and b into these temporary variables. Then, it does the calculation
which results in an int answer and stores that answer into c which is also an int. Finally, it
deletes the memory that was used by these temporary variables. Table 3.3 Data Type Ranking
shows the ranking of the different data types from the worst at the top to the least at the bottom.
long double

double

float

unsigned long

long

unsigned int

int
Table 3.3 Data Type Ranking
Additional Processing Details 84

Data Conversion Rule 2. When an operator joins two values that have different data
types, it converts the one of the lesser rank into a temporary instance of the data type of the
higher rank. Now math can be performed on data of the same data type. Another way of saying
this is that the compiler always converts to the worst data type. Unsigned numbers are worse than
signed numbers; longs are worse than ints; floats are worse than longs and so on.

This conversion is precisely what happens when the compiler encounters the first
example in this section.
total = cost * quantity;
Since quantity is a short and cost is a long, the short is promoted to the higher rank, a long.
The compiler allocates a temporary long variable and converts quantity into that temporary long
variable. The calculation is then long * long yielding a long result which is assigned to total
which is also a long.

Consider this messy conversion problem.

char a;
short b;
int c;
long d;
long result = a * b + c * d;

Since there are no parentheses, the normal precedence of operators applies. First, the
compiler does a times b, but it cannot multiply char times short. Data Conversion Rule 1
applies to both operands. The compiler converts both the char a and the short b into a temporary
int variables. It then does int * int giving an int result. Next, it cannot multiply int times long.
Here, Data Conversion Rule 2 applies. The worse type is long. It converts the int c variable into
a temporary long and does long times long yielding a long result. Now it goes back to add the
two partial results and discovers it cannot add an int result of the first multiplication and a long
result of the second one, so the int intermediate result of a * b is converted into a long. The
compiler can then add long plus long yielding a long final result. At last, it copies the long
answer into the result variable which is also a long and the compiler deletes all of the temporary
variables it used.

Assigning Smaller Sized Integers to Larger Sized Integers

One can always assign a smaller sized integer value to a larger sized integer variable. For
example, a long can easily hold the maximum value that any int, short or char can hold.
Similarly a short can hold any possible value that a char might contain.

The following assignments are always safe.

char a;
unsigned char b;
short c;
unsigned short d;
Additional Processing Details 85

int e;
long f;
c = a;
c = b;
e = c;
e = d;
f = e;

Assigning Larger Sized Integers to Smaller Sized Integer Variables (The

Typecast)

Data conversion is often automatically done by the compiler as needed during calculations and
during assignments. However, the programmer must also force conversions at other times.
Consider the following calculation to find the average number of students per section of a course
that has a lot of sections in it.
char numberOfSections;
short totalStudentsEnrolled;
char studentsPerSection;
studentsPerSection = totalStudentsEnrolled/numberOfSections;

While the total number of students taking a course could be large, the number of course
sections is not likely to exceed +127 so a char is appropriate. The average number of students in
a course section is also not large; a char should hold the result nicely. However, when we
compile this program, the calculation line generates a compile-time warning message — possible
truncation of data.

Data Conversion Rule 3. The compiler always converts the final value of an expression
to the data type of the result variable. However, if there is a chance that data conversion may
result in a loss of accuracy, the compiler also issues a warning message so stating.

In the above example, data conversion rules show that the char numberOfSections is
going to be converted into a temporary int as well as the short variable. Then, the compiler can
do the divide of int by int, yielding an int result. However, the assignment is to a char sized
variable. What happens?

The compiler copies only those bytes that can fit in the answer variable, beginning with
the bytes on the right of the sending field. The following illustrates this; each x represents one bit
of data. An int is four bytes or 16 bits long while a char is one byte or 8 bits.

studentsPerSection = Result
xxxx xxxx <--- 0000 0000 0000 0000 0000 0000 xxxx xxxx

In other words, the high order three bytes of the int result are simply pitched. If part of the
answer were stored there, where the zeros are located above, then the answer variable,
Additional Processing Details 86

studentsPerSection, would have a bogus value in it. Hence, the compiler issues a warning
message, possible loss of precision. Nevertheless, the compiler is going to go ahead and make the
assignment.

In this particular case, we know by the nature of the calculation that the answer is going
to be small and will always fit within the smaller char variable. But the compiler does not.

A program should compile error and warning message free when it is done. Never leave
warning messages in a program. If you leave these warnings in a program, then every time
anyone compiles the program, they have to go back and re-evaluate whether or not those
warnings are significant or not. This is very bad programming style indeed.

So how can we get rid of the warning in this case. True, there is a compiler option to
disable such warning messages. But that is playing Russian Roulette; if you do that, then you will
not be notified of assignments that will cause trouble! The answer is to insert some coding that
tells the compiler that this is ok that we are assuming full responsibility for this particular
assignment statement. This is called a typecast.

A typecast consists of the desired data type surrounded by parentheses. The typecast
applies to what comes immediately after it.

To remove the warning message in the above example, code

studentsPerSection = (char) (totalStudentsEnrolled /
numberOfSections);

The typecast tells the compiler to convert the int final result into a char and that we say it
is ok to do so. Notice that the parenthesis is around the complete result, not just in front of the
total students variable. Coding only this
studentsPerSection = (char) totalStudentsEnrolled /
numberOfSections;
causes the compiler to convert the short into a char, likely truncating the number, causing even
more problems.

There are also times when this warning indicates a fundamental design flaw. Consider the
problem of calculating the total cost of an order.
int quantity = 10;
long cost = 10000;
int total;
total = quantity * cost;
Here the compiler again does data conversion when performing the multiply operation. It
promotes the int quantity to a long temporary value and multiplies long times long yielding a
long result. Next, the assignment is a long to an int. Since an int type could be two bytes not
four, the compiler issues the warning message and then moves what it can of the resulting
product into total. When compiled and run on a 32-bit platform such as Win32 console
Additional Processing Details 87

applications, the int is really four bytes and all is well. When that same code is compiled as an
older DOS application, the results are garbage in total.

Thus, whenever assigning a large sized result into a smaller sized variable, you must
analyze the situation and determine whether the best course is to make the answer variable larger
or put in the typecast because the answer really is not that large.

Assigning smaller sized floating point values to larger sized floating point variables is
always acceptable and totally safe. Obviously, if a float value contains 6 digits, it can be assigned
to a double which has 15 digits. The exponents are safe as well, since the double can handle a
much larger exponent.

Assigning a large sized floating point value to a smaller sized floating point variable
raises the same compiler warning about truncation as with integers. Consider this assignment
float a;
double b = 1.23456789012345;
a = b;
Variable a cannot store more than 6 digits accurately. Thus, there is going to be a loss of
precision. If, however, this is acceptable, then supply the typecast
a = (float) b;
and the warning message is handled.

However, there is an additional consideration, the exponent or power of ten. Consider this
version.
float a;
double b = 1.23456789012345E100;
a = (float) b;
By using the typecast, the warning goes away but when the program actually executes this
assignment line, trouble occurs. A float can only store an exponent of 1038 and the computer
cannot store 10100 in float variable a and promptly issues a floating point overflow error message
and terminates the program.

Thus, always consider both the needed number of digits as well as the possible magnitude
of the number.

Calculations Involving Multiple Floating Point Data Types

Just as with the integer family, the compiler must perform data conversion when an instruction
involves floating point numbers of different data types. Consider the following.
float a;
double b;
float c;
c = b * a;
Additional Processing Details 88

The compiler must convert the contents of float a variable into a temporary double so that it can
perform the multiply operation. It also issues the warning message about possible truncation
when it gets to the assignment portion since the answer variable is a float and the result of the
calculation is a double. If the double’s exponent exceeds that of a smaller float, a math
exception is raised and the program is terminated. Even if the exponent of the result is small
enough to fit in a float, one is trying to place 15 digits of accuracy into a number that can only
hold a little more than 6 digits!

If you knew by the nature of the problem that the exponent was in range and the loss of
precision was not a factor, then the following removes the compiler warning.
c = (float) (b * a);

What happens if integer data types and floating point types are used in the same
calculation? A situation such as this is called mixed mode math.

Mixed Mode Math

Mixed mode math occurs any time both integer data types and floating point types occur in the
same calculation. Any floating point type is worse than any of the integer types. Consider the
following mixed mode calculation.
char a;
float b;
int c;
double d;
double answer;
answer = a / b + c * d;

The computer cannot divide char by float, so the char variable a is converted into a
temporary float and the division is done, yielding a float result. Next, the multiplication is done,
but the int c variable’s contents are converted into a temporary double variable and then the
multiplication is done, yielding a double. Then, the addition of the two intermediate results is
performed, but the float result of the division must first be converted into a double. The final
result is a double. At last, the assignment can be made which goes without a hitch because the
right side result value is a double and the left side answer variable is a double.

When making an assignment from a floating point type to any of the integer types, a new
problem arises. Let’s examine the problem and then see how it can be rectified. Suppose that the
grading scale for the course is 90-up is an A. Suppose that in calculating your final grade the
following was done.
int totalPoints = 899;
int numParts = 10;
int grade;
grade = totalPoints / numParts;
Additional Processing Details 89

If this was your final grade, would you be happy and content getting an 89 or a B? But you say
you actually had an 89.9. Ah, the above was integer division. We can fix that easily using a
typecast.
grade = (double) totalPoints / numParts;
Just typecast to a double either of the two variables and the division now must be done using
floating point numbers. The result is 89.9 and then the assignment is done.

When the computer assigns a floating point number to any of the integer types, it issues
first a warning message about possible loss of data. It is possible for a float value to overflow a
long, not by digits, but by the power of ten involved. Consider a float value of 1x1038 — it would
totally overflow a long. But in this case, student grades should range between 0 and 100 and
would even fit in a char type. So now our typecast fix looks like this.
grade = (int) ( (double) totalPoints / numParts);
Now the warning message is gone. But one major problem remains.

Data Conversion Rule 4. When assigning any floating point type to any of the integer
types, the computer assigns only the digits to the left of the decimal point. It drops all decimal
fractions.

Thus in the above assignment, even though the result is now clearly 89.9, the compiler
places the whole number portion only into grade, the 89. What we want it to do is to round up!

Corollary: when assigning any float data type to any of the integer data types, add +.5
before the assignment to round up.

The correct calculation is

grade = (int) ( (double) totalPoints / numParts + .5);
This line says to convert totalPoints to a double and do the division using doubles; then, add .5;
then, convert the double result into an int and copy it into grade. In the case of 89.9, adding .5
yields 90.4 but the conversion to an int copies only the 90 portion, yielding the desired result.

Constants and Data Types

When numerical constants are coded, such as 10 and 123.45, the compiler assumes that they are
of data types int and double respectively. That is, the compiler assumes you desire the fastest
math possible on integers and the 6 digits of accuracy is not enough with floating point numbers.
In general, these are wise decisions.

Occasionally, you might wish to override those assumed data types to specify that the
constant is a long or is an unsigned int or unsigned long. These are the only overrides available.
Suppose that you are dealing with a graphics problem in which only positive x and y values are
allowed. Further, suppose that you wanted to move the point one pixel to the right along the x-
Additional Processing Details 90

axis. Here, the compiler assumes a data type of int for the constant value of 1. It would be better
if we could force it to be an unsigned int of value 1.
unsigned int x;
unsigned int y;
x = x + 1;
The integer constant data type can be specified by adding a one or two letter suffix to the
number. The possible suffixes are U or u for unsigned and L or l for long or UL or ul for
unsigned long.

We can rewrite the above addition as follows

x = x + 1U;
and now the constant 1 is of type unsigned int.

When making a constant a long value, please do not use the lowercase l for it is easily
confused with a digit of 1. Look at these two attempts to make a 1000 into a long constant.
1000l
1000L
Notice that the first one looks more like ten thousand and one.

Suppose that we needed to add 100,000 to a long total variable. Coding

long total;
total = total + 100000;
often gives the compiler warning message the constant is a long. To remove the compiler
warning, simply append the suffix L to the constant.
total = total + 100000L;

Notice that one major use of constant suffixes is within calculations. Suppose that our
calculation is the number of ounces in a variable number of tons. There are 16 ounces in a pound
and 2000 pounds in a ton. Here is the problem.
int numTons;
long totalOunces;
totalOunces = numTons * 2000 * 16;
Assume that the user has entered something easy to calculate, say 10 tons. What results? Can you
spot the error? The calculation involves all ints and thus can result in an overflow for any
number of tons above one. 2000 times 16 is 32000. If numTons is above one, it may overflow an
int result if the program is compiled to run under old DOS where an int is only 2 bytes in size.
The calculation can be fixed by using a typecast as we have learned earlier.
totalOunces = (long) numTons * 2000 * 16;
But using a long suffix is much easier.
totalOunces = numTons * 2000L * 16;
Notice that I applied the L suffix to the 2000 constant.

What would likely happen if I had coded it this way?

totalOunces = numTons * 2000 * 16L;
If numTons had been the initial 10, all would have been fine, since an int result can hold 20,000.
Additional Processing Details 91

But what if the user entered 20 tons? Overflow. Here is a case where position of the override is
important.

The data type of floating point constants can similarly be altered by two suffixes: f or F
for float and L or l for long double. Coding
123.5f
creates a float constant instead of a double.

The f suffix is quite useful if all of your variables are of type float. Assume that we are
dealing with weather temperatures. Storing such a temperature in a double is overkill, since we
would not need more than say four digits, as in 101.5 degrees. Suppose that we needed to add
one degree to a temperature measurement.
float temp;
temp = temp + 1.;
This again yields a possible loss of precision warning message because the 1.0 is a double
constant, forcing the addition to be done using doubles. By forcing the 1.0 to be of type float, the
calculation is done using float types.
temp = temp + 1.f;

Here’s another common goof. What is wrong with this one?

int x = 10;
int y = 10055;
float ansr;
ansr = (float) x + y / x;
Have you spotted the error? Which operation is done first? The divide operator has precedence
over the add operator. Thus, the compiler performs int divided by int and discards the remainder.
Next, it converts x into a float and then converts the division result into a float so that they can
be added. The fractional portion is not therefore present in the answer variable. Here is a
corrected version.
ansr = x + (float) x / y;

By now you are probably thinking why not make everything in the program a double
variable and forget about it? Master files of data on disk very often use these smaller sized
integer values to conserve disk space. Turning every integer variable into a double is also
wasteful of computer memory within your program. Perhaps even more importantly, math
operations on floating point data is much slower than math on integer data.
Additional Processing Details 92

Additional Operators

The Increment and Decrement Operators

An extremely common operation in programs is adding one to a counter. Perhaps less common is
the need to subtract one from a counter. The increment operator ++ and the decrement
operator – – do exactly these common actions.

int count = 0;
int tally = 42;
count = count + 1;
tally = tally - 1;
These can be replaced with just
count++;
tally--;
After these are executed, count contains 1 and tally contains 41.

The operators can be placed after or before the variable on which it applies. These are
called the prefix inc or postfix inc. The meanings are different. The prefix inc or dec means to
increment/decrement the variable before that variable is used in the statement. The postfix inc or
dec means to go ahead and use the current contents of the variable in the statement and when the
statement is done, go ahead and increment/decrement the variable.

In the case above, both

count++;
tally--;
and
++count;
--tally;
produce the same results because no use of the current contents of either count or tally is made
(other than the increment/decrement operation).

However, that is not always the case. Consider these two

int count1 = 42;
int count2 = 42;
int sum1 = 0;
int sum2 = 0;
sum1 = count1++;
sum2 = ++count2;
In both cases, count1 and count2 contain 43 when the instructions are finished. Variable sum1
contains 42 while sum2 contains 43.
These two calculation lines are equivalent to writing
sum1 = count1;
count1 = count1 + 1;
Additional Processing Details 93

count2 = count2 + 1;
sum2 = count2;

The Compound Assignment Operators

Another common coding action is to add a value to a total. Consider the problem of
accumulating the total cost of an order.
int quantity;
double cost;
double totalCost = 0;
totalCost = totalCost + quantity * cost;
The lengthy calculation line can be shortened by use of the += operator.

The += operator adds the result to the right of the operator to the variable to the left of the
operator. The total cost calculation can be shortened to just
totalCost += quantity * cost;

Here is another example. Suppose we need to accumulate the student scores on a test.
double score;
double sum = 0;
sum += score;
The calculation says to add the contents of score to the current contents of sum and then place
the revised value back into sum. It is equivalent to writing
sum = sum + score;

Please note that in both the above examples, variables sum and totalCost must be
initialized to zero because the first use of those variables is to add something to them.

The use of the += operator is widespread. However, there are also –=, *=, /= and %=
operators available for use. The %= operator only applies to integer types values, of course, since
floating point values already have a decimal point.

Suppose that we wished to calculate 3! — that is 3 * 2 * 1. One way to do this is

int fact = 3;
fact *= (fact - 1);
This says to multiply fact by the quantity fact–1 and place the result back in fact. It is a short cut
for
fact = fact * (fact - 1);

Ok. I admit that I could have determined 3! in my head; however, soon you will get to
find N! in which the user enters a value for N.
Additional Processing Details 94

Suppose that we wished to divide a number by 10. We could code

int num = 123;
num /= 10;
This results in 12 in num when the operation is complete.

Just be alert for possible uses of these short cut type of operators.

Section B: Computer Science Examples

CS03a — Vote Tally Program
Input the vote count received by each of the three candidates running for election. Output the
percentage of the votes received by each candidate. Call the candidates Mr. Jones, Ms. Baker and
Ms. Smith. Test the program using 19345, 23673 and 34128 votes respectively.

Design the solution first. We need three variables to hold the input vote counts. These
should be longs since vote counts can be rather large numbers. Let’s call them votesJones,
votesBaker, and votesSmith. To find the percentage each candidate has, the total vote count
must be found; it also must be a long variable. Let’s call it totalVotes. Given the total votes cast,
the percentage each candidate received can then be calculated. So we need three result variables,
say called percentJones, percentBaker and percentSmith. These must be doubles if they are to
hold a result such as 10.5%. Draw up a picture of main storage with small boxes for each of these
variables. Figure 3.1 shows the main storage diagram.

Figure 3.1 Main Storage for Votes Program

Now using these names for the variables, write out the sequence of processing steps
needed, following the Cycle of Data Processing, IPO.
prompt and input the vote count for Jones placing it in votesJones
prompt and input the vote count for Baker placing it in votesBaker
prompt and input the vote count for Smith placing it in votesSmith
set totalVotes to 0
add votesJones to totalVotes
add votesBaker to totalVotes
add votesSmith to totalVotes
Additional Processing Details 95

set percentJones = votesJones * 100. / totalVotes

set percentBaker = votesBaker * 100. / totalVotes
set percentSmith = votesSmith * 100. / totalVotes
Notice that I force the calculation to use doubles by first multiplying by 100. and not 100!

Finally, output the results nicely formatted

displaying percentJones, percentBaker and percentSmith.

Now desk check it using the input test vote counts. When you have convinced yourself
that there are no errors, go ahead and code it into a C++ program. Here is the completed program.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs03a - Vote Count Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs03a: vote tally program */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 using namespace std; *
* 10 int main () { *
* 11 long votesJones; // total votes for Jones *
* 12 long votesBaker; // total votes for Baker *
* 13 long votesSmith; // total votes for Smith *
* 14 *
* 15 long totalVotes = 0; // total votes cast in the election *
* 16 *
* 17 double percentJones; // results for Jones *
* 18 double percentBaker; // results for Baker *
* 19 double percentSmith; // results for Smith *
* 20 *
* 21 // prompt and input the three vote counts *
* 22 cout << "Enter the vote count for Mr. Jones: "; *
* 23 cin >> votesJones; *
* 24 cout << "Enter the vote count for Ms. Baker: "; *
* 25 cin >> votesBaker; *
* 26 cout << "Enter the vote count for Ms. Smith: "; *
* 27 cin >> votesSmith; *
* 28 *
* 29 // find the total votes cast *
* 30 totalVotes += votesJones; *
* 31 totalVotes += votesBaker; *
* 32 totalVotes += votesSmith; *
* 33 *
* 34 // calculate the percentages *
* 35 percentJones = votesJones * 100. / totalVotes; *
* 36 percentBaker = votesBaker * 100. / totalVotes; *
* 37 percentSmith = votesSmith * 100. / totalVotes; *
Additional Processing Details 96

* 38 *
* 39 // setup floating point format for output of percentages *
* 40 cout << fixed << setprecision (1); *
* 43 *
* 44 // output a title *
* 45 cout << "\nVoting Results\n\n"; *
* 46 cout << "Mr. Jones: " << setw(5) << percentJones << "%\n"; *
* 47 cout << "Ms. Baker: " << setw(5) << percentBaker << "%\n"; *
* 48 cout << "Ms. Smith: " << setw(5) << percentSmith << "%\n"; *
* 49 *
* 50 return 0; *
* 51 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Cs03a - Vote Count Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Enter the vote count for Mr. Jones: 19345 *
* 2 Enter the vote count for Ms. Baker: 23673 *
* 3 Enter the vote count for Ms. Smith: 34128 *
* 4 *
* 5 Voting Results *
* 6 *
* 7 Mr. Jones: 25.1% *
* 8 Ms. Baker: 30.7% *
* 9 Ms. Smith: 44.2% *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Section C: An Engineering Example

Engr03a—Calculating the Power Supplied to a Load
(Electrical Engineering)
An AC power supply of V volts is applied to a circuit load with impedance of Z (Ø) with current
I. Display the real power P, the reactive power R, the apparent power A and the power factor PF
of the load. Test the program with a voltage of 120 volts and an impedance of 8 ohms at 30
degrees.
Additional Processing Details 97

Here are the formulae that define these power values

I = V/Z
P = V I cos Ø
R = V I sin Ø
A = VI
PF = cos Ø
where V is the root mean square (RMS) voltage of the AC power source in volts, Z is the
impedance in ohms, Ø is the angle of impedance in degrees, I is the current in amperes, P is the
real power in watts, R is the reactive power in volt–amperes–reactive (VAR), A is the apparent
power in volt–amperes and PF is the power factor of the load.

The starting point is to design the solution on paper. We begin by identifying the input
variables. Here we must input the volts, impedance and its angle. Let’s call these variables V, Z
and angle. Draw main three storage boxes and label them with these names. Next look at the
output and create main storage boxes for what is needed. Let’s call them P, R, A and PF. Finally,
we must have some intermediate result variables. The angle must be converted to radians and the
current is needed. Let’s call those boxes, radAngle and I. To convert to radians, we need the
constant PI. This is shown in Figure 3.2 below.

Figure 3.2 Main Storage for Power Program

Now follow the Cycle of Data Processing, IPO, and write out the sequences we need to
solve this problem making sure we use the same variable names as we have in the main storage
boxes. First, we need to input the three variables.
prompt and input the voltage and store it in V
prompt and input the impedance and store it in Z
prompt and input the angle and store it in angle

Next, calculate the intermediate values we are going to need in the main power calculations.
radAngle = angle * PI / 180
I=V/Z
Now calculate the various power values.
P = V * I * cos radAngle
R = V * I * sin radAngle
A=V*I
PF = cos radAngle
Additional Processing Details 98

And lastly, display the results

print nicely labeled the input situation: V, Z and angle
print nicely labeled the four results: P, R, A, and PF

Here are the completed program and the sample run.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr03a - Calculate the Power Supplied to a Load *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr03a: Calculate the power delivered to a load */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 const double PI = acos (-1.); *
* 12 *
* 13 int main () { *
* 14 *
* 15 double V; // initial volts of AC power supply *
* 16 double Z; // initial impedance of load *
* 17 double angle; // initial angle of impedance of load *
* 18 *
* 19 double I; // current flow in amperes *
* 20 double radAngle; // angle in radians *
* 21 *
* 22 double P; // real power in watts *
* 23 double R; // reactive power in VAR *
* 24 double A; // apparant power in VA *
* 25 double PF; // power factor *
* 26 *
* 27 // prompt and input the initial values of V, Z and angle *
* 28 cout << "Enter the AC power supply voltage: "; *
* 29 cin >> V; *
* 30 cout << "Enter the impedance of the load in ohms: "; *
* 31 cin >> Z; *
* 32 cout << "Enter the angle in degrees: "; *
* 33 cin >> angle; *
* 34 *
* 35 // calculate intermediate needed values *
* 36 radAngle = angle * PI / 180; *
* 37 I = V / Z; *
* 38 *
* 39 // calculate the four resulting power factors *
* 40 P = V * I * cos (radAngle); *
* 41 R = V * I * sin (radAngle); *
* 42 A = V * I; *
* 43 PF = cos (radAngle); *
Additional Processing Details 99

* 44 *
* 45 // setup floating point format for output *
* 46 cout << fixed << setprecision (3); *
* 49 *
* 50 // echo print the input *
* 51 cout << "\n Power Supplied to a Load Results\n"; *
* 52 cout << "AC voltage supplied: " << setw(8) << V<<" volts\n";*
* 53 cout << "Load impedance of: " << setw(8) << Z <<" ohms\n";*
* 54 cout << " at an angle of: " << setw(8) << angle *
* 55 << " degrees\n"; *
* 56 cout << "Yields these power factors\n"; *
* 57 cout << "Real power supplied: " << setw(8) << P<<" watts\n";*
* 58 cout << "Reactive power supplied: " << setw(8) << R *
* 59 << " volt-amperes-reactive\n"; *
* 60 cout << "Apparant power supplied: " << setw(8) << A *
* 61 << " volt-amperes\n"; *
* 62 cout << "Power factor is: " << setw(8) *
* 63 << PF << endl; *
* 64 *
* 65 return 0; *
* 66 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Engr03a - Calculate the Power Supplied to a Load *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Enter the AC power supply voltage: 120 *
* 2 Enter the impedance of the load in ohms: 8 *
* 3 Enter the angle in degrees: 30 *
* 4 *
* 5 Power Supplied to a Load Results *
* 6 AC voltage supplied: 120.000 volts *
* 7 Load impedance of: 8.000 ohms *
* 8 at an angle of: 30.000 degrees *
* 9 Yields these power factors *
* 10 Real power supplied: 1558.846 watts *
* 11 Reactive power supplied: 900.000 volt-amperes-reactive *
* 12 Apparant power supplied: 1800.000 volt-amperes *
* 13 Power factor is: 0.866 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Additional Processing Details 100

New Syntax Summary

The Typecast: explicitly causes conversion of the item after it to the indicated new data type.
The syntax is (new data type).
double calculatedGrade;
char grade = (char) (calculatedGrade + .5);

The Increment and Decrement Operators: ++ and --

The increment operator adds one to the variable while the decrement subtracts one from it.
The operator can be in a prefix or a postfix position. The prefix increment first
adds one to the variable and then uses the new value in the remaining expression.
The postfix increment uses the current value in the expression calculations and
then afterwards adds one to the variable.
int x = 42;
int y;
y = x++; // stores 42 in y and makes x be 43
y = ++x; // makes x be 43 and stores 43 in y

The Compound Assignment Operators: +=, -=, *=, /=, %= (integer types only)
Samples:
x += y; This adds the contents of y to the contents of x and the
new result replaces the contents of x.

x -= y; This subtracts y from the contents of x and the results

replaces the contents of x.

x *= y; The contents of y are multiplied by the contents of x

and the product replaces the contents of x.

x /= y; The contents of x are divided by y and the result

replaces the contents of x;

Design Exercises
1. Sketch the pseudocode to solve this problem. The user enters some even integer greater than
two, called say number. The program determines the next higher even integer from number and
the next lower even integer from number. Display the sum of the three numbers, the product of
the three numbers, and the average of the three numbers.
Additional Processing Details 101

2. Sketch the pseudocode to solve this problem. The user wishes to enter six temperature
observations taken at four-hour intervals throughout the day. Compute and print the average
temperature for the day.

Stop! Do These Exercises Before Programming

1. What is in the variable result when the calculation is finished?
double result = 123 / 10 + .5;

2. What is in the variable result when the calculation is finished?

double result = 123 / 10. + .5;

3. What is in the variable result when the calculation is finished?

char a = 2;
short b = 3;
long c = 100000L;
double result = b / a + c;

4. What is in the variable result when the calculation is finished?

char a = 2;
short b = 3;
long c = 100000L;
double result = (double) b / a + c;

5. What is in the variable result when the calculation is finished?

char a = 2;
short b = 3;
long c = 100000L;
double result = b / a + (double) c;

On the next two problems, fix the errors by changing the calculation line; do not change the data
types of the variables.

6. Fix the compiler truncation warning message.

int itemsOrdered;
double totalCost;
double unitCost;
itemsOrdered = totalCost / unitCost;
Additional Processing Details 102

7. Repair the equation so that totalBytes contains the correct amount even on old DOS systems.
short k = 1024; // 1k bytes = 1024, 1m bytes = 1024k
short numMegs;
long totalBytes;
totalBytes = numMegs * k * k;

8. What is in sum and count after these instructions complete.

int count = 99;
int sum = 10;
sum += (++count)++;

9. What is in sum and count after these instructions complete.

int count = 99;
int sum = 10;
sum *= count++;

10. What is in sum and count after these instructions complete.

int count = 10;
int sum = 99;
sum /= count++;
Additional Processing Details 103

Programming Problems
Problem Cs03-1 — Height of a Rainbow

How high is a rainbow? Interestingly enough, when light is refracted by the water droplets just
after a storm, the angle between the level of your eye and the top of a rainbow is always the
same, 42.3333333 degrees. From trigonometry, if we know the distance to the rainbow, then
using the tan() function, we can calculate the unknown height.

Thus, given the distance and the angle, the height is distance * tan (angle). Note that the
trig functions of the C++ library all take the angle in radians. Use radians = angle * PI / 180 to
convert.

Sometimes you can see a second rainbow just above the first one. The magic angle for
this secondary rainbow is 52.25 degrees. (I have actually once seen seven rainbows at one time.)

Write a program that inputs the distance in miles from the rainbow and displays the
height of the primary and secondary rainbows. Display the distance and heights with two decimal
digits and appropriately labeled. Test your program with a distance of 2 miles, 1 mile and .5
miles.

Cs03-2 — Area of a Triangle

The perimeter of a triangle is the sum of the lengths of all three sides. The semi-perimeter is ½ of
the perimeter. Given a triangle, the area of the triangle is as follows, where s is the semi-
perimeter.

Write a program that inputs the three sides of a triangle and displays the area of that
triangle. Display the area to three decimal digits. Test the program with these three test runs.
10 15 20
Additional Processing Details 104

10 7.5 12.5
25.25 18.5 21.77

Cs03-3 — Dollar Conversion Program

Input a dollar amount as a double, such as 1.23 for $1.23. Convert it into the number of pennies
in that amount and store it in a long. Next, print out the minimum number of dollars, quarters,
dimes, nickels and pennies in that amount. Prompt the user like this:
Enter the amount of money (such as 1.23 for $1.23): 1.23
Your output should show the following lines.
$1.23 contains
1 dollar
0 quarter(s)
2 dime(s)
0 nickel(s)
3 penny/pennies

Test the program with these values as well as 1.23: 4.18, 8.88, 0.22.

Problem Engr03-1 — Carbon–14 Dating

Radioactive isotopes of elements are not a stable form and they spontaneously decay into another
element over a period of time. This radioactive decay is an exponential function over time and is
given by

where Q0 is the initial quantity of a radioactive substance at time t = 0 and lambda is the
radioactive decay constant.

Now the reverse process is valuable. That is, since the decay is at a known rate, if we
observe a given quantity of radioactive substance in a sample and we know the initial quantity
that was there to begin with, then the time for that initial quantity to have decayed into the
current amount can be calculated, yielding the date of the sample.

Archaeologists use Carbon–14 isotopes to determine the age of samples. Plants and
animals continuously absorb Carbon–14 while they are living. Once they die, nothing new is
absorbed and the slow decay process begins. Thus, assumptions can be made about the initial
Additional Processing Details 105

quantity of Carbon–14 in a sample and the current amount of Carbon–14 in a sample can be
measured in the lab. Samples are commonly trees used as building materials and campfire
remains, for example. The decay constant lambda of Carbon–14 is well known to be 0.00012097
per year. Typical lab measurements report the percentage of Carbon–14 remaining in a sample.

Write a program that inputs the percentage of Carbon–14 in a sample and displays the age
of that sample in years. Note that the formula uses a ratio not a percentage. Test the program with
these percentages: 50%, 25%, and 12.5%. Display the year results to one decimal. Echo print the
original inputted percentage. Label all values appropriately.

Finally, suppose that the percentage ratios are only sufficiently accurate to measure 1.0%
because with concentrations that low, field contamination plays a more significant role. What
would the oldest date that the Carbon–14 process be able to yield?

Engr03-2 — Railroad Track Design (Transportation Engineering)

When a train travels over a straight section of track, it exerts a downward force on the rails. But
when it rounds a level curve, it also exerts a horizontal force outward on the rails. Both of these
forces must be considered when designing the track. The downward force is equivalent to the
weight of the train. The horizontal force, known as centrifugal force, is a function of the weight
of the train, the speed of the train as it rounds the curve, and the radius of the curve. The
equation to compute the horizontal force, in pounds, is:

where weight is the weight of the train in tons, mph is the speed of the train in miles per hour and
radius is the radius of the curve in feet.

Write a program to input the weight, mph and radius. Compute and print the
corresponding horizontal force generated along with the initial three values, appropriately
labeled. Test the program using these situations:
a). use weight = 405.7 tons at a speed of 30.5 mph on a curve of radius 2005.33 feet
b). run again increasing the speed
c). run again increasing the radius
d). run again decreasing the radius
Write two sentences describing what happens to the force when the speed and curve vary. For
example, what happens when either the speed or radius is doubled or is cut in half?
Additional Processing Details 106

Engr03-3 — Period of an Oscillating Pendulum

The period of an oscillating pendulum on the surface of the earth is given by

where the period T is in seconds, L is the length of the pendulum in meters and g is the
gravitational acceleration of the earth on the surface, 9.8 m/s/s.

Write a program that first prompts the user to input the length of the pendulum and then
calculates its period. Display both the length and the period appropriately labeled and with their
units.

Analysis: Suppose that the length of a pendulum was carefully constructed to be

0.24824m long so that the period was one second. Suppose that the space shuttle took that
pendulum into orbit around the earth. What would happen to the period of that pendulum?
Decisions 107

Chapter 4 — Decisions

Section A: Basic Theory

Introduction
A decision asks a question that can be answered true or false, yes or no. Decisions are widely
used in most programs. If a question is true, then often one or more actions are to be performed.
However, if the question is false or not true, then one might have some alternative processing
steps to be performed instead.

A decision can be thought of as having three parts: a test condition to be examined, one or
more instructions to follow when the test condition is true, and one or more instructions to follow
when the test condition is false. When considered from this point of view, the test condition itself
can be used in far more C++ constructs than just a simple decision structure.

In C++, the decision structure is called an If-Then-Else.

The Components of an If-Then-Else Decision Structure

The decision structure is shown below in Figure 4.1. Notice that flow of control comes in at the
top and after branching and doing one of two alternative sets of statements, control leaves out the
bottom. The statements to do when the test is true are called the then-clause. The statements to
do when the test is false are called the else-clause.

The If-Then-Else Syntax

The If-Then-Else basic syntax to implement the decision structure is as follows.

if (test condition) {
0, 1 or more statements to do if the test condition is true
}
else {
0, 1, or more stmts to do if the test condition is false
}
Decisions 108

Figure 4.1 The Decision Structure

The keyword if begins the decision. It is followed by a test condition surrounded by

parentheses. Note that there is no “then” keyword but that there is an else keyword.

The else-clause is strictly optional; if nothing needs to be done when the test condition is
false, the else-clause does not need to be coded.

Notice that the statements to be done are surrounded by a begin-end block { }. I prefer to
place the begin block { on the line that is launching that block. All statements within that block
are uniformly indented. The end block } must align with the start of the line that is launching the
block.

The other commonly used style looks like this.

if (test condition)
{
0, 1 or more statements to do if the test condition is true
}
else
{
0, 1, or more stmts to do when the test condition is false
}
In this style, the begin block { and all statements within that block and the end block } are all
uniformly indented the uniform amount. Choose one style or the other and remain consistent in
its use throughout the program.

The else-clause is optional. If there is nothing to do when the test condition is false, it can
be omitted as shown below.
Decisions 109

if (test condition) {
0, 1 or more stmts to do when the test condition is true
}

Further, if there is only a single statement to do, the begin-end pair {} can be omitted.
if (test condition)
a single statement when true;
else
a single statement when false;
Or if there is nothing to do when the statement is false, it can be simply
if (test condition)
a single statement when true;

Note that a single statement can be a null statement consisting of just a semicolon.
if (test condition)
a single statement when true;
else
;

The Test Condition

Test conditions can be very complex and the rules, likewise. However, let’s start with simple
ones and add onto the complexity as we gain understanding of what they are and how they are
used. A test condition asks some kind of question that can be answered true or false. In its most
basic form, it parallels how we ask a question in English. For example
is the quantity less than or equal to five?
is x greater than y?
is count not equal to zero?
is sum equal to zero?
Notice in these examples that the comparison operators are less than or equal to, greater than, not
equal to, and equal to. Observe that there is some quantity to the left and also to the right of each
comparison operator. In English, the following would make no sense
if quantity greater than
“Greater then what” is the immediate reply. This gives us the basic format of a simple test
condition:

operand1 comparison-operator operand2

In C++, there are six comparison operators. They are

> greater than
>= greater than or equal to
< less than
<= less than or equal to
!= not equal to
== equal to
Decisions 110

Pay particular attention to the comparison equals operator! Notice it is a double equals sign (==);
it is not a single equals sign (=). A single equal sign (=) is always an assignment operator in
C++.

Using these operators, we can translate the above four English comparisons into C++ as
follows.
if (quantity <= 5) {
// do these things if true
}

if (x > y) {
// do these things if true
}

if (count != 0) {
// do these things if true
}

if (sum == 0) {
// do these things if true
}

if (hoursWorked > 40) {

// calculate overtime pay
}

Let’s apply just this much to some programming situations. Suppose that we wanted to
print a message if an employee was eligible for early retirement. That is, if their age was greater
than or equal to 55. One could code the following to do this.
int age;
long employeeID;
cout << "Enter the employee id number and age\n";
cin >> employeeID >> age;
if (age >= 55) {
cout << employeeID
<< " is eligible for early retirement\n";
}

Suppose that we wanted to determine whether or not a person was eligible to vote. We
can input their citizenship status which contains a one if they are a citizen and a zero if they are
not a citizen.
int citizenship;
cout << "Enter citizenship status: ";
cin >> citizenship;
if (citizenship == 1) {
cout << "You are eligible to vote\n";
Decisions 111

}
else {
cout << "Non-citizens are not eligible to vote\n";
}

Ok. So far it looks fairly simple, but complexity can arise swiftly. Decisions can be nested
inside each other.

Nested Decisions

There is no limit to the complexity of statements that can be contained in the then-clause or the
else-clause of a decision. Hence, another decision structure could be found inside of the then-
clause, for example. However, such nested decisions must be entirely contained within the then-
clause. This gives us the ability to choose from among several choices, not just between two.

In the above voting example, a citizen must also be 18 or older to be eligible to vote. So
realistically inside the then-clause, which is executed if the person is a citizen, we need to further
test to see if the person is old enough to vote. And here programmers can get into trouble.

Consider this version in which I have manually added line numbers for reference.
1. if (citizenship == 1)
2. if (age < 18)
3. cout << "You must be 18 to be eligible to vote\n";
4. else
5. cout << "Non-citizens are not eligible to vote\n";
Notice the nice block structure. It “looks” reasonable. However, it is not correct. Suppose that the
citizenship is a one and the age is 50. What actually prints out is “Non-citizens are not eligible to
vote.” Why? White space is the delimiter in C++. Thus, the nice alignment of line 4’s else with
line 1’s if statement makes no difference to the compiler. Line 4’s else-clause actually is the else-
clause of line 2’s if statement!

There are several ways to code the nested if statements correctly. One way is to provide
the missing else-clause for line 2's if statement.
if (citizenship == 1)
if (age < 18)
cout << "You must be 18 to be eligible to vote\n";
else
;
else
cout << "Non-citizens are not eligible to vote\n";
Here the else-clause has been provided and consists of a null statement, that is, a simple
semicolon.
Decisions 112

However, the real genus of the coding error came from not using begin-end braces {}
around the two clauses. If you always use the braces, you will be far less likely to code these
inadvertent errors. Here is perhaps the best way to repair the coding.
if (citizenship == 1) {
if (age < 18) {
cout << "You must be 18 to be eligible to vote\n";
}
}
else {
cout << "Non-citizens are not eligible to vote\n";
}
By using the begin-end braces on the then-clause of the citizenship test, the compiler knows for
certain that there can be no else-clause on the age test because the age decision must be contained
within the then-clause of the citizenship test.

Let’s look at an even more complex decision structure. Suppose that our company is
asked to check up on its hiring practices. After inputting the information on an employee, we are
to display a message if that person is over age 50 or is physically challenged or if their race is not
Caucasian. The input fields consist of age, disability (1 if so), and race (1 if white). We can code
the decisions as follows.
if (age > 50) {
cout << "Over 50 ";
}
else if (disability == 1) {
cout << "Disabled ";
}
else if (race != 1) {
cout << "Non-white ";
}
Notice how the else-clauses use a single statement which is itself another If-Then-Else statement.
It could also be coded this way.
if (age > 50) {
cout << "Over 50 ";
}
else {
if (disability == 1) {
cout << "Disabled ";
}
else {
if (race != 1) {
cout << "Non-white ";
}
}
}
It could also be coded this way
if (age > 50)
Decisions 113

cout << "Over 50 ";

else
if (disability == 1)
cout << "Disabled ";
else
if (race != 1)
cout << "Non-white ";

Probably the first way is the easiest to read. Okay. What does the program display for
output if one enters a 55-year-old African-American who has a limp? The age test is checked first
and out comes the fact that this employee is over 50. There is no mention of the other aspects.
Suppose that we restate the problem to display all of the possible aspects an employee might
have. How would the coding change? Notice that the reason the second test was never performed
with the current employee is that all the other tests began with an else, meaning only check
further if the age was not more than 50. If we just remove the else’s, we are left with three
independent decisions that are not nested in any way.
if (age > 50) {
cout << "Over 50 ";
}
if (disability == 1) {
cout << "Disabled ";
}
if (race != 1) {
cout << "Non-white ";
}
Now the output would be “Over 50 Disabled Non-white.”

When programming decisions, one must be very careful to duplicate precisely the
problem’s specifications.

Suppose that we are running a dating service. A client wishes to see if a specific person
would be a possible match for them. The client wishes to see if this candidate is a single male
between the ages of 20 and 25. Assume that we have input the age, maritalStatus (0 for single),
and sex (1 for male). In this example, to be a possible match, the candidate must satisfy all four
tests, single, male, age greater than or equal to 20 and age less than or equal to 25. Notice that
you cannot write
if (20 <= age => 25) {
It requires two separate test conditions. Notice also that all four of these test conditions must be
true for us to display the message. Here is how this might be coded.
if (age >= 20)
if (age <= 25)
if (maritalStatus == 0)
if (sex == 1)
cout << "Is a potential match\n";
It could also be written this way.
Decisions 114

if (age >= 20) {

if (age <= 25) {
if (maritalStatus == 0) {
if (sex == 1) {
cout << "Is a potential match\n";
}
}
}
}
It could also be written this way.
if (age >= 20)
{
if (age <= 25)
{
if (maritalStatus == 0)
{
if (sex == 1)
{
cout << "Is a potential match\n";
}
}
}
}

Here is another example. Suppose that the month has been input. Display the message
“Summer Vacation” if the month is June, July or August. For any other months, print “School in
session.” Please note that you cannot code
if (month == 6, 7, 8) // does not compile
Each value must be a complete test condition. If the month is 6 or if the month is 7 or if the
month is 8, then print the message.
if (month == 6)
cout << "Summer Vacation\n";
else if (month == 7)
cout << "Summer Vacation\n";
else if (month == 8)
cout << "Summer Vacation\n";
else
cout << "School in session\n";
Notice that an else verb connects each decision after the first one.

Now suppose that we did not have to output the last message if school was in session.
One might be tempted to code the following.
if (month == 6)
cout << "Summer Vacation\n";
if (month == 7)
cout << "Summer Vacation\n";
if (month == 8)
Decisions 115

cout << "Summer Vacation\n";

Yes, it still produces the correct answer. But this raises a serious efficiency concern. If the month
contains a six, then after printing the message, control passes to the next decision. But since it
contained a six, it cannot under any circumstances also contain a seven or an eight! Yet, the
program wastes time retesting the month for a seven and then for an eight. This kind of
programming is wasteful of computer resources and is generally frowned upon in the industry. It
shows a distinct lack of thought on the part of the programmer. Don’t do it.

Compound Test Conditions

The previous test conditions have become rather lengthy and a bit unwieldy. There is a much
better alternative to so much coding. A test condition can be a compound one. That is, two or
more tests can be joined together to form a larger test using either the AND or OR relational
operators. First, let’s define AND and OR logic.

AND logic says that both tests must be true to get a true result. This is often expressed
using boolean (two-valued) logic.
true true false false
AND true AND false AND true AND false
----- ----- ----- -----
true false false false

OR logic says that if either one or both of the tests are true, the result is true. Expressed in
boolean logic, OR logic appears as follows.
true true false false
OR true OR false OR true OR false
----- ----- ----- -----
true true true false

In C++ these two operators are && for AND and || for OR. To either side of these
operators must be a test condition. I often refer to these two operators as the “joiner ops” since
they are used to join two tests together. Figuratively, if we code
if (test1 && test2)
then this is saying that if test1 and test2 are both true, then execute the then-clause. Again,
figuratively if we code
if (test1 || test2)
then this is saying that if either test1 or test2 is true, then execute the then-clause.

We can greatly simplify the previous examples using these compound joiner operators. In
the dating service example, all the tests had to be true before the program printed the potential
match message. This means AND logic is used and the joiner would be &&. Rewriting those
four decisions into one greatly simplifies that decision.
Decisions 116

if (age >= 20 && age <= 25 && maritalStatus == 0 && sex == 1)

cout << "Is a potential match\n";

The summer vacation decision is an example of OR logic, since if either of the three
month tests is true, it is a summer month. It can be simplified as follows using the || operator.
if (month == 6 || month == 7 || month == 8)
cout << "Summer Vacation\n";
else
cout << "School in session\n";

These two operators are actually even more efficient. Take the AND operator && for
example. Suppose that the age of the person in the above dating service problem was 19. When
the very first test condition is executed, the age is not greater than or equal to 20. Thus, a false
results. Since the && operator is joining all of these other test conditions, C++ immediately
knows the final outcome of the compound test, false. Remember, AND logic says that they all
have to be true to get a true result. And that is exactly how C++ behaves. As soon as one of the
joined tests yields a false result, the compiler stops doing the remaining tests and immediately
goes to the else-clause if there is one. The remaining tests are not even executed! C++ is being as
efficient as it can with compound test conditions. Programmers often take advantage of just this
behavior, terminating remaining tests joined with the && operator when a false result is
encountered. We will do just that in Chapter 9.

The same efficiency applies to test conditions that are joined with OR || operators.
Consider the summer vacation tests. If the month is indeed six, then the result of the first test is
true. Since the remaining tests are joined with || operators, the compiler immediately knows the
final result, true, and does not bother to perform the remaining tests, jumping immediately into
the then-clause.

Thus, when we join our related decisions using AND and OR operators to form longer
tests, we gain a measure of efficiency from the language itself.

The AND operator has a higher precedence than does the OR operator and both are lower
than the six relational operators. In the dating service example, we could have used parentheses
as shown below, but because the && is at a lower precedence, they are not needed.
if ( (age >= 20) && (age <= 25) &&
(maritalStatus == 0) && (sex == 1))
cout << "Is a potential match\n";

Both && and || operators can be used in the same decision. If so, the && is done
before the || operation.
if (a > 5 && b > 5 || c > 5)
This is the same as if we had coded
if ( (a > 5 && b > 5) || c > 5)
Sometimes, parentheses can aid readability of a program. No harm in using them in this manner.
Decisions 117

Also note that the six relational operators have a lower precedence than all of the math
operators. Thus, if we coded the following
if (a + b > c + d)
then this would group as if we had used parentheses:
if ((a + b) > (c + d))
However, I tend to use the parentheses anyway because it aids program readability. Without
them, the reader must know that the relational operator > is of lower precedence.

The Logical Not Operator — !

The last of the logical operators is the not (!) operator. This operator reverses the condition to it’s
right. Suppose that one coded the following.
if ( ! (x > y) )
The ! reverses the result. If x contains 10 and y contains 5, then the test x > y is true, but the !
reverses it; the true becomes false and the else branch is taken. However, if x equals y or is
actually less than y, then the test x > y is false; the ! reverses that to true and the then-clause is
executed.

Confusing? Likely so. I have more than 35 years experience in the programming world.
One thing that I have found to be uniformly true among all programmers is confusion over not-
logic. While no one has any trouble with test conditions like x != y, there is uniform non-
comprehension about not-logical conditions, such as the one above. In fact, the chances of mis-
coding a complex not-logical expression are exponential! My advice has always been “Reverse
it; say the test in the positive and adjust the clauses appropriately.”

Here is another example to illustrate what I mean. In the dating service match test
condition, the age is to be between 20 through and including 25. That is, an age that is less than
20 or an age that is more than 25 are not candidates for a match. Thus, one could test for those
using
age < 20 || age > 25
But if these were true, then this person is not a match and we are looking for a match, so not
logic would reverse it.
if ( !((age < 20) || (age > 25)) &&
(maritalStatus == 0) && (sex == 1))
cout << "Is a potential match\n";
This is much more difficult to read for the average programmer. Unless your mathematical
background is well attuned to these kinds of logical expressions, it is much better to reverse the
not logic and say it in the positive as in the following.
if ( (age >= 20) && (age <= 25) &&
(maritalStatus == 0) && (sex == 1))
cout << "Is a potential match\n";
Decisions 118

There are a few times where not logic improves the coding by making it tighter, shorter
and more compact, mostly in the area of controlling the iterative or looping process (next
chapter). I use not logic sparingly throughout this text, preferring to always try to say it in the
positive manner.

Data Type and Value of Relational Expressions — The bool Data

Type
In C++, the result of any test condition or relational expression is always an int data type whose
value is either 0 for false or 1 for true.

When the test condition evaluation is completed by the compiler, the result is an int
whose value is either 0 or 1. Assume that integer variable x contains 10 and integer variable y
contains 5. In the following if statement
if (x > y)
the test condition x > y evaluates to true or an integer 1. The compiler then sees just
if (1)
and so does the then-clause. If however, we reverse the contents of variables x and y, then the
test results in a false and the compiler sees just
if (0)
and takes the else-clause if present.

This means that one could define an integer variable to hold the result of a relational
expression. Consider the following code.
int result = x > y;
At first, it may look a bit bizarre. To the right of the assignment operator is a test condition. After
the test is complete, the result is either 0 or 1 and that int value is what is being assigned to the
variable result. Another way of looking at the variable result is that it is holding a true/false
value.

The bool Data Type

However, there is a far better data type to use if only a true/false value is desired. This is the new
data type called bool which represents boolean data or two-valued logic. A variable of type bool
can have only two possible values, true and false. Some examples of variables that can
effectively utilize this data type include the following.
bool isWindowVisible; // true if this window is visible
bool isMoving; // true if this object is in motion
bool isAvailable; // true if available for work
bool isFoodItem; // true if this item is classified as
// food for tax purposes
Decisions 119

Consider the readability of this section of coding that deals with moving objects.
bool isMoving;
...
if (speed > 0)
isMoving = true;
else
isMoving = false;
...
if (isMoving) {
...
}
If the variable speed contains say 45 miles an hour, then the test is true and a true value is
assigned to isMoving. Notice that the test condition could also have been written
if (isMoving == true) {
...
}
But since a bool already contains the needed 1 or 0 value (true or false), it is not needed.

Using bools can add a measure of readability to a program. Consider using a bool
whenever the variable can be expressed in a true/false manner.

The compiler can always convert an integer relational expression result into a bool data
type. The above coding can be rewritten even simpler as follows.
isMoving = speed > 0;
...
if (isMoving) {
...
}

This leads us to the two vitally important shortcut test conditions that are widespread in
C++ coding. To summarize, if we code
if (x < y)
then this results in an integer value 1 or 0 which is then evaluated to see if the then-clause or
else-clause is taken
if (1)
or
if (0)
Notice that the then-clause is executed if the test result is not 0. The else-clause is taken if the
result is equal to 0. This is commonly extended by coding these two shortcuts
if (x)
or
if (!x)
where x is a variable or expression.
Decisions 120

When the compiler encounters

if (x)
it checks the contents of variable x. If x’s contents are non-zero, the then-clause is executed.
Similarly, when the compiler encounters
if (!x)
it checks the contents of variable x. If x’s contents are zero, then the then-clause is taken. Thus,
if (x)
is a shortcut way of saying
if (x != 0)
and
if (!x)
is a shortcut way of saying
if (x == 0)

I commonly keep track of these two shortcuts by using this scheme. If something does not
exist, it is zero. So if (!apples) means not apples means no apples or that apples is zero. And if
(apples) means if apples exist and apples exist if apples is not zero. However, you choose to
remember these two shortcuts, make sure you understand them for their use is widespread in
advanced C++ programming.

The Most Common Test Condition Blunder Explained

At long last, we can finally understand the most common error made by programmers when
coding test conditions. And that error is shown here.
if (quantity = 0)
or
if (x = y)
The error is coding an assignment operator instead of the conditional equality operator ==. This
is not a syntax error and does compile and execute, but the results are disastrous. Why? It is first
and foremost an assignment operator. When either of the above two if instructions are executed,
the first action is to replace the value in quantity with a 0 and to replace the contents of variable
x with the contents of variable y. This is therefore destructive of the contents of quantity and x!
Secondly, once the value has been copied, the compiler is left with just evaluating
if (quantity)
and
if (x)
But we now know what that actually becomes. If the newly updated quantity is not zero, take the
then-clause. If the newly updated x is not zero, take the then-clause.

This coding action is exceedingly rarely used. Thus, many compilers actually issue a
warning message along the lines of “assignment in a test condition!”

Always be extra careful to make sure you use the equality relational operator == and not
the assignment operator = when making your test conditions.
Decisions 121

The Conditional Expression

The conditional expression operators ? : provide a shortcut to the normal If-Then-Else coding.
Let’s calculate the car insurance premium for a customer. The rates are based on the age of the
insured. The following calculates the insured’s premium using If-Then-Else logic.
double premium;
if (age > 55)
premium = 100.00;
else
premium = 250.00;
Notice in both clauses, something is being assigned to the same variable, premium, in this case.
That is the key that the conditional expression can be used. It would be coded like this
double premium = age > 55 ? 100.00 : 250.00;

The syntax of the conditional expression is

test condition ? true expression : false expression
The test condition is the same test condition we have been discussing this whole chapter. It can
be simple or compound. It is followed by a ? After the ? comes the then or true expression. The
expression can be as simple as a constant as in this case or it can be a variable or an expression
that results in a value to be used. After the true expression comes a : to separate the true and false
expressions and the false expression follows.

Here is a more complex version. Suppose that younger drivers pay a higher rate. We now
have the following.
double premium;
if (age > 55)
premium = 100.00;
else if (age < 21)
premium = 1000;
else
premium = 250.00;
This can be rewritten as follows
double premium = age > 55 ? 100 : (age < 20 ? 1000 : 250);
Here the false portion of the first conditional expression is another entire conditional expression!
But more importantly, we have reduced seven lines of coding into one line.

The conditional expression is not limited to assignments, though it is commonly used in

such circumstances. Suppose that we need to make a very fancy formatted line indicating the
higher temperature for the day. The line is to be shown on the 10 o’clock weather. Assume that
we have already calculated the morning and evening temperatures and need now to display the
larger of the two temperatures as the higher temperature. We could do the following.
if (am_temp > pm_temp)
cout << ...fancy formatting omitted << am_temp << endl;
else
cout << ...fancy formatting omitted << pm_temp << endl;
Decisions 122

While this works well, as you have undoubtedly discovered at this point, making output look
good requires a lot of trial and error, fiddle, fiddle, to get it to look good. Here in this example,
we have precisely the same fancy formatting to do twice! As you tweak the first output, you must
remember to do the same exact things to the second output instruction. I have enough trouble
getting it right once. The conditional expression comes to our rescue. This can be rewritten as
cout << ...fancy formatting omitted
<< (am_temp > pm_temp ? am_temp : pm_temp)
<< endl;
This then displays the larger of the two temperatures. Note the () are required with the insertion
operator.

The Precedence of Operators

Table 4.1 shows the precedence of operators from highest at the top to the lowest at the bottom.
Each row is at a different level. A function call must always be done first so that the value the
function returns is available for the rest of the expression’s evaluation. Assignments are always
last. Of course, parentheses can be used to override the normal precedence.

Notice that the postfix operator (after inc or dec) and prefix operator (before inc and dec)
have high precedence so that their use can be detected early and properly applied after the current
value is used.

The unary - is used in instructions such as

x = - y;

The address operator & returns the memory location of the item that follows it. So if we
coded
&x
this returns where in memory variable x is located. We deal with addresses in a later chapter.
Decisions 123

Table 4.1 The Precedence of Operators

Operator Name Associates
functionName (...) function call left to right
++ and -- postfix increment and left to right
decrement operators
++ and -- prefix increment and right to left
decrement operators
-, +, !, & unary -, unary +, logical not, right to left
and address of operators
(datatype) typecast right to left
*, / and % multiply, divide and left to right
remainder operators
+ and - add and subtract operators left to right
>, >=, <, <= greater than, greater than or left to right
equal, less than, less than or
equal to operators
== and != equal to and not equal to left to right
operators
&& logical and left to right
|| logical or left to right
?: conditional expression right to left
=, +=, -=, *=, /=, %= assignment operators right to left

Testing of Real Numbers

There remains one additional test condition situation that must be understood. This applies only
to floating point or real numbers. Recall that a floating point number is only an approximation of
a specific real number, as close as the computer can get in the finite number of binary decimal
bits. Further, when calculations are done on these floating point numbers, small roundoff errors
and precision effects begin appearing.
Decisions 124

For example, let’s take a variable x of float data type. Suppose that it is initialized to 4.0.
The computer stores this number as close as it can get to 4.0. It might be 3.99999 or it might be
4.00001. Now assume that we do the following to x
x = x - y
where y contains 3.99998. At this point, variable x contains 0.00001 or 0.00003, depending upon
the above values. What happens if we do the following test condition?
if (x == 0)
Clearly, the else-clause is taken because x is not zero. But it sure is close to zero! The question is
“is x sufficiently close to zero to be actually considered zero?” Or wilder still, suppose x was a
double that contained 0.000000000000000123456789012345. A double has fifteen digits of
accuracy and that is what is stored here — .123456789012345E-15. Is this version of x zero?
Nope.

When testing floating point numbers, to avoid this kind of error, always test in such a
manner to see if it is sufficiently close to the desired value. Use the floating point absolute value
function.
if (fabs (x) <= .000001)
This takes the absolute value of x and compares it to the desired degree of closeness. If you have
no idea how close is close enough, try one part in a million or .000001.

Similarly when comparing two floating point values for equality, always compare the
absolute value of their difference to the desired degree of closeness. Instead of coding
if (x == y)
code
if (fabs (x - y) <= .000001)

How close is “close enough” depends on the problem at hand. Suppose that x represents
the cubic yards of concrete to place into a concrete truck to deliver to a construction site.
Probably .1 is highly accurate enough!
Decisions 125

Section B: Computer Science Example

Cs04a — Compute the Total Bill By Finding the Sales Tax Rate
Acme Company sells products in two states. Typically, state codes of 13 for Illinois and 14 for
Iowa are used to determine the tax rates. Assume that the Illinois tax rate is 7.5% and the Iowa
rate is 8%. Additionally, if 10 or more than of the same item are purchased, a discount of 4% is
given on the total cost of those items. If the total sale is $100.00 or more, shipping costs are free.
Otherwise the customer pays shipping which is $4.00 or .5% of the total order before taxes,
whichever is larger. Write a program that inputs one order consisting of the customer number (up
to six digits long), the state code number, the item number of the product ordered, the quantity
ordered and the unit cost. Print out a nice billing form showing the order details and final total
cost to the customer.

The design begins as usual by identifying the input fields. Here we need custNumber,
stateCode, itemNumber, quantity, and cost. Draw a set of main storage boxes for these and
label them with their chosen names. Figure 4.2 shows the complete main storage diagram.

Now using these names, write out the steps to solve this problem.
Prompt and input custNumber, stateCode, itemNumber, quantity, and cost
The first calculation is to find the total cost of the quantity purchased. Then we can apply
discounts. Let’s call this one subTotal; add another box in the main storage diagram for it and
write
subTotal = quantity * cost;

Figure 4.2 Main Storage for Sales Tax Problem

Next, check the quantity ordered and see if there is a discount to be applied.
if the quantity is greater than or equal to 10, then do the following
discount = subTotal * .04;
otherwise
discount is 0;
Decisions 126

Next, figure the total before tax, calling it totalBeforeTax

totalBeforeTax is subTotal minus discount

To figure the tax, we need to get the rate. Let’s call it taxRate; make a box for it and tax.
Then calculate them by
if stateCode is equal to 13 then
taxRate = .075
else check if stateCode is equal to 14 if so then
taxRate = .08
tax = totalBeforeTax * taxRate

To get the shipping costs, we need a field to hold it, say shippingCost and it is calculated
as follows
if totalBeforeTax is greater than or equal to 100 then
shippingCost is 0
otherwise
shippingCost = totalBeforeTax * .005
but if shippingCost < 4.00 then
shippingCost = 4;

Finally, the grand total due, say called grandTotal, is given by the sum of the following
partial totals.
grandTotal = totalBeforetax + shippingCost + tax
now display all these results nicely formatted

One should thoroughly desk check the design. Make up various input sets of data so that
all possible situations can occur and be verified.
customer state item qty cost
number code number
12345 13 1111 5 10.00 // no discounts
12345 13 1111 10 10.00 // only 4%
12345 13 1111 15 10.00 // 4% & free shipping
123456 14 1111 5 10.00 // other state rate

Here are the completed program and the output from the above four test executions.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs04a: Customer Order Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs04a: Customer Order */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
Decisions 127

* 8 #include <iomanip> *
* 9 using namespace std; *
* 10 int main () { *
* 11 *
* 12 // input variables *
* 13 long custNumber; // customer name *
* 14 int stateCode; // state code 13 or 14 *
* 15 long itemNumber; // item number ordered *
* 16 int quantity; // quantity ordered *
* 17 double cost; // cost of one item *
* 18 *
* 19 // prompt and input a set of data *
* 20 cout << "Enter Customer Id number: "; *
* 21 cin >> custNumber; *
* 22 cout << "Enter state code: "; *
* 23 cin >> stateCode; *
* 24 cout << "Enter Item number: "; *
* 25 cin >> itemNumber; *
* 26 cout << "Enter quantity ordered: "; *
* 27 cin >> quantity; *
* 28 cout << "Enter cost of one item: "; *
* 29 cin >> cost; *
* 30 *
* 31 // the calculation fields needed *
* 32 double subTotal; // basic cost of these items *
* 33 double discount = 0; // 4% discount if quantity >= 10 *
* 34 double totalBeforeTax; // total ordered with discount applied *
* 35 double taxRate; // tax rate based on state code *
* 36 double tax; // total tax on totalBeforeTax *
* 37 double shippingCost = 0; // shipping free if totalBeforetax>=100*
* 38 double grandTotal; // total due from customer *
* 39 *
* 40 // the calculations section *
* 41 subTotal = quantity * cost; // figure basic cost *
* 42 *
* 43 if (quantity >= 10) // apply 4% if quantity large enough*
* 44 discount = subTotal * .04; // if not, leave original 0 in it *
* 45 totalBeforeTax = subTotal - discount; // total before taxes *
* 46 *
* 47 if (stateCode == 13) // find the right tax rate to use *
* 48 taxRate = .075; // state Illinois rate *
* 49 else if (stateCode == 14) *
* 50 taxRate = .08; // state Iowa rate *
* 51 else { // oops, not a valid state code *
* 52 cout << "Invalid state code. It was " << stateCode << endl; *
* 53 return 1; *
* 54 } *
* 55 *
* 56 tax = totalBeforeTax * taxRate; // calc the tax owed *
* 57 *
* 58 if (totalBeforeTax < 100) { // need to figure shipping costs *
* 59 shippingCost = totalBeforeTax * .005; *
Decisions 128

* 60 if (shippingCost < 4.00) // if it is less than minimum amt*

* 61 shippingCost = 4.00; // reset shipping to minimum amt *
* 62 } // no else is needed since shippingCost was initialized to 0 *
* 63 *
* 64 grandTotal = totalBeforeTax + shippingCost + tax; *
* 65 *
* 66 // setup floating point format for output of dollars *
* 67 cout << fixed << setprecision (2); *
* 70 *
* 71 // display the results section *
* 72 cout << endl << endl << "Acme Customer Order Form\n"; *
* 73 cout << "Customer Number: " << custNumber << " in State: " *
* 74 << stateCode << endl; *
* 75 cout << "Item Number Quantity Cost Total\n"; *
* 76 cout << setw (8) << itemNumber << setw (11) << quantity *
* 77 << setw(11) << cost << setw (11) << subTotal <<endl <<endl;*
* 78 cout << "Total after discount:"<<setw(20)<<totalBeforeTax<<endl;*
* 79 cout << "Tax: "<<setw(20)<<tax << endl; *
* 80 cout << "Shipping costs: "<<setw(20)<<shippingCost << endl;*
* 81 cout << "Grand Total Due: "<<setw(20)<<grandTotal << endl; *
* 82 *
* 83 return 0; *
* 84 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from 4 test runs of Cs04a: Customer Order Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Test Run # 1 Results *
* 2 *
* 3 Enter Customer Id number: 12345 *
* 4 Enter state code: 13 *
* 5 Enter Item number: 1111 *
* 6 Enter quantity ordered: 5 *
* 7 Enter cost of one item: 10 *
* 8 *
* 9 *
* 10 Acme Customer Order Form *
* 11 Customer Number: 12345 in State: 13 *
* 12 Item Number Quantity Cost Total *
* 13 1111 5 10.00 50.00 *
* 14 *
* 15 Total after discount: 50.00 *
* 16 Tax: 3.75 *
* 17 Shipping costs: 4.00 *
* 18 Grand Total Due: 57.75 *
* 19 *
* 20 ========================================= *
* 21 Test Run # 2 Results *
* 22 *
* 23 Enter Customer Id number: 12345 *
* 24 Enter state code: 13 *
* 25 Enter Item number: 1111 *
Decisions 129

* 26 Enter quantity ordered: 10 *

* 27 Enter cost of one item: 10 *
* 28 *
* 29 *
* 30 Acme Customer Order Form *
* 31 Customer Number: 12345 in State: 13 *
* 32 Item Number Quantity Cost Total *
* 33 1111 10 10.00 100.00 *
* 34 *
* 35 Total after discount: 96.00 *
* 36 Tax: 7.20 *
* 37 Shipping costs: 4.00 *
* 38 Grand Total Due: 107.20 *
* 39 *
* 40 ========================================= *
* 41 Test Run # 3 Results *
* 42 *
* 43 Enter Customer Id number: 12345 *
* 44 Enter state code: 13 *
* 45 Enter Item number: 1111 *
* 46 Enter quantity ordered: 15 *
* 47 Enter cost of one item: 10 *
* 48 *
* 49 *
* 50 Acme Customer Order Form *
* 51 Customer Number: 12345 in State: 13 *
* 52 Item Number Quantity Cost Total *
* 53 1111 15 10.00 150.00 *
* 54 *
* 55 Total after discount: 144.00 *
* 56 Tax: 10.80 *
* 57 Shipping costs: 0.00 *
* 58 Grand Total Due: 154.80 *
* 59 *
* 60 ========================================= *
* 61 Test Run # 4 Results *
* 62 *
* 63 Enter Customer Id number: 123456 *
* 64 Enter state code: 14 *
* 65 Enter Item number: 1111 *
* 66 Enter quantity ordered: 5 *
* 67 Enter cost of one item: 10 *
* 68 *
* 69 *
* 70 Acme Customer Order Form *
* 71 Customer Number: 123456 in State: 14 *
* 72 Item Number Quantity Cost Total *
* 73 1111 5 10.00 50.00 *
* 74 *
* 75 Total after discount: 50.00 *
* 76 Tax: 4.00 *
* 77 Shipping costs: 4.00 *
Decisions 130

* 78 Grand Total Due: 58.00 *

.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Section C: An Engineering Example

Engr04a — Quadratic Root Solver
A major usage of decisions is to avoid doing calculations when one or more variables are out of
range for that calculation. For example, an attempt to divide by zero causes a program crash.
Passing values out of range to the arcsine function cause the asin() function to crash the program.
Commonly, decisions protect programs from such attempts. This example explores these uses.

Write a program that displays the roots of the quadratic equation,

ax2 + bx + c, given any user inputted values for a, b and c.

Analyzing the problem, the equation we need to solve is

But the complicating factor is that the user might enter values such that imaginary roots
occur or even division by zero if a is zero. To make a totally general program, we must handle all
the possibilities. The first consideration is “Is the value the user entered for the a term 0?” If so,
there is no solution. Next if the discriminant, b2–4ac, is negative, then there are two imaginary
roots given by

Further, if b2–4ac is 0, then there are two identical real roots.

Figure 4.3 Main Storage for Quadratic Root Program

Designing our solution first, we must make main storage variables for the input values.
Let’s call them a, b and c. Since the discriminant, b2–4ac, must be evaluated, let’s also make a
variable to hold it, say desc. Finally, the result variables might be root1 and root2. But in the
Decisions 131

case of the imaginary roots, there are going to be an imaginary part, so let’s also define iroot to
hold the imaginary portion. Figure 4.3 shows the main storage diagram.

Now write out the sequence of operations needed to solve this problem using our variable
names. We have
prompt and input a, b and c
if a is 0 then display no solution
otherwise to the following
calculate desc = b2–4ac
if desc is 0 then do the following
find root1 = –b/(2a)
display two roots at root1
otherwise if desc is negative then do the following
root1 = –b/(2a)
iroot = sqrt (|desc|) / (2a)
display one imaginary root as root1 + iroot * i
display the other imag root as root1 – iroot * i
otherwise
root1 = (–b+sqrt (desc))/(2a)
root2 =(–b-sqrt (desc))/(2a)
display root1 and root2
end otherwise
end otherwise

Next, make up some test values to thoroughly check out the program. For example, we
might use these sets
0, 1, 2 - for no solution
3, 4, 5 - for imaginary roots
2, 8, 6 - for two real roots
4, 4, 1 - for multiple roots

Here are the program and the test runs.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr04a - Quadratic Roots Solver *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr04a: Quadratic Equation Roots */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
Decisions 132

* 11 int main () { *
* 12 *
* 13 // input variables *
* 14 double a; *
* 15 double b; *
* 16 double c; *
* 17 *
* 18 // result variables *
* 19 double desc; *
* 20 double root1; *
* 21 double root2; *
* 22 double iroot; *
* 23 *
* 24 // prompt and input the user's coefficients for a, b and c *
* 25 cout << "Quadratic Equation Root Solver Program\n\n"; *
* 26 cout<<"Enter the quadratic equation's coefficients a, b and c\n"*
* 27 << "separated by a blank\n"; *
* 28 cin >> a >> b >> c; *
* 29 *
* 30 // setup floating point format for output of roots *
* 31 cout << fixed << setprecision (4) << endl; *
* 34 *
* 35 // check for division by 0 or not a quadratic case *
* 36 if (fabs (a) < .000001) { *
* 37 cout <<"Since a is zero, there is no solution-not quadratic\n";*
* 38 } *
* 39 // here it is a quadratic equation, sort out roots *
* 40 else { *
* 41 desc = b * b - 4 * a * c; *
* 42 // is desc basically 0 indicating multiple roots at one value? *
* 43 if (fabs (desc) <= .000001) { *
* 44 root1 = - b / (2 * a); *
* 45 cout << "Multiple roots at " << setw (12) << root1 << endl; *
* 46 } *
* 47 // is the desc positive indicating two real roots *
* 48 else if (desc > 0) { *
* 49 root1 = (-b + sqrt (desc)) / (2 * a); *
* 50 root2 = (-b - sqrt (desc)) / (2 * a); *
* 51 cout << "Two real roots at: " << setw (12) << root1 << endl; *
* 52 cout << " " << setw (12) << root2 << endl; *
* 53 } *
* 54 // desc is negative indicating two imaginary roots *
* 55 else { *
* 56 desc = fabs (desc); *
* 57 root1 = -b / (2 * a); *
* 58 iroot = sqrt (desc) / (2 * a); *
* 59 cout << "Two imaginary roots at :" << setw (12) << root1 *
* 60 << " + i * " << setw (12) << iroot << endl; *
* 61 cout << " " << setw (12) << root1 *
* 62 << " - i * " << setw (12) << iroot << endl; *
* 63 } *
* 64 } *
Decisions 133

* 65 *
* 66 return 0; *
* 67 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Four Test Runs of Engr04a - Quadratic Roots Solver *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Results of test tun #1 *
* 2 *
* 3 Quadratic Equation Root Solver Program *
* 4 *
* 5 Enter the quadratic equation's coefficients a, b and c *
* 6 separated by a blank *
* 7 0 1 2 *
* 8 *
* 9 Since a is zero, there is no solution - not a quadratic *
* 10 *
* 11 Quadratic Equation Root Solver Program *
* 12 *
* 13 ======================================================= *
* 14 Results of test tun #2 *
* 15 *
* 16 Enter the quadratic equation's coefficients a, b and c *
* 17 separated by a blank *
* 18 3 4 5 *
* 19 *
* 20 Two imaginary roots at : -0.6667 + i * 1.1055 *
* 21 -0.6667 - i * 1.1055 *
* 22 *
* 23 ======================================================= *
* 24 Results of test tun #3 *
* 25 *
* 26 Quadratic Equation Root Solver Program *
* 27 *
* 28 Enter the quadratic equation's coefficients a, b and c *
* 29 separated by a blank *
* 30 2 8 6 *
* 31 *
* 32 Two real roots at: -1.0000 *
* 33 -3.0000 *
* 34 *
* 35 ======================================================= *
* 36 Results of test tun #4 *
* 37 *
* 38 Quadratic Equation Root Solver Program *
* 39 *
* 40 Enter the quadratic equation's coefficients a, b and c *
* 41 separated by a blank *
* 42 4 4 1 *
* 43 *
* 44 Multiple roots at -0.5000 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Decisions 134

New Syntax Summary

Decision Structures — asks a question that is either true or false
if (test condition) {
0, 1, or more statements to do only when the test
condition is true
}
else { // optional else clause
0, 1, or more statements to do only when the test
condition is false
}

If no else clause is needed,

if (test condition) {
0, 1, or more statements to do only when the test
condition is true
}

If there is only one statement to do, the braces { } can be omitted, but is more error prone.
if (test condition)
1 statement to do only when the test condition is true
else
1 statement to do only when the test condition is false

The test condition often is of the form:

operand1 logical operator operand2 Where the logical operators are
== for equals
!= for not equals
> for greater than
>= for greater than or equal
< for less than
<= for less than or equal
Examples: if (x < 42) or if (x == y)

Compound Test Conditions are formed by joining two or more separate test conditions together
by using AND or OR or NOT operators (&& || !) depending on the desired relationship.
if (x > y && z < y) {
here both tests must be true for the then clause to be taken
further, if x is not greater than y, the else clause is taken immediately without
testing z < y; however, if x is greater than y, it must next test z and y to
find out the actual result
if (x > y || z < y) {
here if either one or both tests are true, the then clause is taken
further, if x > y, it immediately takes the then clause; it does not need to inspect z
Decisions 135

and y; however, if x is not greater than y, it must check z and y to see if the
test ends up true
if (!(x > y)) {
the NOT operator reverses the true/false result. Here if x is greater than y, that test
results in a true; the ! then reverses it to a false result.

All test conditions result in an int data type whose value is 1 for true or 0 for false. Thus,
if (x > y) { results in an integer, either 1 or 0, reducing the statement to
if (1) or if (0) so if it is 0, the else is taken or if it is not 0, the then is taken

Shortcuts:
if (x) is short for if (x != 0)
if (!x) is short for if (x == 0)

The Conditional Expression ? :

test condition ? true item : false item
where the true and false items can be a constant, a variable or an expression. If you use an
expression, it is wise to surround the expression with parenthesis ( ).
x = y > z ? 1 : 2; This is a shortcut way for the following:
if (y > z)
x = 1;
else
x = 2;

Design Exercises
1. Avoiding a Mostly Working Program.

Programs usually accept some kind of user input. Here we are dealing with numerical
input data. When a specific program uses numerical data, the programmer must be alert for
particular numerical values which, if entered, cause problems for the algorithm or method that is
being used. The programmer must check for possible incorrect values and take appropriate
actions. Sometimes the program specifications tell the programmer what to do if those “bad
values” are entered; other times, it is left to the good sense of the programmer to decide what to
do.
For each of the following, assume that the input instruction has been executed. You are to
sketch in pseudocode the rest of the needed instructions.

a. The program accepts a day of the week from one to seven. It displays Sunday when the day is
one; Monday, when two; Saturday, when seven.
Input a dayNumber
Decisions 136

b. Housing Cost Program. When buying a house, the seller specifies the length and width of the
outside of the home, the number of stories it has and the asking price. This program calculates
the actual cost per square foot of real living area within that home. Usually, 25% of the home
area is non-liveable, being occupied by doors, closets, garages and so on. Using this program, a
buyer can evaluate which home offers the best living area value. Using pseudocode, indicate
what should be done to make this program work appropriately for all possible numerical inputs.

input the length, width, numberStories and cost

let grossArea = length * width * numberStories
let nonLivingArea = grossArea * .25
let liveableArea = grossArea – nonLivingArea
let realCostPerLiveableFoot = cost / liveableArea
output the realCostPerLiveableFoot

2. Comparison of Cereal Prices. Grocery store shoppers are often looking for the best value for
their money. For example, a given type of cereal may come in several different sized boxes, each
with a different price. A shopper wants to purchase the most cereal they can for the least money;
that is, they want the best value for their money. A further complexity arises with coupons.
Specific size boxes may have a coupon available to lower the total cost of that box. This program
inputs the data for two different boxes and displays which one has the better value (most cereal
for the least money). Write the rest of the pseudocode to determine for each box, the actual cost
per ounce. Then, display the actual cost per ounce of each box and which is the better value,
box1 or box2.
Input box1Weight, box1Cost, box1CouponAmount
Input box2Weight, box2Cost, box2CouponAmount
Decisions 137

Stop! Do These Exercises Before Programming

1. Given the following variable definitions, what is the result of each of the following test
conditions? Mark each result with either a t (for true or 1) or f (for false or 0).
int x = 10, y = 5, z = 42;
____ a. if (x > 0)
____ b. if (x > y)
____ c. if (x == 0)
____ d. if (x == z)
____ e. if (x + y > z)
____ f. if (x / y == z)
____ g. if (x > z / y)
____ h. if (x > 0 && z < 10)
____ i. if (x > 0 && z >= 10)
____ j. if (x > 0 || z < 10)
____ k. if (x > 0 || z >= 10)
____ l. if (x)
____ m. if (!x)

2. Using the definitions in 1. above, what is the output of the following code?
if (z <= 42)
cout << "Hello\n";
else
cout << "Bye\n";

3. Using the definitions in 1. above, what is the output of the following code?
int t = y > x ? z : z + 5;
cout << t;

4. Correct all the errors in the following coding. The object is to display the fuel efficiency of a
car based on the miles per gallon it gets, its mpg.

if (mpg > 25.0) {

cout << Gas Guzzler\n";
else
cout << "Fuel Efficient\n";

In the next three problems, repair the If-Then-Else statements. However, maintain the spirit of
each type of If-Then-Else style. Do not just find one way to fix it and copy that same “fix” to all
three problems. Rather fix each one maintaining that problem’s coding style.
Decisions 138

5. Correct all the errors in the following coding. The object is to display “equilateral triangle” if
all three sides of a triangle are equal.
if (s1 == s2 == s3);
{
cout << "equilateral triangle\n";
}
else;
cout >> "not an equilateral triangle\n";

6. Correct all the errors in the following coding. The object is to display “equilateral triangle” if
all three sides of a triangle are equal.
if (s1 == s2)
if (s2 == s3)
cout << "equilateral triangle\n";
cout >> "not an equilateral triangle\n";

7. Correct all the errors in the following coding. The object is to display “equilateral triangle” if
all three sides of a triangle are equal.
if (s1 == s2) {
if (s2 == s3) {
cout << "equilateral triangle\n";
}
else {
cout >> "not an equilateral triangle\n";
}
}

8. Correct this grossly inefficient set of decisions so that no unnecessary decisions are made.
if (day == 1)
cout << "Sunday\n";
if (day == 2)
cout << "Monday\n";
if (day == 3)
cout << "Tuesday\n";
if (day == 4)
cout << "Wednesday\n";
if (day == 5)
cout << "Thursday\n";
if (day == 6)
cout << "Friday\n";
if (day == 7)
cout << "Saturday\n";
Decisions 139

9. Correct this non-optimum solution. Consider all of the numerical possibilities that the user
could enter for variable x. Rewrite this coding so that the program does not crash as a result of
the numerical value entered by the user. You may display appropriate error messages to the user.
Ignore the possibility of the user entering in nonnumerical information by accident.
double x;
double root;
double reciprocal;
cin >> x;
root = sqrt (x);
reciprocal = 1 / x;
cout << x << " square root is " << root
<< " reciprocal is " << reciprocal << endl;

10. Correct this inherently unsound calculation.

double x;
double y;
cin >> x;
y = x * x + 42.42 * x + 84.0 / (x * x * x + 1.);
if (!y || y == x) {
cout << "x’s value results in an invalid state.\n"
return 1;
}
Decisions 140

Programming Problems
Problem Cs04-1 — Easter Sunday
Given the year inputted by the user, calculate the month and day of Easter Sunday. When the
program executes, it should produce output similar to this.
Easter Sunday Calculator
Enter the year: 1985
Easter Sunday is April 7, 1985

The formula is a complex one and produces the correct day for any year from 1900
through 2099. I have broken it down into intermediate steps as follows. Frequently, Easter
Sunday is in March, but occasionally it is in April. The following formula calculates the day of
the month in March of Easter Sunday.
let a = year % 19
let b = year % 4
let c = year % 7
now start to put these pieces together
let d = (19 * a + 24) % 30
let e = (2 * b + 4 * c + 6 * d + 5) % 7
finally, the day of the month of Easter Sunday is
let day = 22 + d + e
However, if the day is greater than 31, then subtract 31 days and the resulting value in day is in
April instead. But the equation is off by exactly 7 days if these years are used: 1954, 1981, 2049
and 2076. Thus, when the calculation is finished, if the year is one of these four, you must
subtract 7 days from the day variable. The subtraction does not cause a change in the month.
Test your program on the following years — I have shown the day you should obtain in
parentheses:
1985 (April 7)
1999 (April 4)
1964 (March 29)
2099 (April 12)
1900 (April 15)
1954 (April 18)
1981 (April 19)
2049 (April 18)
2076 (April 19)
1967 (March 26)
Decisions 141

Problem Cs04-2 — Calculating Wages

Calculate a person’s wages earned this week. Prompt and input the person’s social security
number (nine digits with no dashes), their hourly pay rate, the hours worked this week and the
shift worked. The shift worked is 0 for days, 1 for second shift and 2 for the “graveyard shift”.
The company pays time and a half for all hours worked above 40.00. The additional shift bonus
is a 5% for second shift and 15% for the graveyard shift. Format the output as follows:
Employee Number: 999999999
Hours Worked: 99.99
Base Pay: $ 9999.99
Overtime Pay: $ 9999.99
Shift Bonus: $ 9999.99
Total Pay This Week: $99999.99

Make the following test runs of the program.

Employee rate hours shift
123456789 5.00 40.00 0
123456788 5.00 40.00 1
123456787 5.00 40.00 2
123456786 5.00 60.00 0
123456785 5.00 60.00 1
123456784 5.00 60.00 2
123456783 5.00 0.00 2

Problem Cs04-3 — Scholastic GPA Results

The program inputs a student id number that can be nine digits long and their grade point
average, GPA. The program is to display that student’s status which is based only on their GPA.
If the GPA is less than 1.0, the status is Suspended. If the GPA is less than 2.0 but greater than or
equal to 1.0, then the status is Probation. If the GPA is greater than or equal to 2.0 and less than
3.0, the status is Satisfactory. If the GPA is greater than or equal to 3.0 and less than 4.0, then the
status is Dean’s List. If the GPA is 4.0, then the status is President’s List. Display the results as
follows.
Id GPA Status
123456789 3.25 Dean’s List

Make sure that no unneeded tests are made. That is, if you find that the GPA is that for
Suspended, then do not additionally test for the other conditions. Once you have found a match,
when finished displaying the results, do not subject that set of input data to additional test
conditions.

Test your program with several test runs. The following series of values should
thoroughly test the program.
Decisions 142

123456789 0.5
123456788 1.0
123456787 1.1
123456786 2.0
123456785 2.1
123456784 3.0
123456783 3.1
123456782 3.9
123456781 4.0

Problem Engr04-1 — Snell’s Law — Optical Engineering

Snell’s Law gives the angle that light is bent when it passes through a region with an index of
refraction n1 into another region with a different index of refraction n2 . An example is a light ray
that passes through water in a crystal bowl. As the ray passes from the water through the clear
crystal glass sides of the container, it is bent according to Snell’s Law.
n1 sin angle1 = n2 sin angle2

When a ray passes from a region with a low index of refraction n1 into a region with a
higher index n2, the exit angle is smaller than the entrance angle or the light bends toward the
vertical. When passing from a region with a higher index of refraction into a region of lower
index of refraction, the angle of exit is greater than the entrance angle or the angle bends away
from the vertical. This is shown in the above drawing.

Write a program that calculates the exit angle of incidence angle2 , given the entrance
angle of incidence angle1 and the two indices of refractions, n1 and n2 . Prompt the user to enter
these three values; the angle input should be in degrees. Display the original input data along
with the exit angle nicely formatted. The equation to be solved is
Decisions 143

Caution: if n1 > n2, then for some angles, the absolute value passed to the arcsine is greater than
1.0. This means that all light is reflected back in the direction it came from and none goes into
the region two.

Test your program using a crystal bowl of water. The bowl is made of Crown Glass
whose index of refraction is 1.52326. The water has an index of 1.33011.
Test 1 q1 is 30 degrees coming from the glass and going into the water
Test 2 q1 is 90 degrees coming from the glass and going into the water
Test 3 q1 is 90 degrees coming from the water and going into the glass
Test 4 q1 is 30 degrees coming from the water and going into the glass

Problem Engr04-2 — Power Levels

Decibels (dB) are often used to measure the ratio of two power levels. The equation for the
power level in decibels is

where P2 is the power level being monitored and P1 is some reference power. Prompt the user for
the two power levels; then calculate and display the resulting decibels. You must guard against
all ranges of numerical entries for the two power levels. Test your program with these inputs for
P1 and P2.
1.0 5.0
1.0 50.0
496.64 1932.4
0.0 42.

Problem Engr04-3 — Formula Evaluation

Write a program to evaluate the following function for all possible numerical values of x that the
user can input.

Show sufficient test runs to demonstrate that all possible situations are handled by your
program.
Files and Loops 144

Chapter 5 — Files and Loops

Section A: Basic Theory

Introduction
Up to this point, our programs have been severely limited to inputting only one set of data. This
chapter introduces the various iterative instructions that allow a program to perform a series of
instructions repetitively until the ending condition is reached. Iterative instructions are
figuratively called looping instructions. Now programs can input as many sets of data as
required. Programs can perform a series of instructions many times until the desired result is
achieved.

Often the input consists of many sets of input data. In all of the previous programs in the
first four chapters, there was only one set of input data. In those programs it was a simple matter
to key in the input data from the keyboard. However, suppose that there were fifty such sets of
data? Of course, you only needed one test run of your programs, right? They all worked perfectly
on the very first test run, so that you only had to enter the test data one time, right? Okay. Okay. I
am teasing. But this could have happened if one designed on paper and thoroughly desk checked
before coding it in C++. But more than likely, you needed to run the program several times, if
only to get the output looking good. If there are going to be 50 sets of test data, would you like to
enter all that data numerous times?

Of course not! Hence, this chapter begins by showing you how you can use existing files
of data for input as well as writing the program’s output to a result.txt file that can then be
printed from Notepad in its entirety.

How are the input files created in the first place? For those that come with this text, I used
Notepad. You can make your own using Notepad as well. I would suggest that you use the .txt
file extension so that you can just double click on the data file and have Notepad open it up
automatically. Within Notepad, just type the lines as if you were going to enter them into the
program from the keyboard.
Files and Loops 145

Input Files
To input data from a file instead of from the keyboard is extremely easy. It requires minimal
changes on your part. Input files, in C++, make use of the already existing istream class. Recall
that a class in C++ is just a blueprint or model to follow when the compiler needs to make an
object of that kind. The class contains all the data it needs to do its job plus all of the functions
we need to effectively use it. The input file class, called ifstream, is built upon the simple
istream class, extending it to operate on a file of data. Thus, everything you currently know
about an istream instance, cin, applies totally to an instance you make of the ifstream.

To use a file for input, you must include another header <fstream>. This header file
includes the definitions of both input and output files.

Next, you need to define an instance of the ifstream class. Unlike the istream class in
which we the used built-in instance called cin, here we can name the file variable anything
desired. I often call mine infile. One codes just
ifstream infile;
or
ifstream myfile;
Any descriptive name is fine.

Next, you need to have the program actually open the input file by calling the open()
function. The open process is where you provide the actual filename to use. The system then
finds that file, attempts to open it for read operations and get everything ready for your first
actual input operation. Hence, you must know the actual filename of the data file on disk that you
want to use for input — its precise name and file extension, if any. You must also know the drive
and path to the file as well. If you have any doubts, use the Explorer and navigate and find the
actual file you want to use for input operations. The single biggest problem that students have
when using input files instead of directly inputting data from the keyboard is getting the location
and filename correct.

Let’s say that the filename is test1.txt and it is located on disk drive D: in the folder
\ComputerScience\TestData. The full path to the file stored in a literal character string is then
“d:\\ComputerScience\\TestData\\test1.txt”. Did you notice the double \\? In C++, a single
backslash \ indicates an escape sequence character is coming next. We have used the \n escape
sequence to create a new line. There are a number of other escape sequences. Since a single \
means here comes an escape sequence code, we cannot code as a filename string
“d:\ComputerScience\TestData\test1.txt”! It assumes each of the \C, \T and \t are escape
sequences which they are not! (A \t is the escape sequence for a tab code.) Thus, if you forget the
double backslashes, the system will never find your file, ever. In my opinion, this is the single
most klutzy feature of the language. I won’t tell you how many times I have accidentally forgot
the double backslashes in my filenames.
Files and Loops 146

Now there is a point to all this effort to get the filename string coded properly. If the
filename is not found during the open process, then the stream winds up in the fail state and
nothing can be input. Usually, this is simply a case of coding the wrong path, misspelling the
filename, omitting the file extension, omitting the double backslashes and so on.

The open() function is coded as follows.

infile.open ("d:\\ComputerScience\\TestData\\test1.txt");
The parameter is the name of the file and is usually a character string literal until we get to the
chapter on character string variables.

Here is the complete opening sequence.

ifstream infile;
infile.open ("d:\\ComputerScience\\TestData\\test1.txt");

As usual, there is a shortcut to coding both these two lines. The commonly used shortcut
is to define the variable and initialize it at the same time.
ifstream infile("d:\\ComputerScience\\TestData\\test1.txt");

I highly recommend that the test data files be copied into the project’s folder residing
along side of the cpp files. Then the coding is simpler.
ifstream infile ("test1.txt");
This is saying to open the file test1.txt located in the current default disk drive in the current
default subfolder. When running Microsoft Visual C++, for example, that current drive and
folder is the folder that the project containing the cpp file. Seldom are one’s data files in that
folder. But there is nothing to keep you from using Explorer to copy the test data files into the
project folder to simplify the filenames for the opening process.

There is, of course, one small detail that I mentioned and did not elaborate upon when I
said it above. If the file cannot be opened for whatever reason, the file system goes into the bad
state. This means that no input operations can be done on using it. All attempts to input data fail.
Hence, after creating and opening an input file, one must check to see if the file was successfully
opened and ready for input operations.

I/O Stream States

Any I/O stream has several state flags that indicate the current state of affairs within that stream.
There are simple access functions to retrieve these status flags. The functions take no parameters.
The more frequently used ones include
good (); // returns true if all is well
bad (); // returns true if a serious I/O error occurred
eof (); // returns true if it is at the end of the file
fail (); // returns true if bad data has been encountered
To use any of these, place the stream variable and a dot to the left of the function; the following
are all valid test conditions.
Files and Loops 147

if (cin.good ())
if (cin.bad ())
if (cin.eof ())
if (cin.fail ())
if (infile.good ())
if (infile.bad ())
if (infile.eof ())
if (infile.fail ())
Notice we can test cin as well as our files. Each of these must be examined in detail.

Testing for Goodness

After opening the file, if the process has failed because the filename is misspelled, we should
display a message and terminate the program. This can be done as follows.
ifstream infile ("test1.txt");
if (!infile.good ()) {
cout << "Error: cannot open the input file\n";
return 1;
}
Since the good() function returns true if all is ok, we must check for the opposite, hence the ! in
the test condition. Notice that to abort the program, I return back to DOS but this time I returned
a non-zero value. DOS does not care what you return. However, by convention, a value of zero
being returned means all is ok and any non-zero value means the program failed in some way.
Alternatively, one can test the fail bit.
ifstream infile ("test1.txt");
if (infile.fail ()) {
cout << "Error: cannot open the input file\n";
return 1;
}

There is a shortcut way to do the above same coding and this is what most programmers
use.
ifstream infile ("test1.txt");
if (!infile) {
cout << "Error: cannot open the input file\n";
return 1;
}
It is just the expression !infile. Recall that the test condition !x can be interpreted to mean if x
does not exist. Here !infile is true if the input stream is not in the good state for whatever reason.
The reverse test can also be used and shortly becomes a work horse for us.
if (infile)
This is a shortcut for asking if the input file stream is still in the good state.
Files and Loops 148

Any given stream at any given time might not be in the good state for several reasons. It
has encountered bad data or it has reached the end of the file or a serious I/O error occurred, such
as running out of disk space on an output file.

Testing for Bad Data Entry

Suppose that we attempt to input the integer quantity by coding

cin >> quantity;
and assume the user enters
A0<cr>
That is, the A key is pressed instead of the 1 key. The input stream locks up at this point and goes
into the fail state since it was asked to input an integer and it failed to do so. The stream remains
in the fail state and no further input operations are attempted. (In advanced programming
situations, there are ways to reset the state flags, remove the offending data and resume, but these
techniques as far beyond the beginning level.)

So at this point, we must examine in detail how the extraction operator works on the
stream of data coming into the computer. The key to extraction operations is an internal current
position within the stream pointer which keeps track of where the we are within the input
stream of characters. Let’s say that we have coded the following input instruction.
int quantity;
double cost;
cin >> quantity >> cost;
Let’s also imagine all the ways one could correctly and incorrectly enter the data. Recall that
white space (consecutive series of blanks, tabs, carriage returns and line feeds, for example) is
the delimiter between the values. Here is the first way.
10 42.50<cr>

Initially the current position in the stream is pointing to the 1 digit. The extraction
operator first skips over any white space. In this case, there is none. Next, it must extract an
integer. So it inputs successive characters that meet the requirements of an integer (that is, the
digits 0 through 9 and the + and – signs). It stops whenever it encounters any character that
cannot be a part of a valid integer number. In this case, the blank between the 0 and 4 digits
terminates the extraction of quantity. Note that the current position in the input stream is now
pointing to that blank.

The next input operation, the extracting of the double cost, resumes at the current
position in the input stream. It again skips over white space to the first non-white space character,
here the 4 digit. Next, it extracts successive characters that can be part of a double until it
encounters any character that cannot be part of the double. In this case, the CR and LF codes (the
enter key) end the extraction. Again, the current position in the input stream is updated to point
to the CRLF pair. The next input operation resumes here at this location, usually skipping over
white space if an extraction operator is used.
Files and Loops 149

Now consider this erroneous input.

10 A2.45<cr>
After inputting the quantity, the current position in the input stream is pointing to the blank
between the 0 and A characters. When the extraction of the double begins, it skips over white
space to the A character. It then inputs all characters that can be a part of a double. Here the A
character ends it; there are none. The input stream now goes into the fail state and all further
input operations are not done. Clearly, we must check on this state after all of our input
operations are done and point out the error in user input.

But what about this circumstance?

10 4A.45<cr>
When the extraction operator is to begin the process for the double cost, the current position is at
the blank between the 0 and 4 digits. It skips over white space and begins extracting characters
that can be in a double. It successfully inputs the 4 digit. It stops on the A character and the
current position in the input stream is on the A, but it then believes it has successfully input a
double whose value is 4.0. If we perform another input operation that is not asking for a letter,
the stream then goes into the fail state.

The End of File

DOS marks the physical end of a file with a special byte whose value is a decimal 26 or a CTRL-
Z code, ^Z as it is displayed on the screen. (Hold down the control key and press Z.) Most editors
do not display this end of file marker byte, but display all bytes up to that point. On a keyboard
data entry, one could press Ctrl-Z to simulate the end of file. When reading in information from a
file, we must be able to detect when we have reached the end of that file.

Consider the following input operation of quantity and cost.

infile >> quantity >> cost;
and the file contains
10 42.50^Z
The extraction operator retrieves 10 for quantity and 42.50 for cost. The ^Z code is where the
current position in the input stream is pointing. If we do an additional input operation, then the
end of file condition occurs.

Suppose that a premature end of file exists because someone forgot to enter the cost.
10^Z
The instruction was asked to input two values. Clearly the quantity is successfully inputted, but
when the extraction operator attempts to skip over white space and find the first non-white space
character to begin inputting the double cost, the end of file is reached and the stream goes into
the EOF state and the operation fails.

After we discuss the iterative looping instructions, we will see how to actually test for
these circumstances in our programs.
Files and Loops 150

Closing a File

When a program is finished working with a file, it should call the close() function. The
close() function makes the file available for other programs to use. If the file is an output file, any
remaining data not yet physically written to the disk are actually written and the end of file
marker is written.

C++ and DOS automatically close all files should the programmer forget to close them. It
is bad programming style to fail to close the files that you open. On some operating systems,
error messages are generated if you fail to close your files. The close() function is very simple.
infile.close ();

Here is a complete sample program that inputs a single set of data from a file and displays
the total cost of an order. Notice the placement of the new instruction.
#include <iostream>
#include <iomanip>
#include <fstream>
using namespace std;

int main () {
// input variables and total
int quantity;
double cost;
double total;

// define the file and open it

ifstream infile ("Test1.txt");
if (!infile) { // unable to open file, display msg and quit
cout << "Error: cannot open test1.txt file\n";
return 1;
}

// try to input a set of data

infile >> quantity >> cost;
if (!infile) { // check if input was successful
cout << "Error: unable to input quantity and cost\n";
return 2;
}

// calculate total cost of order

total = cost * quantity;

// display the results

cout << "The quantity is: " << quantity << endl
<< "The cost of each is: $ " << cost << endl
<< "The total of the order is: $" << total << endl;
Files and Loops 151

infile.close ();
return 0;
}

The Iterative Instructions

There are three iterative instructions available in C++. They are used to create programming
loops so that a series of instructions can be executed more than once. The typical program
follows the Cycle of Data Processing, Input a set of data, Process that set of data in some manner,
Output that set of data and then repeat the entire process until there are no more sets of data, in
other words, the end of the file has been reached.

The first of the three is the Do While structure which is implemented with the while
instruction. The Do While structure shown in Figure 5.1 illustrates how the looping process
works. First a test condition is checked. While that test condition is true, a series of things to do
are performed. Then the test is checked again. The loop continues until the test condition
becomes false. Of vital importance is that something in the series of things to do must somehow
eventually alter the test condition so that the loop can end. If not, an infinite loop results.

Figure 5.1 The Do While Structure

The syntax of the while statement that implements the Do While Structure is
while (test condition) {
0, 1, or more things to do while the condition is true
}
If there is only one instruction or even a null instruction to do, it can be shortened to
while (test condition)
1 statement;
The alternative indentation would be
while (test condition)
{
0, 1, or more things to do while the condition is true
}
Files and Loops 152

The test condition is exactly the same test condition that was used with If-then-else
instructions. There is no change to it whatsoever. A while instruction actually is rather simple,
yet powerful instruction. Using it, many different kinds of processing loops can be built.

Loops That Are to Be Executed a Known Number of Times

Let’s apply the while statement to one of the simplest forms of looping. Sometimes one must
perform a loop a known number of times. For example, let’s sum all the odd integers from one to
twenty-five. Here is a way it can be done.
int number = 1;
int sum = 0;
while (number <= 25) {
sum += number;
number += 2;
}
cout << "The sum of all odd integers from 1 to 25 is "
<< sum << endl;

This short program contains a number of very key elements of the looping process in
general. First, notice how sum is defined. The sum variable must be initialized to 0 before the
loop begins. Inside the loop, the next number is added into sum; then the next odd integer is
calculated. When the loop is finished, sum contains the answer and is then displayed. This gives
some general guidelines for creating the sum or total of some quantity.

To develop a total or sum of some quantity, follow these steps.

a. Before the loop begins, initialize the sum or total to 0
b. Inside the loop, add the next value to the total or sum
c. When the loop is done, display the contents of that total or sum

The next key point with this loop is that the variable number is used as the loop control
variable which is a variable that is used to control the number of times the loop is executed.
Here are the steps involved.
a. A loop control variable must be initialized before the loop begins.
b. A loop control variable is tested for the ending value in the while clause
c. A loop control variable is incremented or decremented at the very end of the loop.

Consider what would happen if we used this incorrect sequence.

int number = 1;
int sum = 0;
while (number <= 25) {
number += 2; // wrong order
sum += number;
}
Files and Loops 153

It is obvious that we fail to add the initial value of number, a 1 in this case, to sum. But worse
still, when number is incremented the last time to 27, that value is then erroneously added into
sum before the test condition gets a chance to shut the loop down.

When designing a loop, ask yourself “What is going to control the number of times the
loop is to execute?” Here, it was “keep going while number is less than or equal to 25,” the last
odd number to be used. Next, before the while statement, initialize the control variable to its first
or initial value. Finally, at the very end of the loop, do what is necessary to get the loop control
variable ready for its next iteration.

Here is another example. Suppose that we needed the sum of the reciprocals of all
integers from one to twenty. That is, compute the sum of 1 + 1/2 + 1/3 + 1/4 + ... + 1/20. Here is
a way it can be done following the design guidelines. The current integer num controls the
number of times the loop is to be done. So we have this far
int num;
...
while (num < 21) {

What should the initial value for num be? 1, in this case. And at the very end of the loop,
the next value is given by incrementing num. So now we have
int num = 1;
...
while (num < 21) {
...
num++;
}

Finally, write the body of the loop. Here we need to sum the reciprocal of num. We
cannot just write sum += 1/num. Can you spot why? Integer division yields zero for all terms but
the first. The sum must be a double. So here is the final version.
int num = 1;
double sum = 0.;
while (num < 21) {
sum += 1./num;
num++;
}
cout << "Result is " << sum << endl;
Files and Loops 154

Loops to Input All Data in a File

By far the most common while loop in any program is one setup to process all of the sets of data
in a file or entered by hand from the keyboard. The input loops can be constructed in several
different ways, depending upon the circumstances.

Sentinel Controlled Input Loops

A sentinel value is a unique, special value that has been placed in the data to signal that there are
no more sets of data. Suppose that the file contains lines that have the quantity ordered and the
unit cost of them. There are an unknown number of lines. One way to let the program know that
the end has been reached is to enter some unique value for the cost and quantity. Two common
sentinel values are 999 and –999, though they have to be chosen with the problem in mind. If you
do not expect that anyone would return 999 of the items (which would be the meaning of
entering a quantity of –999), then this would work to define the end of the data. In other words,
if you opened the data file with Notepad, you would see
10 42.50
3 10.99
-999 -999
The program should then input sets of quantity and cost until the quantity becomes –999.
How do we structure the program to do this? Following the loop design guidelines, the
while clause is
while (quantity != -999) {
So before the while statement, quantity must have its initial value; this means we must read in
the first set of data so that quantity has a value. Then, at the very end of the loop, we must
attempt to input another set of data. Here is the complete program; notice the locations of the
different statements.
#include <iostream>
#include <iomanip>
#include <fstream>
using namespace std;

int main () {
int quantity; // quantity ordered
double cost; // price of one item
double total; // total cost of this order

// define, open the input file--display error if fails

ifstream infile ("test.txt");
if (!infile) {
cout << "Error: cannot open test.txt\n";
return 1;
}

// setup floating point format for output of dollars

Files and Loops 155

cout << fixed << setprecision (2);

infile >> quantity >> cost; // input first set of data

// a quantity of -999 marks last set of data

while (quantity != -999) {
// calculate this sale and display results
total = quantity * cost;
cout << setw (4) << quantity << setw (10) << cost
<< setw (12) << total << endl;
// attempt to get the next set of data
infile >> quantity >> cost;
}

infile.close ();
return 0;
}

As coded, this is a prime example of mostly working software! Consider what this
program does if the last line of the file contains -99 -99 by accident instead of the expected -999?
Or what happens if we reach the end of the file unexpectedly because the user forgot to insert the
sentinel values line? Or what happens if the file contains bad data, such as A2 instead of a
quantity of 42? Look at the while statement. Under what circumstances is the loop actually
ended? Only when the quantity is -999. And in the above situations, it never will contain that
ending value! So our loop continues to run endlessly, an infinite loop, displaying the same
garbage over and over.

So how could we change the while statement to guard against erroneous situations? In
this example, the file should always be in the good state. If it is ever in the fail state for whatever
reason, in this problem, it is an error. So we could remedy this mostly working program this way.
while (quantity != -999 && infile) {
// calculate this sale and display results
total = quantity * cost;
cout << setw (4) << quantity << setw (10) << cost
<< setw (12) << total << endl;
// attempt to get the next set of data
infile >> quantity >> cost;
}
if (!infile)
cout << "An error occurred processing test.txt\n";

Sentinel controlled loops are often programmed by novice programmers who have not
learned how to check for and handle the end of file condition. The input streams are perfectly
capable of detecting and reporting that there are no more sets of data. Where sentinel controlled
loops shine are in keyboard data entry and in menu processing.
Files and Loops 156

Keyboard Data Entry Sentinel Controlled Loops

Consider this sequence displayed on the screen.

Enter another student grade or -1 to quit: -1<cr>
The loop’s test condition must be
while (grade != -1) {
This also means that before the while statement, we must input a student grade to have one for
which to test in the while statement. This also means that there must be another input a student
grade instruction at the very end of the loop. Here is what the loop looks like.
double grade;
cout << "Enter a student grade or -1 to quit: ";
cin >> grade;
while (grade != -1. && cin) {
...do something with this grade - process & output
cout << "Enter a student grade or -1 to quit: ";
cin >> grade;
}
if (!cin)
cout << "An error was encountered in the input\n";

One aside. Even though grade is a double floating point type, I did not use the fabs()
function to check for equality. Why? Well, if –1 is entered, however it maybe stored in grade,
when comparing it to –1. which is also a double, they are both going to be stored identically.
However, had grade been the result of a calculation, then fabs() would have been prudent.

Menus as Sentinel Controlled Loops

Menus are commonly found in applications. Consider the following screen display and single
digit user enter from the keyboard.
Acme File Services

1. Produce the Daily Sales Report

2. Produce the Weekly Sales Summary Report
3. Produce the Salesperson Ranking Report
4. Quit the program

Enter the number of your choice: 4<cr>

The while test condition is to keep doing the loop as long as choice != 4 where 4 is the
sentinel value. Here is how the menu can be done. In a later chapter, we will see how this large
amount of duplicate coding can be reduced.
cout << "Acme File Services\n\n"
<< " 1. Produce the Daily Sales Report\n"
<< " 2. Produce the Weekly Sales Summary Report\n"
Files and Loops 157

<< " 3. Produce the Salesperson Ranking Report\n"

<< " 4. Quit the program\n\n"
<< "Enter the number of your choice: ";
cin >> choice;
while (choice != 4) {
if (choice == 1) {
...do the daily sales report
}
else if (choice == 2) {
...do the weekly summary report
}
else if (choice == 3) {
...do the salesperson ranking report
}
else {
cout << "Choice is out of range, reenter 1 through 4";
}
cout << "Acme File Services\n\n”
<< " 1. Produce the Daily Sales Report\n"
<< " 2. Produce the Weekly Sales Summary Report\n"
<< " 3. Produce the Salesperson Ranking Report\n"
<< " 4. Quit the program\n\n"
<< "Enter the number of your choice: ";
cin >> choice;
}

Okay. But what happens in the above loop if the user enters the letter A for a choice or
presses Ctrl-Z signaling the end of file? Under what conditions does the while loop terminate?
Only when choice contains a 4 does it end. And if bad data or eof occurs, the loop continues on
endlessly in an infinite loop and the menus fly by on the screen at a rapid rate. It is a nice “light
show,” but not productive. How can we alter the while loop so that this cannot occur?
while (choice != 4 && cin) {
By simply adding an additional check that the cin stream is still in the good state will prevent
silly things occurring.

Primed Input Loops that Detect End of File

This example is the easiest method for beginning programmers to implement. It is called a
primed loop approach because we must input the first set of data before the while statement.
Why? We know that the input streams have a way to check for EOF. Available to us are
while (!infile.eof()) {
and
while (infile) {
In the first while statement, we check directly for the end of file condition; and, if it has not yet
occurred, continue the loop. However, the second while statement is a better choice. It is
Files and Loops 158

checking to see if the input stream is still in the good state. If the end of file has been reached or
if bad data has caused the stream to lock up or freeze on the bad character, the loop ends. Since
bad data can and does occur, it is wiser to use the second test condition for our loops.

Following the loop design guidelines, if while (infile) is the test, then before the while
statement, infile must be initialized. This means we need to attempt to input the first set of data.
Then, we would input the next set of data at the very end of the loop. Here is the main loop
portion of the preceding program.
infile >> quantity >> cost; // input first set of data
while (infile) { // stop at eof or bad data
// calculate this sale and display results
total = quantity * cost;
cout << setw (4) << quantity << setw (10) << cost
<< setw (12) << total << endl;
// attempt to get the next set of data
infile >> quantity >> cost;
}

This is called a primed loop because there is an initial “Get the First Set of Data” that is
done before the loop begins. The loop only continues as long as the input stream is in the good
state. At the very end of the loop, an identical input instruction is coded. This identical second
input instruction is known as the “Get Next Set of Data” instruction.

Here is a common way to mess up the loop coding. Can you spot what is going to happen
at the end of file or if bad data is encountered?
while (infile) { // stop at eof or bad data
infile >> quantity >> cost;
total = quantity * cost;
cout << setw (4) << quantity << setw (10) << cost
<< setw (12) << total << endl;
}
When the end of file is reached, the input stream goes into the not good state and nothing is
inputted for quantity or cost. However, the next two lines ignore this and go right ahead
calculating a total and displaying the data as if it actually had another set of data. This often
results in the last record in the file being processed and output twice! It is even worse if bad data
were encountered as the contents of quantity and cost are unpredictable.

And this brings up the testing that we must perform when the loop actually ends. The
while statement terminates when the file is no longer in the good state. That is, it is the end of file
(eof), it has run into a bad spot on the disk and is unable to read the data (bad), or has detected
bad data in the input lines (the fail bit is on but not eof and not bad). Of the three ways the while
loop terminates, only the eof situation is normal and to be expected. The other two represent an
error situation. We must alert the user to any errors that we encounter.
Files and Loops 159

Error checking can be done in many ways. Here is one simple version that does not
discriminate between a physically bad disk drive and bad data entered.
infile >> quantity >> cost; // input first set of data
while (infile) { // stop at eof or bad data
...
infile >> quantity >> cost;
}
if (infile.eof())
cout << "All data processed successfully\n";
else
cout << "Bad data was encountered, output is incorrect\n";
Or one can check this way.
if (!infile.eof())
cout << "Bad data was encountered, output is incorrect\n";
Or one can check this way.
if (!infile.eof()) {
if (infile.bad()) {
cout << " The disk cannot be read. Use a backup copy\n";
else
cout << "Bad data was encountered\n";
}

I highly recommend using the Primed Loop approach when processing a file of data. It is
an easy one to design, code and test. However, there is a shorter way that most experienced C++
programmers are going to use.

A More Compact Loop That Detects End of File

Seasoned programmers balk at coding the same instructions twice. Here, we take advantage of
the extraction operator’s return value. We know that we can chain extraction operators.
cin >> quantity >> cost;
This means that the extraction operator must be returning the input stream so that it is available
for the next extraction operator to its right. In other words, the first portion
cin >> quantity
when completed must return back cin so that the next extraction would be
cin >> cost;

Since the while clause desires to test the goodness of the stream, the shortcut version
merges the input operation into the while clause test condition!
while (cin >> quantity >> cost) {
This first does the indicated input operations. When that is finished, the last extraction operator
returns cin and the test condition becomes the expected while (cin). Clever, but complex. Here is
the whole looping process, this time using a file of data.
while (infile >> quantity >> cost) {
Files and Loops 160

total = quantity * cost;

cout << setw (4) << quantity << setw (10) << cost
<< setw (12) << total << endl;
}
Notice that the loop is now more compact, not necessarily more readable.

This is one of the most common forms of input loops you will see. It does have one
drawback, besides complexity. And that is, the entire input process must be able to be stated in
one long chain of extraction operators. This is not always possible as we will see later on.

Now that we can write loops, what can we do with them. The next section covers some of
the many uses for loops.

Applications of Loops

Application: The Summation of a Series

Summations are a very common action in numerical analysis. The preceding summation of the
reciprocal (1/num) from one to twenty lacks one very important feature. Just running the loop
through the first twenty terms is not sufficient to guarantee any kind of accuracy. Let’s formalize
the problem as seen in numerical analysis: the summation of a series. Further let’s sum 1/x2
instead.

where N can be infinity.

What is needed is a result that is sufficiently accurate, not just the first twenty terms. Let’s
say that we need the result accurate to .001. That is, keep on adding in the next term until the
value of that next term is less than or equal to .001. Within the loop body, the next term to add
into the sum is 1./x2 using the current value of the loop control variable x. We could change the
while test condition to just
while (term > .001) {

However, one safety factor should always be considered. Certainly 1./x2 is going to
become small rapidly as the value of x increases. If x is 10, the term is .01; if x is 100, the term
is .0001 which is more accuracy than was requested. The while clause certainly would stop long
before x is up to 100.
However, what if we summed this one
Files and Loops 161

Oops! This one would never converge on an answer. Each term gets significantly larger than the
previous one! If our only ending condition was to continue while the term is greater than .001,
we would have an infinite loop that would run forever, until we found a way to abort the
program. Thus, in all summation programs, some kind of alternate loop termination is always
installed, just in case something goes terribly wrong with the process.

In this example, if x ever reached 1000, for example, something must be very wrong. If x
becomes that large, terminate the loop and give it further study. So now our test condition
becomes
while (x < 1000 && term > .001) {
Following the loop construction guidelines, both x and term must be given their starting values
before the while clause. Both must be assigned their next value at the end of the loop. Here is the
complete loop.
int x = 1;
double term = 1; // 1/x/x = 1 for the first term
double sum = 0;
while (x < 1000 && term > .001) {
sum += term;
x++;
term = 1. / (x * x);
}
cout << "Sum yields " << sum << endl;

The summation is a powerful technique that is found both in Computer Science and
Engineering applications.

Counters and Totals — Grand Totals

In many applications, the count of the number of sets of data is needed. If, for example,
each line represents a single sales, then the count of the number of lines input would be the
number of sales. If each line of input represented the data for a single policy holder, then the
count of the number of lines input would be the number of policyholders. In a bowling score-
keeping program, if each line contained the data for one game, then a count of the lines would be
the number of bowling games played.

Commonly, when the end of file is reached, the application is required to display grand
totals. Using the sales example of quantity and cost, for each set of data, we calculate and
display the total of that sales. However, at the end of the file, the grand total sales ought to be
shown along with the number of sales and even perhaps an average sales.

Here is how the user would like the report to appear.

Acme Daily Sales Report
Files and Loops 162

Quantity Cost Total

Sold Per Item Sales

9999 $9999.99 $99999.99

9999 $9999.99 $99999.99
9999 $9999.99 $99999.99
---------
$99999.99
Number sales: 9999
Average Sales: $99999.99

First, identify what is new from the previous version of the program above. Headings
have been added along with some specific columnar alignments and dollar signs. What is really
significant is the ----- line and what comes after it. When are all these final lines displayed?
Clearly, they are displayed after the main loop terminates at the end of the file. A grand total
sales variable is needed, and this is officially called a total. A variable must be added to add up
the number of sales which is really a count of the number of lines inputted, this is known as a
counter.

The rules for counters and totals are simple. Repeating the previous guidelines
a. Before the loop begins, counters and totals must be initialized to their starting
values, usually zero.

b. Within the loop, counters must be incremented and totals added to.

c. When the loop ends, counters and totals are often displayed.

Here are the completed program and the output from a sample run. Notice carefully the
placement of the new instructions.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic05a - Acme Sales Report with Grand Totals *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /************************************************************/ *
* 2 /* */ *
* 3 /* Basic05a Acme Sales Report with grand totals */ *
* 4 /* */ *
* 5 /************************************************************/ *
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 using namespace std; *
* 11 *
* 12 int main () { *
* 13 int quantity; // quantity ordered *
* 14 double cost; // price of one item *
* 15 double total; // total cost of this order *
Files and Loops 163

* 16 *
* 17 int numSales = 0; // total number of sales *
* 18 double grandTotal = 0; // grand total sales *
* 19 double avgSales; // the average sales amount *
* 20 *
* 21 // define, open the input file - display error if fails *
* 22 ifstream infile ("sales.txt"); *
* 23 if (!infile) { *
* 24 cout << "Error: cannot open sales.txt\n"; *
* 25 return 1; *
* 26 } *
* 27 *
* 28 // setup floating point format for output of dollars *
* 29 cout << fixed << setprecision (2); *
* 31 *
* 32 // display headings and column heading lines *
* 33 cout << " Acme Daily Sales Report\n\n" *
* 34 << "Quantity Cost Total\n" *
* 35 << " Sold Per Item Sales\n\n"; *
* 36 *
* 37 // main loop - process all input lines in the file *
* 38 while (infile >> quantity >> cost) { *
* 39 // calculate this sale *
* 40 total = quantity * cost; *
* 41 *
* 42 // increment counters and totals *
* 43 numSales++; *
* 44 grandTotal += total; *
* 45 *
* 46 // display this report line *
* 47 cout << " " << setw (4) << quantity *
* 48 << " $" << setw (7) << cost *
* 49 << " $" << setw (8) << total << endl; *
* 50 } *
* 51 *
* 52 // display grand total lines *
* 53 cout << " --------\n"; *
* 54 cout << " $" << setw (8) *
* 55 << grandTotal << endl; *
* 56 cout << " Number Sales: " << setw (4) *
* 57 << numSales << endl; *
* 58 *
* 59 // find and display the average sales *
* 60 avgSales = grandTotal / numSales; *
* 61 cout << " Average Sales: $" << setw (8) *
* 62 << avgSales << endl; *
* 63 *
* 64 infile.close (); *
* 65 return 0; *
* 66 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
Files and Loops 164

* Output from Basic05a - Sales Report with Grand Totals *

/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Acme Daily Sales Report *
* 2 *
* 3 Quantity Cost Total *
* 4 Sold Per Item Sales *
* 5 *
* 6 10 $ 42.00 $ 420.00 *
* 7 1 $ 10.00 $ 10.00 *
* 8 20 $ 15.00 $ 300.00 *
* 9 2 $ 14.50 $ 29.00 *
* 10 7 $ 30.00 $ 210.00 *
* 11 5 $ 10.00 $ 50.00 *
* 12 -------- *
* 13 $ 1019.00 *
* 14 Number Sales: 6 *
* 15 Average Sales: $ 169.80 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Finding the Maximum and Minimum Values

Often when processing a set of data, the maximum and minimum values are desired. In the
previous sales report, two additional lines could be added at the end of the report.
Highest Sales: $99999.99
Lowest Sales: $99999.99
To produce these, we need another pair of doubles to hold these, say hiSales and lowSales.

How are they found? Within the loop, each time a new total is found by multiplying cost
times quantity, we must compare that new total to what is currently in hiSales and lowSales. If
the new total is larger than the current hiSales, replace hiSales with this new value. Likewise, if
the new total is lower than the current value in lowSales, replace lowSales with this lower value.
When the loop ends, these two fields, hiSales and lowSales, contain the largest and smallest
sales.

But to what do we initialize these two variables? In this problem, the finding of the
maximum and minimum values is overly simplified. Due to the nature of the problem, there can
be no negative values (unless we can expect refunds to be in this file). One might suspect that all
we need do is to initialize both hiSales and lowSales to 0. Wrong. Let’s see what happens after
we input the very first sales line. The total is 420.00. That is certainly larger than the 0 in hiSales,
so hiSales is now updated to contain 420.00. But look what happens to the lowSales; its initial
value of 0 is certainly smaller than 420.00 and thus lowSales is not updated. In fact, none of the
totals are below 0 and thus, lowSales is never updated and ends up being 0!

Rule: when finding the maximum or minimum values, initialize the two variables
that are to contain the maximum and minimum to the actual data contained in the first set
of data.
Files and Loops 165

Since we must have the first set of data to use to get the initial values for the high and low
sales, we should use the primed loop approach. Since much of the program is the same, only
excerpts are shown here. Pay careful attention to the location of the various steps. In the main
processing loop, after the total sales is calculated, the new total is compared to the maxSales and
then to the minSales variables.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic05b - Acme Sales Report with Grand Totals and High/Low Sales *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***********************************************************/ *
* 2 /* */ *
* 3 /* Basic05b Acme Sales with grand totals and max/min sales */ *
* 4 /* */ *
* 5 /***********************************************************/ *
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 using namespace std; *
* 11 *
* 12 int main () { *
* 13 int quantity; // quantity ordered *
* 14 double cost; // price of one item *
* 15 double total; // total cost of this order *
* 16 *
* 17 int numSales = 0; // total number of sales *
* 18 double grandTotal = 0; // grand total sales *
* 19 double avgSales; // the average sales amount *
* 20 double maxSales = 0; // the largest sales - if file is empty *
* 21 double minSales = 0; // the smallest sales - if file is empty *
* 22 *
* 23 // define, open the input file - display error if fails *
* 24 ifstream infile ("sales.txt"); *
* 25 if (!infile) { *
* 26 cout << "Error: cannot open sales.txt\n"; *
* 27 return 1; *
* 28 } *
* 29 *
* 30 // setup floating point format for output of dollars *
* 31 cout << fixed << setprecision (2); *
* 33 *
* 34 // display headings and column heading lines *
* 35 cout << " Acme Daily Sales Report\n\n" *
* 36 << "Quantity Cost Total\n" *
* 37 << " Sold Per Item Sales\n\n"; *
* 38 *
* 39 // get first set of data to initialize max/min values *
* 40 infile >> quantity >> cost; *
* 41 if (infile) // only assign if there was a set of data *
* 42 maxSales = minSales = quantity * cost; *
* 43 *
* 44 // main loop - process all input lines in the file *
Files and Loops 166

* 45 while (infile) { *
* 46 // calculate this sale *
* 47 total = quantity * cost; *
* 48 *
* 49 // check on min and max values *
* 50 if (total > maxSales) { *
* 51 maxSales = total; *
* 52 } *
* 53 else if (total < minSales) { *
* 54 minSales = total; *
* 55 } *
* 56 *
* 57 // increment counters and totals *
* 58 numSales++; *
* 59 grandTotal += total; *
* 60 *
* 61 // display this report line *
* 62 cout << " " << setw (4) << quantity *
* 63 << " $" << setw (7) << cost *
* 64 << " $" << setw (8) << total << endl; *
* 65 *
* 66 // get next set of data *
* 67 infile >> quantity >> cost; *
* 68 } *
* 69 *
* 70 // display grand total lines *
* 71 cout << " --------\n"; *
* 72 cout << " $" << setw (8) *
* 73 << grandTotal << endl; *
* 74 cout << " Number Sales: " << setw (4) *
* 75 << numSales << endl; *
* 76 *
* 77 // find and display the average sales - guard against no data *
* 78 if (numSales) *
* 79 avgSales = grandTotal / numSales; *
* 80 else *
* 81 avgSales = 0; *
* 82 cout << " Average Sales: $" << setw (8) *
* 83 << avgSales << endl; *
* 84 *
* 85 // display max/min sales values *
* 86 cout << " Highest Sales: $" << setw (8) *
* 87 << maxSales << endl; *
* 88 cout << " Lowest Sales: $" << setw (8) *
* 89 << minSales << endl; *
* 90 *
* 91 infile.close (); *
* 92 return 0; *
* 93 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Files and Loops 167

Bulletproofing Programs
Once you have the basics of looping down, the next thing to consider is what about all the things
that can go wrong while inputting data.

The first thing you must always consider is that sometimes a file can have no data in it
yet. Suppose that Acme Company has a Daily Sales File that contains all of the sales data for its
salespeople for one day. What happens to our program if someone runs it before anyone has
made a sale for the day? The file is empty. What does your program do at that point? Certainly it
should not do calculations on nonexistent data!

What does Basic05b program do if there are no data in the file? Look at lines 20, 21 and
40 through 42. I initialized the maximum and minimum sales variables to 0 so that they have a
starting value. On line 40, the input operation encounters the end of file. Was it checked for?
Yes, on line 41, only if the input stream is in the good state does the assignments to the
maximum and minimum sales variables take place. And when line 45 is executed, the while test
condition fails because the stream is not in the good state, rather it is at the end of file. The main
loop is never executed.

Ok. Now look over the display of all the totals and results. First, look at lines 77 through
81. Here, I slid in a bit more protection. The program must calculate the average sales, but the
divisor, numSales could be 0, and is, if there the file is empty. If I did not guard against this
possibility, then, should such occur, the program would crash with a divide exception at this
point! Notice that the maximum and minimum sales correctly display their 0 initial values.

So Basic05b is in good shape if there are no data in the file. Always double check your
programs to be certain that all works if there are no data inputted. A situation of no input data can
and does happen in the real world.

Ok. Next, what happens if bad data is encountered? Ah ha, Basic05b is not ready for bad
data events! If there is bad data on the very first line, the program simply prints all zeros for the
fields and quits without any notice to the user that there was anything wrong! Worse still,
suppose the bad data was on line one hundred of the sales file! Now we get a report that actually
looks like it is a valid one, at least for the first ninety nine sets of data that are shown and used in
the calculations! This can completely mislead the user who is not aware that bad data was
encountered. The report gives no indication whatsoever that the results are not valid. This is
totally unacceptable for a program to do. We must at the least let the user know that bad data
was encountered and the results shown are not valid.

Where can we take care of this detail? The main loop is testing the goodness of the
stream and terminates whenever it is not good, either end of file or bad data. Thus, the proper
place to check is immediately after the main loop terminates and before we begin displaying the
totals. Insert after line 68, the end brace of the main processing loop, the following bulletproofing
Files and Loops 168

code.
if (!infile.eof()) {
cout << "Error: bad data encountered on line: "
<< numSales + 1 << endl;
infile.close();
return 1;
}
Sometimes programmers even go so far as to add one more line after these four.
cout << "All data successfully processed\n";
Thus, whenever you write a program that has an input processing loop, you should always
bulletproof your coding, allowing for files with no data in them and for encountering bad data.

Creating Output Files

When programs process all of the sets of data in a file, very often the number of output lines
exceeds the amount that can be shown on the screen without having the screen scroll. This makes
it much more difficult to verify the output is correct. Instead, programs that produce a lot of
output lines often write the lines to an output file and later that file can be viewed with Notepad
or printed.

Switching from cout to an actual output file is very easy. The output file class is
ofstream. Similar to input files, an instance must be created and the file opened for output. The
only information that is really needed is the filename you wish it to create. My suggestion is to
use a file extension of txt so that simply double clicking the file in Explorer launches Notepad to
open it for viewing or printing.

Similar to an input file, the output file can be defined and opened with a single statement
or explicitly opened with a second instruction. Here are both methods.
ofstream outfile ("results.txt");
or
ofstream outfile;
outfile.open ("results.txt");

If the file does not exist, a new one is built and initially it contains 0 bytes. If the file
exists, it is emptied of its prior contents and now contains 0 bytes. As the program outputs to the
outfile stream, the system stores the data in the file. When the program is done, it should call the
close() function. The close operation on an output file writes any remaining data and places the
end of file marker in it.

One should also check for failure to successfully open the file. If the disk is full, the open
fails. If you should use an invalid path or folder name, the open fails. If you should use an invalid
filename, the open fails. So it is always wise to check. Thus, we follow the opening of the output
file with
Files and Loops 169

if (!outfile) {
cout << "Error: unable to open output file: result.txt\n";
return 2;
}

How do you write data to the file? It’s done exactly the same way you write data to the
cout stream.
outfile << "Hello World\n";
This writes the text Hello World and a new line code to the file just as it does when sent to the
screen with cout.

There is, however, one detail to remember. Just as we must setup the floating point flags
with the cout stream, we must do the same with the output file stream.
// setup floating point format for output of dollars
outfile.setf (ios::fixed, ios::floatfield);
outfile << setprecision (2);

An output file is closed in exactly the same manner as an input file.

outfile.close ();

Basic05c is a rewrite of program Basic05b using an output file instead of using cout. I
have also taken the liberty to insert bulletproofing for bad data this time. To save pages, some of
the coding that is exactly the same as the previous example has been removed.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic05c - Acme Sales - Output Goes to a File *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /************************************************************/ *
* 2 /* */ *
* 3 /* Basic05c Acme Sales Report using an output file */ *
* 4 /* */ *
* 5 /************************************************************/ *
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 using namespace std; *
* 11 *
* 12 int main () { *
* 13 int quantity; // quantity ordered *
* 14 double cost; // price of one item *
* 15 double total; // total cost of this order *
* 16 *
* 17 int numSales = 0; // total number of sales *
* 18 double grandTotal = 0; // grand total sales *
* 19 double avgSales; // the average sales amount *
* 20 double maxSales = 0; // the largest sales - if file is empty *
* 21 double minSales = 0; // the smallest sales - if file is empty *
* 22 *
Files and Loops 170

* 23 // define, open the input file - display an error msg if fails *

* 24 ifstream infile ("sales.txt"); *
* 25 if (!infile) { *
* 26 cout << "Error: cannot open sales.txt\n"; *
* 27 return 1; *
* 28 } *
* 29 *
* 30 // define and open the output file *
* 31 ofstream outfile ("results.txt"); *
* 32 if (!outfile) { *
* 33 cout << "Error: cannot open results.txt for output\n"; *
* 34 infile.close (); *
* 35 return 2; *
* 36 } *
* 37 *
* 38 // setup floating point format for output of dollars *
* 39 outfile << fixed << setprecision (2); *
* 41 *
* 42 // display headings and column heading lines *
* 43 outfile << " Acme Daily Sales Report\n\n" *
* 44 << "Quantity Cost Total\n" *
* 45 << " Sold Per Item Sales\n\n"; *
* 46 *
* 47 // get first set of data to initialize max/min values *
* 48 infile >> quantity >> cost; *
* 49 if (infile) // only assign if there was a set of data *
* 50 maxSales = minSales = quantity * cost; *
* 51 *
* 52 // main loop - process all input lines in the file *
* 53 while (infile) { *
* 54 // calculate this sale *
* 55 total = quantity * cost; *
* 56 *
* 57 // check on min and max values *
* 58 if (total > maxSales) { *
* 59 maxSales = total; *
* 60 } *
* 61 else if (total < minSales) { *
* 62 minSales = total; *
* 63 } *
* 64 *
* 65 // increment counters and totals *
* 66 numSales++; *
* 67 grandTotal += total; *
* 68 *
* 69 // display this report line *
* 70 outfile << " " << setw (4) << quantity *
* 71 << " $" << setw (7) << cost *
* 72 << " $" << setw (8) << total << endl; *
* 73 *
* 74 // get next set of data *
* 75 infile >> quantity >> cost; *
Files and Loops 171

* 76 } *
* 77 *
* 78 // check for bad data in the input file - if found, display an *
* 79 // error message to screen and in the output file - abort pgm *
* 80 if (!infile.eof()) { *
* 81 cout << "Error: bad data encountered in the input file\n" *
* 82 << "The line containing the error is " << numSales + 1 *
* 83 << endl; *
* 84 outfile << "Error: bad data encountered in the input file\n" *
* 85 << "The line containing the error is " << numSales + 1 *
* 86 << endl; *
* 87 infile.close (); *
* 88 outfile.close (); *
* 89 return 3; *
* 90 } *
* 91 *
* 92 // display grand total lines *
* 93 outfile << " --------\n"; *
* 94 outfile << " $" << setw (8) << grandTotal *
* 95 << endl; *
* 96 outfile << " Number Sales: " << setw (4) *
* 97 << numSales << endl; *
* 98 *
* 99 // find and show average sales - guard against empty input file *
* 100 if (numSales) *
* 101 avgSales = grandTotal / numSales; *
* 102 else *
* 103 avgSales = 0; *
* 104 outfile << " Average Sales: $" << setw (8) *
* 105 << avgSales << endl; *
* 106 *
* 107 // display max/min sales values *
* 108 outfile << " Highest Sales: $" << setw (8) *
* 109 << maxSales << endl; *
* 110 outfile << " Lowest Sales: $" << setw (8) *
* 111 << minSales << endl; *
* 112 *
* 113 infile.close (); *
* 114 outfile.close (); *
* 115 return 0; *
* 116 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Files and Loops 172

The Do Until Instruction — An Alternative to the Do While

Figure 5.2 The Do Until Structure

The Do Until iterative structure is different from the Do While in that the body of the loop is
always done once and then the test condition is checked. The Do Until is shown above.

What is the difference between a Do While and a Do Until structure? By way of an

analogy, imagine you are driving down Main Street on a Friday night. You decide to turn left.
The Do Until says to turn left. Now apply the test condition — was it ok to turn left — no
oncoming cars — not a one way street — not a red light — no pedestrians? In contrast, the Do
While says to check to see if it is ok to turn left and if so, then turn left. Ok. This is a bit of an
extreme viewpoint on the two structures, but it serves to drive home the main point of difference,
the Do Until always executes the series of things to do one time. Why is this important? Consider
reading a file of data using a Do Until. If there are no data in the file, the test for that condition
does not occur until after all the processing of the nonexistent set of data is done and the results
of the calculations on the nonexistent set of data are output! Thus, a Do Until is a specialized
form of looping to be used in those circumstances in which one can guarantee in all
circumstances the body of the loop must be done one time.

The syntax of the Do Until is

do {
0, 1 or more statements
} while (test condition);
or
do
{
0, 1 or more statements
} while (test condition);
Files and Loops 173

While one can always use a Do While to solve a programming problem, there are a few
times that a Do Until is more convenient. Here is one such time. Suppose that we need to display
the following on the screen and get the user response.
Enter a numerical choice of 1 through 4: _
Once the user has entered a number, it might not be in the correct range. The program should
then insist on a proper entry be made.
Enter a numerical choice of 1 through 4: 5
Enter a numerical choice of 1 through 4: 6
Enter a numerical choice of 1 through 4: -42
Enter a numerical choice of 1 through 4: 4
The program must provide a loop that repeatedly prompts and gets the user’s choice until it is
within the valid range. While this could be done with a Do While, a Do Until is more convenient.
int choice;
do {
cout << "Enter a numerical choice of 1 through 4: ";
cin >> choice;
} while (choice < 1 || choice > 4);
The corresponding Do While takes a bit more coding.
int choice;
cout << "Enter a numerical choice of 1 through 4: ";
cin >> choice;
while (choice < 1 || choice > 4) {
cout << "Enter a numerical choice of 1 through 4: ";
cin >> choice;
}

The Do Loop or for Statement

Frequently, a program needs to perform a series of instructions a known, finite number of times.
We have seen that a while loop can easily be used to sum the reciprocals of the numbers from
one to ten, for example. Here is the while version.
int num;
double sum;
num = 1;
sum = 0;
while (num < 21) {
sum += 1./num;
num++;
}
cout << "Result is " << sum << endl;
The for loop is a shortcut way to do the same thing. Here is how that same reciprocal sum
program could have been written using a for statement.
int num;
double sum = 0.;
for (num=1; num<21; num++)
Files and Loops 174

sum += 1./num;
cout << “Result is “ << sum << endl;

The for syntax is

for (0, 1 or more initial expressions separated by a comma;
test condition;
0, 1 or more bump expressions) {
body of the loop
}

The initial expressions represent all statements that occur before the while statement. If
there are more than one, separate them with commas. The initial expression is ended with a
semicolon. The test condition is the same test condition we have been using and is ended with
another semicolon. The bump expressions represent the incrementing of the loop control
variables, but are not so limited. We could have written this same for loop more compactly.
double sum;
for (int num=1, sum=0; num<21; num++)
sum += 1./num;
cout << "Result is " << sum << endl;
Here the initialization of sum has been moved into the for statement as one of the initial
expressions. Also note that I am now defining the loop control variable num within the initial
expression. This variable then is technically only available within the loop itself and should not
be used later on after the loop ends.

But we could condense this for loop even further.

double sum;
for (int num=1, sum=0; num<21; sum += 1./num, num++) ;
cout << "Result is " << sum << endl;
Now I have moved the sum calculation into the first bump expression location. Notice that num
is incremented after the sum is calculated using the original value in num before the increment
takes place.

But it could be consolidated even further.

double sum;
for (int num=1, sum=0; num<21; sum += 1./num++) ;
cout << "Result is " << sum << endl;
Here sum is calculated using the current contents of num and the ++ after increment then takes
place. And now we have a very compact line of code.

Here is another example. Suppose that we had the following coding.

double x, y, sum;
x = 1;
y = 10;
sum = 0;
while (x < y) {
Files and Loops 175

sum += x * y;
x++;
y--;
}
This could be rewritten using a for loop as follows.
double x, y, sum;
for (x=1, y=10, sum = 0; x < y; x++, y--) {
sum += x * y;
}
Notice that all statements above the while clause can be considered initialization statements.
Here there are three. There are two bump expressions since both x and y can be considered loop
control variables in this case. This leaves then only one statement in the loop’s body.

However, this can be condensed even more by moving the remaining statement into the
bump expressions.
double x, y, sum;
for (x=1, y=10, sum = 0; x < y; sum += x * y, x++, y--) ;
Notice that the sum instruction must be done before the incrementing and decrementing of x and
y.

This can be condensed even further.

double x, y, sum;
for (x=1, y=10, sum = 0; x < y; sum += x++ * y--) ;
Again notice that this uses the postfix ++ and – – operators so that the current values of x and y
are used in the calculation before they are altered.

Rule. All while loops can be rewritten as a more compact for loop. All for loops can be
rewritten as a more understandable and readable while loop.

By now you are probably wondering why anyone would want to write such a compact,
dense line of code as the for loop with x and y instead of the more readable while version. Job
security, no — just kidding. The reason lies with compiler guidelines and speed of execution. In
all circumstances, the compiler is allowed to create the fastest, best possible machine instruction
sequence for one C++ statement. Take the summation of the x*y example. The while version has
seven separate executable instructions. The condensed for version has one statement. Thus, in all
circumstances the for version is guaranteed to have the best possible, fastest executing set of
machine instructions generated by the compiler. How much faster? In this case the for statement
version is perhaps 5% faster in execution.

However, most compilers have a “global optimize for speed” compiler option. Microsoft
Visual C++ has a Debug build and a Release build option, with Debug as the default. A Debug
version of a program contains lots of debugging information to assist in finding programming
errors. When a Release build is chosen, the compiler by default optimizes for speed. No
debugging information is included in the resulting exe file, which is therefore drastically smaller
Files and Loops 176

in size. The Release version of a program executes substantially faster than the Debug version,
which tends to check for all sorts of internal errors as well as those committed by the program
directly. When global optimize for speed is in effect, the compiler can then do anything it wishes
to any statement in the entire program. Typically, the compiler rearranges program statements
into somewhat different order to gain speed of execution. Of course, when you let the compiler
begin to move lines of coding around, it is entirely possible the compiler may move a line that it
should not have and a new bug that was not there in the Debug version now appears. In large
programs this can happen and sometimes global optimization is disabled for a section of the
program.

By writing these compact for loops, you are guaranteeing that in all circumstances the
compiler creates the fastest possible execution speed for the loop. However, is this really
significant? It all depends. If the entirety of the program was the summation of x*y above, the
amount of time actually saved in measured in nanoseconds — it’s insignificant. However, if this
calculation was being done 10,000,000 times, then that speed increase is observably significant.

Since speed of execution is the primary concern with programs today, programmers
usually code for speed.

By the way, you read a condensed for statement by decomposing it back into the while
loop of which it is equivalent. However, condensed for loops are very hard to read. So many
companies prefer the more verbose while versions because they are more readable and therefore
more maintainable.

Also, for loops have other purposes than consolidating while loops. In subsequent
chapters, we will see that a major use of a for loop is to do a large series of instructions a known
number of times. Typical coding that we will see in later chapters is like this.
for (int i=0; i<limit; i++) {
a significant body of things to do
}
In these cases, the for loops are easy to follow.

Here is a more practical use of a for loop. Suppose we need to calculate a table of 101
values of the sine function from an angle of 0 to PI in uniform steps. Here is a good use of a for
loop.
const double PI = acos (-1.);
const double delta = PI / 100;
double angle = 0;
double sinAngle;
cout << setprecision (4);
for (int i=0; i<101; i++) {
sinAngle = sin (angle);
cout << setw (5) << angle << " "
<< setw (6) << sinAngle << endl;
Files and Loops 177

angle += delta;
}
The uniform increment in the angle is calculated and the loop done 101 times. Each iteration
through the loop, the sine is calculated and the angle and sine are displayed. The angle is
incremented as the last instruction to get it ready for the next loop iteration. While I could also
have moved the angle increment into the bump expressions, keeping it in the loop’s body kept
the for statement more readable.

Efficient Loops
Often loop bodies are executed a large number of times. In such cases, it is important to keep
things that do not need to be in the body out of it. For example, consider this assignment to
variable z.
int x, z;
double y = 0;
for (x=0; x<100; x++) {
y += x * x / 42.;
z = 100;
}

In this case, the assignment of 100 to z is done 100 times. How many times does z need to
be assigned its value of 100? Once. Since variable z is not used within the loop body, it could be
moved either before the loop or after the end } of the loop. In fact, this is precisely one thing that
the compiler does during global optimizations for speed in Release builds of a program. It moves
these loop invariants, as they are called, out of the loop, either before the loop if it is needed
within the loop or after it is done if it is not needed within the loop.

Loop control variables should be of the integer data type whenever possible. The integer
math instruction set on any computer is the fastest executing math types. The floating point math
instruction set is one of the slowest. If the loop has to be done a large number of times, the
difference in speed is noticeable when the loop control variable is an integer type versus a
floating point type.

For example, I could have rewritten the display of 101 sines program above using the
variable angle to control the loop. However, having the loop control variable be a double slows
down the overall speed of execution.
const double PI = acos (-1.);
const double delta = PI / 100;
double angle = 0;
double sinAngle;
cout << setprecision (4);
for (; fabs(angle - PI)> .0001; angle += delta;) {
sinAngle = sin (angle);
Files and Loops 178

cout << setw (5) << angle << " "

<< setw (6) << sinAngle << endl;
}

Notice that if there are no initial conditions, the semicolon must still be coded. The same
is true if there are no bump expressions. What does this loop do?
for (; true; ) {
...
}
It is equivalent to writing
while (true) {
...
}

Nesting of Loops
Loops can be nested. The rules are simple. The inner loop must be entirely contained within the
outer loop. If there are If-Then-Else statements within a loop, the entire If-Then-Else must be
within the loop. Here is a correct nested loop.
int j = 0, k;
while (j < 100) {
// some outer loop statements can be here
for (k=0; k<100; k++) {
// inner loop statements
}
// some more outer loop statements
j++;
}

If you try to incorrectly nest a loop and an If-Then-Else, the compiler catches this and
gives an error messages. Here is an example.
int j, k;
for (j=0; j<10; j++) {
// some loop statements
if (k < 10) {
// then-clause
}
} <---- this ends the for loop and there is no else-clause
else { <---- compiler error cannot find the corresponding if
// else clause of k<10
}
Files and Loops 179

An Example of Nested Loops

Very often programs use nested loops. In this next example, a student’s average grade is
calculated. The sentinel controlled inner loop inputs test scores and accumulates the total points.
When the user enters a score of –99, the inner loop terminates and calculates and displays the
average grade. Wrapped around this process is an outer loop asks the user if there is another
student to grade. If there is, the inner loop is repeated for that student. Assume that floating point
fixed format has been setup on cout and the precision set to 1.
long id;
cout << "Enter Student ID number or -1 to quit: ";
while (cin >> id && id != -1)
{
double sum = 0;
int count = 0;
double score;
cout << "Enter test score or -99 when finished: ";
while (cin >> score && score != -99)
{
sum += score;
count++;
cout << "Enter test score or -99 when finished: ";
}
if (count > 0)
{
cout << "Student " << id << " grade: "
<< setw (4) << sum/count <<endl;
}
cout << "Enter Student ID number or -1 to quit: ";
}

This example illustrates a vital detail when using nested loops. Inside the outer loop and
before the inner loop begins, notice that sum and count must be reinitialized to 0 to get ready for
the next student’s set of test scores. A common error is to code this as follows.
long id;
double sum = 0;
int count = 0;
double score;
cout << "Enter Student ID number or -1 to quit: ";
while (cin >> id && id != -1)
{
cout << "Enter test score or -99 when finished: ";
while (cin >> score && score != -99)
{
sum += score;
count++;
cout << "Enter test score or -99 when finished: ";
Files and Loops 180

}
...
This correctly calculates the first student’s average. But what happens when the second student’s
scores are entered? Since sum and count are not reset to 0, the second student’s scores are added
to the first and so on.

Section B: Computer Science Examples

Cs05a — Acme Ticket Sales Summary Program
Back in Chapter 2, we wrote a program to calculate the price of tickets that a customer purchased
in which a discount is given for children and senior citizens. In such sales applications, it is vital
that some form of a transaction log is also written at the time of purchase documenting that
purchase. Assume that just such a log has been produced, called trans-log.txt. The log file
contains three integers that represent the number of regular tickets purchased, the number of
children’s tickets purchased and the number of senior citizens’ tickets purchased by a single
customer. After those three integers comes a double that contains the total purchase price of all
those tickets.

Management now wishes to have a summary report of the sales. This program inputs the
transaction log file and builds a daily ticket sales summary report which is actually written to a
file called results.txt. The Ticket Sales Summary Report contained in that file looks like this.
Acme Ticket Sales Summary Report

Number of Tickets Sold Total Cost

Adult Child Senior Of Tickets

99 99 99 $ 999.99
99 99 99 $ 999.99
99 99 99 $ 999.99
--- --- --- --------
Totals: 999 999 999 $9999.99
Percents: 99% 99% 99%

Average cost of ticket: $ 999.99

When the end of the sales input occurs, after displaying a line of dashes, the totals of each
of the four columns are shown. Then, the percentage sold in each of the three categories are
displayed. Finally, the average price of a ticket is calculated and presented.

As usual, begin by defining the input fields and the needed calculation and output fields.
Let’s call the three number of tickets variables numAdult, numChild and numSenior. The
input total cost of the tickets is just cost. Make four main storage boxes so labeled with these
Files and Loops 181

names. Next, what are we going to need to calculate the first total line after the line of dashes?
Four total variables are needed, one for each column. Let’s call them totNumAdult,
totNumChild, totNumSenior and totalCost. How do we calculate the percent results? We can
add up the three total tickets’ results to find the grandTotalTicketsSold integer. Knowing that,
the seniors’ ticket percentage is just totNumSenior * 100. / grandTotalTicketsSold. Let’s call
these percentAdult, percentChild, percentSenior. Finally, the average ticket cost, avgCost, is
just the totalCost / grandTotalTicketsSold. After making labeled main storage boxes for all of
these, we can then write the sequence of instruction we need. The final main storage diagram is
shown in Figure 5.3.

Since there is both an input file and an output file, we need two file variables; these could
also be added to the main storage diagram if desired. Following the usual design procedure, now
sketch out the solution in pseudocode or pseudo English using these variable names.

Figure 5.3 Main Storage for Ticket Summary Program

The initial steps and the main loop are as follows.

open the input file, infile
if it fails to open, display an error message and quit
open the output file, outfile
if it fails to open, display an error message, close the input file and quit
setup floating point format with two decimal digits for dollars on outfile
display the heading line on outfile
display the two column heading lines on outfile
set totNumAdult, totNumChild, totNumSenior and totalCost to 0
input numAdult, numChild and numSenior and cost from infile
if there are no data, display a message, close the files and quit
while (input operation is successful) {
add numAdult to totNumAdult
add numChild to totNumChild
add numSenior to totNumSenior
add cost to totalCost
Files and Loops 182

display the numAdult, numChild and numSenior and cost variables on outfile
input numAdult, numChild and numSenior and cost from infile
}

When we get to this point, all the data have been input, if any. We should guard against
bad input data. Thus, we can add the following to handle such eventualities. Then we move onto
the final calculations and display.
if infile is in the bad state, display an error message, close the files and quit
display on outfile the dashes line
display on outfile totNumAdult, totNumChild, totNumSenior and totalCost
let grandTotalTicketsSold = totNumAdult + totNumChild + totNumSenior
let percentAdult = totNumAdult * 100 / grandTotalTicketsSold
let percentChild = totNumChild * 100 / grandTotalTicketsSold
let percentSenior = totNumSenior * 100 / grandTotalTicketsSold
display on outfile the percentAdult, percentChild, percentSenior
let avgCost = totalCost / grandTotalTicketsSold
display on outfile the avgCost
close infile and outfile

With the simple sequence written, make up some test data and thoroughly desk check the
solution to verify it works perfectly on paper. Then, code it into a C++ program. Notice how
easily this one converts into C++. A good design makes programming much easier to do.

Here are the complete program and a test run.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs05a - Acme Ticket Sales Summary Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs05a Acme Ticket Sales Summary Program */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 using namespace std; *
* 11 *
* 12 int main () { *
* 13 *
* 14 // input fields *
* 15 int numAdult; // number adult tickets sold to this customer *
* 16 int numChild; // number child tickets sold to this customer *
* 17 int numSenior; // number senior tickets sold to this customer*
* 18 double cost; // total cost of this customer's tickets *
* 19 *
* 20 // calculation fields *
Files and Loops 183

* 21 int totNumAdult = 0; // total adult tickets sold *

* 22 int totNumChild = 0; // total child tickets sold *
* 23 int totNumSenior = 0; // total senior tickets sold *
* 24 double totalCost = 0; // total *
cost of all tickets sold
* 25 *
* 26 // final totals and results *
* 27 int grandTotalTicketsSold;// total number of all tickets sold*
* 28 double percentAdult; // percent adult of total tickets *
* 29 double percentChild; // percent child of total tickets *
* 30 double percentSenior; // percent senior of total tickets *
* 31 double avgCost; // average cost of one ticket *
* 32 *
* 33 // attempt to open the input file *
* 34 ifstream infile ("trans-log.txt"); *
* 35 if (!infile) { // failed, so display an error message and quit *
* 36 cout << "Error: cannot open file trans-log.txt\n"; *
* 37 return 1; *
* 38 } *
* 39 *
* 40 // attempt to open the output file *
* 41 ofstream outfile ("results.txt"); *
* 42 if (!outfile) { // failed, so display error, close and quit *
* 43 cout << "Error: cannot open the output file results.txt\n"; *
* 44 infile.close (); *
* 45 return 2; *
* 46 } *
* 47 *
* 48 // setup floating point format for output of dollars *
* 49 outfile << fixed << setprecision (2); *
* 51 *
* 52 // display heading line and two column heading lines *
* 53 outfile << " Acme Ticket Sales Summary Report\n\n" *
* 54 << " Number of Tickets Sold Total Cost\n" *
* 55 << " Adult Child Senior Of Tickets\n\n"; *
* 56 *
* 57 // try to get the first set of data *
* 58 infile >> numAdult >> numChild >> numSenior >> cost; *
* 59 if (!infile) { // fails, no data or bad data in the file *
* 60 cout << "Error: file is empty or has bad data in first line\n";*
* 61 infile.close (); *
* 62 outfile.close (); *
* 63 return 3; *
* 64 } *
* 65 *
* 66 // process all the input sets of data *
* 67 while (infile) { *
* 68 // accumulate totals *
* 69 totNumAdult += numAdult; *
* 70 totNumChild += numChild; *
* 71 totNumSenior += numSenior; *
* 72 totalCost += cost; *
* 73 // display this set of data *
Files and Loops 184

* 74 outfile << setw (12) << numAdult << setw (8) << numChild *
* 75 << setw (9) << numSenior << " $" << setw (7) *
* 76 << cost << endl; *
* 77 // input next set of data *
* 78 infile >> numAdult >> numChild >> numSenior >> cost; *
* 79 } *
* 80 if (!infile.eof()) { // oops, bad data encountered *
* 81 cout << "Error: bad data in the input file\n"; *
* 82 infile.close (); *
* 83 outfile.close (); *
* 84 return 4; *
* 85 } *
* 86 *
* 87 // display first totals line *
* 88 outfile << " --- --- --- --------\n"; *
* 89 outfile << "Totals:" << setw (5) << totNumAdult << setw (8) *
* 90 << totNumChild << setw (9) << totNumSenior *
* 91 << " $" << setw (7) << totalCost << endl; *
* 92 *
* 93 // calculate and display the percentages line *
* 94 grandTotalTicketsSold = totNumAdult + totNumChild +totNumSenior;*
* 95 percentAdult = totNumAdult * 100. / grandTotalTicketsSold; *
* 96 percentChild = totNumChild * 100. / grandTotalTicketsSold; *
* 97 percentSenior = totNumSenior * 100. / grandTotalTicketsSold; *
* 98 outfile << setprecision (0); *
* 99 outfile << "Percents:" << setw (3) << percentAdult << "%" *
* 100 << setw (7) << percentChild << "%" << setw (8) *
* 101 << percentSenior << "%" << endl << endl; *
* 102 outfile << setprecision (2); *
* 103 *
* 104 // calculate and display the average cost of a ticket *
* 105 avgCost = totalCost / grandTotalTicketsSold; *
* 106 outfile << "Average cost of ticket: $" << setw (7) *
* 107 << avgCost << endl; *
* 108 *
* 109 // close files *
* 110 infile.close (); *
* 111 outfile.close (); *
* 112 return 0; *
* 113 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* results.txt from Cs05a - Acme Ticket Sales Summary Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Acme Ticket Sales Summary Report *
* 2 *
* 3 Number of Tickets Sold Total Cost *
* 4 Adult Child Senior Of Tickets *
* 5 *
* 6 2 2 2 $ 42.00 *
* 7 2 0 0 $ 20.00 *
* 8 1 8 0 $ 30.00 *
Files and Loops 185

* 9 0 2 2 $ 15.00 *
* 10 2 0 0 $ 20.00 *
* 11 1 0 0 $ 10.00 *
* 12 1 2 0 $ 15.00 *
* 13 6 8 0 $ 80.00 *
* 14 2 0 0 $ 20.00 *
* 15 --- --- --- -------- *
* 16 Totals: 17 22 4 $ 252.00 *
* 17 Percents: 40% 51% 9% *
* 18 *
* 19 Average cost of ticket: $ 5.86 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Since I stored the percentage results in a double and since the specifications called for no
decimal points on the display of the percentages, I did not set the ios::showpoint flag this time.
Thus, on line 98 when the precision is set to 0 digits, no decimal point results and the numbers
are rounded to the nearest whole number. On line 102, the precision is set back to two digits for
the next dollar amount.

Cs05b — Calculating N! (N factorial)

N! is commonly needed in equations, particularly in statistical type applications and probability
calculations. If the user needs 5!, then we must calculate 5*4*3*2*1. In this problem, the user
wishes to enter an integer and we are to display the factorial of that integer. For example, the user
enters 5 and we must calculate 5! Here is the way the screen display is to appear.
Acme Factorial Program
Enter a number or -1 to quit: 5
5! = 120

Enter a number or -1 to quit: 4

4! = 24

Analyzing the problem a bit, two loops are going to be needed. The outer loop prompts
and inputs the user’s number. The inner loop does the actual factorial calculation. When
designing a solution that involves nested looping as this one does, it is sometimes useful to
design the outer loop first and make sure it works and then come back and design the inner loop.

The outer loop is responsible for the user input. Let’s call the input value number and the
result, factor. What kind of data ought factor be? If we make it only an int, then on some
platforms the largest value is 32,767 which is not a very large n! value. So let’s make it a long. If
we knew that very large factorials needed to be calculated, then we could use a double and limit
them to only 15 digits of accuracy. So make two main storage boxes for these two variables.
(Since there are so few variables, I have omitted the figure of main storage this time.) Now let’s
design the program through the outer loop, leaving the actual details of how to calculate the
Files and Loops 186

factorial to last.

No files are required. One time only the title of the program is displayed. Then, a prompt
and input loop is needed.
output the title
prompt “Enter a number or –1 to quit: ”
input number
while (number is not equal to –1) do the following
calculate factor
output number and factor and double space
prompt “Enter a number or –1 to quit: ”
input number
end the while loop

Okay. The main loop is simple enough. Now how do we calculate the factorial? Care
must be taken here. 0! is defined to be 1. 1! = 1. We need to be able to handle all circumstances.
What would the basic working line of this inner loop be? We can try something like
factor = factor * term
where term is the next number by which to multiply. Add a main storage box for term. Suppose
we initialize factor to 1. Then, a loop can be used to calculate all terms above one until we have
done number of them.
let factor = 1
let term = 2
while term is less than or equal to number do the following
factor = factor * term
term = term + 1
end the while loop
Will this work for all numbers whose results do not exceed what can fit in a long integer?
Suppose the user enters a 0. Then, factor is set to 1 and term is 2, but the while test is false
because term is not less than or equal to number. The answer in factor is 1 which is correct.

Now test it further. What is the result if the user enters a 1 or 2 or 3 or 4? Does it work
correctly? It does. When we convert this inner loop into C++, a for loop can be used. We might
have this short loop.
for (term=2; term <= number; term++)
factor = factor * term;
This could be shortened to just
for (factor=1, term=2; term <= number; factor = factor * term++);
Files and Loops 187

Here is the completed program. Notice the coding of the for loop.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs05b - Calculating N! *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs05b Calculation of N! */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 using namespace std; *
* 11 int main () { *
* 12 *
* 13 long number; // the number to use to calculate its factorial *
* 14 long factor; // the factorial of the number *
* 15 *
* 16 long term; // next term to use in the calculation *
* 17 *
* 18 // prompt and get the first number *
* 19 cout << "Enter a number or -1 to quit: "; *
* 20 cin >> number; *
* 21 *
* 22 while (number >= 0 && cin) { *
* 23 // calculate number factorial *
* 24 factor = 1; *
* 25 for (term=2; term <= number; term++) { *
* 26 factor *= term; *
* 27 } *
* 28 // the following is the short-cut version *
* 29 //for (term=2, factor=1; term <= number; factor *= term++); *
* 30 *
* 31 // output number and its factorial *
* 32 cout << number << "! = " << factor <<endl << endl; *
* 33 *
* 34 // get another number to do *
* 35 cout << "Enter a number or -1 to quit: "; *
* 36 cin >> number; *
* 37 } *
* 38 *
* 39 return 0; *
* 40 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Sample Run of Cs05b - Calculating N! *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Enter a number or -1 to quit: 0 *
* 2 0! = 1 *
* 3 *
* 4 Enter a number or -1 to quit: 1 *
Files and Loops 188

* 5 1! = 1 *
* 6 *
* 7 Enter a number or -1 to quit: 2 *
* 8 2! = 2 *
* 9 *
* 10 Enter a number or -1 to quit: 3 *
* 11 3! = 6 *
* 12 *
* 13 Enter a number or -1 to quit: 4 *
* 14 4! = 24 *
* 15 *
* 16 Enter a number or -1 to quit: 5 *
* 17 5! = 120 *
* 18 *
* 19 Enter a number or -1 to quit: -1 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Section C: Engineering Examples

Engr05a — Summation of Infinite Polynomials
One major use of loops is to evaluate the summation of infinite series. Examine first the Cs05b
N Factorial program just above. Sometimes summations are to be done over a finite range. For
example, one might be asked what is the sum of the square roots of all the numbers from one to
fifty.

However, often the summation is an infinite one, or rather it is an infinite series of terms
or polynomials. From mathematical text books, the series expansion for the exponential function
is

Suppose that we needed to write a program to calculate ex by adding up the sum of the
terms. How would it be done? We need to formulate this into something that can be done inside a
loop. What we need is to be able to say sum = sum + term within the loop. Thus, we need to
find how to calculate the next term in the series. But wait; if we have just calculated say the x3/3!
term and are going on to the x4/4! term, we are redoing nearly all the calculations! While this
would work, it is horribly inefficient and wasteful of computer time. Instead, is there a way that
we can calculate the next term based on the previous term?
Files and Loops 189

Yes, there is. Examine the ratio of the n+1 term to the n term. It is

And since (n+1)! = (n+1)n!, the ratio becomes just x/(n+1). In other words, the next term is equal
to the previous term times x/(n+1). Okay. Let’s see how this would work in a loop to calculate ex.
Assuming that x is a double, we can sketch
input x
let sum = 0
let term = 1
let n = 0
while (not sure what ending condition is yet) do the following
sum = sum + term
ratio = x / (n+1)
term = term * ratio
n=n+1
end the while
Since this looks good so far, make up main storage boxes for x, sum, term and n. All we have to
do is determine how to end the loop. Main Storage is shown in Figure 5.4.

Figure 5.4 Main Storage for Summation Program

Now mathematically speaking, this process must be carried out to infinity to produce the
precise identity of ex. However, nothing on the computer can go to infinity — that would be an
infinite loop. Here is where numerical analysis on a computer diverges from pure mathematics. If
n is sufficiently large, the divisor (n!) becomes so large that from that term onwards no
appreciable amount is added into the sum. This is called a converging series. The opposite is
called a diverging series, such as the sum from one to infinity of x; in this case, x just keeps on
getting bigger and bigger.

Since the series is converging, there will come a point at which the next term is so small
that it can be neglected and we are done. The question is “what is the desired degree of accuracy
that we need for ex?” The answer is that it depends on the problem we are solving. If the desired
Files and Loops 190

degree of accuracy, often called the error precision or eps, is not specified, .000001 is
commonly assumed.

Realize that numerical methods are nearly always going to give an approximate answer or
rather it gives an answer sufficiently accurate for our needs. The ending condition in this case is
given by the following.
while (term > .000001)

However, one should always bulletproof coding to guard against unexpected events, such
as a slight mis-coding of the series in this example. If we make an error in calculating the ratio or
the next term, then it is possible that by accident we now have a diverging series. That means,
our ending test condition would never be met and our program would execute forever until we
manually abort it.

Rule. In numerical analysis, always provide a backdoor way for a loop to end if it does
not find an answer.

In this problem, n begins at 0 and works its way steadily up by 1 through each iteration of
the loop. The significance of n is that we are evaluating n! and for large values of n, the term
must become smaller and smaller as n increases. A backdoor shut down might be to also stop the
loop if n becomes sufficiently large. If n was say 100, then 100! is quite large.

100! =
933262154439441526816992388562667004907159682643816214685929638952175999932299
156089414639761565182862536979208272237582511852109168640000000000000000000000
00

If we divide by that number (100!), that term has got to be infinitesimal in this case. So
the loop now should have two ways to terminate
while (n < 100 && term > .000001)

Here are the completed program and a sample run.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr05a - Finding e to x by Summation *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr05a Calculate e to the x power using summation technique*/*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 *
Files and Loops 191

* 12 int main () { *
* 13 *
* 14 double x; // the number to use *
* 15 double sum; // holds the result of e to x *
* 16 double term; // next term in the series *
* 17 double ratio; // multiplicative factor to get next term *
* 18 int n; // the current term to do *
* 19 *
* 20 // setup floating point output for 6 digits of accuracy *
* 21 cout << fixed << setprecision (6); *
* 23 *
* 24 // prompt and input the user's value *
* 25 cout << "Enter the number to use or Ctrl-Z to quit: "; *
* 26 cin >> x; *
* 27 *
* 28 // loop through all the user's values *
* 29 while (cin) { *
* 30 // reset to initial starting point *
* 31 sum = 0; *
* 32 term = 1; *
* 33 n = 0; *
* 34 *
* 35 // permit 100 tries to get it accurate to .000001 *
* 36 while (n < 100 && term > .000001) { *
* 37 sum += term; // add in this term *
* 38 ratio = x / (n + 1); // find next term to use *
* 39 term = term * ratio; *
* 40 n++; *
* 41 } *
* 42 *
* 43 // display results *
* 44 if (n >= 100) { // check for diverging result *
* 45 cout << "Error: after " << n *
* 46 << " tries, the result is not sufficiently accurate\n" *
* 47 << "The series might be diverging. The result so far is\n"*
* 48 << sum << endl << "The built-in function yields\n" *
* 49 << exp (x) << endl << endl; *
* 50 } *
* 51 else { // converged result *
* 52 cout << "e to x = " << sum << " and was found after " *
* 53 << n << " iterations\nThe built-in function yields " *
* 54 << exp (x) << endl << endl; *
* 55 } *
* 56 *
* 57 // get the next user's value to calculate *
* 58 cout << "Enter the number to use or Ctrl-Z to quit: "; *
* 59 cin >> x; *
* 60 } *
* 61 *
* 62 return 0; *
* 63 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Files and Loops 192

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Sample Run of Engr05a - Finding e to x by Summation *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Enter the number to use or Ctrl-Z to quit: 2 *
* 2 e to x = 7.389056 and was found after 14 iterations *
* 3 The built-in function yields 7.389056 *
* 4 *
* 5 Enter the number to use or Ctrl-Z to quit: 20 *
* 6 e to x = 485165195.409790 and was found after 65 iterations *
* 7 The built-in function yields 485165195.409790 *
* 8 *
* 9 Enter the number to use or Ctrl-Z to quit: 42 *
* 10 Error: after 100 tries, the result is not sufficiently accurate *
* 11 The series might be diverging. The result so far is *
* 12 1739274941520462800.000000 *
* 13 The built-in function yields *
* 14 1739274941520501000.000000 *
* 15 *
* 16 Enter the number to use or Ctrl-Z to quit: ^Z *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Notice that I also displayed the value given by the built-in function exp(). In the sample
run, I purposely entered an x value of 42 to generate a giant result. In that case, we have 13 digits
correct after 100 iterations, clearly more are needed when the value of x is large.

Engr05b — Artillery Shell Trajectory

An artillery shell is fired from a howitzer at a velocity of V at some angle. If we ignore air
friction and the curvature of the earth, the path of the projectile is a parabola. At any point in its
flight, the shell’s coordinates with respect to the firing point are

where

Here, g is the gravitational acceleration or –32 feet/sec/sec in this coordinate system. The
problem is to plot the trajectory until the shell hits. That is, display successive values of x, y and t
until the projectile hits. Since the number of lines can be lengthy, write the results to a file called
results.txt.
Files and Loops 193

In this problem, could easily be solved mathematically. However, let’s do it iteratively to

illustrate some additional looping techniques. The variables are x, y, V, angle, Vx, Vy and g
(gravitational acceleration). To convert the angle into radians, we need PI and a variable rangle
to hold it along with t for time and outfile for the file. Main Storage is shown in Figure 5.5.

Figure 5.5 Main Storage for Shell Trajectory Program

Calculating the current coordinates as a function of time is the iterative approach that I
use in this problem. Time begins at 0; with each iteration, time is incremented by one second. A
new position is calculated using this new time and written to the output file. The beginning
design is
open outfile using the filename “results.txt” and display an error msg if it fails
prompt and input V and angle
while there is a set of values to use
display to outfile V and angle
let t = 1
rangle = angle * PI / 180
calculate the Vx as V cos (rangle) and Vy as V sin (rangle) components
while (some ending condition as yet unknown) do the following
calculate x and y
display to outfile x, y, and t
increment t by one second
end the inner while
prompt and input the V and angle
end the outer while
close outfile

This is a simple design. But how do we know when to end the loop? It must be when the
projectile has landed. When a shell has landed, the y value becomes zero or negative. This
suggests that we could use while (y>0). However, at the starting point in the above solution, y
has not yet been calculated. We could set y to some positive value initially just to force the test
condition to permit the loop to be entered the first time.

Since some problems that we may want to solve have very complex ending criteria which
are hard to express in a single test condition, I use a different approach here. Suppose that we
also define a variable called done and set it to 0. Let the test condition be while (!done). It does
read well. Inside the loop after y is found, we can then check the ending criteria. If y is zero or
less, set done to 1. On the next iteration, the loop ends. This is a useful technique if the ending
criteria are complex. Note that done could also be a bool variable.
Files and Loops 194

Here are the completed program and a sample test run.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr05b - Plotting the Trajectory of a Projectile *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1/****************************************************************/*
* 2 /* */*
* 3 /* Engr05b Plotting the trajectory of an artillery shell */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 #include <fstream> *
* 11 using namespace std; *
* 12 int main () { *
* 13 *
* 14 double V; // velocity of the shell as it leaves the howitzer*
* 15 double angle; // the initial angle of the firing *
* 16 *
* 17 double x; // position of the shell as a function of time *
* 18 double y; *
* 19 long t; // time in seconds since the firing *
* 20 *
* 21 double Vx; // velocity along the x axis *
* 22 double Vy; // velocity along the y axis *
* 23 double rangle; // angle in radians *
* 24 *
* 25 const double PI = acos (-1.); *
* 26 const double g = -32.2; *
* 27 *
* 28 ofstream outfile ("results.txt"); *
* 29 if (!outfile) { *
* 30 cout << "Error: cannot open output file\n"; *
* 31 return 1; *
* 32 } *
* 33 // setup floating point output for 2 digits of accuracy *
* 34 outfile << fixed << setprecision (2); *
* 37 *
* 38 // prompt and input the initial velocity and angle *
* 39 cout << "Enter the initial velocity (feet/sec)\n" *
* 40 << "and the angle in degrees (0-90)\n" *
* 41 << "or Ctrl-Z to quit\n"; *
* 42 cin >> V >> angle; *
* 43 *
* 44 // loop through all the user's test firings *
* 45 while (cin) { *
* 46 // display initial settings *
* 47 outfile << endl << "Trajectory of a shell fired at " *
* 48 << setw (5) << angle << " degrees\n" *
* 49 << "With initial velocity of " *
* 50 << setw (10) << V << " feet/sec\n\n" *
Files and Loops 195

* 51 << " Time X Y\n\n"; *

* 52 *
* 53 // calculate initial Vx and Vy *
* 54 rangle = angle * PI / 180; *
* 55 Vx = V * cos (rangle); *
* 56 Vy = V * sin (rangle); *
* 57 *
* 58 t = 1; // initialize time *
* 59 int done = 0; // done will be non-zero when shell lands *
* 60 while (!done) { // repeat until shell lands *
* 61 // calculate new position *
* 62 x = Vx * t; *
* 63 y = Vy * t + .5 * g * t * t; *
* 64 // display new position on report *
* 65 outfile <<setw(8)<< t << setw(12) << x << setw(12) << y<<endl;*
* 66 // check for ending criteria *
* 67 if (y <= 0) *
* 68 done = 1; // will terminate loop *
* 69 else *
* 70 t++; // add one second for next position *
* 71 } *
* 72 // prompt for next attempt *
* 73 cout << "Enter the initial velocity (feet/sec)\n" *
* 74 << "and the angle in degrees (0-90)\n" *
* 75 << "or Ctrl-Z to quit\n"; *
* 76 cin >> V >> angle; *
* 77 } *
* 78 outfile.close (); *
* 79 return 0; *
* 80 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Sample output from Engr05b - Plotting the Trajectory of a Projectile*
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 *
* 2 Trajectory of a shell fired at 10.00 degrees *
* 3 With initial velocity of 1000.00 feet/sec *
* 4 *
* 5 Time X Y *
* 6 *
* 7 1 984.81 157.55 *
* 8 2 1969.62 282.90 *
* 9 3 2954.42 376.04 *
* 10 4 3939.23 436.99 *
* 11 5 4924.04 465.74 *
* 12 6 5908.85 462.29 *
* 13 7 6893.65 426.64 *
* 14 8 7878.46 358.79 *
* 15 9 8863.27 258.73 *
* 16 10 9848.08 126.48 *
* 17 11 10832.89 -37.97 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Files and Loops 196

New Syntax Summary

Files
Input:
ifstream infile (“myfile.txt”);
or a deferred open until later on in the program
ifstream infile;
infile.open (“myfile.txt”);
When done,
infile.close ();

Output:
ofstream outfile (“results.txt”);

State Function to Call Returns State Flags Set

good infile.good () true if all okay None
eof infile.eof () true if eof bit is on Eof and Fail End of the file found
bad infile.bad () true if bad bit is on Bad and Fail Corrupt or bad disk
fail infile.fail () true if fail bit is on Fail Bad data inputted

Testing:
if (infile.good()) {
if (cin.fail ()) {
Shortcuts:
if (infile) or while (infile)
is short for if (infile.good() == true)

if (!infile)
is short for if (infile.good() == false)

Do While Loops:
while (test condition is true) {
0, 1, or more statements to do
}
or
while (test condition)
1 statement
or
while (test condition) ;
Files and Loops 197

Loops Done a Known Number of Times

I = 1;
while (I < 26) {
... // do something
I++;
}

Loops Ended By Inputting a Sentinel Value

cin >> quantity;
while (quantity != -1 && cin) {
... // do something with this quantity
cin >> quantity;
}

Valid Menu Choice

int choice = 5; // valid ones are from 1 through 4
cin >> choice;
while (( choice < 1 || choice > 4) && cin) {
cout << “user prompt”;
cin >> choice;
}

Inputting a File of Data

infile >> a >> b >> c;
while (infile) {
// process this set
// output this set
infile >> a >> b >> c;
}
or more compactly
while (infile >> a >> b >> c) {
// process this set
// output this set
}

Checking For Errors When a Loop Is Finished (many possible ways to check)
if (infile.eof ()) {
cerr << “All data successfully processed\n”;
}
else if (infile.bad ()) {
cerr << “Corrupt File or Bad Hard Disk\n”;
}
else {
cerr << “Bad data input\n”;
}
Files and Loops 198

Counters and Totals

1. Initialized to starting values before the loop starts
2. Incremented or added to within the loop body
3. Outputted or used once the loop finishes
Example: Find the average cost from a set of costs
double total = 0;
double cost;
int count = 0;
while (infile >> cost) {
total += cost;
count++;
}
if (!infile.eof()) {
cerr << “Bad data on line ” << count + 1 << endl;
}
else if (count) {
cout << “Average cost is $” << total / count << endl;
}
else {
cout << “No costs entered\n”;
}

Finding Maximum and Minimum Values

Set the max and min variables to the data found in the first set of data
Example: find the max and min costs
double maxCost= 0;
double minCost= 0;
double cost;
int count = 0;
infile >> cost;
if (infile)
maxCost = minCost = cost;
else {
cerr << “No costs entered\n”;
return 1;
}
while (infile) {
// use this set of costs if needed
if (maxCost < cost)
maxCost = cost;
else if (minCost > cost)
minCost = cost;
count++;
infile >> cost;
}
if (!infile.eof()) {
cerr << “Bad data on line ” << count + 1 << endl;
Files and Loops 199

}
else {
cout << maxCost << “ ” << minCost << endl;
}

Do Until Loops
do {
0, 1, or more statements to do
} while (test condition is true);

Example: enter a valid menu choice from 1 through 4

int choice;
do {
cout << “a nice prompt of what to enter”;
cin >> choice;
} while ( (choice < 1 || choice > 4) && cin);

For Loops
for (0, 1 or more initial expressions separated by commas;
test condition; 0, 1 or more bump expressions) {
// body of loop
}
Common Usage
for (j=0; j<max; j++) {
// body of loop
}

Design Exercises
1. Write a short loop that inputs integers from the user and sums them until the user enters any
negative number. When the loop ends, display the sum of the numbers and the average value of
the numbers that were entered.

2. The migration flight of a flock of birds has been observed by scientists. From data returned by
electronic tracking devices attached to the birds’ legs, they have created a file consisting of the
number of miles traveled each day by the birds. The first line in migrate.txt is the number of
days of observations. Each subsequent line consists of the miles traveled that day. For example, if
the first number in the file is 30, then there are 30 lines after that first line representing the
number of miles traveled each day. Write the pseudocode to input the file and compute the
average daily distance the birds have traveled.
Files and Loops 200

Stop! Do These Exercises Before Programming

Correct all errors in these programs. The first six illustrate different basic methods of creating the
input looping process. However, they contain one or more errors. When you correct them, do
NOT just convert them all to a single method — retain the intended input method. In other
words, on the first three problems, the while loop was intended to not only input the values but
also to control the looping process. Problems 4, 5 and 6 use a different method; they should not
be rewritten as duplicates of Problem 1.

1. This program is to input pairs of integers from a file called integers.txt.

#include <iostream>
#include <fstream>
using namespace std;
int main(){
ifstream infile;
if (infile) {
cout << "Error cannot open file integers.txt\n"
return 1;
}
int i, j;
while (infile << i << j) {
process them and output results
...
}
infile.close ();
return 0;
}

2. This program is to input pairs of integers from a file called d:\testdata\data.txt.

#include <iostream>
#include <fstream>
using namespace std;
int main(){
ifstream infile ("d:\testdata\data.txt");
if (!infile) {
cout << "Error cannot open file data.txt\n"
return 1;
}
int i, j;
while (infile >> i >> j) {
process them and output results
...
}
infile.close ();
return 0;
}
Files and Loops 201

3. This program is to input pairs of integers from a file called inputdata.txt.

#include <iostream>
#include <fstream>
using namespace std;
int main(){
ifstream infile ("inputdata.txt");
if (!infile) {
cout << "Error cannot open file inputdata.txt\n"
return 1;
}
int i, j;
while (cin >> i >> j) {
process them and output results
...
}
infile.close ();
return 0;
}

4. This program is to input pairs of integers from a file called filedata.txt.

#include <iostream>
#include <fstream>
using namespace std;
int main(){
ifstream infile ("filedata.txt"
if (!infile) {
cout << "Error cannot open filedata.txt\n"
return 1;
}
int i, j;
infile >> i >> j;
while (infile) {
infile >> i >> j
process them and output results
...
}
infile.close ();
return 0;
}
Files and Loops 202

5. This program is to input pairs of integers from a file called filedata.txt.

#include <iostream>
#include <fstream>
using namespace std;
int main(){
ifstream infile ("filedata.txt");
if (!infile) {
cout << "Error cannot open filedata.txt\n"
return 1;
}
int i, j;
while (cin) {
process them and output results
...
infile >> i >> j
}
infile.close ();
return 0;
}

6. This program is to input pairs of integers from a file called filedata.txt.

#include <iostream>
#include <fstream>
using namespace std;
int main(){
ifstream infile ("filedata.txt");
if (!infile) {
cout << "Error cannot open filedata.txt\n"
return 1;
}
int i, j;
while (infile.good());{
infile >> i >> j
process them and output results
...
}
infile.close ();
return 0;
}
Files and Loops 203

The next four questions refer to this short program. Note the user enters CTL-Z to signal the end
of file or input.
#include <iostream>
using namespace std;
int main(){
int i, j;
while (cin >> i >> j)
cout << i << " " << j << " ";
return 0;
}

7. What is the output if this is the input: 5 6 7 8 CTRL-Z

8. What is the output if this is the input: 5 6 A 7 8 CTRL-Z

9. What is the output if this is the input: 1 2 3.4 5 6 CTRL-Z

10. What is the output if this is the input: 1 2 3 A 5 6 CTRL-Z

11. A programmer wrote this program to input five numbers from the user. What is wrong with
this program? How can it be fixed?
#include <iostream>
using namespace std;
int main(){
double number; // number from user
int count; // stop after 5 numbers inputted
while (count <= 5) {
cout << "Enter a number: ";
cin >> number;
count++;
}
Files and Loops 204

12. Since the previous version did not work, he rewrote it believing this version would input five
numbers from the user. What is wrong with this program? How can it be fixed?
#include <iostream>
using namespace std;
int main(){
double number; // number from user
int count; // stop after 5 numbers inputted
for (count=1; count<6; count++) {
cout << "Enter a number: ";
cin >> number;
count++;
}

13. Since the first two versions did not work, he then tried to write this program to input five
numbers from the user. What is wrong with this program? How can it be fixed?
#include <iostream>
using namespace std;
int main(){
double number; // number from user
int count; // stop after 5 numbers inputted
do {
cout << "Enter a number: ";
cin >> number;
} while (count < 6);

14. This program is supposed to write the sum of the odd integers from one to fifty to the file
sum.txt. What is wrong with it? How can it be fixed?
#include <fstream>
using namespace std;
int main(){
ofstream outfile ("sum.txt"
if (!outfile) {
cout << "Error cannot open sum.txt\n"
return 1;
}
int j = 1, sum;
while (j < 50) {
sum += j;
j++;
}
cout << sum << endl
return 0;
}
Files and Loops 205

15. What is printed by this program?

#include <iostream>
#include <iomanip>
using namespace std;
int main(){
int a = 5;
while (a) {
cout << a << endl;
--a;
}
cout << a << endl;
return 0;
}

16. What is printed by this program where the decrement has been moved into the test condition
as a postfix decrement?
#include <iostream>
#include <iomanip>
using namespace std;
int main(){
int a = 5;
while (a--) {
cout << a << endl;
}
cout << a << endl;
return 0;
}

17. What is printed by this program where the decrement has been moved into the test condition
as a prefix decrement?
#include <iostream>
#include <iomanip>
using namespace std;
int main(){
int a = 5;
while (--a) {
cout << a << endl;
}
cout << a << endl;
return 0;
}
Files and Loops 206

Programming Problems
Problem Cs05-1 — Roots Table
The user wishes to make a table of square roots and cube roots for whole numbers from one to
some upper value that the user chooses. Write a program that prompts and inputs the user’s
choice of an upper limit. Then, from one to that number, display the current number, square root
and cube root of that current number. Display the results in columns as shown:
Number Square Root Cube Root
1 1.0000 1.0000
2 1.4142 .....
and so on through the number entered by the user. Repeat the process until the user enters a
negative number or a zero.

Problem Cs05-2 — Series Summation — Fibonacci Sequence

Write a program to sum the series given by
1, 1, 2, 3, 5, 8, 13, 21, 34, 55, ...
This is called the Fibonacci sequence. The first two numbers are one. Each one after that is the
sum of the two preceding numbers. Thus, the fourth number is 2 + 1; the fifth number is 2 + 3;
the eighth number is 8 + 13.

Prompt the user to enter an integer to which to find the Fibonacci number. Then display
the result. Continue until the user enters a zero or negative number to quit. Test your program
with this test run.
Enter a number (0 or negative to quit): 10<cr>
Fibonacci (10) = 55
Enter a number (0 or negative to quit): 3<cr>
Fibonacci (3) = 2
Enter a number (0 or negative to quit): 9<cr>
Fibonacci (9) = 34
Enter a number (0 or negative to quit): 1<cr>
Fibonacci (1) = 1
Enter a number (0 or negative to quit): 20<cr>
Fibonacci (20) = 6765
Enter a number (0 or negative to quit): 0<cr>
Files and Loops 207

Problem Cs05-3 — Programming Your CD Changer

You are holding a dance party next Friday night and you need to get 60 minutes of music lined
up to play. The CD player can be programmed with a sequence of songs to play one after another.
The objective is to write a program to generate a play list to be used to program the CD player.
The input file consists of one line per song you have chosen. Each song line contains the song
number and the song’s total play time as two integers, minutes and seconds.

As each song is input, display its information on the output report along with the total
accumulated play time after that song is finished. However, should that song force a total play
time to exceed 60 minutes, print an error message and terminate the program.
The output should appear something like this.
Proposed CD Play List
Song Song Time Total Play Time
Number Minutes Seconds Minutes Seconds
------ ------- ------- ------- -------
1 2 44 2 44
2 3 16 6 00
Test the program with the two test data files provided, SongTest1.txt and SongTest2.txt.

Problem Cs05-4 — Acme Sales Report

Write a program to input and display a sales file. Each input line contains an integer ID number
and a sales amount. When the end of the file occurs, print the high sales amount and the number
of times that amount occurred. Print also the second highest sales amount and the number of
times that one occurred. Print the results like this:
Acme Sales Report
ID Sales

9999 $9999.00
9999 $9999.00
9999 $9999.00
9999 $9999.00
High sales amount: $9999.99 occurs: 99 times.
Second highest sales amount: $9999.99 occurs: 99 times.

Test your program with the provided data files: sales1.txt, sales2.txt and sales3.txt. Be
sure to verify the accuracy of your program output. The program should make only one pass
through the file to determine the results. Do not close and reopen or otherwise read in the file a
second time in order to find the second high sales and count. One pass through the file should
result in both sets of results. Hint: pseudocode out the logic needed to determine the highest and
second highest vales. Then thoroughly desk check it before coding the solution.
Files and Loops 208

Problem Engr05-1 — The Great Motorcycle Jump

A well known daredevil stunt man is planning a spectacular canyon jump on his motorcycle.
After staring at the canyon and its walls, he has asked you for help in determining how to set up
the jump. Assume that the air drag is proportional to his velocity, where k is the proportional
constant. The following equations determine his position (x = horizontal distance, y = height) as
a function of time, t.

where g is acceleration due to gravity of 32.2 ft/sec/sec, V0 is the initial takeoff speed, a is the
angle of takeoff from the ramp and k is the air drag constant of 0.15.

The canyon is 1000 feet across and 100 feet deep. The ramp is 20 feet high at the takeoff
point. Note the trig functions require the angle to be in radians.

Write a program that prompts and inputs the initial values for the takeoff speed and angle
(in degrees) at time zero as he leaves the top of the 20-ft. high jump ramp. Print heading lines
identifying the input values for this case and then a column heading for distance, height, and
seconds. Then, print out three values, x, y, and t in formatted columns of data, as t varies from
zero by .5 second intervals. Stop the loop when either he makes it successfully to the other side
displaying “Successful Jump” or he crashes displaying “Crash.”

Then, allow the user to restart by entering another set of takeoff parameters and repeat the
process. When the user enters a takeoff speed of –1, terminate the program. Test the program
with these three test cases.
v = 330 ft/sec at a = 45 degrees
v = 200 ft/sec at a = 45 degrees
v = 260 ft/sec at a = 40 degrees

Hint: Draw a figure illustrating the problem; note carefully the origin of the coordinate
system that the equations are using. Then, determine the ending conditions; there are more than
one.

Extra Credit: determine whether the stuntman lands at the bottom of the canyon or hits
the side of the canyon. (I'm not sure which would be preferable!)
Files and Loops 209

Problem Engr05-2 — A Summation of an Infinite Series

The value of e can be found from the series

Determine the ratio of the (n+1)th term to the nth term. Then write a program that uses this
equation to calculate the value of e. Prompt and input the desired degree of accuracy desired. The
user might enter the desired accuracy or error precision as 1.0E-10. After finding the value of e
to that desired degree of accuracy, repeat the process until the user enters a 0 or a negative
number. Also quit that specific summation if there have been 3000 iterations with no solution as
yet. Display the value of e and the number of iterations it took like this.
Enter the Accuracy of e or 0 to quit: 1.0e-5<cr>
e = 2.71828 (with 9999 iterations required)

Enter the Accuracy of e or 0 to quit: 1.0e-5<cr>

e = 2.718281828 (with 9999 iterations required)

Enter the Accuracy of e or 0 to quit: 0<cr>

Problem Engr05-3 — Diode Current Flow (Electrical Engineering)

The current flow through a semiconductor diode is given by

where id is the current flow through the diode in amps, vd is the voltage across the diode in volts,
I0 is the leakage current of the diode in amps, q is the charge on an electron of 1.602x10–19, k is
Boltzmann’s constant of 1.38x10–23 J/K and T is the temperature of the diode in kelvin. Assume
that the leakage current is 2.0 microamps (10–6 amps). The temperature in kelvin is given by

Write a program that prompts and inputs the temperature in Fahrenheit of the diode. Using a
columnar form of output showing voltage and current, calculate and display the current flowing
through the diode for voltages ranging from –1.0 through +0.8 volts in .02 volt steps. Repeat the
entire process until the user enters a –1 for the temperature. Test your program with these
temperatures: 75, 100 and 125 degrees Fahrenheit. Note the results are very small values.
Writing Your Own Functions 210

Chapter 6 — Writing Your Own Functions

Section A: Basic Theory

Introduction
As programs become longer and do more sophisticated actions, main() functions tend to exceed
even a printed page. One must scroll many screens just to view the entirety of the program. The
more one has to scroll to view the function, the greater the chance for errors and the more
difficult it is for someone to read the code. It is my own observation based on over thirty years in
this business that the chance for errors is non-linear as the number of lines in a function increase.
With new programmers, my guess is that it is nearly exponential.

The industry has adopted the guideline that no function, or module as they are sometimes
called, should ever exceed fifty lines of code. Of course, programmers immediately ask: does that
include variable definitions or even comment lines? My answer is “try to keep the whole
function, counting every line, to one screen — no scrolling, in other words.” I have found that
when a programmer can see the entirety of a function on the screen, the chance for errors drops
significantly.

We have been using many of C++’s built-in functions, such as sqrt(), sin(), acos(),
pow(), and so on. C++ encourages you to develop your own functions to break the more complex
activities down into more manageable units. In fact, this idea of breaking the total work load
down into smaller pieces is a fundamental design principle of modern programming. The general
term to describe this process of breaking a complex problem down into more manageable units is
functional decomposition. One of the simplest design tool to assist functionally decomposing a
problem is Top-down Design.

This chapter begins with a discussion of the principles of Top-Down design. Once a
problem can be broken down into functional modules or just functions, then the principles of
how to write C++ functions are thoroughly presented. Since the topic is extremely broad, two
chapters are devoted to writing our own functions.
Writing Your Own Functions 211

Principles of Top-Down Design

The Cycle of Data Processing has been our workhorse for designing programs up to this point.
Nearly every program you write inputs a set of data, processes that set of data in some way,
outputs that set of data and/or results and then repeats the process until there are no more sets of
data. However, as problems become more complex, the volume of instructions involved
increases rapidly. Sometimes the process a set of data operation can involve a huge number of
instructions. Top-Down Design provides a logical method for breaking complex problems down
into more manageable units, called modules or functions.

The basic principle of Top-Down Design, or functional decomposition as it is sometimes

called, begins with a statement of the problem to be solved. Let’s take a simple non-data
processing problem to illustrate the principles. The problem to solve is to “Bake a Loaf of
Bread.” Draw a top box around this statement. Notice that each statement must contain one and
only one verb and one and only one object of that verb. The problem to solve is not “Bake a Loaf
of Bread and Vacuuming the House and Tuning Up the Car.” That is one common mistake —
each statement must contain one action to do on one thing. Confusion results with a program that
tries to do the daily sales report and update the master file and print monthly bills. These
represent three separate programs.

You can have all the adjectives, adverbs, and prepositional phrases desired. For example,
“Bake a loaf of pumpernickel bread quickly.” Use only one verb and object of that verb. You are
after one specific function, one action to do per box.

Next, ask yourself “What broad, large scale actions must I do to accomplish the task in
the top box?” In this case, I would have Gather Materials, Mix Ingredients and Bake Bread as the
major steps needed to Bake a loaf of bread. Draw a box for each of these and connect them to the
top box. Our solution so far appears as shown in Figure 6.1.

Figure 6.1 Initial Top-Down Design to Bake a Loaf of Bread

One level of the break down is complete when you ask “If I Gather Materials, Mix
Ingredients and Bake Bread, have I accomplished ‘Bake a Loaf of Bread’?” and the answer is
Writing Your Own Functions 212

yes. Then, focus on one of the subordinate boxes, Gather Materials. Now ask, “What do I need to
do in order to accomplish that?” My solution requires a breakdown into two functions: Get
Ingredients and Get Utensils. Mix Ingredients requires three functions: Make milk mixture, Mix
dry ingredients, Do a final mix. The solution now is shown in Figure 6.2.

Figure 6.2 Complete Top-Down Design for Make a Loaf of Bread

At this point we have decomposed Gather Materials and Mix Ingredients into the
subfunctions that are needed to accomplish those tasks. When do you stop decomposing,
breaking a box or function down into smaller steps? The answer is simple. The time to stop
breaking a function down further occurs when you can envision in your head the simple sequence
of steps that a box or function requires.

The Bake Bread function is not broken down because it represents the simple sequence of
actually doing the baking, such as: turn on the oven, set to 350º, when preheated, insert pans, set
timer for 45 minutes and so on. Likewise, as we look at each of the remaining lower level boxes,
each one represents a simple sequence. Well, perhaps your solution might need another function
below Get Ingredients, such as Go to the Store.

A Top-Down Design solution represents an optimum solution to the problem at hand.

Each of the boxes represents a function or module in the program. Each box should be a simple
sequence of instructions or steps to accomplish that smaller task. Certainly each box or function
should be significantly less than the industry guideline of 50 lines of code per module.

There are two common errors one can make when functionally decomposing a problem.
The first is omitting a function. Actually, this is rather common. Sometimes when the design is
being converted into the actual program coding, changes need to be made to the design because
of aspects that were not thought about initially. This is really not an error, just add onto the
design as needed. Often the design can be evolutionary.
Writing Your Own Functions 213

The second error is the single most commonly made error even by experienced
programmers. The design error is shown in Figure 6.3.

Figure 6.3 A Common Top-Down Design Error

The error is jumping from a clean statement of the problem immediately into ALL of the
details needed to solve the problem, omitting all of the higher level functions or abstractions.
You can spot this person a mile away. You give them a problem to solve and at once they are
totally enmeshed in all of the details all at once. Ah, to solve this one, you need to do this and
this and that and that and this and on and on. They see all the trees in the forest at once and the
problem sure looks huge and unconfrontable to them. What is missed is taking a more general
look first. What are the major steps needed? It is rather like peeling an onion; you design layer by
layer.

Let’s do another non-data processing example. Suppose on your way to class you have a
flat tire. So the problem to solve is Change a Flat Tire. There are numerous solutions to this
problem including ignoring it and continuing to drive on to class. However, let’s design an
optimum solution. Begin with a top box which contains a clean statement of the problem,
Change a Flat Tire. Now peel the onion. Ask what major functions need to be done? Get the
Materials (such as a spare tire and tools), Change the Tire, Put Tools Away, Clean Self Up and
Take Tire To Be Fixed might represent the first layer of the design shown in Figure 6.4.

Figure 6.4 Initial Top-Down Design to Change a Flat Tire

Writing Your Own Functions 214

Next, concentrate on just one of the subfunctions and break that one down. For example,
Get Materials involve Get Tools and Get Spare Tire. Similarly, Change Tire can be broken down
into these functions: Jack Up the Car, Remove Flat, Install New Tire and Lower Car. The other
functions would not need to be further broken down, assuming that you have a spare tire in the
car. The final design is shown in Figure 6.5.

Figure 6.5 Complete Top-Down Design to Fix a Flat Tire

Try your hand at creating a Top-Down Design for these problems.

1. Cook dinner for four, two of which are your parents.
2. Tune-up the engine of a car.
3. Photograph a model for a magazine cover.
4. Prepare an income tax return.
5. Do laundry for a family of five.

Writing your own functions

Each box in the Top-Down Design represents one function in the program. The top box is the
main() function. For the first example of writing our own functions, let’s take an overly simple
one in which the coding should be obvious. Suppose that the main() function had defined two
doubles, x and y. Further, main() wishes to store the larger of the two numbers in a double
called bigger. It should be obvious that this could be done with a simple If-Then-Else statement.
But to illustrate how to write functions, let’s say that we need a function called higher() whose
purpose is to return the larger of two numbers. The Top-Down Design for this program is shown
in Figure 6.6.
Writing Your Own Functions 215

Figure 6.6 Top-Down Design for Higher Program

Here is the main program up to the point where the new function higher() is needed.
#include <iostream>
#include <iomanip>
using namespace std;

int main () {
double x; // a number entered by user
double y; // a second number entered by user
double bigger; // a place to store the larger of x and y

cout << "Enter two numbers: ";

cin >> x >> y;

bigger = ....

Writing your own functions is a simple task if you follow the procedure step by step.

Step A. Define the Function’s Prototype

Every function in C++ must have a prototype or model or blueprint for the compiler to use when
it needs to call or invoke that function. Knowing a function’s prototype enables the compiler to
handle any necessary data conversions. For example, in this problem, both the two numbers are
doubles. Our function should expect to receive a pair of doubles as its parameters. If we passed
a pair of integers instead, the compiler can automatically convert the integers into temporary
doubles and pass the correct data.

If one did not use a prototype, then the compiler would have no choice but to pass the pair
of integer values which would become a disaster. When the function accesses what it believes to
be a pair of doubles, the memory is actually integers. Since a double occupies 8 bytes and an
integer takes up 4 bytes (on a 32-bit platform), clearly the function is going to access data beyond
the boundaries of each integer value. Wildly unpredictable results occur. Hence, C++ requires
every function to have a prototype so that such goofs can be avoided.

To create a function prototype, first invent a good name for the function and place
parentheses after it along with a semicolon. In this case, a good name for the function is higher().
Writing Your Own Functions 216

...higher (...);
Next, determine what items must be passed to the function so that it can do its job. List
the items, their data types and the order in which you want to pass them to the function. In this
example, there are two items that higher() must have in order to do its job: the doubles x and y.
Place the items in the parameter list in the order you want them to be passed, code their data type
first, then the name of the parameter and separate them with commas. So now we have
...higher (double x, double y);

The final step is to determine what the function is to return — what kind of data. In this
case, it must return the larger number which must be a double as well. Place the return data type
before the name of the function. Here is the complete prototype for our higher() function.
double higher (double x, double y);

All that remains is where to place the prototype in our program. Obviously, if the
prototype is to be used by the compiler as a model to follow when invoking the function, it must
be physically before the actual call to the higher() function. However, the vast majority of the
time, function prototypes are placed after the #includes and before the start of the main()
function. This way, you never have to worry about if you have the prototype ahead of the
function call that uses it. Here is the revised beginning of the program.
#include <iostream>
#include <iomanip>
using namespace std;

double higher (double x, double y);

int main () {
double x; // a number entered by user
double y; // a second number entered by user
double bigger; // a place to store the larger of x and y

cout << "Enter two numbers: ";

cin >> x >> y;

bigger = ....
return 0;
}

In a function prototype, the names of the parameters are optional. Names are provided for
good documentation and convenience.

In other words, the compiler could also use this as the prototype.
double higher (double, double);
If names are provided for good documentation, they are ignored by the compiler.
However, always provide good names for the parameters. Why?
Writing Your Own Functions 217

Consider the prototype for a calctax() function whose purpose is to calculate the sales tax
depending upon the state code. A reasonable prototype might be
double calctax (double cost, int quantity, int statecode);
However, the compiler is content with just
double calctax (double, int, int);
This shortened form with no parameter names can cause trouble. When you are going to call the
function, do you pass cost, quantity and statecode or do you pass cost, statecode and quantity?
From the short form prototype above, you cannot tell. Imagine the tax returned if the purchaser
bought one new car for $20,000 in state 13 (which is often Illinois) and the main() function
passed the cost, statecode and the quantity. The tax calculated would be for 13 cars in state 1
(often Alabama)!

With no parameter names coded on the prototypes, you have no choice but to find where
the actual function is coded and see what is the real order; this is no fun if the source file is a
large one.

Step B. Define the Function Header

The actual coding of the function begins with a function header. The function header follows
the exact same format as the prototype except that the ending semicolon is replaced with a
beginning { and ending } indicating here come the actual statements that the function represents.

The function header for the higher() function is

double higher (double x, double y) {
...
}
Notice that the only difference is the ending semicolon is replaced with a begin-end block set of
braces.

Thus, it is highly recommended that you simply copy the prototype and paste it where the
function is to be coded. That way, fewer errors can occur.

Where in the program do the function headers get placed? Since the function header is the
start of the actual instructions of that function, the real question is where does the code for the
functions go?

Rule: The function coding must be outside of any other block of coding.

In other words, each function in a program including main() must be outside of any other
function. While there are no limits on where you can call or invoke functions, their actual
definitions cannot be nested within other functions like an inner while loop contained within an
outer loop. Each function definition must be by itself. There are several possibilities. Here is the
order that I prefer, main() comes first and then the functions that main() directly calls and then
the functions that those functions call and so on.
Writing Your Own Functions 218

#includes
const ints go here
prototypes go here
double higher (double x, double y);

int main () {
...
}

double higher (double x, double y) {

...
}
This is a Top-down point of view. The reader sees the overall main() function first so that the big
picture of the program is the starting point.

Here is an alternative. The functions are coded first and main() last.
#includes
const ints go here
prototypes go here

double higher (double x, double y) {

...
}

int main () {
...
}
This is called the Bottom-up Style. The programming language, Pascal, must be coded in this
style in which the main() function is last. Notice one small detail. Since the entire body of the
function occurs before main() and before any other call to higher(), the function header can
serve as the prototype because the compiler has now seen the entirety of the function and knows
how to call or invoke it later on when it encounters references to higher() in main().

The following is illegal because the function definition itself is within the body of the
main() function. If compiled, it often generates an error message that local functions are not
supported.
#includes
const ints
prototypes
int main () {
...
double higher (double x, double y) { // illegal
...
}
...
}
Writing Your Own Functions 219

A third possibility is that one or more functions are contained in their own separate cpp
files. This is examined in the next chapter.

Step C. Code the Function’s Body

With the prototype coded at the top and the function header coded after the main() function, the
next step is to code the actual instructions that the function is to perform.

In the case of the higher() function, here is where the coding to determine which number
is the larger is written. There are a number of ways that this function can be implemented. Let’s
examine the simplest and then see some variations. Define another double, big, to hold the larger
number. Then, a simple If-Then-Else can place the larger of x and y into it.
double higher (double x, double y) {
double big;
if (x > y)
big = x;
else
big = y;
...
}

The last step is to return what is now in higher()’s big variable back to the calling
program. This is done with the return instruction. The return instruction syntax is
return;
return constant;
return variable;
return expression;
We have already been using the return of a constant in all of our main() functions.
return 0;
or if the file could not be opened
return 1;
Here we need to code
return big;
This instruction copies the current contents of variable big and returns that value back to the
calling or invoking function.

Here is the complete higher() function.

double higher (double x, double y) {
double big;
if (x > y)
big = x;
else
big = y;
return big;
}
Writing Your Own Functions 220

We can use the ?: shortcut to reduce the amount of coding we need. Notice in both
clauses of the if statement a value is assigned to variable big. Thus, we can also implement
higher() this way.
double higher (double x, double y) {
double big;
big = x > y ? x : y;
return big;
}

However, we can get it even shorter by returning an expression. Here is the ultimate
version of the higher() function.
double higher (double x, double y) {
return x > y ? x : y;
}
We have gotten the function body reduced to a one-liner!

Are there any benefits to using the one-line version of higher() versus the original one
that stored the larger in big which was then returned? Yes and no. Yes, in that the compiler can
create the fastest possible machine instructions for one line of code in all circumstances. The
longer version also wastes memory for the variable big, 8 bytes in this case. The fewer the lines,
the more of the program that can be seen on the screen at one time and the lower the chance for
errors.

No, in that the one line version is harder to read. Debugging a one line function body is
difficult at best. The one line is either right or wrong. The debugger can trace through your
program one line at a time; after each line has been executed, it can show you the contents of
each variable. In other words, the longer versions allow you to inspect intermediate results along
the way toward the final value the function is to return. When it is all jammed into one line of
code, the debugger cannot assist much.

C++ programs in the real world tend to have numerous functions reduced to one line of
coding. So be prepared to read them when they occur. However, for beginning programers, I
highly recommend coding the function one step at a time and not trying to produce one-liners. By
the way, the technique used to create the one line function is the same that I used here. Begin
with a straightforward implementation; get it working producing the correct answer. Then come
back and see if any of the shortcuts you have learned can be applied to your coding, just as was
done here with higher().

Step D. Invoke or Call the Function

With the prototype and function coded, now go back to the calling function, main() in this case,
and write the line of code that is to call or invoke the function, line 20 in the Basic06a figure
below. In main() variable bigger is to hold the larger value, which is the returned value from
higher().
Writing Your Own Functions 221

bigger = higher (x, y);

Here is the complete program, Basic06a.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic06a - Finding the larger of two numbers - function higher *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Basic06a Finding the larger of two numbers - function higher*/*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 using namespace std; *
* 10 double higher (double x, double y); *
* 11 *
* 12 int main () { *
* 13 double x; // a number entered by user *
* 14 double y; // a second number entered by user *
* 15 double bigger; // a place to store the larger of x and y *
* 16 *
* 17 cout << "Enter two numbers: "; *
* 18 cin >> x >> y; *
* 19 *
* 20 bigger = higher (x, y);//call higher &store return val in bigger*
* 21 *
* 22 cout << bigger << endl; *
* 23 return 0; *
* 24 } *
* 25 *
* 26 /***************************************************************/*
* 27 /* */*
* 28 /* higher: a function to return the larger of two numbers */*
* 29 /* */*
* 30 /***************************************************************/*
* 31 *
* 32 double higher (double x, double y) { *
* 33 double big; *
* 34 big = x > y ? x : y; *
* 35 return big; *
* 36 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Notice that main() could also make use of higher() this way:
cout << higher (x, y);
just as we could write
cout << sqrt (number);
In other words, when the function call is done, the coding higher (x, y) is replaced by the value it
returns. That value can then be used in any manner desired, such as assigning it to bigger or
sending it to the output stream on the screen.
Writing Your Own Functions 222

You could also use that returned value in a calculation such as

double z = 42 * higher (x, y) / sqrt ( higher (x, y));
In this weird line, the larger value of x and y is multiplied by 42 and then divided by the square
root of that larger value.

When you call a function, the function call in the invoking statement is replaced by the
value that the function returns; then the remaining actions of that original line are done.

Here is a common mistake beginners can make. Can you spot the error here in main()?
higher (x, y);
This is a function call. The compiler passes the contents of x and y to the higher() function
which calculates the larger value and returns or gives it back to the calling program. What does
the calling program then do with that answer? Nothing. It is simply pitched or ignored. It is the
same thing as doing the following
sqrt (x);
acos (-1.);
The first function call computes the square root of variable x and then no use is made of that
answer. The second computes PI and then no use is made of it. Both answers are pitched by the
compiler. You need to make some use of the value being returned from the functions, such as
cout << sqrt (x) / 42.;
const double PI = acos (-1.);
bigger = higher (x, y);

A Second Example, calcTax()

Let’s review the steps needed to create our own functions by working another problem.
Suppose that our company is currently selling items in two states but that we shortly plan to go
nationwide. Tax on the sales is dependent upon each state’s rate. Thus, in the main() function if
we calculated the tax on the sales, main() would be cluttered up with many lines of code making
it harder to read and follow. So a function, calcTax(), is to be written migrating all of the
detailed work of actually calculating the tax into a separate function, thereby streamlining the
main() function. The main() function coding goes as follows.
#includes go here
const ints go here
prototypes go here

int main () {
double cost; // cost of one item
int quantity; // number of items purchased
int statecode; // state of purchase
double subtotal; // total of items purchased
double tax; // total tax on this purchase
double total; // total of order
Writing Your Own Functions 223

...input a set of data

subtotal = cost * quantity;

tax = calcTax (...
total = subtotal + tax;
output the results
return 0;
}

At the boldfaced location above, we need to call our new function. Applying Step A,
invent a good name and surround it with parenthesis and end it with a semicolon.
... calcTax (...);
Next, determine the number of parameters that must be passed to calcTax() so that it may fulfill
its purpose. Note their data types and decide upon the order you wish to pass them. Here the
subtotal and the statecode must be passed, a double and an integer in that order. So our
prototype is now
... calcTax (double total, int statecode);
What, if anything, should calcTax() return to the caller? It must return the tax which is a double.
Here is the complete prototype.
double calcTax (double total, int statecode);
Place it above the start of main().
#includes go here
const ints go here

double calcTax (double total, int statecode);

int main () {

Step B. Now, copy it and paste it below the end of the main() function and change the
semicolon to a set of begin-end braces. Now we have the function header coded.
#includes go here
const ints go here

double calcTax (double total, int statecode);

int main () {

double calcTax (double total, int statecode) {

}

Step C. Here we must code the body of the function. I’ve allowed for just two initial state
codes. Presumably later on all fifty states will be represented in this function.
Writing Your Own Functions 224

double calcTax (double total, int statecode) {

double rate;
if (statecode == 13)
rate = .075;
else if (statecode == 1)
rate = .065;
return total * rate;
}

Step D. Back in the main() function, code the call for calcTax(). Here is the complete
shell.
#includes go here
const ints go here

double calcTax (double total, int statecode);

int main () {
double cost; // cost of one item
int quantity; // number of items purchased
int statecode; // state of purchase
double subtotal; // total of items purchased
double tax; // total tax on this purchase
double total; // total of order

...input a set of data

subtotal = cost * quantity;

tax = calcTax (subtotal, statecode);
total = subtotal + tax;
output the results
return 0;
}

double calcTax (double total, int statecode) {

double rate;
if (statecode == 13)
rate = .075;
else if (statecode == 1)
rate = .065;
return total * rate;
}

Again, we have mostly working software. Consider what the function returns when the
client program passes it a state code that is not 1 or 13. As it stands, rate is uninitialized. So
wildly unpredictable things occur. If that garbage is not decipherable as a floating point number,
a runtime error and program crash results. If it is interpretable as a number, a wildly wrong tax is
returned. How can it be repaired? Here is one way.
Writing Your Own Functions 225

double calcTax (double total, int statecode) {

double rate = 0;
if (statecode == 13)
rate = .075;
else if (statecode == 1)
rate = .065;
else
cout << "Error: bad state code: " << statecode << endl;
return total * rate;
}

How Parameters Are Passed to Functions

Just how do the values get passed to the invoked function? Look again at line 20 in the Basic06a
program above. In C++, all parameters are passed by value, never by address. Passing by value
means that a copy of the contents of the caller’s variable is made and the copy passed to the
function’s parameter variable. Some languages pass parameters by address which means the
memory location or address of the caller’s variable is what is passed to the function’s parameter
variable. What is the difference? Figure 6.7, Passing by Value Versus Passing by Address, shows
the major differences.

When a variable is passed by value, a copy of its contents is made and given to the
function. If the function should try to alter it, only the function’s copy is changed. However,
when passing by address, if the function should attempt to change it, the calling program’s
variable is actually changed.

Figure 6.7 Passing by Value Versus Passing by Address

Writing Your Own Functions 226

Let’s step through the sequence of events that occur in Basic06a’s line 20 call to
higher(). Suppose that the user has input 10 for x and 42 for y. Figure 6.8 shows the
circumstances in memory as C++ executes line 20,
bigger = higher (x, y);

A copy of main()’s x and y values are placed into higher()’s parameters x and y. Next,
higher() places the larger value, 42, into its big variable and returns big’s contents back to the
main() function. The returned value 42 is then copied into main()’s bigger variable as shown in
the next figure. Notice that the function call to higher() is replaced with the return value of 42
and then the assignment is done to bigger.

Figure 6.8 Invocation of Function higher

The Types, Scope and Storage Classes of Variables

C++ has three main types of variables: local, parameter and global. A local variable is any
variable that is defined within a function. A parameter variable is any variable in the parameter
list of a function header. A global variable is any variable defined outside of any other block of
coding; global variables are examined in the next chapter.

The type of variable determines the scope of that variable. The scope of a variable is that
portion of the program in which it is available for use by using its name. In other words, the
scope of a variable is the area within a program in which that variable can be referenced just by
coding its name.

Scope Rule 1. For local type variables, the scope is from the point of their definition to
the end of the defining block of code.

The scope of x and y in the main() function is from the point of their definitions to the
end of the defining block or the end brace } of main(). The scope of big in the higher() function
is from the point of its definition in higher() to the end brace of higher().

Scope Rule 2. For parameter type variables, the scope is from the point of their definition
within the function header to the end brace of the function.
Writing Your Own Functions 227

Thus, the scope of higher()’s parameters x and y is from their definition points in the
function header to the end of the defining block, or }, of higher().

A variable’s name can only be used when it is in scope. This means that once the end
brace } of main() is reached, all of main()’s variables go out of scope and cannot be accessed by
using their names. Thus, within higher() we could not write
bigger = big;
because main()’s bigger is now out of scope.

Another way of looking at this is that a local or parameter type of variable belongs
exclusively to the function in which it is defined. Its name is not known outside of that function
or earlier in the same function before it is defined. The following Figure 6.9 illustrates the scope
of both local and parameter types of variables.

Figure 6.9 The Scope of Local and Parameter Variables

Sometimes variables are defined within smaller blocks. Consider this example.
int main () {
...
if (x == y) {
int z = 0;
...
}
Whenever the then-clause is executed, variable z comes into scope and can be used. It is the then-
clause’s variable z. It goes out of scope when the end brace } of the then-clause is reached.
Writing Your Own Functions 228

Here is another example.

for (j=0; j<count; j++ ) {
int sum = 0;
...
}
The scope of the for loop’s sum is from its point of definition to the end brace } of the for loop.

Scope rules can also cause variables to become hidden within blocks of coding. In this
example, main() defines a local variable named x and initializes it to 0. Within a while clause,
another local variable, local to the body of the while loop, defines a variable x. Note that the
assignment of 42 goes to the while loop’s x and not main()’s x because main()’s x is hidden by
the local block’s variable x.
int main () {
int x = 0;
int j = 0;
while (j < 10) {
int x; // hides main’s x from this point on to the }
x = 42; // assignment is to while loop’s x
j++;
}

This hiding effect can create problems. Consider this block of coding.
int main () {
int i = 0;
while (i<10) {
int i = 0;
i++;
}

When this code is executed, an infinite loop is created. main()’s local variable i is
initialized to zero and the while loop test condition is testing main()’s i. However, within the
while loop, a local variable to the while block is created also called i, hiding main()’s i from this
point through the end of the while block. Thus, the wrong variable is incremented and main()’s i
is never incremented and the loop repeats endlessly.

The C++ storage classes define further properties of variables in terms of memory usage
methods. The storage class of a variable determines the lifetime of a variable and its initialization
method, if any. The lifetime of a variable represents the duration of its existence, from the point
that its memory is created to the point of its destruction.

The two storage classes that have been used in the book thus far are automatic and
parameter storage. The storage class automatic is the default; all variables that we have defined
thus far in our programs are automatic storage variables, with the exception of the parameter
variables within function headers. Variables of automatic storage have their memory allocated
for them by the compiler upon entry to the defining block of code. When the program execution
Writing Your Own Functions 229

leaves that block (reaches the end brace }), the storage for that variable is destroyed.

All parameter variables in a function header are of the parameter storage class.
Parameter class variables have their memory allocated for them by the compiler when the
function is invoked and before actual entry into to the function’s body. When the program
execution leaves that function (i.e., reaches the end brace }), the storage for parameter variables
is destroyed.

Notice that automatic and parameter storage lifetimes are nearly identical, with parameter
storage coming into existence just before the automatic storage variables defined within a
function.

The storage class affects a variable’s initialization as well. Parameter storage variables
are initialized by the compiler when the function is invoked. It first creates memory for the
parameter variables and then copies the values to be passed from the calling function into the
newly created variables. Thus, we say that parameter storage variables are automatically
initialized by the compiler with a copy of the variables to be passed as the function is invoked.

Similarly, automatic storage variables are initialized by the compiler at run time every
time the block is entered. In the following loop, the local variable sum is held in automatic
storage and is initialized 100 times, each time the body of the loop is executed.
j = 0;
while (j<100) {
int sum = 0; // created and initialized 100 times
...
j++;
} // sum is destroyed here 100 times

Registers and the Stack — a Bit of Computer Architecture

Most beginners find that the storage classes are a bit of an abstract concept — just something to
memorize and file away for future use. However, if you understand a bit of computer architecture
that underlies these storage classes, their rules become totally obvious.

A register is a high speed circuit element, usually capable of holding 32 bits or 4 bytes of
information. When you hear that a computer is a “a 32-bit” computer, they are describing the size
of these high speed registers. The original PCS were 16-bit computers, meaning that the registers
were only 2 bytes in size (16 bits). Computer instructions operating upon data held within one of
these registers is done in a few nanoseconds (10–9 sec). Unfortunately, there are only a few of
these registers available, so registers are a scarce resource not to be squandered. When the
compiler is creating the machine instructions for your programs, it tries to have your program’s
instructions make effective use of the registers so that your program runs fast.
Writing Your Own Functions 230

How a Function Returns a Value

One way the program can run more quickly is to place the values being returned by functions into
a known register; then, when back at the invoking point in the calling function, that return value
is sitting in a high speed register ready for the usage that is to be made of it. Typically on a PC,
one register known as EAX, extended accumulator register, is used to hold return values from
functions in C++ and in many other languages as well. Register EAX is 4 bytes in size. Within
this 4-byte register, the lower 2-byte portion or the rightmost two bytes is called register AX.
Original PCS only had register AX available, which is 16 bits in size. The rightmost byte of
register AX or EAX is also accessible or subdividable as register AL, accumulator lower byte
register. From these three portions of register EAX, C++ handles the high speed return of values
as follows.
size of the value return value
to be returned is in register
4 EAX
2 AX
1 AL
any other size special on the stack

A stack is a last-in-first-out save area, abbreviated LIFO. That is, the last item placed
onto the stack is the first item that can be taken off the stack. An analogy is a stack of plates on a
stainless steel serving cart at a local buffet. When a new diner enters, he or she can take only the
top plate off of the stack — that plate was the last one placed on top of the stack. Often a spring
then pushes the remaining plates upwards. When the next diner enters, he or she then takes the
next plate from the stack, which is on top. Only the top plate can be removed. When a clean plate
is placed back onto the stack, it is placed on the top of the stack; the others are “pushed down”;
this new plate now becomes the new top of the stack.

Every C++ program has a stack save area. In fact, most all programs that run on the
computer have a stack save area.

Rule 3. All automatic and parameter storage class variables are stored on the stack save
area.

To see how this works, let’s return to the higher program, Basic06a, and follow how the
computer actually handles the function call to higher() from main(). When the main() function
is called at the start of the program, the automatic storage variables of main() are allocated on the
stack. Figure 6.10 represents the stack as it appears at the time of the call to function higher().

In this example, a register cannot be used to return a double since the double is larger
than 4 bytes. Hence, immediately at the top of the stack, the compiler reserves some space for the
return value that higher() will eventually be returning back to main(). Next, it places the
parameters for higher() onto the stack and copies main()’s x and y values into higher()’s x and y
parameters. So that the computer knows where within the calling function to return back to when
Writing Your Own Functions 231

Figure 6.10 The Stack Before the Call to Higher

higher() issues its return instruction, the return address to go back to within main() is stored on
the stack next.

Figure 6.11 The Stack Upon Entry to Higher

Finally, it allocates space for the automatic storage variables of higher(), here variable
big. Figure 6.11 represents what the stack looks like when the program has entered higher() and
is about to test higher()’s x and y values to place the larger into variable big. The new top of the
stack is now as shown in the figure.
Writing Your Own Functions 232

When the higher() function executes, the return big instruction, a copy of the contents of
higher()’s big is placed into that special save area just above automatic storage for main(). Now,
one by one in reverse order, the various items are removed from the stack and the top of the stack
is lowered back to where it was before higher() was called. Thus, this is how all automatic and
parameter class variables are destroyed when the end brace } is executed. They are removed from
the stack. Well, they are actually not removed. Rather the top of the stack pointer is adjusted
downward, effectively making that space available for another function to use. Figure 6.12 shows
what the stack looks like upon the return back to the main() function.

The return value of 42 is sitting right at the top of the stack ready for the assignment into
bigger. It also shows what happens if no use is made of the return value from a function. The
memory it occupies is overlaid by the next function call.

Figure 6.12 The Stack Upon the Return to main

Thus, storage on the stack is a very dynamic affair. When functions get called, the stack
grows in size as parameters, return address and automatic variables for that function are placed
on the stack. When the end brace } of the function is executed, it is all removed from the stack.

Automatic storage and parameter class variables are always stored on the stack. Can you
see now why the lifetime of automatic and parameter class variables is just for the duration of the
defining block of code? It makes sense when you know what is going on with the stack.
Writing Your Own Functions 233

More on the bool Data Type and Functions that Return a bool
Sometimes a variable is meant to hold a simple yes/no true/false type answer. We saw in Chapter
4 that there is boolean data type, bool. A variable of the bool type can contain only the values
true and false. Here are some variables that may easily be declared as bool.
bool isWindowVisible;
bool isLeapYear;
bool isAvailable;
bool isVoter;
bool passingGrade;
bool ok;
bool error;

Sometimes a function needs to return a simple yes/no or true/false result. A function may
also be passed a bool parameter and can return a bool as well. Sometimes this makes good sense.

Consider a function that is to determine if a person is a potential voter. It might be called

isVoter(). It is passed the person’s age and citizenship status. While the age might be an integer,
the citizenship ought to be a bool as well. The prototype is then
bool isVoter (int age, bool citizenship);
The main() function and the isVoter() function can be coded this way. Of course, the test
condition in the function could be shortened into one line if desired.
int main () {
bool citizenship;
int age;
bool potentialVoter;
potentialVoter = isVoter (age, citizenship);
if (potentialVoter) { // here: a possible voter
...
}

bool isVoter (int age, bool citizenship) {

if (age < 18)
return false;
if (!citizenship)
return false;
return true;
}

Also main() does not need to actually store the bool return value from isVoter() if it is
not going to be needed beyond the single test. In such a case main() could be done this way.
int main () {
bool citizenship;
int age;
if (isVoter (age, citizenship)) {
...
Writing Your Own Functions 234

}
bool isVoter (int age, bool citizenship) {
return (age < 18 ||!citizenship) ? false : true;
}

The Shipping Cost Function

Look back at program Cs04a Customer Order on pages 120-121. We could create a function to
calculate the shipping cost. The function CalcShippingCost would need only the total before tax
amount in order for it to find the correct shipping charges. It can return the calculated shipping
cost. Its prototype is coded this way.
double CalcShippingCost (double total);

It is implemented as follows.
double CalcShippingCost (double total) {
double shippingCost = 0;
if (total < 100) { // need to figure shipping costs
shippingCost = total * .005;
if (shippingCost < 4.00) // if it is less than minimum amt
shippingCost = 4.00; // reset shipping to minimum amt
}
return shippingCost;
}

Then, back in main() we would replace lines 58 through 62 with a simple function call.
shippingCost = CalcShippingCost (totalBeforeTax);

Functions that Return No Value

Sometimes a function does its work and has no value to return at all. Suppose that a program
needed a function to print headings. Reports typically have one or more heading lines that
identify the company and the report type. Also, homework assignments typically require an initial
top line with the assignment number and student name on it. A headings function might not be
passed any variables nor does it have anything to return; it just prints headings.

When a function has no parameters, nothing is coded in the prototype or header — just
the pair of parentheses (). When a function returns nothing, the keyword void must be used for
the return data type.

The prototype for a headings function that took no parameters and returned no value is as
follows.
void headings ();
Writing Your Own Functions 235

The prototype of an outputLine() function that took several variables and displayed them might
be
void outputLine (double cost, int quantity,
long itemNumber);

A void function can return back to the caller by coding

return;
or by reaching the end brace } of the function. Either of the following is acceptable.
void headings () {
cout << "Assignment 6 by J.J. Student\n\n";
cout << " Acme Sales Report\n\n";
}
or
void headings () {
cout << "Assignment 6 by J.J. Student\n\n";
cout << " Acme Sales Report\n\n";
return;
}

Where Should Error Messages Be Displayed?

When a program detects an error situation while it is running, it usually must display an
appropriate error message for the user. Often those errors are fatal, meaning the program cannot
continue to run and must terminate.

Up to this point, the error messages have been displayed on the cout output stream.
However, C++ and DOS provide another output stream specifically designed for displaying error
messages. This is the output stream cerr.

Any message displayed to the stream cerr always goes to the screen. In contrast, the
normal output stream, cout, can be redirected to the printer or a disk file using DOS output
redirection
C:\pgm1\debug>pgm1 >> results.txt
(In Visual C++ 6.0, set the program arguments edit control found under Project — Settings —
Debug tab to >> results.txt. In Visual C++7.0, choose properties on the project in Project View
and go to the Debug section.)

Error messages must be seen by the user. Thus, it is always a wise practice to display
error messages to cerr instead of cout because cout is often redirected to an output file.
cerr << "Error: invalid data in the input file on line 42\n";

Frequently the program’s output is directly written to an output file. In such a case, it is
prudent to display the error message to both cerr and the output file streams. The user of the
Writing Your Own Functions 236

program then sees the error message on the screen. If the user opens the output result file, the
error message is there as well, indicating the set of results being viewed are not correct. For
example,
if (!infile.eof() && infile.fail ()) {
cerr << "Error: bad data in the input file; results are not correct\n";
cout << "Error: bad data in the input file; results are not correct\n";
...
}

Note that there is another error stream instance, clog, which also displays only on the
DOS error device, the screen. However, clog messages are held in an internal buffer area and
only displayed when the buffer is full, the program ends, or a flush the buffer instruction is
executed. Thus, error messages might not appear as soon as they happen. You should normally
use the cerr stream for error messages.

Controlling Leading 0's on Output — the setfill() Function

Sometimes the output format requires fields to have leading zeros instead of blanks. This is quite
common when dealing with dates. For example, month numbers are commonly displayed as 01,
02, 09, 10, and 12. The following does not work.
int month = 2;
cout << setw (2) << month;
The output would be a blank followed by the ‘2’ digit. The fill character for leading zeros is a
blank. That is, all leading zeros in a number are replaced by blanks on output. The fill character
can be changed using the setfill() function which takes one parameter, the character to be used as
the fill. The following produces the correct output of 02 for the month.
int month = 2;
cout << setw (2) << setfill ('0') << month << setfill (' ');

Once the fill character is set, it remains in force for all numerical fields until it is changed
back to a blank. Thus, when you are done using a ‘0’ as a fill character, change it back to the
default of ‘ ’, a blank as shown above. A fill character of ‘*’ is often used when printing
numerical values on checks: $*****1.42 for example.

Inputting Integers that have Leading Zeros — The dec Manipulator

Function
Sometimes the input data contains leading zeros. Dates are commonly entered with leading
zeros; for example, 01 09 2000, representing January 9, 2000. One might expect that the
sequence to input just such a date would be given as follows.
int month;
int day;
Writing Your Own Functions 237

int year;
cin >> month >> day >> year;
or
infile >> month >> day >> year;
While the month is input successfully, garbage results in the day variable. Why?

C++ supports the input of numbers in three number schemes, decimal, octal and
hexadecimal. Decimal numbers are base 10 and valid digits range from 0 through 9. Octal
numbers are base 8; that is, valid digits range from 0 through 7. Hexadecimal numbers are base
16. The input stream uses a clever method to detect the difference between the input of a
decimal-based number and an octal number. The rule that the input streams use is: if the number
begins with a 0 digit and is not just the number 0, then that number is considered to be an octal
number. In the above example, when the extraction operator encounters the 01 for January, it
assumes that it is an octal number. However, 1 in octal is equivalent to 1 in decimal and all is
well. But when it extracts the day, it encounters 09 and therefore assumes an octal number. Since
a 9 digit is an invalid octal digit, trouble arises.

To input or specify a number in hexadecimal, prefix it with 0x.

There is a function available that tells the input stream to change its assumptions about
the inputting of numbers with leading zeros. It is the dec manipulator function. Within the input
stream there is a flag that controls this assumption regarding leading zeros. In order for us to
input decimal numbers that contain leading zeros, we only need to change this flag to accept
decimal numbers. Thus, one time only usually at the start of the program, we change the flag by
coding
cin >> dec;
or
infile >> dec;
Notice that you must set the flag on the input stream you are actually using to input those decimal
numbers and that you only need to do it one time.
Writing Your Own Functions 238

Section B: Computer Science Example

Cs06-1 — Employee Payroll Program
Acme Company wants a program to compute the monthly payroll for its employees. The input
file, employee.txt, contains the employee’s social security number, the hours worked, the pay
rate, the pay type and the shift worked. The hours worked is in the format of hh:mm. The pay
type contains a one if this employee is salaried which means their pay is just the flat pay rate
independent of hours worked; a zero means this is an hourly worker. The shift worked is a
number between one and three. For all employees, there is a 5% bonus for working 2nd shift and
a 12% bonus for working 3rd shift. Overtime is paid at time and a half. For each employee,
display their social security number, hours worked, pay rate, pay type, shift and pay.

Begin by examining the output of the program, the report itself.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Cs06a Employee Payroll Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Acme Employee Payroll Report *
* 2 *
* 3 Social Hours Pay Salaried Shift Gross *
* 4 Security Worked Rate ? Worked Pay *
* 5 Number *
* 6 *
* 7 333333333 40.00 $ 10.00 No 1 $ 400.00 *
* 8 333445555 40.00 $ 10.00 No 2 $ 420.00 *
* 9 231444556 40.00 $ 10.00 No 3 $ 448.00 *
* 10 334534456 35.50 $ 10.00 No 1 $ 355.00 *
* 11 234560934 35.17 $ 450.00 Yes 1 $ 450.00 *
* 12 959649449 60.50 $ 554.00 Yes 2 $ 581.70 *
* 13 683433344 40.00 $ 350.00 Yes 3 $ 392.00 *
* 14 434433242 60.00 $ 10.00 No 1 $ 700.00 *
* 15 423425699 60.00 $ 10.00 No 2 $ 735.00 *
* 16 534432832 60.00 $ 10.00 No 3 $ 784.00 *
* 17 ---------- *
* 18 $ 5265.70 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Notice that the Salaried column is Yes or No. It makes sense to convert this input value of
zero or one into a bool data type for our convenience. Note that a bool data type cannot be input
or output directly. Now look at what one line of input contains.
333333333 40:00 10.00 0 1
The hours can be input directly and the colon ends the extraction of the hours. By extracting a
single character, we can input the colon itself. Then, the minutes can be extracted.

Next, design a solution. Since this problem has some involved calculations, let’s have a
calcPay() function. Also, let’s have a print headings() function to handle setting up the floating
point formats and printing the heading and column headings; these are all done one time at the
Writing Your Own Functions 239

start of the program. The Top-Down Design is shown in Figure 6.13.

Figure 6.13 Top-Down Design for Payroll Program

Now design the main() function that represents the top box in the design. For input, we
need these variables: ssno, hoursWorked (which will be created from variables hh and mm),
payRate, paytype to hold the 0 or 1, salaried (the bool conversion from pay type), and
shiftWorked. We also need a single character, c, to hold the colon separator between the hours
and minutes integers. The calculation result field is pay. Finally, add a box for the grand total
pay, grandTotal. Figure 6.14 shows the main storage diagram for the program.

Figure 6.14 Main Storage for Payroll Program

The sketch of the sequence of instructions for main() is as follows.

print headings
open input file, displaying an error message if it cannot be opened and abort program
while the input of ssno, hh, c, mm, payRate, paytype and shiftWorked is ok do these
convert paytype to a bool and store in salaried
convert hh:mm to hoursWorked (hh + mm / 60.)
let pay = calcPay (payRate, hoursWorked, shiftWorked, salaried);
accumulate grandTotal (add in this pay)
display this employee's pay line: ssno, hoursWorked, payRate,
salaried ? “ Yes” : “ No ”, shiftWorked and pay
end while
Writing Your Own Functions 240

check for bad input data and abort with an error message if so
display grandTotal payroll amount
infile.close ();

The headings() function is next. It is passed nothing and consists of a simple sequence.
Function: headings()
setup floating point output for two digits of accuracy
print headings and column headings

Finally, let’s write the sequences needed for the calcPay function. It needs one variable
besides those passed to it, the pay answer variable. Assume the names that we intend to call the
parameters are: payRate, hoursWorked, shiftWorked and salaried. The sequence of steps is as
follows.
Function: calcPay is given: payRate, hoursWorked, shiftWorked and salaried
if (salaried) then pay = payRate;
else do the following:
if (hoursWorked > 40.)
pay = payRate * (40 + 1.5 * (hoursWorked - 40.));
else pay = payRate * hoursWorked;
end else do
// apply the shift bonus to all employees
if (shiftWorked == 2) then pay *= 1.05;
else if (shiftWorked == 3) then pay *= 1.12;
return pay to caller

After thoroughly desk checking the program, code it into C++. When making up your
own test data to thoroughly test the program, be sure to test all If-Then-Else paths. The test data
set I provided does just that. Here is the completed program, Cs06a.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs06a Employee Payroll Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs06a Employee Payroll Program */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 using namespace std; *
* 11 *
* 12 void headings (); *
* 13 double calcPay (double payRate, double hoursWorked, *
* 14 int shiftWorked, bool salaried); *
* 15 *
Writing Your Own Functions 241

* 16 int main () { *
* 17 *
* 18 // input data *
* 19 long ssno; // social security number *
* 20 double hoursWorked; // hours worked converted from hh:mm *
* 21 int hh, mm; // work fields to input time hh:mm *
* 22 char c; // to hold the : separator *
* 23 double payRate; // hourly pay rate or flat rate *
* 24 int paytype; // 0 or 1, converted into salaried bool *
* 25 bool salaried; // true if salaried *
* 26 int shiftWorked; // the shift worked 1, 2 or 3 *
* 27 double pay; // employee's pay *
* 28 *
* 29 double grandTotal = 0; // grand total sales *
* 30 *
* 31 headings (); // print the headings of the report *
* 32 *
* 33 // open input file *
* 34 ifstream infile ("EmployeeData.txt"); *
* 35 if (!infile) { *
* 36 cerr << "Error: cannot open EmployeeData.txt file\n"; *
* 37 return 1; *
* 38 } *
* 39 *
* 40 // process all employee data *
* 41 while (infile >> ssno >> hh >> c >> mm >> payRate *
* 42 >> paytype >> shiftWorked) { *
* 43 // convert paytype to a bool *
* 44 salaried = paytype ? true : false; *
* 45 // convert hh:mm to double total hours worked *
* 46 hoursWorked = hh + mm / 60.; *
* 47 *
* 48 // calculate employee pay *
* 49 pay = calcPay (payRate, hoursWorked, shiftWorked, salaried); *
* 50 *
* 51 // accumulate grand total payroll *
* 52 grandTotal += pay; *
* 53 *
* 54 // display this employee's pay line *
* 55 cout << setw (10) << ssno << setw (7) << hoursWorked << " $" *
* 56 << setw (7) << payRate *
* 57 << (salaried ? " Yes" : " No ") *
* 58 << setw (8) << shiftWorked << " $" *
* 59 << setw (8) << pay << endl; *
* 60 } *
* 61 *
* 62 // check for bad input data - abort with an error message if so *
* 63 if (!infile.eof()) { *
* 64 cerr << "Error: bad data on input\n"; *
* 65 infile.close (); *
* 66 return 2; *
* 67 } *
Writing Your Own Functions 242

* 68 *
* 69 // display grand total payroll *
* 70 cout << *
* 71 " ----------\n";*
* 72 cout << " $" *
* 73 << setw (10) << grandTotal << endl; *
* 74 *
* 75 infile.close (); *
* 76 return 0; *
* 77 } *
* 78 *
* 79 /***************************************************************/*
* 80 /* */*
* 81 /* headings: print headings for the report */*
* 82 /* */*
* 83 /***************************************************************/*
* 84 *
* 85 void headings () { *
* 86 // setup floating point output for 2 digits of accuracy *
* 87 cout << fixed << setprecision (2); *
* 89 *
* 90 // print headings and column headings *
* 91 cout << " Acme Employee Payroll Report\n\n" *
* 92 << " Social Hours Pay Salaried Shift Gross\n"*
* 93 << " Security Worked Rate ? Worked Pay\n" *
* 94 << " Number\n\n"; *
* 95 } *
* 96 *
* 97 /***************************************************************/*
* 98 /* */*
* 99 /* calcPay: calculate the actual pay for this employee */*
* 100 /* */*
* 101 /***************************************************************/*
* 102 *
* 103 double calcPay (double payRate, double hoursWorked, *
* 104 int shiftWorked, bool salaried) { *
* 105 double pay; // will hold the person's pay *
* 106 *
* 107 if (salaried) { // salaried are paid flat rate *
* 108 pay = payRate; *
* 109 } *
* 110 else { // hourly are paid overtime *
* 111 if (hoursWorked > 40.) { *
* 112 pay = payRate * (40 + 1.5 * (hoursWorked - 40.)); *
* 113 } *
* 114 else { // hourly but no overtime *
* 115 pay = payRate * hoursWorked; *
* 116 } *
* 117 } *
* 118 *
* 119 // apply the shift bonus to all employees *
* 120 if (shiftWorked == 2) *
Writing Your Own Functions 243

* 121 pay *= 1.05; *

* 122 else if (shiftWorked == 3) *
* 123 pay *= 1.12; *
* 124 *
* 125 return pay; *
* 126 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Notice how on line 44 the bool salaried is given its more convenient value. See how it is
passed to the calcPay() function and subsequently used on line 107. Also notice how the
corresponding Yes and No are displayed on line 57.

Section C: An Engineering Example

Introduction to Numerical Analysis
Numerical analysis is the study of procedures which give an approximate answer to problems
that cannot be expressed in simple algebraic expressions. Frequently, real world problems do not
have a simple algebraic answer.

A study of mathematics leads one to believe that there is always a neat precise
mathematical answer to the problem at hand, such as summing a series from one to infinity.
While this is mathematically sound and precise, real world applications cannot sum to infinity or
the bridge would never get built. Thus, we must settle for an approximate answer that is as
accurate as we need for the job at hand.

The approximation often has three aspects.

a. Done over a limited range
b. Done to the desired degree of accuracy
c. With some estimate of the error from the precise answer
Let’s look at these points one by one.

Computer programs cannot sum to infinity, but they can sum a finite number of terms.
When looking for the roots of an equation, such as sin x, while there are an infinite number of
values of x where sin x becomes zero, a program can be written to find one of these roots within
a certain range of x values. Thus, the first step in numerical analysis often involves some kind of
range reduction.

Critical to any numerical analysis is obtaining an answer that is sufficiently accurate

enough for our needs at hand. This is frequently expressed as obtaining an answer to a desired
number of decimal places. Recall in the last chapter when evaluating a polynomial, if the next
term in the series did not appreciably add any further accuracy, we could terminate the summing
Writing Your Own Functions 244

process with a sufficiently accurate answer. Again, if one has no idea of the needed accuracy, use
.000001 as a beginning point for the analysis.

If our task was to calculate the volume of cubic yards of concrete to place into one of our
cement mixer trucks to deliver to a construction site, an accuracy of .1 cubic yard is totally
sufficient. If you have even assisted in pouring a concrete driveway, recall how it comes out of
the truck. It is nearly impossible for the truck delivery system to unload mixed cement in any
more accurate volume. Always keep in mind how accurate the results need to be.

Finally, when given the approximate numerical analysis answer, sometimes it is wise to
know some estimate of just how accurate that answer really is. Often it is possible to apply
mathematical procedures to calculate an estimate of the error. However, since this is just a
beginning examination of numerical analysis, no estimates of roundoff errors are presented in
this text. They are completely covered in a text devoted to the study of numerical analysis.

When attempting to create numerical analysis tools on the computer, there is one other
factor that may be important to the solution.

d. The rapidity of convergence to the approximate answer over the desired range.
In other words, does the method or algorithm being implemented yield the result sufficiently
rapidly? Usually, this is measured by the number of iterations of the main processing loop needed
to obtain the result. This also translates into the actual amount of time required to obtain that
answer.

For this class, rapidity of convergence is negligible. In the real world, such is not always
the case. Back when I began my computer career in astronomy, a major problem to be solved on
the computer was the calculation of the intensities of hydrogen spectral lines. When a hydrogen
atom’s electron is hit by a photon of light, the electron can be energized into a higher orbit about
the proton, in the Bohr model. Over time, that unstable orbit decays and as the electron drops
back toward its original normal orbit, it releases light energy of a precise frequency. The series of
frequencies emitted is called the hydrogen spectral lines. In astronomy back then, it was vital that
we know the theoretical intensities. Knowing the laboratory values, astronomers could then
determine many properties of stars and gaseous clouds throughout the universe. In 1964, the first
numerical methods were applied to the exceedingly complex equations needed to predict the
intensities of the first seven lines or frequencies of hydrogen. The solution produced the results
but the program ran on the computer for 48 hours nonstop! The problem was that the lines or
frequencies that were most needed were not the first seven but those around 50! Here is an
extreme case where faster convergence was needed.

Another example is NASA’s real-time flight simulator for the space shuttles. When a
shuttle is in orbit, one major concern is collision with space debris. On earth, if you are driving a
car down the road and debris is in the way, you just swerve out of the way. In orbit about the
earth, one cannot just swerve out of the way because that swerve might put the shuttle into orbit
Writing Your Own Functions 245

about the sun instead without sufficient fuel to ever return to earth! Thus, ground control has a
flight simulator that can be used to determine just how to best swerve out of the way. That
program, then, must have rapid convergence or by the time that the solution is relayed to the
shuttle, the collision has already occurred. Hence, in the real world, sometimes the rapidity of
convergence is extremely important.

Our first numerical analysis procedure will be to find the root(s) of a function. In this
chapter, the simplest method is employed. However, in the next chapter, alternative methods will
be seen to converge far more rapidly to the answer.

Numerical Analysis: Root Solving, the Bisection Method

Engineering, science and math problems frequently need a method to find the roots of an
equation. That is, find one or more values of x such that f(x) = 0. The restriction we must place
on this problem is that the user knows or suspects that there is a root of f(x) that lies somewhere
between points A and B along the x-axis. In other words, given the left and right endpoints along
the x-axis (A and B) and given the function f(x) and desired degree of accuracy, we can
determine if there is a root within that interval and find it, if it exists. Figure 6.15 shows the
general situation.

Figure 6.15 The Root of f(x)

The root lies about 1/3 the way from point A to B. The key concept that allows us to find
that root is that the function evaluated at A and the function evaluated at B changes sign. That is,
f (A) has a negative value, while f (B) has a positive value. If point A was moved to about 1/2 of
the current distance from point B, then f (at that A) would be positive. If the user asked for a root
between that new A and B, then there is no root in that interval. Likewise, if the user specified a
point B that was about 1/4 of the distance to the right of the current point A, then f (at that B)
would be negative, and again no root is in that interval. The fact that the sign of the f (x) changes
between points A and B is the basis of our ability to find the desired root.
Writing Your Own Functions 246

The Bisection Method of root solving requires only f (x), points A and B, and the desired
degree of accuracy of that root. Thus, it requires minimal knowledge of the situation. The
procedure is as follows. Let’s call the left x endpoint, x1 and the right endpoint x3; the value of f
(x1) is called f1 while the value of f (x3) is called f3. Now how can we determine if there is any
root within that interval to be found? If we multiply f1 by f3 and that product is negative, then
there is a root to be found — the sign changed. If both f1 and f3 were positive or both negative,
the product would be positive indicating no root in the interval.

Now divide the interval x1 to x3 in half and call that new point x2. It is found by taking
(x3 + x1) / 2. The function evaluated at x2 is called f2. Figure 6.16 shows the solution thus far.

Figure 6.16 Bisection — Picking x2

Next, examine the two intervals, x1–x2 and x2–x3. In which half does the root lie? From
the figure, the root lies in the left half, between x1–x2. How can a computer program tell which
half the root is in? If we check the products f1f2 and f2f3, one of these will be negative, the other
positive; the root is in the interval whose product is negative.

Now we need to formulate this into an iterative procedure or a series of steps that we can
repetitively execute. Since the root is in the left interval this time, reassign the right endpoint x3
to be x2 and f3 to be f2. And repeat the process by computing a new x2 that is halfway between
the x1 and new x3. This is shown in Figure 6.17.
Writing Your Own Functions 247

Figure 6.17 Readjusting x3 with New x2

Repeat the process. This time the root is in the x2x3 interval or the right half because
f2*f3 is negative. We move the left x1 point; x1 becomes x2 and f1 becomes f2. Again calculate
a new x2 halfway between the new x1–x3 interval. This is shown in Figure 6.18.

Figure 6.18 The Third Iteration Results

How do we know when to quit? When the new width of the interval is less than or equal
to the desired accuracy, it is time to quit and the root is that value of x2.
Writing Your Own Functions 248

Engr06a — Root Solving, the Bisection Method

The f(x) equation whose root we wish to find is

The program should prompt the user to enter the left and right endpoints (A and B), the
desired degree of accuracy and the maximum number of iterations to attempt before aborting the
attempt should something go wrong.

The Top-Down Design shown in Figure 6.19 includes two functions besides main(): the
function f(x) and bisect().

The main() function needs variables a, b, eps for the desired degree of accuracy and it
needs imax, which is the maximum number of iterations bisect() should attempt to do. Variable
root holds the answer. The design is simple for main().
prompt and input user choices for a, b, eps and imax
let root = bisect passing it a, b, eps and imax.

Figure 6.19 Top-Down Design for Bisect

The f(x) function is also simple. While one could use a one-line return expression, I store
the result in a variable ansr and return ansr. This way, if something went wrong, you could
either insert some cout instructions within the function to display values useful in finding the
bug. If it was written as a one line return expression, then only the parameters could be displayed.
Function f is given x
let ansr = exp (–x) – sin (.5 PI x)
return ansr

The bisect() function raises some new details. Initially, we have this line.
Function bisect is given a, b, eps and imax parameters
Now determine what additional variables are going to be needed. From the algorithm discussed
above, we need to define variables for x1, x2, x3, f1, f2, f3 and the distance or width of the
current interval, d. The function counts the number of iterations, I, that it takes to find the root.
This is done for two reasons. First, users always like to know how many iterations it took to find
Writing Your Own Functions 249

the answer. Second, in case something goes wrong, we can test the current number of iterations
against imax and avoid an infinite loop by shutting down. Figure 6.20 shows the main storage
diagram.

Figure 6.20 Main Storage for Root Solver

Next, using the variable names in the main storage diagram, write out the bisect()
sequence. The initial steps of bisect() are
let x1 = a;
let x3 = b;
let x2 = (b + a) * .5;
let f1 = f (x1);
let f2 = f (x2);
let f3 = f (x3);
let d = (x3 – x1) * .5 which is the width of the interval
let I = 0; which is the current number of iterations

The first action must be to verify that there is a root within this interval.
is f1*f3 >= 0 ? If so do
display using cerr “Error: no root in the interval:”, a to b
display also some reasons why such as the function may be miscoded or
the values of the end points are incorrect
return back to the caller a 0

Here we have a slight problem. We really need to be able to abort the program since no
solution is possible. A way to do that is covered in the next chapter, so for now, return zero to the
caller. The error messages should suffice to alert them.

The main loop must check both for a sufficiently accurate result and for too many
iterations.
do the following while I < imax and d > eps
Writing Your Own Functions 250

check to see if root is in left half (f1 * f2 < 0) if so

Let x3 = x2;
Let f3 = f2;
end then clause
otherwise it is in right half so do the following
Let x1 = x2;
Let f1 = f2;
end otherwise
calculate new midpoint and width of interval by
let x2 = (x3 + x1) * .5;
let f2 = f (x2);
let d = (x3 – x1) * .5;
I++; increment number of iterations
end main do while loop
is I == imax meaning it has not converged? If so then do
display Warning: after imax iterations, bisection has not converged
display the results so far which is x2
end then
otherwise x2 is the answer so do
display Root is: x2 and was found after I iterations
end otherwise
return x2 back to caller

Next, desk check the solution thoroughly and then convert it into the program. Here are
the complete program and a couple test executions.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr06a - Bisect Root Solver *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr06a - Bisection to find root of f(x)=exp(-x)-sin(.5PIx) */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 double f (double x); *
* 12 double bisect (double a, double b, double eps, int imax); *
* 13 *
* 14 const double PI = acos (-1.); *
* 15 *
* 16 int main () { *
* 17 *
* 18 double a; // left x endpoint - root is between a and b *
* 19 double b; // right x endpoint *
Writing Your Own Functions 251

* 20 double eps; // the desired degree of accuracy *

* 21 int imax; // maximum number of iterations to try *
* 22 double root; // the answer, the root of the equation *
* 23 *
* 24 // setup floating point output for 5 digits of accuracy *
* 25 // on cout and cerr since bisect is sending error msgs to cerr *
* 26 cout << fixed << setprecision (5); *
* 29 cerr << fixed << setprecision (5); *
* 32 *
* 33 cout << "Finding roots of exp (-x) - sin (.5 PI x)\n\n"; *
* 34 cout << "Enter the interval in which to search for a root\n" *
* 35 << "Enter the left endpoint: "; *
* 36 cin >> a; *
* 37 cout << "Enter the right endpoint: "; *
* 38 cin >> b; *
* 39 cout<<"Enter the desired degree of accuracy, such as .000001\n";*
* 40 cin >> eps; *
* 41 cout << "Enter the maximum number of iterations to attempt: "; *
* 42 cin >> imax; *
* 43 cout << endl; *
* 44 *
* 45 root = bisect (a, b, eps, imax); *
* 46 *
* 47 return 0; *
* 48 } *
* 49 *
* 50 /***************************************************************/*
* 51 /* */*
* 52 /* bisect: function to find the root of function f(x) */*
* 53 /* */*
* 54 /***************************************************************/*
* 55 *
* 56 double bisect (double a, double b, double eps, int imax) { *
* 57 double x1 = a; *
* 58 double x3 = b; *
* 59 double x2 = (b + a) * .5; *
* 60 double f1 = f (x1); *
* 61 double f2 = f (x2); *
* 62 double f3 = f (x3); *
* 63 *
* 64 double d = (x3 - x1) * .5; // the width of the interval *
* 65 int i = 0; // the current number of iterations *
* 66 *
* 67 // verify there is a solution *
* 68 if (f1*f3 >= 0) { *
* 69 cerr << "Error: no root in the interval: " *
* 70 << a << " to " << b << endl *
* 71 << "The function may be miscoded or\n" *
* 72 << "the values of the end points are incorrect\n"; *
* 73 return 0; *
* 74 } *
* 75 *
Writing Your Own Functions 252

* 76 // find the root, stop when the root is sufficiently accurate *

* 77 // or imax iterations have been done *
* 78 while (i < imax && d > eps) { *
* 79 *
* 80 if (f1 * f2 < 0) { // is root in left half? *
* 81 x3 = x2; // yes, so move right end point leftwards *
* 82 f3 = f2; *
* 83 } *
* 84 else { // root in right half *
* 85 x1 = x2; // so move left end point rightwards *
* 86 f1 = f2; *
* 87 } *
* 88 *
* 89 // calculate new midpoint and width of interval *
* 90 x2 = (x3 + x1) * .5; *
* 91 f2 = f (x2); *
* 92 d = (x3 - x1) * .5; *
* 93 i++; *
* 94 } *
* 95 *
* 96 // check if it converged or not - either way, display results *
* 97 if (i == imax) { *
* 98 cerr << "Warning: after " << imax *
* 99 << " iterations, bisection has not converged\n" *
* 100 << "Root thus far is: " << x2 << endl; *
* 101 } *
* 102 else { *
* 103 cout << "Root is: " << x2 *
* 104 << " found after " << i << " iterations\n"; *
* 105 } *
* 106 *
* 107 return x2; // return root of function *
* 108 } *
* 109 *
* 110 /***************************************************************/*
* 111 /* */*
* 112 /* f(x): exp(-1) - sin (.5 PI x) */*
* 113 /* */*
* 114 /***************************************************************/*
* 115 *
* 116 double f (double x) { *
* 117 double ansr; *
* 118 ansr = exp (-x) - sin (.5 * PI * x); *
* 119 return ansr; *
* 120 // return exp (-x) - sin (.5 * PI * x); *
* 121 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Two Sample Test Runs - Engr06a - Bisect Root Solver *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Finding roots of exp (-x) - sin (.5 PI x) *
Writing Your Own Functions 253

* 2 *
* 3 Enter the interval in which to search for a root *
* 4 Enter the left endpoint: .4 *
* 5 Enter the right endpoint: .5 *
* 6 Enter the desired degree of accuracy, such as .000001 *
* 7 .000001 *
* 8 Enter the maximum number of iterations to attempt: 100 *
* 9 *
* 10 Root is: 0.44357 found after 16 iterations *
* 11 *
* 12 ==== second test run ================================= *
* 13 *
* 14 Finding roots of exp (-x) - sin (.5 PI x) *
* 15 *
* 16 Enter the interval in which to search for a root *
* 17 Enter the left endpoint: .1 *
* 18 Enter the right endpoint: .2 *
* 19 Enter the desired degree of accuracy, such as .000001 *
* 20 .000001 *
* 21 Enter the maximum number of iterations to attempt: 100 *
* 22 *
* 23 Error: no root in the interval: 0.10000 to 0.20000 *
* 24 The function may be miscoded or *
* 25 the values of the the end points are incorrect *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Some of the awkwardness of the solution will be alleviated with the new C++ features
covered in the next chapter. The largest limitation to this solution is that the function f(x) must be
self contained. In other words, main() cannot input values for some other variables that are
needed in the function f(x). If it did, then with just what we know now, we would have to pass
those variables to bisect() which would in turn pass them onto f(x). And if we did that, we would
not have a generalized bisect() function that can find the root of any function, we would have to
write lots of specialized bisect() functions that took all sorts of other parameters just to pass on to
the more specialized f(x) function. This situation is handled by the new features covered in the
next chapter.
Writing Your Own Functions 254

New Syntax Summary

Writing Functions
1. Define a function prototype
2. Code the function header
3. Write the function body
4. Invoke the function from as many places as needed

Defining the Function Prototype

1. Invent a good name and place ( ); after it
Higher ( );
2. Determine the number of parameters the function must be passed for it to do its job,
then determine their data types and finally the order they are to be passed. Code them within the
( ) separated by commas
Higher (double x, double y);
3. Determine what the function is to return if anything. Place the data type of the returned
value to the left of the name.
double Higher (double x, double y);
If a function is to return nothing, use the keyword void as the return data type.
void Headings ();
4. Place the header before its first usage. Typically, function headers occur after the const
ints and #includes and before int main () {

Code the Function Header

Copy the function prototype and paste it below the end brace } of the main function.
Remove the ending semi-colon and replace it with the left brace { and add an ending brace }

Note if the function header does not match precisely the function prototype, an
unresolved external reference to the function linker error results.

Invoke the Function

bigger = Higher (x, y);

Note this does nothing:

Higher (x, y);
The function is called but no use of the returned larger value is made.

The return Statement

return; // nothing is returned; found in void functions
return 0; // returning a constant
return big; // returning a variable
return x > y ? X : y; // returning an expression
Writing Your Own Functions 255

The setfill Input Function

This manipulator function changes the character used to replace leading 0's. By default it
is a blank, ' '. However, a '0' is often used with numbers. Once set, it remains set until it is
changed by another call to setfill.
cout << setfill('0') << setw(2) << month << “-” << setw(2)
<< day << “-” << setw(4) << year << setfill (' ');
The above displays 01-02-2003 if the month, day and year contain 1, 2 and 2003
respectively.

The dec Input Manipulator Function

When inputting numbers with leading 0's, the compiler assumed those numbers are octal
(base 8). If the numbers are really decimals, as with a date or time, the input stream default of
numbers with leading 0's are octal must be changed to decimal.
cin >> dec;
infile >> dec;
This is only done once; it remains this way until changed back to octal.

Design Exercises
1. Acme Company is planning to create a Shapes Helper Package that consists of a number of
helpful graphical drawing functions. Write the pseudocode for each of these functions. The
rectangle functions are passed a length and width. The circle functions are passed the radius.
getRectangleArea() returns the area of the rectangle.
getRectanglePerimeter() returns the perimeter of the rectangle.
getCircleArea() returns the area of the circle (PI * radius * radius)
getCircleCircum() returns the circumference (2 * PI * radius)

2. A billing application needs to frequently determine the number of days between a pair of dates.
A single date is stored as three integers for month, day and year. The billing program needs to
know the number of days between pairs of dates for every set of data in the master file. Thus, it
would be desirable for the main program to do something like
if (numDays (.....) > 30) { // print a bill

In order to calculate the number of days between two dates, it is usually more convenient
to convert the date into a Julian day number, which is the number of days since noon on January
1, 4713 B.C. The following sequence converts a date (month, day, year) into the corresponding
Julian day.
if the month is 1 or 2, add 12 to month and subtract 1 from the year
Julian day = 1720994.5 + day + (int) ((month + 1) * 30.6001) +
Writing Your Own Functions 256

(int) (365.25 * year)

if the original date is greater than October 15, 1582, then add one more term to the
above Julian day equation: (int) (2 – year / 100 + year / 400).

Since the numDays() function needs to convert two dates into the Julian day equivalent,
it makes sense to also have a toJulian() function. Write out the pseudocode for each of the two
functions: numDays() and toJulian(). Be sure to indicate what variables are being passed to your
functions and what is being returned.

Stop! Do These Exercises Before Programming

Correct the syntax and logical goofs in the following problems one through seven.

1. The compiler issues an error on this one. Why? How can it be fixed?
#include <iostream>
#include <iomanip>
using namespace std;

int main () {
double x = 99;
double y;
y = fun (x);
cout << y;
return 0;
}
double fun (double x) {
return x * x * x;
}

2. The linker issues an error on this one. Why? How can it be fixed?
#include <iostream>
#include <iomanip>
using namespace std;

fun1 (double x);

int main () {
double x = 99;
double y;
y = fun1(x);
cout << y << endl;
return 0;
}
double fun1 (double x) {
return x * x * x;
}
Writing Your Own Functions 257

3. This one does not work right. Why? How can it be fixed?
#include <iostream>
#include <iomanip>
using namespace std;

int fun2 (double x);

int main () {
double x = 99;
double y;
fun2 (x);
cout << y << endl;
return 0;
}

int fun2 (double x) {

return x * x * x;
}

4. This one creates a compiler error. Why? How can it be fixed?

#include <iostream>
#include <iomanip>
using namespace std;

double fun3 (double x);

int main () {
double x = 99;
double y;
y = fun3 (x);
cout << y << endl;
return 0;
}
double fun3 (double x) {
double sum = x * x * x;
}
Writing Your Own Functions 258

5. This one does not produce the correct output. Why? How can it be fixed?
#include <iostream>
#include <iomanip>
using namespace std;

double fun4 (double x);

int main () {
double x = 99;
double y;
y = double fun4 (double x);
cout << y << endl;
return 0;
}

double fun4 (double x) {

return x * x * x;
}

6. This one produces the wrong results. Why? How can it be fixed?
#include <iostream>
#include <iomanip>
using namespace std;

double fun5 (double x, int power);

int main () {
double x = 99;
double y;
y = fun5 (x);
cout << y << endl;
return 0;
}

double fun5 (double x, int power) {

double ansr = pow (x, power);
return power;
}
Writing Your Own Functions 259

7. This one creates a compiler error about local functions not supported. Why? How can it be
fixed?
#include <iostream>
#include <iomanip>
using namespace std;

double fun6 (double x);

int main () {
double x = 99;
double y;
y = fun6 (x);
double fun6 (double x) {
return = x * x * x;
}
cout << y << endl;
return 0;
}

8. What will the cout display for the variables a, b and c?

#include <iostream>
#include <iomanip>
using namespace std;

int fun (int a, int b);

int main () {
int a = 1;
int b = 2;
int c = 3;
c = fun (a, b);
cout << c << " " << b << " " << a << endl;
return 0;
}
int fun (int a, int b) {
a = 42;
b = 42;
return 42;
}
Writing Your Own Functions 260

9. What will the cout display for the variables a, b and c?

#include <iostream>
#include <iomanip>
using namespace std;

int fun (int a, int b);

int main () {
int a = 1;
int b = 2;
int c = 3;
if (a == 1) {
int d = 42;
int c;
c = fun (d, a);
}
cout << c << " " << b << " " << a << endl;
return 0;
}

int fun (int a, int b) {

return a + b;
}

Programming Problems
Problem Cs06-1 — Leap Year Determination Program
Write a function called isLeap() that takes one integer parameter, the year, and returns a bool
value of true if the parameter year is indeed a leap year and false if it is not a leap year. No data
is either input or output from within this function.

A year IS a leap year if it satisfies EITHER of these rules:

(1) the year is evenly divisible by 4 but not by 100
(2) the year is evenly divisible by 400

These years are leap years: 1600, 1988, 1992, 2000, 2400.
These years are not leap years: 1900, 1991, 1994, 2200.

Write a main() function to input a year, call the isLeap() function and output that year
plus the text either “Leap Year” or “Not a Leap Year.” It should loop and repeat until the end of
file is signaled by the user pressing the ^Z key (CTRL-Z). Test with the above nine values.
Writing Your Own Functions 261

Problem Cs06-2 — Determining the Number of Days in a Month

Write a function, daysInMonth(), that is passed a pair of integers, the year and the month, and
returns an integer representing the number of days in that month. The function does not read any
input nor write any output. Note that February has 28 days except in a leap year when it has 29
days.
The daysInMonth() function must call isLeap() function when it is necessary to
determine if this is a leap year or not. Use the exact same isLeap() function coded in Problem
Cs06-1; no changes to it are needed or wanted.
Write a main() function to input the month and year, call daysInMonth() to get the
number of days in that month and output the number of days in that month. It should repeat the
process until the end of file is signaled.
Test with the following runs:
02 1600
02 1991
12 1990
01 1976
06 1776

Problem Cs06-3 — The Valid Date Program

Write a function, named isDate(), that takes three integer arguments representing a calendar date
as month, day, and year. The function returns a bool — true if the date is valid and returns false
if the date is invalid. The function does not read any input or write any output.

The date is valid if the year is between 1600 and 2400, inclusive, the month is between 1
and 12, inclusive, and the day is within the appropriate range for the month. The isDate()
function must call the daysInMonth() function from Problem Cs06-2 to determine the number
of days in a month. The function, isDate(), does not have any input or output statements in it.
Note, you should use the exact same daysInMonth() and isLeap() functions copied exactly
as they are in the previous two problems with NO changes.

Write a main() function to input a date (as three integers, month, day, and year), call the
isDate() function and write the output. Do not use any prompts; all input comes from the file
dates.txt.
Write the output in this format:
02-09-1995
06-31-1993 INVALID DATE
10-01-2000
13-15-1999 INVALID DATE

Always print two digits for the month and day, and separate the month, day, and year with
hyphens. To produce the leading 0's, you may use the setfill (‘0') function. If the date is invalid,
Writing Your Own Functions 262

write “INVALID DATE” on the same line after the date, but if the date is valid, leave the rest of
the line blank.

Problem Cs06-4 — Checking Account Monthly Report

Write a Top-Down Design and a C++ program with functions to create a checking account
monthly balance report. The first line of input contains the account number and the beginning
balance. The on subsequent lines are a variable number of checking account transactions. There
are two types, a deposit entry and a check entry. The first character is 1 for a check or 2 for a
deposit. Next on the line is the monetary amount of that transaction.

A simple test run might consist of the following

11234152 879.46
1 400.00
2 100.00

For each transaction, service charges must be calculated and applied to the running
account balance. For a deposit, there is a 10-cent charge. For a check, there is a 15-cent charge.
Also, should the balance ever drop below $500, then there is a one-time monthly $5.00 service
charge. If the current balance should ever fall below $50, display a warning message. If the
current balance should ever become negative (overdrawn), then apply a $10 service charge per
each check written until the balance becomes positive again.

Finally when the end of file occurs, display the ending monthly balance, the total monthly
deposits, total monthly checks, and the total service charges applied that month. Test the program
with file account.txt.

The above simple set of transactions would be displayed as follows.

Beginning balance: $879.46

Check amount: $400.00

Service charge: $ 0.15
Service charge below $500: $ 5.00
Total service charges: $ 5.15
Current Balance: $474.31

Deposit amount: $100.00

Service charge: $ 0.10
Total service charges: $ 0.10
Current Balance: $574.21

Beginning balance: $879.46

Total checks: $400.00
Total deposits: $100.00
Writing Your Own Functions 263

Total service charges: $ 5.25

Ending balance: $574.21

Problem Cs06-5 — The Factorial Function

Write a function called factorial() that is passed a long integer number and returns that number’s
factorial. That is, if factorial() is passed 5, it should return 5! or 5 * 4 * 3 * 2 * 1.

Next, write a main() function that prompts the user for a number and then displays its
factorial. Repeat the process until the user enters a –1 to quit. The output should be formatted
similar to this.
5! = 120

Problem Engr06-1 — Function Evaluation, Min/Max and Files

Part A. Given the function:

Write a program to evaluate the function beginning at x = –2.0 through +4.0 in steps of
.2. On each iteration, call an f(x) function to calculate the function at the current x value. Then,
display x and f(x) both to the screen and also to a file called result.txt. The screen output only
should contain column headings, nicely centered above the columns of data. The result.txt file
should just contain the two numerical values separated by blanks on each line. Use a precision of
five decimal places for both the screen and file output.

Part B. Given the file result.txt, write another program that reads in the file of results; each line
of that file should contain a pair of values x and f(x). For each pair of values, display them in a
columnar format as was done in Part A.

The main program should determine both the maximum and minimum values of f(x). Do
not hard code the number of lines to input; we might change the evaluation range in Part A.
When the end of file is reached, display the results similar to the following:
The function was evaluated from x = aaaaa to x = bbbbb
The function has a minimum value of nnnnn at x = mmmmm
The function has a maximum value of rrrrr at x = qqqqq
Note that you will need to save both the first and last values of x to create the first line of output.
Writing Your Own Functions 264

Problem Engr06-2 — Probability Theory

In probability theory, one problem that often arises is how many combinations of n objects taken
p at a time are there? The combinatorial function is C (n,p) and is given by

Write a function C() that is passed the two integers, n and p, and returns the number of
combinations based on the formula. Do not just calculate the equation as is. Instead, first work
out the ratio of successive C’s for a given n — that is on paper work out the ratio of
C(n,p+1)/C(n,p). Your implementation of the function C() should make use of this ratio. Note
that C(n,0) is defined to be 1.

Write a main() function that prompts the user for values of n and p and then displays the
number of combinations. Repeat the process until the user enters ^Z for end of file.

Test your program with these entries and any others that you deem appropriate to
demonstrate that the C(n,p) function works properly in all cases.
N P C(n,p)
0 0 1
1 0 1
1 1 1
4 2 6

Problem Engr06-3 — Prime Number Function

Write a function isPrime() that is passed a long integer number and returns a bool whose
value is true if that number is a prime number and whose value is false if it is not a prime
number. A number x is a prime number if it is not evenly divisible by any other number between
2 and sqrt (x). Note that 1 is a prime number as is 2 and 3 but 4 is not a prime number.

Write a main() function that prompts the user to input a number and then display that
number along with a message indicating whether or not it is a prime number. Your output should
look something like this.
1 is prime
4 not prime
5 is prime
6 not prime
Test your program with these values:
1, 4, 5, 6 100, 101, 94274, 93743, 2147000, 2147001
Writing Your Own Functions 265

Problem Engr06-4 — Root Solving

Modify the sample program Engr06a to use bisection to find the roots of these functions. In both
cases, five digits of accuracy are required. Limit the maximum number of iterations to 100.

A. f (x) = x2 + 2x – 15, where a = 2.8 and b = 3.1

B. f (x) = sin (x), where a = 170 degrees and b = 190 degrees

Problem Engr06-5 — Re-forming of Cyclohexane into Benzene

(Chemical Engineering)
Chemical engineers often encounter difficult equations to solve when calculating the mass
balances of chemical reactions. For example, the conversion of hydrocarbons to other
hydrocarbons with higher octane numbers is called re-forming. For a simple example,
Cyclohexane re-forms into benzene. The reaction equation based on lb-moles is given by

where n is the volume of benzene produced in lb-mole.

Solving for n is difficult. However, we can use our bisect root solving method to find n.
Rearrange the equation into the format of f(n) = 0, by moving the 14.7 to the other side.

Create a function f(n) that computes the above formula. Then, modify the Engr06a
sample program. Look for a root between .5 and 1.0 using an accuracy of .000001 and a
maximum number of iterations of 100. You should get the result of .64345 for the resulting value
of n.
More on Functions 266

Chapter 7 — More on Functions

Section A: Basic Theory

Introduction
When one starts using functions, more often than not, one encounters circumstances in which a
function needs to return more than one result. Yet, a function can only return a single value. A
reference variable, which is basically the memory address of the passed argument, is used to
allow a function to directly modify the contents of a calling function’s variable. Reference
variables are a powerful feature of the language.

The remaining storage classes of data are discussed along with more applications for
functions.

Reference Variables
A reference variable is the memory address of the argument being passed to a function. The
compiler is responsible for the implementation details. Instead of passing a copy of the contents
of the variable, C++ passes a copy of its memory location so that the function may change it
directly. Before we dive into the details, let’s see why they are needed.

The Need for Reference Variables

Suppose that the program needed to input a starting time. Often, people encode a time in the
format of hh:mm:ss, such as 10:42:15 meaning 10 hours, 42 minutes and 15 seconds. How can
these integers be input when there is a colon separating them instead of a blank? We must define
a character to hold that single colon character and input that character. The input stream’s get()
function can input one single character. Assume that we have defined three integers to hold this
starting time.
int shrs;
int smin;
int ssec;
Consider where the input stream current position is when the extraction of the hours is
completed:
More on Functions 267

cin >> shrs;

Remember that the extraction operator skips over white space (there is none initially), then inputs
successive characters that can be part of an integer, including a sign and the digits 0 through 9. It
terminates the extraction when it reaches the end of file, white space or a character that cannot be
in an integer. In this case, the colon cannot be part of an integer and thus the 10 hours is inputted
into shrs and the current position in the input stream is pointing to the colon.

Thus, if we defined a character c as

char c;
then we can input that colon as follows.
cin.get (c);
The get() function inputs the next character no matter what it is, blank, a letter, a colon, a <CR>.
If you input the <CR>, it is stored as the new line code, ‘\n’.

Alternatively, we can input that character using the extraction operator

cin >> c;
because the current character is not white space. Thus, the complete sequence to input the time
could be either of these.
cin >> shrs;
cin.get (c);
cin >> smin;
cin.get (c);
cin >> ssec;
or
cin >> shrs >> c >> smin >> c >> ssec;

One caution. When requesting that the user use something other than a blank to separate
values on input, please include very specific details in your prompt to the user. Otherwise, users
get very frustrated trying every imaginable possibility except the right one. Suppose for a
moment that the user accidentally entered instead
10 42 15
What would the lengthy sequence of input instructions using the get() function produce for
input? Precisely the same results because the get() inputs the blank. However, if we used the
single chained series of extraction operators, what would result? Disaster, because the extraction
of the first character c skips over white space and inputs the 4 digit into c. The 2 digit then is
input for the minutes and so on.

Now if there is a starting time, there is also likely to be a finish time. So we can add
int fhrs;
int fmin;
int fsec;
and input those
cin >> shrs >> c >> smin >> c >> ssec;
Let’s say the user entered 11:15:08 for the finish time.
More on Functions 268

And if you have a finish time and a start time, what do you suppose that the user would
like to know next? The elapsed time. Okay, so how do you do the subtraction?
11:15:08
-10:42:15
---------

This would not be a fun thing to do! While we could write a block of coding to do so,
there is a far easier way. The technique often used is to convert the times into the total number of
seconds since midnight. Since we need to do this conversion at least twice, a function is desired.
Always use a function when facing coding the same thing twice.

What would be the data type of the total number of seconds? If we have 10 hours, this
would be 10 * 60 * 60 or 36000 seconds. This exceeds an int type under old DOS. Thus, the
platform independent method would be to make these total seconds variables be longs. Hence,
define three longs as follows.
long stotsec; // start total seconds
long ftotsec; // finish total seconds
long etotsec; // elapsed total seconds

At this point, we need to write a function call to convert the times.

stotsec = function . . .
Let’s follow the procedure developed in the last chapter. First we define the prototype by
inventing a good name for the function, by creating the parameter list and then by adding the
return data type. The function’s name can be hms_to_totsec() and it is passed three integers
representing the hours, minutes and seconds; it must return the long total seconds since midnight.
long hms_to_totsec (int hrs, int min, int sec);

Next, define the function header and then code its body. As a first approximation, we can
write
long hms_to_totsec (int hrs, int min, int sec) {
return hrs * 3600L + min * 60 + sec;
}
Why is the constant written as 3600L? Remember that if the suffix type cast to a long is not used,
then the constant would be an int type. The multiplication would then be int * int which yields
overflow errors if this program were compiled under old DOS, since on that platform an int can
only hold 32,767.

Next, we can tentatively write the function invocation as

stotsec = hms_to_totsec (shrs, smin, ssec);

However, since we are going to be asking a user to enter a time, this is absolutely not the
way to do this. What would our function do if the user enters
100:-88:99
10:99:61
24:00:10
More on Functions 269

00:00:00
-10:42:22
Naturally, all of these are not valid times on a twenty-four hour basis. If the user enters such, we
need instead to reject those. Certainly we can do the following within hms_to_totsec().
if (hrs > 24 || hrs < 0 ||
min > 59 || min < 0 ||
sec > 59 || sec < 0) {
but then what do we do if it is true, the time is in error? We can return an error code. No valid
total number of seconds since midnight could ever be negative. So to indicate an error, let’s
return a negative number. Which one? It is up to you; it is an arbitrary; any negative value works.
Usually, a programmer would pick –1 for the first error type. But we should not just write
return -1;
Instead, we should return a constant integer that is defined once and used throughout the program
as needed. Let’s call it TIME_ERROR. Its definition is
const int TIME_ERROR = -1;
and our return instruction is now
if (hrs > 24 || hrs < 0 ||
min > 59 || min < 0 ||
sec > 59 || sec < 0)
return TIME_ERROR;

Is this sufficient testing to be done? Reexamine the above four incorrect user entries. It
handles all of them except two, 24:00:10 and 00:00:00. Additional testing is needed. We can add
a test for all zeros as follows.
if (hrs == 0 && min == 0 && sec == 0) // disallow 0:0:0
return TIME_ERROR;
However, the 24:00:10 is more difficult to test. Taking it bit by bit, we can write
straightforwardly
if (hrs == 24)
if (min == 0 && sec == 0)
; // all is ok
else
return TIME_ERROR;
else
; // all is ok

The final else ; can be eliminated — I used it solely for clarity. The other test for min and
sec being 0 is also verbose with a needed null statement for the then-clause. We can eliminate the
null then-clause by reversing the test condition result; this is done using the ! operator.
if (hrs == 24)
if (!(min == 0 && sec == 0))
return TIME_ERROR;
Finally, the two test conditions can be merged into one, yielding the following streamlined
version
if (hrs == 24 && !(min == 0 && sec == 0))
return TIME_ERROR;
More on Functions 270

Our completed function is

long hms_to_totsec (int hrs, int min, int sec) {
if (hrs > 24 || hrs < 0 || min > 60 || min < 0 ||
sec > 60 || sec < 0)
return TIME_ERROR;
if (hrs == 0 && min == 0 && sec == 0)
return TIME_ERROR;
if (hrs == 24 && !(min == 0 && sec == 0))
return TIME_ERROR;
return 3600L * hrs + 60 * min + sec;
}

Now back to the calling function, main(). Once we make the assignment to stotsec, we
must check for the error value.
stotsec = hms_to_totsec (shrs, smin, ssec);
if (stotsec == TIME_ERROR) {
cout << "Error: start time is invalid.\n";
return 1;
}

Once again, C++ programmers are likely to take a shortcut on this sequence by coding the
assignment and test into one instruction.
if ((stotsec =hms_to_totsec (shrs, smin,ssec))==TIME_ERROR){
cout << "Error: start time is invalid.\n";
return 1;
}
Notice the boldfaced ( ) parentheses. These are mandatory if this code is to function properly.
Consider what it would look like without the extra parenthesis.
if (stotsec = hms_to_totsec (shrs, smin, ssec)== TIME_ERROR)
Recall the precedence of operators. What is done first?

Always function calls must be evaluated first so that it has the value it represents to use in
the rest of the statement. hms_to_totsec() is going to be called first. What is always done dead
last? Assignments. Thus, the return value from the hms_to_totsec() is compared against the
TIME_ERROR and then the result of that test, either true (1) or false (0) is copied into stotsec!
The parenthesis forces the assignment to be done first and the test of the value copied into stotsec
to be done last.

Applying the same technique to the finish time, we have

if ((ftotsec =hms_to_totsec (fhrs, fmin,fsec))==TIME_ERROR){
cout << "Error: finish time is invalid.\n";
return 1;
}
More on Functions 271

And now we can easily calculate the elapsed time as

etotsec = ftotsec - stotsec;

I think you can see that we could also do any number of other actions to times in this
format, such as adding 30 minutes to one time, subtracting an hour from another time, comparing
two times to see if they were the same and so on. All can be very easily implemented.

Let’s say that the result was 3601 total seconds. Notice one detail. You cannot just
display the result as 3601 total seconds difference! The user would like the result to be 01:00:01
for example. No problem. We’ll just apply our standard function creation steps. Invent a good
name, totsec_to_hms() and pass it only the long total seconds to convert.
... totsec_to_hms (long totsec);
Then figure out what it is to return. Here it must return the hours, minutes and seconds that the
long total seconds represents. Oops. A function can return nothing or one single value, not three
values! Here we run into the brick wall.

Two solutions exist to enable us to get around the barrier that a function can only return a
single value. The most optimum method is discussed first, the use of reference variables.

The Reference Variable Solution

When a reference variable is used as a parameter, the compiler passes the memory location of the
calling function’s variable to the called function. Within the called function all accesses of that
parameter reference variable are actually accessing directly the calling function’s variable’s
contents. Thus, if the called function assigns a value to the reference parameter variable, that new
value is really being stored in the calling function’s variable. Figure 7.1 illustrates this.

Figure 7.1 Reference Variable

More on Functions 272

The calling function’s total is located at memory location 500 and contains 0 when it
calls function fun(). Function fun’s parameter tot is a reference to total and contains that
memory location of 500. Note that tot itself is located at some other memory location, 700.
When the assignment of 42 to tot occurs, the value 42 is actually placed into the calling
function’s total as shown. The compiler handles all of the details for us.

To make a reference variable, simply add the & symbol after the data type and before the
name. Both of the following are correct.
void fun (int& tot);
void fun (int &tot);

When invoking a function and passing a reference to a variable, the calling program is
giving the called function total access to its variable. The called function can alter its contents if
it desires. Normally, when designing functions, only pass references to variables that the function
must alter. In the last chapter’s higher() function, duplicate copies of the main() function’s x and
y variables were passed. It is a design error to pass references to main()’s x and y variables to
higher(); higher() has no business altering main()’s x and y contents! Only pass by reference
when the called function must logically alter the original values of the calling function.

Reference variables handle the problem of how to return three integers from our
totsec_to_hms() function. In addition to passing the long total seconds, we must pass three
references to the three integer answer variables of main(). The prototype is
void totsec_to_hms (long totsec,
int& hrs, int& min, int& sec);

And main() calls the function like this.

totsec_to_hms (etotsec, ehrs, emin, esec);

The actual conversion back into hours, minutes and seconds is very easy.
void totsec_to_hms (long totsec,
int &hrs, int &min, int &sec) {
hrs = totsec / 3600;
min = totsec % 3600 / 60;
sec = totsec % 60;
}

Let’s say that the totsec contained 3601 or one hour and one second. The result of the
division of totsec by 3600 seconds per hour is 1. The 1 is stored in hrs which is a reference to
main()’s ehrs; thus the 1 is stored in main()’s ehrs variable. The 0 minutes is then stored in min
which is a reference to main()’s emin; the 0 is stored in main()’s emin. Finally, the 1 second is
stored in sec which is a reference to main()’s esec and is stored there.
More on Functions 273

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic07a - Using Reference Variables to Find the Elapsed Time *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 #include <iostream> *
* 2 #include <iomanip> *
* 3 using namespace std; *
* 4 /**************************************************************/ *
* 5 /* */ *
* 6 /* Reference Variables Example: finding the elapsed time */ *
* 7 /* */ *
* 8 /**************************************************************/ *
* 9 *
* 10 // global error value *
* 11 const int TIME_ERROR = -1; *
* 12 *
* 13 // function prototypes *
* 14 long hms_to_totsec (int hrs, int min, int sec); *
* 15 void totsec_to_hms (long totsec, int &hrs, int &min, int &sec); *
* 16 *
* 17 *
* 18 int main () { *
* 19 *
* 20 int shr, smin, ssec, // starting hours, minutes, seconds *
* 21 fhr, fmin, fsec, // finish hours, minutes, seconds *
* 22 ehr, emin, esec; // elapsed hours, minutes, seconds *
* 23 *
* 24 long stotsec, ftotsec, etotsec; // total seconds *
* 25 char c; *
* 26 *
* 27 // input the starting time *
* 28 cin >> dec; // allow for 08:09:09 leading 0's *
* 29 cout << "Enter starting time (hh:mm:ss): "; *
* 30 cin >> shr; *
* 31 cin.get (c); *
* 32 cin >> smin; *
* 33 cin.get (c); *
* 34 cin >> ssec; *
* 35 *
* 36 // input the ending time *
* 37 *
* 38 cout << "Enter finish time (hh:mm:ss): "; *
* 39 cin >> fhr >> c >> fmin >> c >> fsec; *
* 40 *
* 41 // validate both times and convert valid times to total seconds *
* 42 *
* 43 if ((stotsec = hms_to_totsec (shr, smin, ssec)) == TIME_ERROR) {*
* 44 cout << "Error: start time is invalid.\n"; *
* 45 return 1; *
* 46 } *
* 47 *
* 48 if ((ftotsec = hms_to_totsec (fhr, fmin, fsec)) == TIME_ERROR) {*
* 49 cout << "Error: finish time is invalid.\n"; *
More on Functions 274

* 50 return 1; *
* 51 } *
* 52 *
* 53 // calculate the elapsed total seconds *
* 54 *
* 55 etotsec = ftotsec - stotsec; *
* 56 *
* 57 // convert elapsed total seconds back to hms *
* 58 *
* 59 totsec_to_hms (etotsec, ehr, emin, esec); *
* 60 *
* 61 // display the elapsed hms-each part is 2 characters zero filled*
* 62 *
* 63 cout << "Elapsed time is: " *
* 64 << setfill('0') << setw(2) << ehr << ':' *
* 65 << setw(2) << emin << ':' *
* 66 << setw(2) << esec << endl << setfill(' '); *
* 67 return 0; *
* 68 } *
* 69 *
* 70 /**************************************************************/ *
* 71 /* */ *
* 72 /* hms_to_totsec: converts hh:mm:ss to total seconds */ *
* 73 /* */ *
* 74 /**************************************************************/ *
* 75 *
* 76 long hms_to_totsec (int hrs, int min, int sec) { *
* 77 // verify time is valid *
* 78 if (hrs>24 || hrs<0 || min>60 || min<0 || sec>60 || sec<0) *
* 79 return TIME_ERROR; *
* 80 if (hrs == 0 && min == 0 && sec == 0) // disallow 0:0:0 *
* 81 return TIME_ERROR; *
* 82 if (hrs == 24 && !(min == 0 && sec == 0)) // disallow > 24:0:0 *
* 83 return TIME_ERROR; *
* 84 *
* 85 // return total seconds *
* 86 return 3600L * hrs + 60 * min + sec; *
* 87 } *
* 88 *
* 89 /**************************************************************/ *
* 90 /* */ *
* 91 /* totsec_to_hms: converts total seconds back to */ *
* 92 /* hh:mm:ss in reference flds */ *
* 93 /* */ *
* 94 /**************************************************************/ *
* 95 *
* 96 void totsec_to_hms (long totsec, int &hrs, int &min, int &sec) { *
* 97 hrs = totsec / 3600; *
* 98 min = totsec % 3600 / 60; *
* 99 sec = totsec % 60; *
* 100 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
More on Functions 275

The Static Storage Class

Sometimes we would like to have a function’s variable remember the last value it had from the
previous call of that function. To see why we might like this kind of behavior, let’s examine a
typical program application that needs just such a behavior.

When producing reports that fill more than one page, just displaying a title and column
headings on the first page is insufficient. Every page of a multiple page report must contain the
report title, column headings and a page number. Obviously, page numbers must begin with page
1 and increase sequentially page by page. We know how to display headings and column
headings on the first page of a report. These principles have to be extended to multiple paged
reports. Commonly, a headings() function is used to encapsulate all of the details of printing the
headings, column headings and page numbers.

For the main program to know when it is time to call the headings() function, it must
know when a page is full and can contain no further lines. This is done by counting the total
number of lines written on a page. The variable is often called lineCount. Before writing any line
to the report, the program must check lineCount to see if there are sufficient lines remaining.
The number of lines that can be written on a page depends upon several factors including the font
size used and the height of the paper. Typical reports have around 55 lines per page; the
remaining space is used for margins. Every time the program displays a line on the report, it must
add the number of lines just written to lineCount. Also, headings must be written on the first
page as the program begins. Thus, lineCount is often initialized to 99, indicating a full page so
that headings are printed at once on the first page. Here is the main() function’s logic to handle
multiple paged reports.
int main () {
int lineCount = 99;
...
// input a set of data
while (there is more data) {
if (lineCount > 55) {
headings (.....);
}
// process this record
// output one line of this set of data and results
lineCount++;
// input next set of data
}

Now desk check this coding through the main loop. Initially, lineCount contains 99 so
that the first pass through the loop, it is larger than 55 lines and headings() is called to generate
the various heading lines. Then the first set of data and results are displayed and lineCount is
incremented by one. What is its value now? 100! Oops. Clearly, either main() must reset
lineCount back to the number of heading and column heading and blank lines written or else
More on Functions 276

headings() must do so. It is lousy design to have main() reset lineCount because main() has no
idea how many lines that headings() actually has written. It is much better to let headings() reset
lineCount. Thus, again, a reference parameter variable must be used. The prototype for
headings() is
void headings (int& lineCount);

How is headings() implemented? The first action must be to eject to a new page.
Frequently, the reports are printed and thus with a full page, the printer must be told to eject to a
new page. This is done using the escape sequence \f for formfeed as it is known. Wherever there
is a \f code, the printer does a page eject and the characters after the \f code are then printed at the
top of the next page. To see how this is done, let’s assume that the report is to appear as follows.
Acme Daily Sales Report Page: 1

Item Quantity Unit Total

Number Sold Cost Cost

1234 10 42.00 420.00

...

To produce the first line at the top of a new page, headings() could do the following.
cout << "\f Acme Daily Sales Report Page: "

Next, comes the page number. Let’s have headings() define and initialize an integer to
represent the page number. Each time headings() is entered, the page number is incremented.
void headings (int& lineCount) {
int pageNum = 0;
pageNum++;
cout << "\f Acme Daily Sales Report Page: "
<< setw (3) << pageNum << endl << endl;
cout << " Item Quantity Unit Total\n"
<< "Number Sold Cost Cost\n\n";
lineCount = 5;
}

After displaying the five lines, lineCount is set to five total lines used so far. The first
page now looks great.

However, after the main() function fills up the page with 50 more lines of sales data,
main() calls headings() to eject to the second page and print headings once more. Now what
prints for the page number? What is the storage class of headings()’s pageNum variable? What
happens to its storage when the end block } is executed? What happens when headings() is
reentered the second time?

The pageNum variable is automatic storage. As soon as the end brace } is executed, the
storage for pageNum is destroyed. When headings() is entered the second time, storage is
More on Functions 277

reallocated and the new pageNum is reinitialized to 0 and then incremented to 1. This version
displays page number 1 for every page in the report.

What is needed is a kind of variable which remembers its previous value so that when
headings() is reentered, the previous value is available to be incremented to 2 and so on. This is
the behavior of data of the static storage class. The scope of a static variable is the same as
automatic, from the point of definition to the end of the defining block of code. That is, it is
headings()’s pageNum. No other function can directly access it by using its name. (Note that a
function can pass it as a reference parameter to a function and then the function can alter it, but
that called function is doing so using its own separate parameter reference variable.)

The lifetime of a static variable is totally different from the automatic and parameter
classes. The lifetime of a static variable is the duration of the whole program! The storage for a
static variable is not destroyed until the entire program is finished and returns to DOS.

Initialization is done one time by the compiler at compile time. It is never initialized at
run time when the function is invoked. Contrast this behavior with that of automatic storage
which gets initialized at run time every time the block that contains the variable is entered.

Static variables then can be thought of as remembering the previous values that they had
from the last call to the function that defines them. To create a static variable, just add the
keyword static before the data type. In this example,
static int pageNum = 0;
pageNum++;

Now the headings() function operates properly. Page number 1 prints on the first page.
When the function is called to produce headings on the second page, pageNum now contains the
last value it held which was the 1. The increment instruction sets pageNum to 2, and so on for
additional calls to headings(). Here is the correct version of the function.
void headings (int& lineCount) {
static int pageNum = 0;
pageNum++;
cout << "\f Acme Daily Sales Report Page: "
<< setw (3) << pageNum << endl << endl;
cout << " Item Quantity Unit Total\n"
<< "Number Sold Cost Cost\n\n";
lineCount = 5;
}

There are very few variables that must be made static in programs. The page number is
one such case. They are handy in the right circumstances.
More on Functions 278

The Global/External Storage Class

The fourth storage class is called global/external storage. To make a variable global, simply
define that variable outside of any block of coding. The scope of a global variable is from the
point of its definition to the end of that cpp file and even beyond into additional cpp files
(external). Generally, global variables are defined near the beginning of the cpp file containing
the main() function so that main() and all other functions can get access to that variable. Here is
the only type of variable that can be referenced directly using its name anywhere in the program
from its point of definition onwards.

The lifetime of a global/external variable is the duration of the entire program. The
storage for global variables, like static variables, is not destroyed until the program terminates
and returns to DOS. Also like static variables, initialization is done one time at compile time and
never at execution or run time.

We have been using global constant items since the beginning. However, non-constant
global variables can also be used. Using global variables represents the second way that the
elapsed time problem could have been solved. Recall that the totsec_to_hms() function could not
return three integers for the hours, minutes and seconds. Another way to handle the problem is to
create three global variables defined above the main() function. Then have the totsec_to_hms()
store the results in them and have the main() function retrieve the results from them.

Here is how they could be defined and positioned within the cpp file. Note that global
variables ought to be readily identifiable; I have uppercased their names.
#include <iostream>
#include <iomanip>
using namespace std;

// global error value

const int TIME_ERROR = -1;

// function prototypes
long hms_to_totsec (int hrs, int min, int sec);
void totsec_to_hms (long totsec);

// global variables
int GHR, GMIN, GSEC; // global hours, minutes and seconds

int main () {

From this point onward, all functions can directly access these three global variables.
Specifically, totsec_to_hms() can directly place new results in them.
void totsec_to_hms (long totsec) {
GHR = totsec / 3600;
GMIN = totsec % 3600 / 60;
More on Functions 279

GSEC = totsec % 60;

}
Back in main(), the contents can be copied to other automatic storage variables.
ehr = GHR;
emin = GMIN;
esec = GSEC;
Or they may be displayed directly.
cout << GHR << ':' << GMIN << ':' << GSEC << endl;

It is usually a wise idea to copy the results from the global results variables into some
local automatic variables. In this case, the program might need to more elapsed times. The next
call to totsec_to_hms() erases the current values in these global variables and places the new
results in them. Here is the complete program Basic07b.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic07b - Using Global Variables to Find the Elapsed Time *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /**************************************************************/ *
* 2 /* */ *
* 3 /* Basic07b: using global variables to find the elapsed time */ *
* 4 /* */ *
* 5 /**************************************************************/ *
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 using namespace std; *
* 10 // global error value *
* 11 const int TIME_ERROR = -1; *
* 12 *
* 13 // function prototypes *
* 14 long hms_to_totsec (int hrs, int min, int sec); *
* 15 void totsec_to_hms (long totsec); *
* 16 *
* 17 // global variables *
* 18 int GHR, GMIN, GSEC; // global hours, minutes and seconds *
* 19 *
* 20 *
* 21 int main () { *
* 22 *
* 23 int shr, smin, ssec, // starting hours, minutes, seconds *
* 24 fhr, fmin, fsec, // finish hours, minutes, seconds *
* 25 ehr, emin, esec; // elapsed hours, minutes, seconds *
* 26 *
* 27 long stotsec, ftotsec, etotsec; // total seconds *
* 28 char c; *
* 29 *
* 30 // input the starting time *
* 31 *
* 32 cout << "Enter starting time (hh:mm:ss): "; *
* 33 cin >> shr; *
* 34 cin.get (c); *
More on Functions 280

* 35 cin >> smin; *

* 36 cin.get (c); *
* 37 cin >> ssec; *
* 38 *
* 39 // input the ending time *
* 40 *
* 41 cout << "Enter finish time (hh:mm:ss): "; *
* 42 cin >> fhr >> c >> fmin >> c >> fsec; *
* 43 *
* 44 // validate both times and convert valid times to total seconds *
* 45 *
* 46 if ((stotsec = hms_to_totsec (shr, smin, ssec)) == TIME_ERROR) {*
* 47 cout << "Error: start time is invalid.\n"; *
* 48 return 1; *
* 49 } *
* 50 *
* 51 if ((ftotsec = hms_to_totsec (fhr, fmin, fsec)) == TIME_ERROR) {*
* 52 cout << "Error: finish time is invalid.\n"; *
* 53 return 1; *
* 54 } *
* 55 *
* 56 // calculate the elapsed total seconds *
* 57 *
* 58 etotsec = ftotsec - stotsec; *
* 59 *
* 60 // convert elapsed total seconds back to hms *
* 61 *
* 62 totsec_to_hms (etotsec); *
* 63 *
* 64 // save a local copy of global values in case we wish to recall *
* 65 // totsec_to_hms for another set of elapsed times *
* 66 *
* 67 ehr = GHR; *
* 68 emin = GMIN; *
* 69 esec = GSEC; *
* 70 *
* 71 // display the elapsed hms-each part is 2 characters zero filled*
* 72 *
* 73 cout << "Elapsed time is: " *
* 74 << setw(2) << setfill('0') << ehr << ':' *
* 75 << setw(2) << setfill('0') << emin << ':' *
* 76 << setw(2) << setfill('0') << esec << endl; *
* 77 return 0; *
* 78 } *
* 79 *
* 80 /**************************************************************/ *
* 81 /* */ *
* 82 /* hms_to_totsec: converts hh:mm:ss to total seconds */ *
* 83 /* */ *
* 84 /**************************************************************/ *
* 85 *
* 86 long hms_to_totsec (int hr, int min, int sec) { *
More on Functions 281

* 87 // verify time is valid *

* 88 if (hr>24 || hr<0 || min>60 || min<0 || sec>60 || sec<0) *
* 89 return TIME_ERROR; *
* 90 if (hr == 0 && min == 0 && sec == 0) // disallow 0:0:0 *
* 91 return TIME_ERROR; *
* 92 if (hr == 24 && !(min == 0 && sec == 0)) // disallow > 24:0:0 *
* 93 return TIME_ERROR; *
* 94 *
* 95 // return total seconds *
* 96 return 3600L * hr + 60 * min + sec; *
* 97 } *
* 98 *
* 99 /**************************************************************/ *
* 100 /* */ *
* 101 /* totsec_to_hms: converts total seconds back to hh:mm:ss */ *
* 102 /* in Global fields */ *
* 103 /* */ *
* 104 /**************************************************************/ *
* 105 *
* 106 void totsec_to_hms (long totsec) { *
* 107 GHR = totsec / 3600; *
* 108 GMIN = totsec % 3600 / 60; *
* 109 GSEC = totsec % 60; *
* 110 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Using Global Variables in Other Cpp Files — the extern Keyword

In larger programs, all of the coding is not placed into one cpp file. Rather than create a scrolling
nightmare, each function or series of related functions is placed into a separate cpp file and the
project is adjusted to include not only the main() cpp file, but also all of the other cpp files. This
poses a problem for global variables. Suppose that Basic07b.cpp was broken into two separate
cpp files. That is, let’s remove the totsec_to_hms() function that main() calls and place it into a
separate cpp file called totsec_to_hms.cpp.

When the main() function compiles, the compiler encounters the definitions for the three
global variables and creates space for them. However, when totsec_to_hms.cpp compiles, errors
occur because GHR, GMIN and GSEC are not defined anywhere in this cpp file. So one could
remove the compile time errors by copying their definitions from the main cpp file as shown.
// global variables
int GHR, GMIN, GSEC; // global hours, minutes and seconds

void totsec_to_hms (long totsec) {

GHR = totsec / 3600;
GMIN = totsec % 3600 / 60;
GSEC = totsec % 60;
}
More on Functions 282

Now it compiles without errors. The compiler also creates the three global variables.
However, when the linker executes combining all the parts into the final executable file, errors
occur. The linker is responsible for actually placing the global variables into their final locations
within the data segment. However, after it creates space for the three globals, GHR, GMIN and
GSEC, along comes the second cpp file asking the linker to make space for three more global
variables with the exact same names! There can only be one global variable with a specific name.

The answer is to tell the compiler and linker what we are really wanting to do inside of
the totsec_to_hms.cpp file. We really want this cpp file to be using the three global variables
that are defined in the main.cpp file. Hence, append the extern keyword before their definitions.
The extern stands for external reference. We are saying that these three are external references to
the global variables of the same name located in another cpp file.
extern int GHR, GMIN, GSEC;

void totsec_to_hms (long totsec) {

GHR = totsec / 3600;
GMIN = totsec % 3600 / 60;
GSEC = totsec % 60;
}

Where are Global and Static Variables Actually Stored?

Again, knowing a bit about how the computer is really operating helps immensely one’s
understanding of how the storage classes operate. We have seen that automatic and parameter
variables are stored in the program’s stack save area. Global and static variables are stored in the
program’s data segment. The following Figure 7.2 represents what a program’s exe file looks
like as well as what the program looks like when it is loaded into memory to actually execute.

Figure 7.2 Memory Layout of a Program

The Data and Stack segments are combined into one large segment. The stack is located
at the higher memory address while the data segment begins at an offset address 0 within the
segment. All global variables are stored in the global variable section of the data segment
beginning at address 0. They are followed by the static variables section. The remainder of the
segment that is not part of the global, static or stack is called the local heap, meaning large pile
of unused memory available for our use. Note that all of the actual program instructions are
More on Functions 283

located in the code segment along with the code for our functions and those that are built into the
language, such as sqrt().

Now you can see why the lifetime of static and global variables is for the duration of the
entire program. The compiler reserves space in the data segment for these variables and they
remain there until DOS removes the program from memory when the program terminates.
Storage on the stack is very dynamic in stark contrast to global and static types. Table 7.1 shows
a summary of the main storage classes.

Storage Where Scope Lifetime Initialization

Class Located
Automatic on the from the point of definition to the end of the run time at
stack to the end of the defining defining block of block entry
block of code code
Parameter on the from the point of definition to the end of the run time when
stack in the function header to the function function is
end of the function invoked
Static in the from the point of definition duration of whole one time at
data to the end of the defining program compile time
segment block of code
Global in the from the point of definition duration of whole one time at
data to the end of the entire program compile time
segment program
Table 7.1 The Main Storage Classes

Philosophy on the Use of Global Variables

Using global constants is highly recommended, such as

const double PI = 3.14159;

Using global variables is highly NOT recommended. Using global variables is generally
very bad program design and should be avoided at all costs if possible. Let’s see why it is such a
bad design idea. When global data are used, there are no longer any controls over who can access
and change the data or when and so on.

At a company that remains forever nameless, the accountants wanted every report to be
stored on a reel of magnetic tape and kept in the destruction proof bunkers for ten years just in
case the government wished to audit their policies and practices. On the printed version of the
report, in the title had to be the volume serial number of that reel of tape that the report was also
More on Functions 284

being written along with the date. IBM tape open programmers supplied a revised tape open
function that returned the date and the serial number of the tape volume so that the information
could be printed in the heading. All went well until one February 29. On that date, the report said
the date was March 1. IBM had an error in their date/time routines and did not calculate leap
years correctly.

Within that operating system program, the date/time was stored as global data so that
every function of the operating system could get at it. Thus, the tape open programmers quickly
fixed the date and sent out a fix to this nameless company and the world. Several months later, a
new version of the operating system was installed and the accountants complained once again
that the date was wrong, it was now ahead an entire day. What had happened is that IBM’s date
programmers discovered their error and applied a fix to the date so that it was now correct. They
did not know about the previous fix applied to the same global date done by the tape open
programmers. Thus, the date ended up being doubly fixed! Hence, in the next version of the
operating system the tape open programmers then removed their fix.

When using global data, all control over who can use it and when and why and where are
lost entirely. The result can be chaos. Had the date been passed by reference, then the caller can
tightly control who has the rights to actually alter that data. There are many, many more horror
stories that have occurred because of global data. Try to avoid using global variables as much as
possible. Until the object oriented classes are covered, there can be some circumstances in which
global variables are the lesser of several evils and are therefore used. The Engineering
application’s section illustrates this situation.

How to Pass iostreams to Functions

Commonly, instances of the iostreams or file versions of them must be passed to functions and
perhaps even returned from functions. How must they always be passed? Let’s explore this a bit.

Suppose that we pass cin to a function called fun() as follows.

int main () {
int x, y;
fun (cin);
cin >> x >> y;
}
void fun (istream input) { // a disaster...
int a, b;
input >> a >> b;
}

Notice that a copy of the input stream is passed. Also assume that the other variables are
defined to be integers. What occurs? When main() calls fun() and executes the extraction of the
integers a and b from the input stream, all is well. The first two integers are input from the
More on Functions 285

keyboard. However, remember what happens with an input stream. There is a current position in
the input stream that is updated as characters are extracted from the stream. There are many other
variables that are also updated such as the end of file state and bad data and so on. But what
happens to the copy of cin that is passed and used within fun() when fun() returns? It is pitched!
Now what is the state of cin back in main() just as it is about to extract the values for x and y? It
believes cin has not even been used yet to obtain any data! Oops! It was a copy of the cin object
that was used within function fun().

This is a recipe for disaster when passing iostreams to functions. And yet, very frequently
ifstreams and ofstreams need to be passed to input and output type functions. The answer is to
pass them by reference. If they are passed by reference, no copy is made. Rather, any changes
made to their contents are made to the real, original stream. The correct way to code function
fun() above is as follows.
void fun (istream& input) {
int a, b;
input >> a >> b;
}

Rule: Always pass iostreams and file streams by reference to functions; if a function is
to return one, make it return a reference to the passed iostream.

Input type functions are often written to unclutter the main() function. Similarly, output
type functions are written to also streamline the main() function. If you review many of the
previous programs in this text, the largest percentage of instructions in the main() function have
dealt with input and output.

Suppose that the program needed to conversationally prompt the user to enter four values
from the keyboard, such as
cout << "Enter item number: ";
cin >> itemNum;
cout << "Enter quantity: ";
cin >> qty;
cout << "Enter unit cost: ";
cin >> cost;
cout << "Enter state code: ";
cin >> stateCode;

This sequence tends to clutter up the main() function making it harder to read. It is even
more cluttered if a primed loop approach is taken in which the above sequence is done once
before the loop and then again at the very bottom of the loop.
More on Functions 286

Consider the simplification that the following function call to inputData() has on the
main() function’s processing loop.
inputData (cin, itemNum, qty, cost, stateCode);
while (cin) {
... process and output this set of data
inputData (cin, itemNum, qty, cost, stateCode);
}
Now main() is much more readable and streamlined.

How would the inputData() function be written? First, develop its prototype. In this case,
since it is inputting the actual values to be stored in main()’s variables, all of the parameters must
be passed by reference.
void inputData (istream& in, long& itemNum, int& qty,
double& cost, int& stateCode);
The body of the function is just the original sequence, assuming the prompts go to the screen.
void inputData (istream& in, long& itemNum, int& qty,
double& cost, int& stateCode){
cout << "Enter item number: ";
in >> itemNum;
cout << "Enter quantity: ";
in >> qty;
cout << "Enter unit cost: ";
in >> cost;
cout << "Enter state code: ";
in >> stateCode;
}

However, using an input file is far more useful and practical. Since a file stream,
ifstream, is derived from an istream, it may be passed either as a reference to an ifstream or as a
reference to an istream. Often it is passed using an istream& so that more flexibility can be
gained. In other words, either an opened input file stream can be passed to the function or cin can
be passed. Also, from a design of the function point of view, if the function returns a reference to
the input stream it was given, then the calling functions gain more flexibility. If our prototype is
istream& inputData (istream& in, long& itemNum, int& qty,
double& cost, int& stateCode){
then the user can do any of the following to invoke our input function.
inputData (infile, itemNum, qty, cost, stateCode);
or
if (inputData (infile, itemNum, qty, cost, stateCode)) {
or
while (inputData (infile, itemNum, qty, cost, stateCode)) {
The while clause is most often coded by programmers.
More on Functions 287

If one is using a file for input, then there is no need for the prompts and the function can
be coded as follows.
istream& inputData (istream& in, long& itemNum, int& qty,
double& cost, int& stateCode){
in >> itemNum >> qty >> cost >> stateCode;
return in;
}
Yet when the function is reduced to just this, it does not seem worthwhile to even do all this,
since it is replacing only one input instruction that could have been coded more easily in the
main() function! True at this point. However, in a later chapter the input operation of one set of
data will become much more complex and expand to fill several instructions. At that point, this
type of an input function is very valuable in reducing complexity.

The same principles apply to output functions. Suppose that a program’s report was
supposed to be written to a file and not to the screen. The headings() function above must now
be passed the output file upon which to write the headings. Here is a revised version that displays
headings but displays them on the passed output stream.
ostream& headings (ostream& out, int& lineCount) {
static int pageNum = 0;
pageNum++;
out << "\f Acme Daily Sales Report Page: "
<< setw (3) << pageNum << endl << endl;
out << " Item Quantity Unit Total\n"
<< "Number Sold Cost Cost\n\n";
lineCount = 5;
return out;
}

Once again, by making the parameter out be a reference to an ostream, then the caller
can pass either cout or an instance of an ofstream class, such as outfile. By returning that same
ostream reference, we give the user the option of chaining a function call to headings() with
some other output actions.

Section B: Computer Science Examples

Cs07c — Acme Ticket Sales Report — a Multi-page Report

Let’s return to Cs05a, the Ticket Sales program. This time because of the rather large volume of
tickets sold, we need to print a multi-page detail report of the sales. Recall that each input line
consists of the number of adult tickets sold, the number of child tickets, the number of senior
tickets sold and the basic cost of a ticket.
More on Functions 288

Assuming that each page of the report can hold 41 lines, here is the output of the
program.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Cs07c Acme Ticket Sales Multi-page Report *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Acme Ticket Sales Summary Report Page: 1 *
* 2 *
* 3 Number of Tickets Sold Total Cost *
* 4 Adult Child Senior Of Tickets *
* 5 *
* 6 2 2 2 $ 42.00 *
* 7 2 0 0 $ 20.00 *
...
* 41 2 0 0 $ 20.00 *
* 42 Acme Ticket Sales Summary Report Page: 2 *
* 43 *
* 44 Number of Tickets Sold Total Cost *
* 45 Adult Child Senior Of Tickets *
* 46 *
* 47 2 2 2 $ 42.00 *
...
* 82 2 0 0 $ 20.00 *
* 83 Acme Ticket Sales Summary Report Page: 3 *
* 84 *
* 85 Number of Tickets Sold Total Cost *
* 86 Adult Child Senior Of Tickets *
* 87 *
* 88 2 2 2 $ 42.00 *
...
* 123 2 0 0 $ 20.00 *
* 124 --- --- --- -------- *
* 125 Totals: 204 264 48 $3024.00 *
* 126 Percents: 40% 51% 9% *
* 127 *
* 128 Average cost of ticket: $ 5.86 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

The Top-down design that I have chosen is shown below.

Top-down Design for Program Cs07c

More on Functions 289

My objective is to migrate as much of the processing details out of the main function as
possible. In order to implement the OpenFiles function, we must pass a reference to the two file
streams, infile and outfile. This function then attempts to open both files. If a file cannot be
opened, an error message is shown. OpenFiles returns a bool. If both files are successfully
opened, it returns true. If one could not be opened, it returns false.

As you study this example, notice how the actual workload is spread out among the
various functions. Each function is short and simple. By designing programs this way, the entire
problem is greatly simplified and far more easily written and debugged.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs07c Acme Ticket Sales Multi-page Report *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 #include <iostream> *
* 2 #include <iomanip> *
* 3 #include <fstream> *
* 4 using namespace std; *
* 5 *
* 6 // prototypes *
* 7 bool OpenFiles (ifstream& infile, ofstream& outfile); *
* 8 *
* 9 istream& InputData (istream& is, int& numAdult, int& numChild, *
* 10 int& numSenior, double& cost); *
* 11 void Headings (ostream& os, int& lineCount); *
* 12 ostream& OutputLine (ostream& os, int numAdult, int numChild, *
* 13 int numSenior, double cost, int& lineCount);*
* 14 ostream& DisplayTotals (ostream& os, int totAdult, int totChild, *
* 15 int totSenior, double totalCost); *
* 16 *
* 17 /***************************************************************/*
* 18 /* */*
* 19 /* Cs07c: Acme Ticket Sales Multiple Page Report */*
* 20 /* */*
* 21 /***************************************************************/*
* 22 int main () { *
* 23 *
* 24 // input fields *
* 25 int numAdult; // number adult tickets sold to this customer *
* 26 int numChild; // number child tickets sold to this customer *
* 27 int numSenior; // number senior tickets sold to this customer*
* 28 double cost; // total cost of this customer's tickets *
* 29 *
* 30 // calculation fields *
* 31 int totNumAdult = 0; // total adult tickets sold *
* 32 int totNumChild = 0; // total child tickets sold *
* 33 int totNumSenior = 0; // total senior tickets sold *
* 34 double totalCost = 0; // total cost of all tickets sold *
* 35 *
* 36 int lineCount = 99; // number of lines written on the page *
* 37 int lines = 1; // line count of total lines for errors*
* 38 *
More on Functions 290

* 39 ifstream infile; // the input file *

* 40 ofstream outfile; // the output file *
* 41 *
* 42 // attempt to open the files *
* 43 if (!OpenFiles (infile, outfile)) *
* 44 return 1; *
* 45 *
* 46 // process all the input sets of data *
* 47 while (InputData (infile, numAdult, numChild, numSenior, cost)){*
* 48 lines++; // increment count of total lines in case of errors *
* 49 // accumulate totals *
* 50 totNumAdult += numAdult; *
* 51 totNumChild += numChild; *
* 52 totNumSenior += numSenior; *
* 53 totalCost += cost; *
* 54 *
* 55 // eject to new page if needed *
* 56 if (lineCount > 40) { *
* 57 Headings (outfile, lineCount); *
* 58 } *
* 59 *
* 60 // output this line *
* 61 OutputLine (outfile, numAdult, numChild, numSenior, cost, *
* 62 lineCount); *
* 63 } *
* 64 *
* 65 // check for bad data on input *
* 66 if (!infile.eof()) { *
* 67 cerr << "Error: bad data in the input file on line " *
* 68 << lines << endl; *
* 69 outfile << "Error: bad data in the input file on line " *
* 70 << lines << endl; *
* 71 infile.close (); *
* 72 outfile.close (); *
* 73 return 4; *
* 74 } *
* 75 *
* 76 // display the grand totals and results *
* 77 DisplayTotals (outfile, totNumAdult, totNumChild, totNumSenior, *
* 78 totalCost); *
* 79 *
* 80 infile.close (); *
* 81 outfile.close (); *
* 82 return 0; *
* 83 } *
* 84 *
* 85 /***************************************************************/*
* 86 /* */*
* 87 /* OpenFiles: attempt to open input and output files */*
* 88 /* returns false if one could not be opened */*
* 89 /* */*
* 90 /***************************************************************/*
More on Functions 291

* 91 bool OpenFiles (ifstream& infile, ofstream& outfile) { *

* 92 // attempt to open the input file *
* 93 infile.open ("ticket-sales.txt"); *
* 94 if (!infile) { // failed, so display an error message and quit *
* 95 cerr << "Error: cannot open file ticket-sales.txt\n"; *
* 96 return false; *
* 97 } *
* 98 *
* 99 // attempt to open the output file *
* 100 outfile.open ("results.txt"); *
* 101 if (!outfile) { // failed, so display error, close and quit *
* 102 cerr << "Error: cannot open the output file results.txt\n"; *
* 103 infile.close (); *
* 104 return false; *
* 105 } *
* 106 *
* 107 // setup floating point format for output of dollars *
* 108 outfile << fixed << setprecision (2); *
* 110 return true; *
* 111 } *
* 112 *
* 113 *
* 114 /***************************************************************/*
* 115 /* */*
* 116 /* InputData: input a set of data */*
* 117 /* */*
* 118 /***************************************************************/*
* 119 istream& InputData (istream& is, int& numAdult, int& numChild, *
* 120 int& numSenior, double& cost) { *
* 121 is >> numAdult >> numChild >> numSenior >> cost; *
* 122 return is; *
* 123 } *
* 124 *
* 125 /***************************************************************/*
* 126 /* */*
* 127 /* OutputLine: output a detail line */*
* 128 /* */*
* 129 /***************************************************************/*
* 130 ostream& OutputLine (ostream& os, int numAdult, int numChild, *
* 131 int numSenior, double cost, int& lineCount){*
* 132 os << setw (12) << numAdult << setw (8) << numChild *
* 133 << setw (9) << numSenior << " $" << setw (7) *
* 134 << cost << endl; *
* 135 lineCount++; *
* 136 return os; *
* 137 } *
* 138 *
* 139 /***************************************************************/*
* 140 /* */*
* 141 /* Headings: eject to new page and display headings and pagenum*/*
* 142 /* */*
* 143 /***************************************************************/*
More on Functions 292

* 144 void Headings (ostream& os, int& lineCount) { *

* 145 static int pageNumber = 0; *
* 146 pageNumber++; *
* 147 *
* 148 os << "\f Acme Ticket Sales Summary Report Page: " *
* 149 << setw (3) << pageNumber << "\n\n" *
* 150 << " Number of Tickets Sold Total Cost\n" *
* 151 << " Adult Child Senior Of Tickets\n\n"; *
* 152 lineCount = 5; // reset number of lines written on this page *
* 153 return; *
* 154 } *
* 155 *
* 156 /***************************************************************/*
* 157 /* */*
* 158 /* DisplayTotals: calcs and displays the grand total values */*
* 159 /* */*
* 160 /***************************************************************/*
* 161 ostream& DisplayTotals (ostream& os, int totAdult, int totChild, *
* 162 int totSenior, double totalCost) { *
* 163 int grandTotalTicketsSold;// total number of all tickets sold*
* 164 double percentAdult; // percent adult of total tickets *
* 165 double percentChild; // percent child of total tickets *
* 166 double percentSenior; // percent senior of total tickets *
* 167 double avgCost; // average cost of one ticket *
* 168 *
* 169 // display first totals line *
* 170 os << " --- --- --- --------\n"; *
* 171 os << "Totals:" << setw (5) << totAdult << setw (8) *
* 172 << totChild << setw (9) << totSenior *
* 173 << " $" << setw (7) << totalCost << endl; *
* 174 *
* 175 // calculate and display the percentages line *
* 176 grandTotalTicketsSold = totAdult + totChild +totSenior; *
* 177 percentAdult = totAdult * 100. / grandTotalTicketsSold; *
* 178 percentChild = totChild * 100. / grandTotalTicketsSold; *
* 179 percentSenior = totSenior * 100. / grandTotalTicketsSold; *
* 180 os << setprecision (0); *
* 181 os << "Percents:" << setw (3) << percentAdult << "%" *
* 182 << setw (7) << percentChild << "%" << setw (8) *
* 183 << percentSenior << "%" << endl << endl; *
* 184 os << setprecision (2); *
* 185 *
* 186 // calculate and display the average cost of a ticket *
* 187 avgCost = totalCost / grandTotalTicketsSold; *
* 188 os << "Average cost of ticket: $" << setw (7) *
* 189 << avgCost << endl; *
* 190 return os; *
* 191 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
More on Functions 293

Cs07a — Multiple Level Control Break Processing

The Policies Sold by State and by Agent Report

Commonly, reports need to be broken down into smaller groupings. Consider a national
insurance company which has many agents in many states selling insurance policies. The
company needs the very lengthy report broken down into smaller portions. The input data has
already been sorted into increasing state code order and then into agent number order within each
state. Here is what the report looks like.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Cs07a - Policy Report by State and by Agent Number *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Acme Insurance Policies Sold Page 1 *
* 2 *
* 3 State Code: 1 Agent: 1 *
* 4 *
* 5 Policy Number Premium *
* 6 *
* 7 1234567 $ 250.00 *
* 8 1234346 $ 345.44 *
* 9 1233455 $ 454.45 *
* 10 *
* 11 Total Policies for this agent: 3 *
* 12 Total Premiums for this agent: 1049.89 *
* 13 --------------------------------------- *
* 14 Acme Insurance Policies Sold Page 2 *
* 15 *
* 16 State Code: 1 Agent: 2 *
* 17 *
* 18 Policy Number Premium *
* 19 *
* 20 2343233 $ 433.44 *
* 21 3453453 $ 345.22 *
* 22 3453455 $ 356.99 *
* 23 *
* 24 Total Policies for this agent: 3 *
* 25 Total Premiums for this agent: 1135.65 *
* 26 *
* 27 Total Policies for this state: 6 *
* 28 Total Premiums for this state: 2185.54 *
* 29 --------------------------------------- *
* 30 Acme Insurance Policies Sold Page 3 *
* 31 *
* 32 State Code: 2 Agent: 1 *
* 33 *
* 34 Policy Number Premium *
* 35 *
* 36 3245674 $ 564.33 *
* 37 4375645 $ 334.55 *
* 38 *
More on Functions 294

* 39 Total Policies for this agent: 2 *

* 40 Total Premiums for this agent: 898.88 *
* 41 --------------------------------------- *
* 42 Acme Insurance Policies Sold Page 4 *
* 43 *
* 44 State Code: 2 Agent: 2 *
* 45 *
* 46 Policy Number Premium *
* 47 *
* 48 3476557 $ 235.55 *
* 49 2453534 $ 456.88 *
* 50 *
* 51 Total Policies for this agent: 2 *
* 52 Total Premiums for this agent: 692.43 *
* 53 *
* 54 Total Policies for this state: 4 *
* 55 Total Premiums for this state: 1591.31 *
* 56 *
* 57 Company Total Policies: 10 *
* 58 Company Total Premiums: 3776.85 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

When printing the data for a specific agent within a state, if the number of lines exceeds
55, the program must eject the printer to a new page. Whenever the agent within a state changes,
the program must print that agent’s total number of policies and the total premium. Begin the
new agent on the next page. Whenever the state changes, also print the state total policies and
total premiums. At the end of the file, print the company total policies and total premiums.

The state code and the agent number are called control fields because their contents are
going to control the way the report is created.

A state code change is referred to as the major level break and an agent number change
is called the minor level break. The detection of a full page is sometimes called a page break.
In this problem there are three events that must be detected and handled as we input each line of
input data: a state break, an agent break and a page break. We can use a variable called
lineCount to check for a full page. But to check for a change in the other two input variables, we
must keep track of the previous state and previous agent number. Whenever the previous value
does not equal the current one just input, a break occurs in which alternative processing is done,
such as printing the totals for the now finished agent or state.

When checking for the occurrence of the breaks, the rule is always check for breaks from
major to minor. That is, after inputting a new set of data, first check if the state code has changed.
If not, then check to see if the agent number has changed. If not, then check to see if we need a
new page because this one is full.

When processing breaks, such as the current state is not equal to the previous state, the
rule is always process the breaks from minor to major. That is, if the state changed, then first
More on Functions 295

display the totals for the last agent we were working on when the state code changed. Then the
state totals can be done.

The processing steps usually found when a break in a control field has occurred include
the following.
1. Rolling any totals into the next higher set of totals, such as adding this agent’s totals
into the state totals.
2. Displaying this set of totals
3. Resetting this set of totals back to 0
4. Resetting the previous value of this control field
5. Ejecting to a new page to begin this new group on a new page.

Rolling totals makes good sense. Suppose that there are 1,000,000 policies (lines in this
case) in the file. If after inputting each set of data, we added that premium to the agent’s total
premium, the state’s total premium and the company total premium, then 3,000,000 additions
need to be done. On the other hand, if there are only 50 states and 1,000 agents involved, then by
rolling totals, only 1,001,050 additions need to be done. That is, after inputting a set of data, add
the premium to the agent’s total premium. When the agent number changes, add that agent’s total
premium into the state total premium. Later on when the state changes, add that state’s total
premium into the company total premium. This becomes a significant factor when there are a
large number of sets of data involved.

The Top-Down Design shows a function for handling the state break, agent break and
headings. These functions are going to be called from several points in the program.

Figure 7.3 Top-Down Design for Policy Program

Next, design the solution. The key variables needed to hold the input set of data include:
policyNumber, stateCode, agentNumber and premium. A pair of variables are needed to store
the previous values of the state code and agent number: previousStateCode and
previousAgentNumber. Three sets of totals are needed, one for the agent, one for the state and
one for the company. So that these can easily be kept straight in the program, very descriptive
names are chosen: totalPoliciesAgent, totalPremiumsAgent, totalPoliciesState,
totalPremiumsState, totalPoliciesCompany and totalPremiumsCompany. Finally, lineCount
is used to track the number of lines written. Figure 7.4 shows main storage for function main().
More on Functions 296

Now sketch the sequences of instructions needed in the main() function.

open the input and output files, displaying an error message and quitting if they cannot
be opened
initialize all totals
totalPoliciesAgent = 0
totalPremiumsAgent = 0
totalPoliciesState = 0
totalPremiumsState = 0

Figure 7.4 Main Storage for Function main()

totalPoliciesCompany = 0
totalPremiumsCompany = 0
input the first set of data from infile, aborting if the file is empty
set previousAgentNumber to agentNumber
set previousStateCode to stateCode
call headings() to get headings on the first page
while infile is still ok, do the following
if previousStateCode is not the same as stateCode, then do
do an agent break to display agent totals
do a state break to display state totals
reset previousStateCode to the new stateCode
reset previousAgentNumber to the new agentNumber
do headings() again
end do
else if the previousAgentNumber is not the same as agentNumber, do these
do an agent break to display agent totals
reset previousAgentNumber to the new agentNumber
do headings() again
end do
else if lineCount is > 55, then do these
More on Functions 297

do headings() again
end do
increment totalPoliciesAgent
add premium to totalPremiumsAgent
display this set of data, policyNumber and premium
increment lineCount by one
input next set of data from infile
end while loop
do an agent break to display agent totals
do a state break to display state totals
display company totals
close the files and return to dos

Continue to sketch the sequences needed for the other three functions. Here are the steps
needed for doAgentBreak(). For brevity, I have not shown the main storage diagrams for these.

Function doAgentBreak() is given the output stream out, totalPoliciesAgent,

totalPremiumsAgent, totalPoliciesState and totalPremiumsState.
totalPoliciesState += totalPoliciesAgent;
totalPremiumsState += totalPremiumsAgent;
output totalPoliciesAgent and totalPremiumsAgent
set totalPoliciesAgent = 0;
set totalPremiumsAgent = 0;

Function doStateBreak() is given the output stream out, and totalPoliciesState,

totalPremiumsState, totalPoliciesCompany and
totalPremiumsCompany
totalPoliciesCompany += totalPoliciesState;
totalPremiumsCompany += totalPremiumsState;
using file out, display totalPoliciesState and totalPremiumsState
set totalPoliciesState = 0;
set totalPremiumsState = 0;

Here is the completed program. Notice how closely it follows our design.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs07a - Policy Report by State and by Agent Number *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs07a: Policy Report By State Code and By Agent Number */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
More on Functions 298

* 9 #include <fstream> *
* 10 using namespace std; *
* 11 *
* 12 const int LinesPerPage = 55; // the maximum lines on a page *
* 13 *
* 14 void doStateBreak (ostream& out, long& totalPoliciesState, *
* 15 double& totalPremiumsState, long& totalPoliciesCompany,*
* 16 double& totalPremiumsCompany); *
* 17 void doAgentBreak (ostream& out, long& totalPoliciesAgent, *
* 18 double& totalPremiumsAgent, long& totalPoliciesState, *
* 19 double& totalPremiumsState); *
* 20 void headings (ostream& out, short& lineCount, char stateCode, *
* 21 short agentNumber); *
* 22 *
* 23 int main () { *
* 24 // input data *
* 25 long policyNumber; *
* 26 char stateCode; *
* 27 short agentNumber; *
* 28 double premium; *
* 29 *
* 30 // previous control field values *
* 31 char previousStateCode; *
* 32 short previousAgentNumber; *
* 33 *
* 34 // totals *
* 35 long totalPoliciesAgent = 0; *
* 36 double totalPremiumsAgent = 0; *
* 37 long totalPoliciesState = 0; *
* 38 double totalPremiumsState = 0; *
* 39 long totalPoliciesCompany = 0; *
* 40 double totalPremiumsCompany = 0; *
* 41 *
* 42 short lineCount; *
* 43 *
* 44 // attempt to open input master file *
* 45 ifstream infile ("PolicyFile.txt"); *
* 46 if (!infile) { *
* 47 cerr << "Error: cannot open PolicyFile.txt file\n"; *
* 48 return 1; *
* 49 } *
* 50 *
* 51 // get first set of data - be alert for an empty file *
* 52 infile >> policyNumber >> stateCode >> agentNumber >> premium; *
* 53 if (!infile) { *
* 54 cerr << "Error: no data or bad data\n"; *
* 55 infile.close (); *
* 56 return 2; *
* 57 } *
* 58 *
* 59 // open output report file *
* 60 ofstream report ("PolicyReport.txt"); *
More on Functions 299

* 61 if (!report) { *
* 62 cerr << "Error: cannot open PolicyReport.txt file\n"; *
* 63 infile.close (); *
* 64 return 3; *
* 65 } *
* 66 // setup floating point output format for dollars *
* 67 report << fixed << setprecision (2); *
* 70 *
* 71 // set previous values to the initial data *
* 72 previousStateCode = stateCode; *
* 73 previousAgentNumber = agentNumber; *
* 74 *
* 75 // print headings for first page *
* 76 headings (report, lineCount, stateCode, agentNumber); *
* 77 *
* 78 while (infile) { *
* 79 // has the state code changed? *
* 80 if (stateCode != previousStateCode) { *
* 81 doAgentBreak (report, totalPoliciesAgent, totalPremiumsAgent, *
* 82 totalPoliciesState, totalPremiumsState); *
* 83 doStateBreak (report, totalPoliciesState, totalPremiumsState, *
* 84 totalPoliciesCompany, totalPremiumsCompany); *
* 85 previousStateCode = stateCode; *
* 86 previousAgentNumber = agentNumber; *
* 87 headings (report, lineCount, stateCode, agentNumber); *
* 88 } *
* 89 // has the agent number changed? *
* 90 else if (agentNumber != previousAgentNumber) { *
* 91 doAgentBreak (report, totalPoliciesAgent, totalPremiumsAgent, *
* 92 totalPoliciesState, totalPremiumsState); *
* 93 previousAgentNumber = agentNumber; *
* 94 headings (report, lineCount, stateCode, agentNumber); *
* 95 } *
* 96 // is the current page full? *
* 97 else if (lineCount > LinesPerPage) { *
* 98 headings (report, lineCount, stateCode, agentNumber); *
* 99 } *
* 100 *
* 101 // accumulate agent totals *
* 102 totalPoliciesAgent++; *
* 103 totalPremiumsAgent += premium; *
* 104 *
* 105 // display this set of data *
* 106 report << setw (10) << policyNumber << " $" *
* 107 << setw (7) << premium << endl; *
* 108 lineCount++; *
* 109 *
* 110 // get next set of data *
* 111 infile >> policyNumber >> stateCode >> agentNumber >> premium; *
* 112 } *
* 113 *
* 114 // process totals for the last agent *
More on Functions 300

* 115 doAgentBreak (report, totalPoliciesAgent, totalPremiumsAgent, *

* 116 totalPoliciesState, totalPremiumsState); *
* 117 // process totals for the last state code *
* 118 doStateBreak (report, totalPoliciesState, totalPremiumsState, *
* 119 totalPoliciesCompany, totalPremiumsCompany); *
* 120 // display final company totals *
* 121 report << "\nCompany Total Policies: " << setw (5) *
* 122 << totalPoliciesCompany << endl *
* 123 << "Company Total Premiums: " << setw (8) *
* 124 << totalPremiumsCompany << endl; *
* 125 *
* 126 infile.close (); *
* 127 report.close (); *
* 128 return 0; *
* 129 } *
* 130 *
* 131 /***************************************************************/*
* 132 /* */*
* 133 /* headings: display all heading lines */*
* 134 /* */*
* 135 /***************************************************************/*
* 136 *
* 137 void headings (ostream& out, short& lineCount, char stateCode, *
* 138 short agentNumber) { *
* 139 static int pageNum = 0; *
* 140 pageNum++; *
* 141 out << "\fAcme Insurance Policies Sold Page " << setw (3) *
* 142 << pageNum << endl << endl; *
* 143 out << "State Code: " << setw (2) << stateCode *
* 144 << " Agent: " <<setw(3) << agentNumber<<endl<<endl;*
* 145 out << " Policy Number Premium\n\n"; *
* 146 lineCount = 6; *
* 147 } *
* 148 *
* 149 /***************************************************************/*
* 150 /* */*
* 151 /* doAgentBreak: process and display agent totals */*
* 152 /* */*
* 153 /***************************************************************/*
* 154 *
* 155 void doAgentBreak (ostream& out, long& totalPoliciesAgent, *
* 156 double& totalPremiumsAgent, *
* 157 long& totalPoliciesState, *
* 158 double& totalPremiumsState) { *
* 159 totalPoliciesState += totalPoliciesAgent; *
* 160 totalPremiumsState += totalPremiumsAgent; *
* 161 out << "\nTotal Policies for this agent: " << setw (5) *
* 162 << totalPoliciesAgent << endl *
* 163 << "Total Premiums for this agent: " << setw (8) *
* 164 << totalPremiumsAgent << endl; *
* 165 totalPoliciesAgent = 0; *
* 166 totalPremiumsAgent = 0; *
More on Functions 301

* 167 } *
* 168 *
* 169 /***************************************************************/*
* 170 /* */*
* 171 /* doStateBreak: process and display state totals */*
* 172 /* */*
* 173 /***************************************************************/*
* 174 *
* 175 void doStateBreak (ostream& out, long& totalPoliciesState, *
* 176 double& totalPremiumsState, *
* 177 long& totalPoliciesCompany, *
* 178 double& totalPremiumsCompany) { *
* 179 totalPoliciesCompany += totalPoliciesState; *
* 180 totalPremiumsCompany += totalPremiumsState; *
* 181 out << "\nTotal Policies for this state: " << setw (5) *
* 182 << totalPoliciesState << endl *
* 183 << "Total Premiums for this state: " << setw (8) *
* 184 << totalPremiumsState << endl; *
* 185 totalPoliciesState = 0; *
* 186 totalPremiumsState = 0; *
* 187 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

There remains one small detail that can turn this program into a “mostly working”
program. Can you spot that detail? Here is a hint. Suppose that we have just displayed the last set
of data for some state and that lineCount now is 55. At the bottom of the loop, the next set of
data is input and it is for the next state. What happens next? Both break functions are called to
display the additional total lines as they should. But have you spotted the problem? The page is
full, so where are those totals going to be displayed? On a new page — one that has no headings,
page numbers or column headings! Oops. One really ought to check on lineCount each time to
see if there are going to be enough lines to print totals and if not, call headings() again.

Cs07b — Summary Reports Based upon Control Break Processing

The Summary Report by State Program

Summary reports are often closely related to control break programs. The difference is that a
summary report program only prints the totals when the control field(s) change. Consider the
following output from program Cs07b.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Cs07b - Summary Report by State *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Acme Insurance Summary Report Page 1 *
* 2 *
* 3 State Total Total *
* 4 Code Policies Premium *
* 5 *
* 6 1 6 2185.54 *
More on Functions 302

* 7 2 4 1591.31 *
* 8 ----- -------- *
* 9 10 3776.85 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

This time, the program is not printing anything from each set of input data. Rather, it
simply accumulates the totals. When the state code changes, then, in the state break function, a
total line is written, lines 6 and 7 in the above report.

The agent break function is removed along with the agent totals. The main processing
loop now accumulates the state totals instead. Nothing is printed within the main loop. The
headings() function becomes simpler since the state code and agent number no longer need to be
passed. However, the state break function now needs the state code in order to display the total
line. When a change in state occurs, we must pass the previousStateCode because the current
stateCode is that of the new state. Also at the end of the file when the main loop terminates, the
state break function is called one more time to produce totals for that last state. Notice that
previousStateCode is also passed since nothing was input into stateCode when the end of file is
reached.

Here is the variation program Cs07b Summary Report by State. To save some pages, I
have removed some of the lines that are duplicates of program Cs07a.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs07b - Summary Report by State *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
...
* 3 /* Cs07b: Summary Report By State Code and By Agent Number */*
...
* 14 void doStateBreak (ostream& out, long& totalPoliciesState, *
* 15 double& totalPremiumsState, long& totalPoliciesCompany,*
* 16 double& totalPremiumsCompany, char stateCode); *
* 17 void headings (ostream& out, short& lineCount); *
* 18 *
* 19 int main () { *
* 20 // input data *
* 21 long policyNumber; *
* 22 char stateCode; *
* 23 short agentNumber; *
* 24 double premium; *
* 25 *
* 26 // previous control field value *
* 27 char previousStateCode; *
* 28 *
* 29 // totals *
* 30 long totalPoliciesState = 0; *
* 31 double totalPremiumsState = 0; *
* 32 long totalPoliciesCompany = 0; *
* 33 double totalPremiumsCompany = 0; *
* 34 *
More on Functions 303

* 35 short lineCount; *
* 36 *
* 37 // attempt to open input master file *
...
* 44 // get first set of data - be alert for an empty file *
* 45 infile >> policyNumber >> stateCode >> agentNumber >> premium; *
...
* 64 // set previous values to the initial data *
* 65 previousStateCode = stateCode; *
* 66 *
* 67 // print headings for first page *
* 68 headings (report, lineCount); *
* 69 *
* 70 while (infile) { *
* 71 // has the state code changed? *
* 72 if (stateCode != previousStateCode) { *
* 73 doStateBreak (report, totalPoliciesState, totalPremiumsState, *
* 74 totalPoliciesCompany, totalPremiumsCompany, *
* 75 previousStateCode); *
* 76 previousStateCode = stateCode; *
* 77 lineCount++; *
* 78 } *
* 79 // is the current page full? *
* 80 else if (lineCount > LinesPerPage) { *
* 81 headings (report, lineCount); *
* 82 } *
* 83 *
* 84 // accumulate agent totals *
* 85 totalPoliciesState++; *
* 86 totalPremiumsState += premium; *
* 87 *
* 88 // get next set of data *
* 89 infile >> policyNumber >> stateCode >> agentNumber >> premium; *
* 90 } *
* 91 *
* 92 // process totals for the last state code *
* 93 doStateBreak (report, totalPoliciesState, totalPremiumsState, *
* 94 totalPoliciesCompany, totalPremiumsCompany, *
* 95 previousStateCode); *
* 96 // display final company totals *
* 97 report << " " *
* 98 << " ----- --------\n" *
* 99 << " " *
* 100 << setw (10) << totalPoliciesCompany *
* 101 << setw (12) << totalPremiumsCompany << endl; *
...
* 114 void headings (ostream& out, short& lineCount) { *
* 115 static int pageNum = 0; *
* 116 pageNum++; *
* 117 out << "\fAcme Insurance Summary Report Page " << setw (3) *
* 118 << pageNum << endl << endl; *
* 119 out << "State Total Total\n" *
More on Functions 304

* 120 << "Code Policies Premium\n\n"; *

* 121 lineCount = 5; *
* 122 } *
...
* 130 void doStateBreak (ostream& out, long& totalPoliciesState, *
* 131 double& totalPremiumsState, long& totalPoliciesCompany,*
* 132 double& totalPremiumsCompany, char stateCode) { *
* 133 totalPoliciesCompany += totalPoliciesState; *
* 134 totalPremiumsCompany += totalPremiumsState; *
* 135 out << setw (3) << stateCode *
* 136 << setw (10) << totalPoliciesState *
* 137 << setw (12) << totalPremiumsState << endl; *
* 138 totalPoliciesState = 0; *
* 139 totalPremiumsState = 0; *
* 140 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Section C: Engineering Examples

Bisection Revisited — Writing a Generic Bisection Function
In the last chapter, we examined the bisection method for root solving. The bisect() function we
wrote then was severely limited. Now that references and global variables are understood, a
much better bisection function can be written. The ideal scene is to have a generic bisect()
function that can be given any function what so ever and have it find the root. In other words,
once we have coded the bisect() function, we should be able to just copy and paste bisect() as is
with no alterations into any other program that needs a root of some equation found.

The first deficiency of bisect() from Chapter 6 is that it needs really to return the root and
the number of iterations needed by way of reference variables. That then frees up the function’s
return data type. An improved bisect() function ought to return true when a root is found or false
when a root is not found.

The second deficiency is that the parameters to the function whose root is to be found
may have only one parameter, the double x. In reality, functions are much more complex, often
requiring a number of other parameters as well. These additional parameters, while constant in
terms of the root solving process, are variables because the user might wish to enter their specific
values. For example, a function might also be a function of time as well as x, that is f (x, t). The
user enters the specific value desired for t and then bisect() finds the x value. How does the
main() function get that value of t to the function?

One poor method would be to pass the value of t to bisect(). That requires changing the
prototype and function header of bisect() as well as finding every function call to function f() and
More on Functions 305

also relaying parameter t to that function. This defeats the idea of a generic function in which we
write it once and use it over without changes.

A far better approach is to make use of global variables. The benefits of using global
variables to allow bisect() to remain a generic function (code it once — then use over and over)
outweigh their disadvantages, in my opinion. Thus, the time variable would be defined as a
global variable, assigned its current value from user input and used directly in the f (x, t)
function.

The third deficiency of the old bisect() function is that the name of the function must be
f() and nothing else. If we wanted to bisect the sin() function or any other function that is not
called f(), then all calls to that function f() within bisect() must be altered to use the new name of
the function to be used this time. Clearly, this is not a generic way to do things. The solution is
actually a simple one — just pass to bisect() the name of the function to be used. Let’s say that
the name of the function to be used was passed as the last parameter to bisect(). Here are several
possible calls to bisect(). (The . . . indicates the omission of the other usual parameters to
bisect().)
bisect (. . . , f);
bisect (. . . , sin);
bisect (. . . , tan);
bisect (. . . , meltingPoint);
bisect (. . . , heatFlow);
While the function might be called f (x), it could also be one of the built-in functions such as
sin() or tan(). It could also be any other function we might write, such as heatFlow(). This gives
bisect() maximum flexibility.

But it brings up a new fundamental rather advanced C++ issue. What is the data type of
the name of a function? The name of a function is the memory address of where that code or
instructions are located in memory. Further, the data type of a function used as a parameter is its
prototype. And at this point the syntax is awful as well as advanced. In the next chapter we will
see how a programmer can invent a new name for some more cumbersome data types. So for
now, just accept the magical line of coding that generates a reasonable data type that we can use
for our function headers and prototypes in which a function is to be passed.

Here are the magical line and the new complete prototype for a generic bisect() function.
// creates a data type for a parameter function value for bisect
typedef double (*function) (double);

bool bisect (double a, double b, double eps, int imax,

double& root, int& I, function f);
The typedef is short for type define. The typedef line is defining a more convenient data type
called function that we can use as a data type for a function being passed as a parameter. We
know that the f (x) function has the prototype of
double f (double);
More on Functions 306

The * in the definition means pointer to or memory address of. Given that typedef line, then the
last parameter to bisect(), which is the function to be used, has a data type of function.

Notice that bisect() now returns a bool, true if a root is found, false if not. The two
answers are root and i, which contain the root and the number of iterations needed to find that
root.
With these changes understood, examine Engr07a, which is a rewrite of our previous
program from Chapter 6, Engr06a.

Engr07a — Using a Generic bisect() Function

The function now called fun() is
fun (x) = e-x - sin (.5 PI x)
The main() function still prompts the user for the interval a to b along with the desired accuracy
and the maximum number of iterations to try. If bisect() returns true, the root is displayed.

Here is the competed new version of bisect(). I have highlighted in bold face the
significant changes of our improved program. Notice that in main() the function to be bisected is
called fun() and is defined beginning on line 124. However, on line 47 where bisect() is invoked,
fun is passed as the last parameter. Within bisect(), that parameter is known as f (x) as before.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr07a - Generic bisect Function *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr07a - Finding roots using a generic function bisect */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 // creates a data type for a parameter function value for bisect *
* 12 typedef double (*function) (double); *
* 13 *
* 14 bool bisect (double a, double b, double eps, int imax, *
* 15 double& root, int& i, function f); *
* 16 *
* 17 double fun (double x); *
* 18 *
* 19 const double PI = acos (-1.); *
* 20 *
* 21 int main () { *
* 22 *
* 23 double a; // left x endpoint - root is between a and b *
More on Functions 307

* 24 double b; // right x endpoint *

* 25 double eps; // the desired degree of accuracy *
* 26 int imax; // maximum number of iterations to try *
* 27 int i; // number of iterations actually done *
* 28 double root; // the answer, the root of the equation *
* 29 *
* 30 // setup floating point output for 5 digits of accuracy *
* 31 cout << fixed << setprecision (5); *
* 34 *
* 35 cout << "Finding roots of exp (-x) - sin (.5 PI x)\n\n"; *
* 36 cout << "Enter the interval in which to search for a root\n" *
* 37 << "Enter the left endpoint: "; *
* 38 cin >> a; *
* 39 cout << "Enter the right endpoint: "; *
* 40 cin >> b; *
* 41 cout<<"Enter the desired degree of accuracy, such as .000001\n";*
* 42 cin >> eps; *
* 43 cout << "Enter the maximum number of iterations to attempt: "; *
* 44 cin >> imax; *
* 45 cout << endl; *
* 46 *
* 47 if (bisect (a, b, eps, imax, root, i, fun)) { *
* 48 cout << "Root is: " << root *
* 49 << " found after " << i << " iterations\n"; *
* 50 } *
* 51 *
* 52 return 0; *
* 53 } *
* 54 *
* 55 /***************************************************************/*
* 56 /* */*
* 57 /* bisect: function to find the root of function f(x) */*
* 58 /* parameter function f must be of the form */*
* 59 /* double funct (double); */*
* 60 /* */*
* 61 /***************************************************************/*
* 62 *
* 63 bool bisect (double a, double b, double eps, int imax, *
* 64 double& root, int& i, function f) { *
* 65 double x1 = a; *
* 66 double x3 = b; *
* 67 double x2 = (b + a) * .5; *
* 68 double f1 = f (x1); *
* 69 double f2 = f (x2); *
* 70 double f3 = f (x3); *
* 71 *
* 72 double d = (x3 - x1) * .5; // the width of the interval *
* 73 i = 0; // the current number of iterations *
* 74 *
* 75 // verify that there is a solution *
* 76 if (f1*f3 >= 0) { *
* 77 cerr << "Error: no root in the interval: " *
More on Functions 308

* 78 << a << " to " << b << endl *

* 79 << "The function may be miscoded or\n" *
* 80 << "the values of the end points are incorrect\n"; *
* 81 return false; *
* 82 } *
* 83 *
* 84 // find the root, stop when the root is sufficiently accurate *
* 85 // or imax iterations have been done *
* 86 while (i < imax && d > eps) { *
* 87 *
* 88 if (f1 * f2 < 0) { // is root in left half? *
* 89 x3 = x2; // yes, so move right end point leftwards *
* 90 f3 = f2; *
* 91 } *
* 92 else { // root in right half *
* 93 x1 = x2; // so move left end point rightwards *
* 94 f1 = f2; *
* 95 } *
* 96 *
* 97 // calculate new midpoint and width of interval *
* 98 x2 = (x3 + x1) * .5; *
* 99 f2 = f (x2); *
* 100 d = (x3 - x1) * .5; *
* 101 i++; *
* 102 } *
* 103 *
* 104 // check if it converged or not - either way, display results *
* 105 if (i == imax) { *
* 106 cerr << "Warning: after " << imax *
* 107 << " iterations, bisection has not converged\n" *
* 108 << "Root thus far is: " << x2 << endl; *
* 109 root = x2; *
* 110 return false; *
* 111 } *
* 112 *
* 113 // here we have found a root, i is set to number of iterations *
* 114 root = x2; // store x2 in the caller's answer variable *
* 115 return true; *
* 116 } *
* 117 *
* 118 /***************************************************************/*
* 119 /* */*
* 120 /* fun(x): exp(-1) - sin (.5 PI x) */*
* 121 /* */*
* 122 /***************************************************************/*
* 123 *
* 124 double fun (double x) { *
* 125 double ansr; *
* 126 ansr = exp (-x) - sin (.5 * PI * x); *
* 127 return ansr; *
* 128 // return exp (-x) - sin (.5 * PI * x); *
* 129 } *
More on Functions 309

.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Engr07b — Molar Volume of Non-Ideal Gases

(Chemical Engineering)
The chemical and physical interactions between gases and liquids are commonly encountered in
chemical engineering. For a specific substance, the mathematical description of the transition
from gas to liquid is vital. The basic ideal gas equation for one mole of gas is

where p is the pressure, V is the volume of one mole, n is the number of moles, T is the
temperature in degrees K (kelvin) and R is the ideal gas constant of .082054 L-atm/(mol-K).

This ideal gas equation assumes low pressures and high temperatures such that the liquid
state is not present at all. However, this assumption often is not a valid one; many situations exist
where there is a combination of a substance in both its gaseous and liquid state present. This
situation is called an imperfect gas. Empirical formulas have been discovered that model this
behavior. One of these is Van der Waal's equation of state for an imperfect gas. If the formula is
simplified, it is

where v is the molar volume or V/n and the values of a and b are empirical values or constants
that are dependent upon the gas at hand. These critical measurements correspond to that point
where equal masses of the gas and liquid phase have the same density. The critical values are
tabulated for many substances. See, for example, the Critical Constants for Gases section in the
Handbook of Chemistry and Physics.

Sending space probes to other planets with higher gravities than the earth poses problems
in gas containment of the onboard fuels. In this problem, the user wishes to enter the temperature
in Kelvin and the pressure in atmospheres. The program must display the volume as given by the
ideal gas equation and then by Van der Waal's equation. While the ideal gas formula is a simple
calculation, root solving must be used to determine v. By moving the RT term to the left side, we
get the usual f (x) format needed for bisection.

For oxygen, the a value is 1.360 and the b value is 0.03183.

More on Functions 310

The bisect() function is copied exactly as-is from program Engr07a. A pair of global
variables, p and t, hold the pressure and temperature respectively. The main() function prompts
the user to enter the two end points, a and b, and then the pressure in atmospheres and the
temperature in kelvin.

If the bisection function returns true, the results are displayed along with the volume
result given from the ideal gas equation.

Here are the complete program Engr07b and a set of sample results.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr07b - Volume of Oxygen (van der Waal) using bisect *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr07b - Finding roots of van der Waal's gas equation */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 // creates a data type for a parameter function value for bisect *
* 12 typedef double (*function) (double); *
* 13 *
* 14 bool bisect (double a, double b, double eps, int imax, *
* 15 double& root, int& i, function f); *
* 16 *
* 17 double van (double x); *
* 18 *
* 19 const double R = 0.082054; // ideal gas constant *
* 20 *
* 21 double p; // pressure in atmospheres *
* 22 double t; // temperature in K *
* 23 *
* 24 int main () { *
* 25 *
* 26 double a; // left x endpoint - root is between a and b*
* 27 double b; // right x endpoint *
* 28 double eps = .000001; // the desired degree of accuracy *
* 29 int imax = 1000; // maximum number of iterations to try *
* 30 *
* 31 int i; // number of iterations actually done *
* 32 double v; // the answer, the volume of gas *
* 33 *
* 34 // setup floating point output for 5 digits of accuracy *
* 35 cout << fixed << setprecision (5); *
* 38 *
* 39 cout << *
* 40 "Finding roots of van der Waal's gas equation for Oxygen\n\n";*
More on Functions 311

* 41 cout << *
* 42 "Enter the left endpoint, right endpoint, pressure in atm\n" *
* 43 << "and temperature in Kelvin - separated by blanks\n" *
* 44 << "Press ^Z to quit\n"; *
* 45 while (cin >> a >> b >> p >> t) { *
* 46 cout << endl; *
* 47 if (bisect (a, b, eps, imax, v, i, van)) { *
* 48 cout << "At pressure of: " << p << endl *
* 49 << "At temperature of: " << t << endl *
* 50 << "The volume is: " << v *
* 51 << " (" << i << " iterations)\n" *
* 52 << "Ideal Gas volume is: " << R * t / p << endl; *
* 53 } *
* 54 cout << endl; *
* 55 cout << "Enter another set or ^Z to quit\n"; *
* 56 } *
* 57 *
* 58 return 0; *
* 59 } *
* 60 *
* 61 /***************************************************************/*
* 62 /* */*
* 63 /* bisect: function to find the root of function f(x) */*
* 64 /* parameter function f must be of the form */*
* 65 /* double funct (double); */*
* 66 /* */*
* 67 /***************************************************************/*
* 68 *
* 69 bool bisect (double a, double b, double eps, int imax, *
* 70 double& root, int& i, function f) { *
* 71 double x1 = a; *
* 72 double x3 = b; *
* 73 double x2 = (b + a) * .5; *
* 74 double f1 = f (x1); *
* 75 double f2 = f (x2); *
* 76 double f3 = f (x3); *
* 77 *
* 78 double d = (x3 - x1) * .5; // the width of the interval *
* 79 i = 0; // the current number of iterations *
* 80 *
* 81 // verify that there is a solution *
* 82 if (f1*f3 >= 0) { *
* 83 cerr << "Error: no root in the interval: " *
* 84 << a << " to " << b << endl *
* 85 << "The function may be miscoded or\n" *
* 86 << "the values of the end points are incorrect\n" *
* 87 << "F(a) = " << f1 << endl *
* 88 << "F(b) = " << f3 << endl; *
* 89 return false; *
* 90 } *
* 91 *
* 92 // find the root, but when the root is sufficiently accurate *
More on Functions 312

* 93 // or imax iterations have been done *

* 94 while (i < imax && d > eps) { *
* 95 *
* 96 if (f1 * f2 < 0) { // is root in left half? *
* 97 x3 = x2; // yes, so move right end point leftwards *
* 98 f3 = f2; *
* 99 } *
* 100 else { // root in right half *
* 101 x1 = x2; // so move left end point rightwards *
* 102 f1 = f2; *
* 103 } *
* 104 *
* 105 // calculate new midpoint and width of interval *
* 106 x2 = (x3 + x1) * .5; *
* 107 f2 = f (x2); *
* 108 d = (x3 - x1) * .5; *
* 109 i++; *
* 110 } *
* 111 *
* 112 // check if it converged or not - either way, display results *
* 113 if (i == imax) { *
* 114 cerr << "Warning: after " << imax *
* 115 << " iterations, bisection has not converged\n" *
* 116 << "Root thus far is: " << x2 << endl; *
* 117 root = x2; *
* 118 return false; *
* 119 } *
* 120 *
* 121 // here we have found a root, i is set to number of iterations *
* 122 root = x2; // store x2 in the caller's answer variable *
* 123 return true; *
* 124 } *
* 125 *
* 126 /***************************************************************/*
* 127 /* */*
* 128 /* van(x): van der waals gas equation */*
* 129 /* */*
* 130 /***************************************************************/*
* 131 *
* 132 double van (double v) { *
* 133 double a = 1.360; *
* 134 double b = 0.03183; *
* 135 double ansr; *
* 136 ansr = (p + a / (v * v)) * (v - b) - R * t; *
* 137 return ansr; *
* 138 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Sample run - Engr07b - Volume of Oxygen (van der Waal) using bisect *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Finding roots of van der Waal's gas equation for Oxygen *
* 2 *
More on Functions 313

* 3 Enter the left endpoint, right endpoint, pressure in atm *

* 4 and temperature in Kelvin - separated by blanks *
* 5 Press ^Z to quit *
* 6 24 26 1 300 *
* 7 *
* 8 At pressure of: 1.00000 *
* 9 At temperature of: 300.00000 *
* 10 The volume is: 24.59280 (20 iterations) *
* 11 Ideal Gas volume is: 24.61620 *
* 12 *
* 13 Enter another set or ^Z to quit *
* 14 2.3 2.6 10 300 *
* 15 *
* 16 At pressure of: 10.00000 *
* 17 At temperature of: 300.00000 *
* 18 The volume is: 2.43840 (18 iterations) *
* 19 Ideal Gas volume is: 2.46162 *
* 20 *
* 21 Enter another set or ^Z to quit *
* 22 .2 .3 100 300 *
* 23 *
* 24 At pressure of: 100.00000 *
* 25 At temperature of: 300.00000 *
* 26 The volume is: 0.22636 (16 iterations) *
* 27 Ideal Gas volume is: 0.24616 *
* 28 *
* 29 Enter another set or ^Z to quit *
* 30 ^ *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Faster Alternative Root Solving Methods

The major problem with the bisection technique is that it can take a relatively large number of
iterations to get the answer. If the f (x) was a complex one and if performance was a design
factor, then alternative methods can be used to obtain the root in far fewer iterations.
Performance can be a crucial design issue in real-time applications. For example, in a space flight
simulator program, roots may need to be found repetitively as the path of the space ship is
plotted. The alternative methods generally require knowing more about the behavior of the
function around the interval in question. Let’s look at three alternative methods.
More on Functions 314

The Regula Falsi Root Solving Method

The Regula Falsi method is a refinement of the bisection approach. Instead of arbitrarily dividing
the interval x3–x1 in half, let’s interpolate where the line between f (x1) and f (x3) crosses the x-
axis.

Figure 7.5 The Regula Falsi Method

From Figure 7.5, we have the following relationship

And solving for x2,

Thus, all that must be done to change bisect into Regula Falsi is to alter how x2 is chosen.

Engr07c — Molar Volume of Non-Ideal Gases — Using Regula Falsi

Method
In this version of Regula Falsi, I added a slight change to better assist in detection of the ending
conditions. If by chance f1 or f3 ever becomes 0, then x1 or x3 is the precise answer. At the start
of each iteration, I check for this possibility and quit if true. All coding in the main() function
and the van() function remain the same, except for the new name of the function, falsi().
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr07c - Volume of Oxygen (van der Waal) using Regula Falsi Method *
More on Functions 315

/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr07c - Regula Falsi Method of root solving */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 // creates a data type for a parameter function value for bisect *
* 12 typedef double (*function) (double); *
* 13 *
* 14 bool falsi (double a, double b, double eps, int imax, *
* 15 double& root, int& i, function f); *
* 16 *
* 17 double van (double x); *
* 18 *
* 19 const double R = 0.082054; // ideal gas constant *
* 20 *
* 21 double p; // pressure in atmospheres *
* 22 double t; // temperature in K *
* 23 *
* 24 int main () { *
* 25 *
* 26 double a; // left x endpoint - root is between a and b*
* 27 double b; // right x endpoint *
* 28 double eps = .000001; // the desired degree of accuracy *
* 29 int imax = 1000; // maximum number of iterations to try *
* 30 *
* 31 int i; // number of iterations actually done *
* 32 double v; // the answer, the volume of gas *
* 33 *
* 34 // setup floating point output for 5 digits of accuracy *
* 35 cout << fixed << setprecision (5); *
* 38 *
* 39 cout << *
* 40 "Finding roots of van der Waal's gas equation for Oxygen\n\n";*
* 41 cout << *
* 42 "Enter the left endpoint, right endpoint, pressure in atm\n" *
* 43 << "and temperature in Kelvin - separated by blanks\n" *
* 44 << "Press ^Z to quit\n"; *
* 45 while (cin >> a >> b >> p >> t) { *
* 46 cout << endl; *
* 47 if (falsi (a, b, eps, imax, v, i, van)) { *
* 48 cout << "At pressure of: " << p << endl *
* 49 << "At temperature of: " << t << endl *
* 50 << "The volume is: " << v *
* 51 << " (" << i << " iterations)\n" *
* 52 << "Ideal Gas volume is: " << R * t / p << endl; *
* 53 } *
More on Functions 316

* 54 cout << endl; *

* 55 cout << "Enter another set or ^Z to quit\n"; *
* 56 } *
* 57 *
* 58 return 0; *
* 59 } *
* 60 *
* 61 /***************************************************************/*
* 62 /* */*
* 63 /* falsi: function to find the root of function f(x) */*
* 64 /* parameter function f must be of the form */*
* 65 /* double funct (double); */*
* 66 /* */*
* 67 /***************************************************************/*
* 68 *
* 69 bool falsi (double a, double b, double eps, int imax, *
* 70 double& root, int& i, function f) { *
* 71 double x1 = a; *
* 72 double x3 = b; *
* 73 double x2; *
* 74 double f1 = f (x1); *
* 75 double f2; *
* 76 double f3 = f (x3); *
* 77 *
* 79 double d = x3 - x1; // width of the current interval *
* 80 i = 0; // the current number of iterations *
* 81 *
* 82 // verify that there is a solution *
* 83 if (f1*f3 >= 0) { *
* 84 cerr << "Error: no root in the interval: " *
* 85 << a << " to " << b << endl *
* 86 << "The function may be miscoded or\n" *
* 87 << "the values of the end points are incorrect\n" *
* 88 << "F(a) = " << f1 << endl *
* 89 << "F(b) = " << f3 << endl; *
* 90 return false; *
* 91 } *
* 92 *
* 93 // find the root, but when the root is sufficiently accurate *
* 94 // or imax iterations have been done *
* 95 while (i < imax && d > eps) { *
* 96 // alternate done checks *
* 97 if (fabs (f1) < eps) { *
* 98 root = x1; *
* 99 return true; *
* 100 } *
* 101 if (fabs (f3) < eps) { *
* 102 root = x3; *
* 103 return true; *
* 104 } *
* 105 *
* 106 x2 = x1 - (x3 - x1) * f1 / (f3 - f1); *
More on Functions 317

* 107 f2 = f (x2); *
* 108 *
* 109 if (f1 * f2 < 0) { // is root in left half? *
* 110 x3 = x2; // yes, so move right end point leftwards *
* 111 f3 = f2; *
* 112 } *
* 113 else { // root in right half *
* 114 x1 = x2; // so move left end point rightwards *
* 115 f1 = f2; *
* 116 } *
* 117 *
* 118 // calculate new width *
* 119 d = (x3 - x1); *
* 120 i++; *
* 121 } *
* 122 *
* 123 // check if it converged or not - either way, display results *
* 124 if (i == imax) { *
* 125 cerr << "Warning: after " << imax *
* 126 << " iterations, Regula Falsi has not converged\n" *
* 127 << "Root thus far is: " << x2 << endl; *
* 128 root = x2; *
* 129 return false; *
* 130 } *
* 131 *
* 132 // here we have found a root, i is set to number of iterations *
* 133 root = x2; // store x2 in the caller's answer variable *
* 134 return true; *
* 135 } *
* 136 *
* 137 /***************************************************************/*
* 138 /* */*
* 139 /* van(x): van der waals gas equation */*
* 140 /* */*
* 141 /***************************************************************/*
* 142 *
* 143 double van (double v) { *
* 144 double a = 1.360; *
* 145 double b = 0.03183; *
* 146 double ansr; *
* 147 ansr = (p + a / (v * v)) * (v - b) - R * t; *
* 148 return ansr; *
* 149 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

The Regula Falsi method finds the roots much more quickly than bisection. Here are the
results from a similar test run as was done for the bisection version above.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Sample Run - Engr07c - Volume using Regula Falsi Method *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Finding roots of van der Waal's gas equation for Oxygen *
More on Functions 318

* 2 *
* 3 Enter the left endpoint, right endpoint, pressure in atm *
* 4 and temperature in Kelvin - separated by blanks *
* 5 Press ^Z to quit *
* 6 24 26 1 300 *
* 7 *
* 8 At pressure of: 1.00000 *
* 9 At temperature of: 300.00000 *
* 10 The volume is: 24.59280 (2 iterations) *
* 11 Ideal Gas volume is: 24.61620 *
* 12 *
* 13 Enter another set or ^Z to quit *
* 14 2.3 2.6 10 300 *
* 15 *
* 16 At pressure of: 10.00000 *
* 17 At temperature of: 300.00000 *
* 18 The volume is: 2.43840 (3 iterations) *
* 19 Ideal Gas volume is: 2.46162 *
* 20 *
* 21 Enter another set or ^Z to quit *
* 22 .2 .3 100 300 *
* 23 *
* 24 At pressure of: 100.00000 *
* 25 At temperature of: 300.00000 *
* 26 The volume is: 0.22636 (5 iterations) *
* 27 Ideal Gas volume is: 0.24616 *
* 28 *
* 29 Enter another set or ^Z to qui *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Newton’s Method of Root Solving

Newton’s method is another approach to finding the roots of an equation that yields an accurate
and fast solution. However, to enable it to obtain the more accurate root more quickly, more must
be known about the function and its behavior.

Newton’s method begins with an initial guess for the root, x0. A tangent line is passed
through the function at point f (x0). Where that tangent line crosses the x-axis becomes the new
guess of the root. This is shown in Figure 7.6.

The slope of the tangent line at point f (x0 ) is f', the first derivative of f (x) with respect to
x. Using the equation of a straight line, we have f (x0 ) + f'(x0) (x – x0 ) = 0. Solving for the new
guess, x, we get
x = x0 – f (x0) / f' (x0)

The actual mathematical derivation begins with rewriting the function as a Taylor series.
Assuming |x – x0 | is small, only the first two terms of the series are used.
More on Functions 319

f(x) = 0 = f (x0) + f' (x0) (x – x0)

Figure 7.6 Newton’s Method of Root Solving

The complete derivation is found in most numerical analysis texts. For our use, it is sufficient to
say that we are approximating the function by a straight line passing through f (x0) which has the
same slope as a tangent line through the point (x0 , f0).

Thus, Newton’s method requires an initial guess and the first derivative function instead
of the interval a to b. However, care must be exercised when using Newton’s method. A poor
initial guess can lead to finding the wrong root or worse. The next series of three figures
illustrates some of the results of making a poor initial guess.

Figure 7.7 Newton — Bad Initial Guess Figure 7.8 Newton — Poor Initial Guess
More on Functions 320

Figure 7.9 Newton — Poor Initial Guess

Another factor is needed for effective use of Newton’s method, the multiplicity factor.
Suppose that the function f(x) was this one.
f(x) = x3 – 5x2 + 7x –3 = (x – 3) (x – 1) (x – 1)
Here there are two identical roots at x = 1. A plot of this function around x=1 is shown in Figure
7.10.

Figure 7.10 Newton — Multiple Roots

Engr07d — Molar Volume of Non-Ideal Gases — Using Newton’s

Method
To use Newton’s method, the first derivative function must be found. It is

Another consideration, when coding an implementation of Newton’s method is divergence. If the

next increment to be added to x to get the next guess for the root is greater than some value, say
called divergeAmt, then the process is diverging. Quite often, that value divergeAmt is twice
the width of the interval in which a root might lie. In other words, commonly, the left endpoint,
More on Functions 321

a, is passed as the initial guess and 2*(b – a) is passed as the divergence value. If |delta x|, the
next amount to be added to the current guess to obtain the next guess at the root, is larger than
this divergence amount, then we should display a warning message. It is also quite likely that
within an iteration or two, Newton’s method will fail anyway.

This implementation uses the Do Until loop because always one iteration must be done
before checking upon ending conditions. The steps within the loop consist of
do {
calculate the value of dfdx (x)
if that value is sufficiently close to zero then
display an error message to avoid division by zero
calculate the next amount to add to x to get the next guess
if that next amount > diverge amount then
display a diverging warning message
root += that next amount
I++
} while (I<imax && the next amount > eps);

Here are the complete program and the sample test runs using the same three test cases of
oxygen.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr07d - Newton's Method to Find Roots of van der Wall's Equation *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr07d - Newton's Method of root solving */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 // creates a data type for a parameter function value for bisect *
* 12 typedef double (*function) (double); *
* 13 *
* 14 bool Newton (double x, int mult, double eps, int imax, *
* 15 double divergeAmt, double& root, int& i, *
* 16 function fx, function dfdx); *
* 17 *
* 18 double van (double x); *
* 19 double vandfdx (double x); *
* 20 *
* 21 const double R = 0.082054; // ideal gas constant *
* 22 *
* 23 double p; // pressure in atmospheres *
* 24 double t; // temperature in K *
* 25 *
More on Functions 322

* 26 int main () { *
* 27 *
* 28 double x; // initial guess for root, often a *
* 29 double divergeAmt; // maximum amount to add to get to next *
* 30 // guess, usually 2 * (b-a) *
* 31 int mult; // number of multiple roots at this root *
* 32 double eps = .000001; // the desired degree of accuracy *
* 33 int imax = 1000; // maximum number of iterations to try *
* 34 *
* 35 int i; // number of iterations actually done *
* 36 double v; // the answer, the volume of gas *
* 37 *
* 38 // setup floating point output for 5 digits of accuracy *
* 39 cout << fixed << setprecision (5); *
* 42 *
* 43 cout << *
* 44 "Finding roots of van der Waal's gas equation for Oxygen\n\n";*
* 45 cout << "Enter the initial guess, the multiplicity factor\n" *
* 46 << "the divergence check amount, the pressure in atm\n" *
* 47 << "and the temperature in Kelvin - separated by blanks\n" *
* 48 << "Press ^Z to quit\n"; *
* 49 while (cin >> x >> mult >> divergeAmt >> p >> t) { *
* 50 cout << endl; *
* 51 if (Newton (x, mult, eps, imax, divergeAmt, v,i,van,vandfdx)) {*
* 52 cout << "At pressure of: " << p << endl *
* 53 << "At temperature of: " << t << endl *
* 54 << "The volume is: " << v *
* 55 << " (" << i << " iterations)\n" *
* 56 << "Ideal Gas volume is: " << R * t / p << endl; *
* 57 } *
* 58 cout << endl; *
* 59 cout << "Enter another set or ^Z to quit\n"; *
* 60 } *
* 61 *
* 62 return 0; *
* 63 } *
* 64 *
* 65 /***************************************************************/*
* 66 /* */*
* 67 /* Newton: function to find the root of function f(x) */*
* 68 /* parameter function f must be of the form */*
* 69 /* double funct (double); */*
* 70 /* */*
* 71 /***************************************************************/*
* 72 *
* 73 bool Newton (double x, int mult, double eps, int imax, *
* 74 double divergeAmt, double& root, int& i, *
* 75 function fx, function dfdx) { *
* 76 double df; // the value of dfdx (x) *
* 77 double dx; // the next increment *
* 78 *
* 79 root = x; *
More on Functions 323

* 80 i = 0; *
* 81 do { *
* 82 df = dfdx (root); *
* 83 // avoid division by zero *
* 84 if (fabs (df) < .0000001) { *
* 85 cerr << "Error: fatal error in Newton.\n" *
* 86 << "The dfdx function returned a value nearly 0\n" *
* 87 << "dfdx was: " << df << " at x: " << root << endl *
* 88 << "On iteration number " << i << endl; *
* 89 return false; *
* 90 } *
* 91 // calculate the next increment to the root *
* 92 dx = - mult * fx (root) / df; *
* 93 // display warning if starting to diverge *
* 94 if (fabs (dx) > divergeAmt) *
* 95 cerr << "Warning: function is diverging.\n" *
* 96 << "The current delta x to be added to get the next guess\n"*
* 97 << "is greater than the specified diverge amount\n" *
* 98 << "|delta x|: " << fabs (dx) << " and diverge amount: " *
* 99 << divergeAmt << endl *
* 100 << "On iteration number " << i << endl; *
* 101 // obtain new guess for the root *
* 102 root += dx; *
* 103 i++; *
* 104 } while (i < imax && fabs (dx) > eps); *
* 105 *
* 106 // check and display a message if it did not converge in time *
* 107 if (i == imax) { *
* 108 cerr << "Warning: after " << imax *
* 109 << " iterations, Newton has not converged\n" *
* 110 << "Root thus far is: " << root << endl; *
* 111 return false; *
* 112 } *
* 113 *
* 114 // here we have found a root, i is set to number of iterations *
* 115 return true; *
* 116 } *
* 117 *
* 118 /***************************************************************/*
* 119 /* */*
* 120 /* van(x): van der waals gas equation */*
* 121 /* */*
* 122 /***************************************************************/*
* 123 *
* 124 double van (double v) { *
* 125 double a = 1.360; *
* 126 double b = 0.03183; *
* 127 double ansr; *
* 128 ansr = (p + a / (v * v)) * (v - b) - R * t; *
* 129 return ansr; *
* 130 } *
* 131 *
More on Functions 324

* 132 /***************************************************************/*
* 133 /* */*
* 134 /* vandfdx(x): van der waals gas equation -first deriv.function*/*
* 135 /* */*
* 136 /***************************************************************/*
* 137 *
* 138 double vandfdx (double v) { *
* 139 double a = 1.360; *
* 140 double b = 0.03183; *
* 141 double v2 = v * v; *
* 142 double v3 = v2 * v; *
* 143 double ansr; *
* 144 ansr = p - a / v2 + 2 * a * b / v3; *
* 145 return ansr; *
* 146 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Sample Run - Engr07d - Using Newton's Method to Find Roots *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Finding roots of van der Waal's gas equation for Oxygen *
* 2 *
* 3 Enter the initial guess, the multiplicity factor *
* 4 the divergence check amount, the pressure in atm *
* 5 and the temperature in Kelvin - separated by blanks *
* 6 Press ^Z to quit *
* 7 24 1 4 1 300 *
* 8 *
* 9 At pressure of: 1.00000 *
* 10 At temperature of: 300.00000 *
* 11 The volume is: 24.59280 (3 iterations) *
* 12 Ideal Gas volume is: 24.61620 *
* 13 *
* 14 Enter another set or ^Z to quit *
* 15 2.3 1 .6 10 300 *
* 16 *
* 17 At pressure of: 10.00000 *
* 18 At temperature of: 300.00000 *
* 19 The volume is: 2.43840 (3 iterations) *
* 20 Ideal Gas volume is: 2.46162 *
* 21 *
* 22 Enter another set or ^Z to quit *
* 23 .2 1 .2 100 300 *
* 24 *
* 25 At pressure of: 100.00000 *
* 26 At temperature of: 300.00000 *
* 27 The volume is: 0.22636 (3 iterations) *
* 28 Ideal Gas volume is: 0.24616 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
More on Functions 325

The Secant Method of Root Solving

The major problem with Newton’s method is the first derivative function. Sometimes the
derivative of f (x) is very hard to create. The Secant method is an alternative that does not require
that derivative function.

The secant method is passed an initial guess and a delta x to add to the initial guess to get
to the next guess. Often this translates into the left end point, a, and the width of the interval,
b – a, that has been used with bisection and Regula Falsi methods. Let’s call that delta x value
the correction to add to the guess to obtain the next guess at the root.

The process starts with the basic idea behind Newton’s method.

If delta xi is sufficiently small then an approximation for f'(xi) is given by

f'(xi) = ( f(xi-1) - f(xi) ) / (xi-1 - xi)
yielding

or simpler

Secant is given the initial guess for the root x and the correction to add to it to get to the
next x. Given those two values and the function f(x), we can calculate the next guess for the root.
Pictorially, secant method gets its name from the fact that what we are in effect doing is replacing
the f(x) with a line through the points (x0 , f0) and (x1 , f1) which is a secant line. This is shown in
Figure 7.11.

The secant method suffers from poor choices of the initial guess just as does Newton’s
method. If the secant line should ever become nearly horizontal, for example, the next delta x to
add to the guess is gargantuan, clearly diverging. Also any implementation must guard against
division by zero should f(x1) become 0, which indicates that x1 is the precise root.
More on Functions 326

Figure 7.11 The Secant Method

Engr07e — Molar Volume of Non-Ideal Gases — Using the Secant

Method
This version uses the secant method to find the roots. The function requires the initial guess x
and the delta x value to be added to the initial guess to get to the next guess. Usually, the left end
point, a, is passed as the guess and the width of the search interval, b – a, is passed for the delta x
value. Additionally, the function is passed a maximum amount for the delta x values that are
calculated. During any iteration, should the next delta x value ever exceed this maximum
amount, then secant displays a diverging message and terminates. Usually, that maximum
amount is set to two times the initial search interval.

Since there are four criteria to check during an iteration, I left the four checks within the
loop body. Before a new delta x value can be calculated, f1 must be checked for zero to avoid
division by zero errors. Also, if f1 is zero, we have landed exactly upon the root. If the next delta
x value is less than the error precision, we have found the root. If that delta x value is greater than
the maximum the user has passed, the diverging error message is displayed. Finally, if that new
delta x is twice the previous delta x, then a warning message is displayed about possible
divergence. It is very likely that such a warning message will be shortly followed by the failure
message when it diverges even more and goes beyond the maximum allowed. Here are the secant
program and test run.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr07e - Using the Secant Method to Find Roots *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr07e - Secant Method of root solving */*
* 4 /* */*
More on Functions 327

* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 // creates a data type for a parameter function value for secant *
* 12 typedef double (*function) (double); *
* 13 *
* 14 bool Secant (double x, double deltaX, double eps, int imax, *
* 15 double divergeAmt, double& root, int& i, function fx);*
* 16 *
* 17 double van (double x); *
* 18 *
* 19 const double R = 0.082054; // ideal gas constant *
* 20 *
* 21 double p; // pressure in atmospheres *
* 22 double t; // temperature in K *
* 23 *
* 24 int main () { *
* 25 *
* 26 double x; // initial guess for root, often a *
* 27 double divergeAmt; // maximum amount to add to get to next *
* 28 // guess, usually 2 * (b-a) *
* 29 double dx; // amount to add to x to get to next guess*
* 30 double eps = .000001; // the desired degree of accuracy *
* 31 int imax = 1000; // maximum number of iterations to try *
* 32 *
* 33 int i; // number of iterations actually done *
* 34 double v; // the answer, the volume of gas *
* 35 *
* 36 // setup floating point output for 5 digits of accuracy *
* 37 cout << fixed << setprecision (5); *
* 40 *
* 41 cout << *
* 42 "Finding roots of van der Waal's gas equation for Oxygen\n\n";*
* 43 cout << *
* 44 "Enter the initial guess, the amount to add to get to the\n" *
* 45 << "next guess, the divergence amount, pressure in atm\n" *
* 46 << "and the temperature in Kelvin - separated by blanks\n" *
* 47 << "Press ^Z to quit\n"; *
* 48 while (cin >> x >> dx >> divergeAmt >> p >> t) { *
* 49 cout << endl; *
* 50 if (Secant (x, dx, eps, imax, divergeAmt, v, i, van)) { *
* 51 cout << "At pressure of: " << p << endl *
* 52 << "At temperature of: " << t << endl *
* 53 << "The volume is: " << v *
* 54 << " (" << i << " iterations)\n" *
* 55 << "Ideal Gas volume is: " << R * t / p << endl; *
* 56 } *
* 57 cout << endl; *
* 58 cout << "Enter another set or ^Z to quit\n"; *
More on Functions 328

* 59 } *
* 60 *
* 61 return 0; *
* 62 } *
* 63 *
* 64 /***************************************************************/*
* 65 /* */*
* 66 /* Secant: function to find the root of function f(x) */*
* 67 /* parameter function f must be of the form */*
* 68 /* double funct (double); */*
* 69 /* */*
* 70 /***************************************************************/*
* 71 *
* 72 bool Secant (double x, double deltaX, double eps, int imax, *
* 73 double divergeAmt, double& root, int& i, function f) {*
* 74 double x0 = x; // *
* 75 double dx0 = deltaX; // the next increment *
* 76 double f0 = f (x0); *
* 77 double x1; *
* 78 double f1; *
* 79 double dx1; *
* 80 *
* 81 i = 0; *
* 82 while (i < imax) { *
* 83 x1 = x0 + dx0; *
* 84 f1 = f (x1); *
* 85 if (f1 == 0) { *
* 86 root = x1; *
* 87 return true; *
* 88 } *
* 89 // calculate the next increment to the root *
* 90 dx1 = dx0 / (f0/f1 -1); *
* 91 if (fabs (dx1) < eps) { *
* 92 root = x1 + dx1; *
* 93 return true; *
* 94 } *
* 95 // display an error message and abort if diverging *
* 96 else if (fabs (dx1) > divergeAmt) { *
* 97 cerr << "Error: function is diverging.\n" *
* 98 << "The current delta x to be added to get the next guess\n"*
* 99 << "is greater than the specified diverge amount\n" *
* 100 << "|delta x|: " << fabs (dx1) << " and diverge amount: " *
* 101 << divergeAmt << endl *
* 102 << "On iteration number " << i << endl; *
* 103 return false; *
* 104 } *
* 105 else if (fabs (dx1) > 2 * fabs (dx0)) { *
* 106 cerr << "Warning: function is diverging.\n" *
* 107 << "The current delta x to be added to get the next guess\n"*
* 108 << "is greater than twice the previous amount\n" *
* 109 << "|delta x|: " << fabs(dx1) <<" and the previous amount: "*
* 110 << dx0 << endl *
More on Functions 329

* 111 << "On iteration number " << i << endl; *

* 112 } *
* 113 x0 = x1; *
* 114 dx0 = dx1; *
* 115 f0 = f1; *
* 116 i++; *
* 117 } *
* 118 // display a message that it did not converge in time *
* 119 root = x0 + dx0; *
* 120 cerr << "Warning: after " << imax *
* 121 << " iterations, Secant has not converged\n" *
* 122 << "Root thus far is: " << root << endl; *
* 123 return false; *
* 124 } *
* 125 *
* 126 /***************************************************************/*
* 127 /* */*
* 128 /* van(x): van der waals gas equation */*
* 129 /* */*
* 130 /***************************************************************/*
* 131 *
* 132 double van (double v) { *
* 133 double a = 1.360; *
* 134 double b = 0.03183; *
* 135 double ansr; *
* 136 ansr = (p + a / (v * v)) * (v - b) - R * t; *
* 137 return ansr; *
* 138 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Test Run - Engr07e - Using the Secant Method to Find Roots *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Finding roots of van der Waal's gas equation for Oxygen *
* 2 *
* 3 Enter the initial guess, the amount to add to get to the *
* 4 next guess, the divergence amount, pressure in atm *
* 5 and the temperature in Kelvin - separated by blanks *
* 6 Press ^Z to quit *
* 7 24 4 8 1 300 *
* 8 *
* 9 At pressure of: 1.00000 *
* 10 At temperature of: 300.00000 *
* 11 The volume is: 24.59280 (2 iterations) *
* 12 Ideal Gas volume is: 24.61620 *
* 13 *
* 14 Enter another set or ^Z to quit *
* 15 2.3 .6 1.2 10 300 *
* 16 *
* 17 At pressure of: 10.00000 *
* 18 At temperature of: 300.00000 *
* 19 The volume is: 2.43840 (3 iterations) *
* 20 Ideal Gas volume is: 2.46162 *
More on Functions 330

* 21 *
* 22 Enter another set or ^Z to quit *
* 23 .2 .2 .4 100 300 *
* 24 *
* 25 At pressure of: 100.00000 *
* 26 At temperature of: 300.00000 *
* 27 The volume is: 0.22636 (3 iterations) *
* 28 Ideal Gas volume is: 0.24616 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Summary of Root Solving Techniques

Table 7.2 shows a summary of the results of the different root solving methods. Notice
that each method yields the correct results. They differ in how many iterations it took to arrive at
the result.
1 atm @ 300 10 atm @ 300 100 atm @ 300
Ideal Gas 24.61620 2.46162 0.24616
Bisection 24.59280 20 iterations 2.43840 18 iterations 0.22636 16 iterations
Falsi 24.59280 2 iterations 2.43840 3 iterations 0.22636 5 iterations
Newton 24.59280 3 iterations 2.43840 3 iterations 0.22636 3 iterations
Secant 24.59280 2 iterations 2.43840 3 iterations 0.22636 3 iterations
Table 7.2 Results of the Root Solving Methods

All methods require the f(x) function, the degree of accuracy and the number of iterations
to attempt. The more advanced methods require greater knowledge of the behavior of the
function. The Bisection and Regula Falsi Methods require only the left and right end points
enclosing a root. The Secant Method requires an initial guess for the root and an interval to add
to get the next guess at the root. Newton’s Method requires an initial guess, an interval, the
number of roots at this location and the first derivative function. Secant and Newton’s Methods
are both very sensitive to the initial guess; a poor choice can yield a total miss of the root.

Looking over the results and the function requirements, it is easy to see why the Secant
Method is one of the most popular methods for root solving applications.
More on Functions 331

New Syntax Summary

Reference Variables
The compiler passes the memory address of the caller’s variable so that the function can
changed the data within the caller’s variable.
void Fun (int& count, double& total) {
count = 42;
total = 88;
}
Here Fun can store a value in the caller’s passed variables for count and total. The caller
may invoke Fun this way.
Fun (tally, sum);
and tally now contains 42 while sum holds 88

All I/O streams must be passed by reference.

void Fun (istream& is, ostream& os);
and main calls it
Fun (cin, outfile);
Fun (infile, cout);

The parameters, whenever possible should be istream& and ostream& instead of

ifstream& and ofstream& so that the caller gains maximum flexibility.

Storage Where Scope Lifetime Initialization

Class Located
Automatic on the from the point of definition to the end of the run time at
stack to the end of the defining defining block of block entry
block of code code
Parameter on the from the point of definition to the end of the run time when
stack in the function header to the function function is
end of the function invoked
Static in the from the point of definition duration of whole one time at
data to the end of the defining program compile time
segment block of code
Global in the from the point of definition duration of whole one time at
data to the end of the entire program compile time
segment program
More on Functions 332

Design Exercises
1. Design an function called inputData(). From the passed input file stream, it is to extract a set
of data. The data include the employee Id number (9 digits long), the employee’s age in years,
and the date that the employee was hired. The hired date is input in the form of mm:dd:yyyy.
However, if the year is less than 100, it is not “year 2000 compliant.” In that case, add 1900 to
the year. The user expects to use the function as follows.
while (inputData (infile, id, age, month, day, year)) {

2. Design a function called circleProperties(). It is passed the circle’s radius. It must calculate
the area of the circle and the circumference of the circle. However, if the radius is within .001
from being 0, then return false along with zeros for the two properties. If the radius is not 0, then
return true. The calling function expects to use it similar to this.
if (circleProperties (radius, circum, area)) {

Stop! Do These Exercises Before Programming

1. Consider the following program to input a date in the format of mm-dd-yyyy, such as 10-22-
2001. What happens if the user inputs by accident 10 22-2001? What happens if the user
enters 10-22 2001? What happens if the user enters 10a22a2001?
int month, day, year, c;
cin >> month >> c >> day >> c >> year;

What would be the results of these three user entries if the input operation had been coded this
way?
cin >> month;
cin.get (c);
cin >> day;
cin.get (c);
cin >> year;

2. Why does this program create a linker error when building the program? How should it be
fixed?
#include <iostream>
using namespace std;

int sumFunction (int& counter);

int main () {
int x = 42;
int sum = sumFunction (x);
More on Functions 333

return 0;
}

int sumFunction (int counter) {

int j;
int sum = 0;
for (j=0; j<counter; j++)
sum += j;
return sum;
}

3. What is wrong with the first if statement? How can it be made to work correctly so that result
contains the correct value, ignoring leap year troubles?
#include <iostream>
using namespace std;
const int ERROR = -1;
int validateDate (int month, int day, int year);
int main () {
int mon = 10;
int day = 32;
int yr = 2000;
int result;
if (result = validateDate (mon, day, yr) == ERROR)
cout << "Bad Date\n"
...
int validateDate (int m, int d, int y) {
if (m < 1 || m > 12 || d < 1 || d > 31 || y < 1900)
return ERROR;
else
return 1;
}

4. When this program is executed, why do very unpredictable actions occur? How can it be fixed
so that it works properly?
#include <iostream>
using namespace std;

void inputDate (iostream in, int& month, int& day,

int& year);
int main () {
int month, day, year, quantity;
double cost;
inputDate (cin, month, day, year);
cin >> quantity >> cost;
...
More on Functions 334

void inputDate (iostream in, int& month, int& day,

int& year) {
char c;
in >> month;
in.get (c);
in >> day;
in.get (c);
in >> year;
}

5. When this program is executed, why do very unpredictable actions occur? How can it be fixed
so that it works properly?
#include <iostream>
using namespace std;

iostream inputDate (iostream& in, int& month, int& day,

int& year);
int main () {
int month, day, year, quantity;
double cost;
while (inputDate (cin, month, day, year)) {
cin >> quantity >> cost;
...
iostream inputDate (iostream& in, int& month, int& day,
int& year) {
char c;
in >> month;
in.get (c);
in >> day;
in.get (c);
in >> year;
return in;
}

6. When this program runs, why does main()’s cout produce erroneous results? How can it be
fixed?
#include <iostream>
using namespace std;

void inputDate ();

int month, day, year;
int main () {
int month, day, year;
inputDate ();
cout << month << '-' << day << '-' << year;
More on Functions 335

...
void inputDate () {
char c;
in >> month >> c >> day >> c >> year;
}

7. When this program runs, why does main()’s cout produce erroneous results? How can it be
fixed?
#include <iostream>
using namespace std;

void inputDate ();

int month, day, year;
int main () {
int month, day, year;
inputDate ();
cout << month << '-' << day << '-' << year;
...
void inputDate () {
int month, day, year;
char c;
in >> month >> c >> day >> c >> year;
}

8. This outputDate() function is badly designed. Why? How can it be repaired?

#include <iostream>
#include <iomanip>
using namespace std;

ostream& outputDate (ostream& out, int& m, int& d, int& y);

int main () {
int month, day, year;
...
outputDate (cout, month, day, year);
...
}

ostream& outputDate (ostream& out, int& m, int& d, int& y){

out << setw (2) << setfill ('0') << m << '-'
<< setw (2) << d << '-' << setw (4) << y
<< setfill (' ');
return out;
}
More on Functions 336

9. When this program is run, a strange, often large, number appears after the correct date on the
screen. Why? How can it be removed?
#include <iostream>
#include <iomanip>
using namespace std;

ostream& outputDate (ostream& out, int& m, int& d, int& y);

int main () {
int month, day, year, quantity;
...
cout << outputDate (cout, month, day, year) << " "
<< setw (10) << quantity;
...
}

ostream& outputDate (ostream& out, int& m, int& d, int& y){

out << setw (2) << setfill ('0') << m << '-'
<< setw (2) << d << '-' << setw (4) << y
<< setfill (' ');
return out;
}

10. What will cout display for the variable c? for b? for a?
#include <iostream>
#include <iomanip>
using namespace std;

int fun (int a, int &b);

int main () {
int a = 1;
int b = 2;
int c = 3;
c = fun (a, b);
cout << c << " " << b << " " << a << endl;
return 0;
}

int fun (int a, int &b) {

a = 42;
b = 42;
return 42;
}
More on Functions 337

11. What are the contents of main()’s variables a and b after the first call to fun()? What are the
contents of a and b after the second call to fun()?
#include <iostream>
using namespace std;

void fun (double &x, double y);

int main() {
double a = 1, b = 2;
fun (a, b);
fun (b, a);
}

void fun (double &x, double y) {

x += y;
return;
}

12. What values are printed for x, y and z in some_fun() the first time that function is called?
What values are printed the second time it is called?
#include <iostream>
#include <iomanip>
using namespace std;

void some_fun ();

int x = 1;

int main(){
int y = 2;
static int z = 3;
some_fun();
x +=10;
y +=10;
z +=10;
some_fun();
return 0;
}

void some_fun () {
int y = 2;
static int z = 3;
cout << x << " " << y << " " << z << endl;
x++;
y++;
z++;
}
More on Functions 338

Programming Problems
Problem Cs07-1 — The Grande Pumpkin Shoot (Game Design)
In fall, amid our pumpkin festivals, pumpkin tossing contests are held. Each contestant spends
the year designing the best contraption to lob pumpkins the farthest. This year, the target is set up
at a specific distance from the shooters. Each contestant gets up to five attempts to hit the target
with a pumpkin toss. There are two variables that the contestants can change in order to hit the
target: the initial velocity of the launch and the angle of the launch.

The distance, dist, to the landing point of a fired pumpkin is given by

Note that we are ignoring air friction effects and other aerodynamics of flying pumpkins.
The angle is in radians which is given by
radians = degrees * PI / 180

Write a program that first, one time only, inputs the distance to the target. Then, notify the
user that he or she has five attempts to hit the target and to enter a pair of values for the velocity
and the angle in degrees. All distances are in feet.

For each set of input data, the main() function calls a function, toRadians(), that returns
the angle converted into radians. Then, the main() function calls another function,
computeDistance(), that calculates both the distance this shot travels and the difference between
this distance and the target’s distance. Next, the main() function displays the distance this shot
traveled and by how much it missed the target. If the absolute value of the distance traveled
minus the target distance is less than .1% of the target distance, display a message that this shot
hit the target and that they win the game; then quit the program. If it misses the target, prompt for
another set of initial velocity and angle values. Repeat the process five times. If they have not hit
the target after five tries, display a message that they have lost and quit.

Problem Cs07-2 — Marketing Sales Summary (Control Break

Processing)
Acme Telemarketing Sales has produced a summary sales information file. Each line in the file
contains the telemarketer’s id number and the total sales they made on that transaction. The file,
summary.txt, contains all of the summarized data for one week and has been sorted into id
number order.

Write a program that produces the marketing sales summary report. The report should be
More on Functions 339

laid out as follows.

Acme Marketing Sales Summary Report Page 1

Id Number Weekly
Number of Sales Sales

123345 10 $ 1234.00
...
---- ---------
999 $ 99999.99
The program should accumulate a count of the number of sales and the total weekly sales for a
specific id. When the id changes, display the summary line as shown above. At the end of the
file, display a dashed line and the total number of sales for the company and the grand weekly
sales.

The program should count the number of lines displayed and allow for a maximum of 10
lines per page (very small paper). You should have at least a headings() function and a
doSalesBreak() function. Test the program with the provided data file, summary.txt.

Problem Cs07-3 — A swap() Function

Write a function called swap() that swaps the contents of two doubles. The function prototype is
void swap (double& x, double& y);
Then, write a main() function that prompts the user to enter two numbers, storing them in x and
y. It then displays these original contents of x and y. The main() function then calls swap() and
displays the new contents of x and y when it returns. Do not define any global variables. The
display should appear similar to the following. (Note numbers are shown with two decimal
digits.)
Before swap. x = 999.99
y = 111.11
After swap. x = 111.11
y = 999.99

Problem Cs07-4 — A sortTwo() Function

Write a function called sortTwo() that is passed two doubles, x and y. The function places the
smaller of the two into x and the larger into y. Use the swap() function written in Problem
Cs07-3 within sortTwo() if the contents of x and y need to be switched. The prototype of the
sortTwo() function is
void sortTwo (double& x, double& y);

Then write a main() function to input a pair of values for x and y. Display the original
More on Functions 340

contents of x and y. Then, call the sortTwo() function and redisplay the values of x and y when
the function returns. Repeat for additional pairs of numbers until the user presses ^Z. The display
results should appear as follows.
Before sorting: x = 999.99 y = 111.11
After sorting: x = 111.11 y = 999.99

Problem Cs07-5 — A sortThree() Function

Write a function called sortThree() that is passed three doubles, x, y and z. The function places
the smallest of the three values into x, the next smallest into y and the largest value into z. Use
three calls to the sortTwo() function written in Problem Cs07-4 within sortThree(). Do not
define any global variables. The prototype of the sortThree() function is
void sortThree (double& x, double& y, double& z);

Then write a main() function to input a trio of values for x, y and z. Display the original
contents of x, y and z. Then, call the sortThree() function and redisplay the values of x, y and z
when the function returns. Repeat for additional trios of numbers until the user presses ^Z. The
display results should appear as follows.
Before sorting: x = 999.99 y = 111.11 z = 88.99
After sorting: x = 88.99 y = 111.11 z = 999.99

Problem Cs07-6 — A quadratic() Function

Write a function called quadratic() that finds both roots of the quadratic equation. The main()
function prompts and inputs the values for a, b and c.

If a is zero, display an error message to cerr and return false. If the discriminant is
negative, display an error message to cerr and return false. Otherwise, calculate the two roots
and return true. The function prototype is
bool quadratic (double a, double b, double c,
double& root1, double& root2);

Next write a main() function that inputs a trio of values for a, b and c. Then, main() calls
quadratic(). If there are roots, main() displays the two roots. Finally, main() inputs another trio
of values until the user signals the end of file by pressing ^Z. Note that the only displaying that is
done within the quadratic() function is to cerr for error messages.
More on Functions 341

Problem Engr07-1 — Vibration Studies (Mechanical Engineering)

In the Midwest, after the spring thaw, pot holes form in many roads. When a car hits a pot hole,
the springs and shock absorbers get a workout dampening out the sudden jolt. Of course, if the
shock absorbers need replacing, the oscillations continue for a much longer time. The basic idea
is that the spring-shock absorber combination should rapidly dampen out that sudden jolt.

Given a car of mass m in grams, we can find the equation that defines the vertical motion
as a function of time, x (t). At any point in time, the forces operating on the mass of the car are
the resistance of the spring and the dampening force of the shock absorber. The spring force is -
kx, where k is the spring constant and it is negative indicating it is returning the car toward
equilibrium. The dampening force is given by -c x' where x' is the first derivative of x with
respect to time or the vertical velocity. Using Newton’s second law (f=ma), and using x'' for the
second derivative, the equation becomes
m x'' = -cx' + (-kx)
Rewriting it into a form that can be used for root solving,
x'' + c x' / m + k x / m = 0

This is a second order differential equation. If the car hits a hole in the road at time 0 and
is displaced x0 then the equation we need to solve is given below.

Assume that the spring constant k = 1.25E9 g/s/s and that the dampening constant c =
1.4E7 g/s. The mass and the initial depth of the pothole, m and x0, are input values from the user.

The program should ask the user to enter the weight of the car in pounds and the depth of
the pothole in inches. The program must convert those into grams and meters.

The program is to display the first three roots that represent the amount of time that the
car takes to get back to the equilibrium point. Figure 7.12 shows a graph of the equation using a
3000-pound car hitting a 10-inch pothole.
More on Functions 342

Figure 7.12 Shock Absorber Displacement as a Function of Time

The program should call bisect() three times to find the first three roots. Then, call falsi()
to find the same three roots and finally call secant() to find the same roots. The program should
display the three roots found by each of the three methods. Design the output to look something
like this. Test your program with a 3000-pound car hitting a 10-inch pothole.
Root 1 Root 2 Root 3
Bisect 0.05xxxx 11 0.15xxxx 10 .25xxxx 9
Falsi 0.05xxxx 11 0.15xxxx 10 .25xxxx 9
Secant 0.05xxxx 11 0.15xxxx 10 .25xxxx 9

Show six digits of accuracy and show the number of iterations the method required to find the
root.

Problem Engr07-2 — Electrical Circuit Design

(Electrical Engineering)
Consider a capacitor, resistor and an inductor in a circuit. If the capacitor alone is given some
initial charge from a power supply, and then that supply is cut off and the circuit is activated by
some switching mechanism, the circuit undergoes a series of oscillations until a new steady state
is reached. The length of the oscillations is related to the charge-storing properties of the
capacitor and the energy storage by the inductor; the resistance in the circuit dissipates the
magnitude of the oscillations. The problem to solve is to find that resistance required to dissipate
More on Functions 343

the energy in a given circuit at a specified rate.

The equations involved begin with the basics. The flow of current through a resistor creates a
voltage drop given by V = iR, where V is in volts, i is in amperes and R is in ohms. The inductor
voltage drop is given by V = Li' where L is the inductance in henrys and i' is the first derivative
of the current flow with respect to time. The voltage drop over the capacitor is V = q/C where C
is the capacitance in farads and q is the charge in coulombs. Kirchoff’s second law says that the
sum of the voltage drops in a closed circuit is zero. So we have
L i' + R i + q/C = 0

The current is related to the charge by i = q' where q' is the first derivative of q with
respect to time. The equation is really a second order differential equation. The calculus solution
then yields

In order to find a root, we need an equation f(R) = 0. Moving q(t) to the right side and dividing
by q0, we get our function.

Write a program that repeatedly prompts the user to input the four values: time t, L, C
and the dissipation ratio q/q0. Then use one of the root solving techniques to find the resistance
needed to make the circuit work. Repeat the process until the end of the file occurs, that is, the
user presses ^Z. Test your program with these test cases.
t (sec) L C q/q0 A B
Case 1: .05 5 H 1E-4 .01 0 400
Case 2: .05 5.1 H 1E-4 .01 0 400
Case 3: .05 8 H .5E-4 .05 0 400
Case 4: .1 10 H 1E-5 .1 0 400
If Case 3 or Case 4 should fail to find a root, explain what has occurred. Can you devise a better
method for determining the end point B. Hint, examine the contents of the square root term in the
equation.

Problem Engr07-3—Doppler Effects—Using Radar to Find a

Vehicle’s Speed
The radar guns used by the state police to monitor a vehicle’s speed uses the principles of
Doppler shift in frequency. The gun emits a beam at a given frequency f0. The beam bounces off
the moving vehicle which then adds or subtracts its speed from the frequency, yielding either an
More on Functions 344

increased or decreased frequency when measured by the radar gun. This same effect can be heard
with approaching sirens. The frequency increases in pitch as the vehicle approaches and then
decreases as it moves away from the observer. The Doppler shift formula is

where c is the speed of light, n is the index of refraction of radar microwaves in the air and f is
the final received frequency. The radar gun emits microwaves at a frequency f0 of 2E10 cycles
per second. The speed of light c is 2.99792458E8 meters per second. The index of refraction n is
1.00031. Finally, the radar gun’s manufacturer claims that the gun can measure an absolute value
of |f – f0| to one part in 104 or 0.01% accuracy.

The speed limit widely adopted is now just 55 miles an hour. An overzealous officer,
from your point of view, has been ticketing motorists whose speed is greater than or equal to
55.01 miles per hour. Your claim is that the gun is not sufficiently accurate to measure such a
small change in speed as 0.01 miles per hour.

To prove or disprove your claim, find the two frequencies, f55.00 and f55.01. Use one of the
root solving methods to find these frequencies. Use .0001 as the desired accuracy and let A be
.99 f0 and B be 1.01 f0. Note in order to create the velocity (f) function to solve, you must move
the velocity term to the other side of the equation.

Calculate the frequency differences df1 = |f55.00 – f0| and df2 = |f55.01 – f0|. Finally if
|df1 – df1| < .0001, then the radar gun is not sufficiently accurate. Display the two roots found
and the final |df1 – df2| result along with a message “pay the ticket” or “Judge dismisses the
case.”
Character Processing and Do Case 345

Chapter 8 — Character Processing and Do Case

Section A: Basic Theory

Introduction
This chapter begins with an in depth look at the processing of character data. Exactly how a
character is stored in memory is shown. Some of the more frequently used character processing
functions are discussed. Next, the Do Case decision making structure is presented as an
alternative to lengthy If-Then-Else constructs. Finally, enumerated data types are presented as a
way to better handle numerical quantities that are used in some kind of control situation.

The Processing of Character Data

This chapter examines the processing of single characters of data. Already we have used
character constants or literals such as ‘A’. Note that coding “A” is not the same; the double quote
marks indicate this is a string of characters. “A” is not the same thing as ‘A’. Character strings
are covered after the chapter on array processing because a string is stored as a series of
characters and includes a special byte to denote the end of the string.

Defining Variables to Hold a Character of Data

A character variable is often defined using the char data type. Here coding
char c;
defines c to be a variable capable of holding a single character of data. Similarly, one can define
other commonly used variables such as
char sex; // a person’s sex: M or F
char payType; // a person’s pay type: S or H
// salaried or hourly
char maritalStatus; // S single, M married, D divorced,
// W widowed
char grade; // a person’s grade in a course
Character Processing and Do Case 346

A character variable can be initialized when it is defined. Suppose that we wished to

define a variable to hold an employee’s pay type and initialize it to the letter H for hourly worker.
The following accomplishes this.
char payType = 'H';
Or one might wish to define a variable to hold a person’s grade and initialize it to the letter A.
char grade = 'A';

Inputting Character Data

There are two methods for inputting a character of data: the extraction operator >> and the get()
function. The effects of these two methods differ. Let’s begin with the extraction operator.

Using the Extraction Operator to Input a Character

A character of data can be input using the extraction operator; the rules parallel those for the
numerical data types. Assume that one has coded the following.
char grade;
cin >> grade;
The extraction operator first skips over white space to the first nonwhite space character. Then
that next nonwhite space character is input and stored in the grade variable. The input stream is
positioned to the next character in the input stream after the grade letter. If the end of file is
reached instead of a nonwhite space character, then the input action fails and the end of file flag
is turned on. Thus, the extraction of a character of data is no different from the extraction of a
numerical value.

Assume that one needs to input the item number, quantity, tax type and the cost values
from an input file. The tax type contains an F for food item and N for non-food item; it is used to
calculate the tax on the item purchased. The following could be the main loop coding.
long itemNumber;
int quantity;
char taxType;
double cost;
...
while (infile >>itemNumber >> quantity >> taxType >> cost) {
...
}
And an input line might look like this.
1234567 10 F 14.99<CR>
In other words, there are no surprises with the extraction operator.

However, please note that the input line could also have been written as follows.
1234567 10F14.99<CR>
Notice that there are no blanks or white space on either side of the tax type value F in this case.
Character Processing and Do Case 347

The data would still be inputted correctly. Remember that the extraction of an integer, quantity,
is ended by the detection of the F in the input stream; the 10 is input for the quantity. Since the
next item to be extracted is a character, the F is inputted into the taxType variable. However,
running the data together is not a good idea because a person would have difficulty reading that
line of input when doing a visual check of their data entry typing.

Running the data together does occur with both dates and times. For example as we have
already seen a time expressed as hours, minutes and seconds frequently contains a colon (:) to
separate the three values, 10:14:42. Dates expressed as three integers, month, day and year, might
appear as 10-01-2000. When dates are in this format, both the month and day integers have
leading zeros present if needed.

Inputting values in this format brings up an additional consideration. In chapter 6, we saw

that the default for input streams is to assume numbers beginning with a 0 digit are in the octal
number system. One time only we must call the dec manipulator function to notify the input
stream that numbers with leading 0's are in fact decimal numbers.
char c;
int month, day, year;
ifstream infile ("aFileOfDates.txt");
if (!infile) {
... output an error message and abort the program
}
infile >> dec;

If one is inputting dates that have leading 0’s before some month and day values, then the
input stream must have the decimal flag turned on for them to be properly inputted. Failure to do
so results in 08 and 09 are not being properly input.

Hexadecimal Numbers

As long as we are on the topic of number systems, the third number system in C++ is the
hexadecimal number system or base 16. In hex, valid digits range from 0 through F. There is the
system
hex decimal
0 0
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
Character Processing and Do Case 348

A 10
B 11
C 12
D 13
E 14
F 15

C++ uses a 0x identifier to signal that the following number is actually a hexadecimal
number: 0x20 defines a hexadecimal value of 32 in decimal. 0x10 defines a hex value of 16
decimal. 0x1F defines a decimal 31. Just as in the decimal system when you add 9 + 1 you get
10, in the hexadecimal system adding 0xF + 0x1 gives 0x10. Fortunately, hexadecimal values are
seldom needed in programming. They are found heavily in system programs, such as code that is
a part of the operating system programs.

Using the get() Function

The get() function represents an alternative method to input a character. Its operation differs
significantly from the extraction operator. The get() function inputs the next character no matter
what it might be. It only fails if the end of file is reached. The syntax is simple
cin.get (c);
or
infile.get (c);
Notice that it is a function call and is therefore not chainable as the extraction operator is. The
following is illegal.
infile >> month >> infile.get (c) >> day;
If one wanted to input the date using the get() function to input the dashes, the sequence would
be
infile >> month;
infile.get (c);
infile >> day;
infile.get (c);
infile >> year;
Clearly, the extraction operator is superior in this case, unless you are being paid for the number
of lines of coding you produce.

There are, however, some circumstances in which the use of the get() function is
mandatory. Suppose that you needed to input the answers to a multiple choice test. There are say
five questions whose answers are A through E. The following coding is a potential disaster for
the grading program.
char answer;
for (int j=0; j<5; j++) {
infile >> answer;
... use answer
}
Character Processing and Do Case 349

Suppose one student’s answer line was as follows.

ABCDE<CR>
Then the above coding would correctly input each question’s letter choice. However, what can
occur on a test? Sometimes a student omits a question for one reason or another. What would the
above coding input if the following two lines were entered?
B D <CR>
ABCDE<CR>
The first line shows that the student failed to answer the first and third and fifth question. But
what does the extraction code actually input for the five answers? B, D, A, B, and C! Remember
that the extraction operator skips over white space and a blank is white space. Here is a classic
opportunity in which the get() function must be used to properly input the data. The following
sequence is a proper way to input the five answers.
char answer;
char endOfLine;
for (int j=0; j<5; j++) {
infile.get (answer);
... use answer
}
infile.get (endOfLine);

Within the loop, five characters are input. Thus, the first character would contain the
blank, then the B, then the blank, then the D and then the blank. However, at this point, the input
stream is now pointing to the newline code, the <CR> carriage return and line feed bytes. If we
assume that the program soon will input the next student’s answers as represented by the next
line of input, that <CR> must be input. Failure to do so would mean that it would be input as the
first answer for the next student!

Hence, the following guideline might help you decide whether or not a get() function is
needed to input a character.

Rule: If a valid character to be input could contain white space, then the get() function
must be used. Otherwise, the extraction operator can be used.

Output of Character Data — the put() Function

The insertion operator can be used to output character data. The setw() function operates as usual
with the character being left aligned by default within the specified width. Assume a person’s
grade variable contains an ‘A’; the following code
cout >> setw (5) >> grade >> endl;
produces
bbbbA
where the b represents a blank.
Character Processing and Do Case 350

The put() function can also be used to output a character. It is given a character to output
and it outputs that character to the indicated stream. For example, the following code displays the
letter A. For more on the put() function, see the Plot() function below.
cout.put (grade);
outfile.put (grade); // where outfile = an ofstream instance

How Are Character Data Stored?

Character data are stored using their ASCII values. ASCII stands for the American Standard
Codes for Information Interchange. The basic idea behind the storing of character data is
“agreement.” For example, suppose that you are designing a computer system and wish to find a
way to store a character of data, such as the letter A. A byte contains 8 bits that are on or off, 1 or
0. The possible bit patterns of 1's and 0's range from
0000 0000
0000 0001
0000 0010
0000 0011
0000 0100
...
1111 1111
For fun, you might write out all the unique possibilities that there are in one byte. There are, in
fact, 256 possible unique bit patterns in a byte.

So to store the letter A, pick one, say for example

0100 0001
And then get everyone else to agree with you so that, when they encounter that bit pattern, they
display, print and so on a letter A. That is all that the ASCII scheme really is — an agreed upon
scheme of bit patterns to represent the letters. Here are the first few ASCII bit patterns for some
letters.
A 0100 0001
B 0100 0010
C 0100 0011
D 0100 0100
and so on.

Of course, with any agreed upon pattern, there are always those that do not agree. On
some computers, such as IBM mainframes, there is a completely different agreed upon scheme
called EBCDIC, Extended Binary Coded Decimal Interchange Codes. On an IBM mainframe, the
letter A is encoded as
A 1100 0001

Thus, when a PC is communicating with a mainframe computer, some form of translation

Character Processing and Do Case 351

must occur as they are speaking different languages, so to speak. This ASCII to EBCDIC
translation can be done using a hardware circuit board placed between the two computers or it
can be done using software.

Another interesting point is that the bit pattern holding the letter A is 0100 0001. That bit
pattern is also the decimal number 65 (1x20 + 1x26 = 1 + 64 = 65). Here is the complete ASCII
table for my PC running Windows NT, Table 8.1.

Table 8.1 The ASCII Codes

When IBM invented the very first PC, they completely filled up all possible ASCII table
values. The values corresponding to decimal numbers below 32 are known as the control codes.
For example, if you press ^C (Ctrl + C key), the actual decimal value input and stored is 3. The
Carriage Return code is 13 and the Line Feed code is a 10. Remember that C++ inputs both
bytes, 13 and 10, but it stores only the line feed code, the 10. On output of a new line code, C++
also outputs the CR or 13 first and then outputs the LF code of 10. The formfeed code, a 12,
causes a printer to eject to a new page.
Character Processing and Do Case 352

A blank or space is stored as a decimal 32. Above the blank are the numbers 0 through 9
stored as a 49 through 57. Beginning at a decimal 65 are the uppercase letters and the lowercase
letters begin at 97. Interspersed are the special characters, such as + – , $ % and the period.

All characters whose decimal values are above 127 are known as the upper ASCII codes
or the graphics codes. Some of these can be used to draw text boxes on the screen in either a
single line or double line version. Additionally, there are many foreign language characters stored
in the remaining values. In short, when the first PC appeared, all 256 possible values were filled
and used. Of course, this then led to a major problem when the 101 special keys were developed,
such as the arrow keys, function keys, insert, delete, page up and down for example. IBM had
used all the slots so some other scheme had to be devised to input these keys.

The Escape Sequences

The escape sequences allow us to enter some of these unusual codes by directly entering their
ASCII code numbers. Table 8.2 shows the complete escape sequences.

Table 8.2 The Escape Sequences

Escape ASCII Meaning
Sequence Code
\a 7 alert — sounds the bell making a beep sound
\b 8 backspace-rub out
\f 12 form feed — printer eject to new page
\n 10 newline code (a line feed)
\r 13 carriage return code
\t 9 tab — horizontal
\v 11 vertical tab
\\ 92 backslash
\’ 39 single quote mark
\” 34 double quote mark
\? 63 question mark
\nnn nnn the character whose code number is specified by this octal number
Character Processing and Do Case 353

\xnnn 0xnnn the character whose code number is specified by this hexadecimal
number

Sometimes these codes are needed. When creating the full path of a filename, the \\ is
used, C:\\Cmpsc125\\test.dat. When a double quote mark is needed within a string, the \" is used,
“He said, \"Hello.\".” When a character literal is to contain a single quote mark, the \' is used, ‘\'‘
creates a byte containing a single quote mark.

When the program needs to display an important error message, including anywhere in
the message string the \a code causes the PC to also beep to catch the user’s attention. For
example
cerr << "The data contains invalid codes!\a\n";

Numbers and Letters

The char data types, including char, signed char and unsigned char, cause much confusion for
everyone including the computer itself. Suppose that one had defined the following fields (recall
that char is short for signed char)
char letter;
char num1; // which is the same as a signed char num1;
unsigned char num2;
letter = 'A';
num1 = +65;
num2 = 65;

How are these three values, the letter A, the +65 and the unsigned 65, actually stored in
memory? All three variables occupy just one byte. Here are their bit patterns
letter 0100 0001
num1 0100 0001
num2 0100 0001
For signed numbers, the sign is the very first bit of the field; a 0 bit means + and a 1 bit means –.
For unsigned numbers the first bit is part of the number. Had the leftmost bit of num2, the
unsigned number, been a 1 bit, it would represent 1x27 or 128 added into the value represented
by the bits.

What is startling is that all three completely different values appear absolutely identical in
memory! In other words, if a byte contains 0100 0001, no PC can tell whether that value is the
letter A, a +65 or an unsigned 65! Therefore, the computer depends upon the context of the
instructions to give it a clue as to the meaning of the value.

When you use the extraction or insertion operators or the get() and put() functions, the
computer assumes that the value is to be the ASCII code or letter. Thus, if you write
cin >> letter;
and the user enters
Character Processing and Do Case 354

A
then the value stored is considered to be the ASCII value of the letter A, or the 65. This makes
sense if we are to input character data.

However, it can sometimes get a programmer into trouble. Consider this user entry.
1
Now what is stored in the variable letter? It stores the ASCII value for the character ‘1’ which is
a decimal 49. It does not store the numerical binary equivalent of 0000 0001! In other words, the
following two inputs give very different results.
char x;
int y;
cin >> x >> y;
where the user enters
1 1
The contents of x contain 0011 0001 while y contains 0000 0000 0000 0001 (Assuming an int
occupies 2-bytes.) The variable x holds the ASCII value of ‘1’ while the variable y holds the
numerical binary value of 1.

On output, if you want the computer to consider the contents of the char to be a
numerical value, use a typecast. I produced the ASCII table by
cout << (int) letter << " " << letter;
where letter is defined as a char. If you want the ASCII letter displayed from an integer type, use
a typecast.
cout << num1 << " " << (char) num1;

The computer takes its clue from your coding. If you perform any math operations on the
char variable, the computer assumes that the value is a numerical one. Thus, if we wrote
letter += 1;
when done, letter contains 0100 0010 or the letter ‘B’;

Occasionally in older coding, you may run across math operations on obviously character
data.
letter += 32;
letter -= 32;
If letter contained ‘A’, then adding 32 to the ASCII value of 65 yields 97 which is the ASCII
code for the letter ‘a’. Similarly, if the contents of letter was an ASCII 97, the second line above
produces a 65 or the letter ‘A’. These two lines are changing the case of a letter. However, these
two lines to change case are NOT platform independent. Rather, they only work correctly on a
computer that is using the ASCII encoding scheme. There are some built-in character processing
functions that properly change the case of a letter independent of the encoding scheme.

Suppose that the program needed to verify that a letter only was input. It could be
checked using some ranges.
if ( (letter >= 'A' && letter <= 'Z') ||
Character Processing and Do Case 355

(letter >= 'a' && letter <= 'z')) {

// here it is a letter
}
But there are also some character processing functions to do this for us that are not dependent
upon the data being encoded in ASCII.

The Character Processing Functions

All character processing function prototypes are in the header file <ctype.h> or in <cctype>
when using the namespace std. Some functions convert case independent of the platform. Some
report whether or not the character is a letter, number, and so on. Table 8.3 shows many of the
more frequently used functions.

Table 8.3 Some Commonly Used Character Processing Functions

function returns
toupper (c) returns the uppercase letter equivalent of the lowercase letter in c, otherwise
returns the character it was given
tolower (c) returns the lowercase letter equivalent of the uppercase letter in c, otherwise
returns the character it was given
isupper (c) returns non-zero value if c contains an uppercase letter; otherwise, returns 0
islower (c) returns non-zero value if c contains a lowercase letter; otherwise, returns 0
isalpha (c) returns non-zero value if c is a letter; otherwise, returns 0
isdigit (c) returns non-zero value if c contains a digit (0-9); otherwise, returns 0

Basic08a — A Word Counter Program

To illustrate these character processing functions as well as character I/O, let’s write a program
that inputs a text file and find and display each word. At the end of file, display a total of all
words found. By definition, a word consists of a series of letters. Any non-letter separates the
words.

In terms of design, let’s have two functions: skipToStartOfWord() and

getRestOfWord(). The skipToStartOfWord() is to input characters until it finds a letter; then it
returns the first letter of a new word. The getRestOfWord() inputs and displays each successive
letter until the end of the word is found. Figure 8.1 shows the main storage drawing for the three
functions.
Character Processing and Do Case 356

Figure 8.1 Main Storage for Word Counter

Here is the design sketch for the main() function.

define infile and open the file
if it fails, display error message and quit
set wordCount to 0
let letter = skipToStartOfWord (infile);
while (infile) do the following
display letter on a new line
wordCount++
getRestOfWord (infile);
let letter = skipToStartOfWord (infile);
end while
display wordCount
close infile

The sequence for skipToStartOfWord() that is passed infile is

infile.get (c)
while infile and c is not a letter do the following
infile.get (c)
end do while
if infile is good, return c
otherwise return 0

The sequence for getRestOfWord() that is passed infile is

infile.get (c)
while infile and c is a letter do the following
output c
infile.get (c)
end do while
output a newline code
Character Processing and Do Case 357

Here are the completed program and a sample test run.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic08a - displays words in a file and counts them *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /**************************************************************/ *
* 2 /* */ *
* 3 /* Basic08a: displays words in a file and counts them */ *
* 4 /* */ *
* 5 /**************************************************************/ *
* 6 *
* 7 #include <iostream> *
* 8 #include <fstream> *
* 9 #include <cctype> *
* 10 using namespace std; *
* 11 char skipToStartOfWord (istream& infile); *
* 12 void getRestOfWord (istream& infile); *
* 13 *
* 14 int main () { *
* 15 ifstream infile ("sample.txt"); *
* 16 if (!infile) { *
* 17 cerr << "Error: cannot open the input file sample.txt\n"; *
* 18 return 1; *
* 19 } *
* 20 *
* 21 char letter; *
* 22 long wordCount = 0; *
* 23 letter = skipToStartOfWord (infile); *
* 24 while (infile) { *
* 25 cout << letter; *
* 26 wordCount++; *
* 27 getRestOfWord(infile); *
* 28 letter = skipToStartOfWord (infile); *
* 29 } *
* 30 cout << endl << "Total Words: " << wordCount << endl; *
* 31 infile.close(); *
* 32 return 0; *
* 33 } *
* 34 *
* 35 /**************************************************************/ *
* 36 /* */ *
* 37 /* skipToStartOfWord: skips over non-letters until it */ *
* 38 /* finds 1st letter */ *
* 39 /* */ *
* 40 /**************************************************************/ *
* 41 *
* 42 char skipToStartOfWord (istream& infile) { *
* 43 char c; *
* 44 infile.get (c); *
* 45 while (infile && !isalpha (c)) { *
* 46 infile.get (c); *
* 47 } *
* 48 return infile ? c : 0; *
Character Processing and Do Case 358

* 49 } *
* 50 *
* 51 /**************************************************************/ *
* 52 /* */ *
* 53 /* getRestOfWord: get and display rest of a word */ *
* 54 /* */ *
* 55 /**************************************************************/ *
* 56 *
* 57 void getRestOfWord (istream& infile) { *
* 58 char c; *
* 59 infile.get (c); *
* 60 while (infile && isalpha (c)) { *
* 61 cout << c; *
* 62 infile.get (c); *
* 63 } *
* 64 cout << endl; *
* 65 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Basic08a - displays words in a file and counts them *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Define *
* 2 infile *
* 3 and *
* 4 open *
* 5 the *
* 6 file *
* 7 if *
* 8 it *
* 9 fails *
* 10 display *
* 11 error *
* 12 message *
...
* 110 end *
* 111 do *
* 112 while *
* 113 output *
* 114 a *
* 115 newline *
* 116 code *
* 117 *
* 118 Total Words: 116 *
* 119 Press any key to continue *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Character Processing and Do Case 359

The Do Case Structure

Sometimes a variable must be tested for a number of possible values and alternative processing
done for each possibility. For example, assume that a company has five departments numbered 1
through 5. The cost of supplies must be charged back to the specific department. Thus, such a
program would have five department total variables and the charge back coding might be as
follows.
if (deptno == 1) {
total1 += cost;
}
else if (deptno == 2) {
total2 += cost;
}
else if (deptno == 3) {
total3 += cost;
}
else if (deptno == 4) {
total4 += cost;
}
else if (deptno == 5) {
total5 += cost;
}
else {
cerr << "Error: incorrect department number. It was: "
<< deptno << endl;
}
Such decision making logic works fine but is cumbersome when the number of possibilities is
large. The Do Case structure represents a convenience method to shorten this.

The circumstances that allow a Do Case structure to be used are twofold: one variable,
deptno, is being checked for values and that variable is an integer. Figure 8.2 shows the industry
standard Do Case structure. However, the C++ default implementation does not follow that
standard but introduces more flexibility. Figure 8.3 shows the C++ Do Case structure.

Notice that by default, when the special processing for a specific case is completed, the
C++Do Case falls into the next case! When the special processing for a specific case is
completed, the normal industry Do Case leaves and goes to the next instruction after the Do
Case. C++ introduces a break statement to enable a program to break out of a loop or a Do Case
structure so that we can implement the industry normal Do Case if desired. Normally, that is
precisely what is desired. When we are finished with the special processing for a given case, we
do not wish to fall through and do all the other cases’ special processing.
Character Processing and Do Case 360

Figure 8.2 Industry Do Case Figure 8.3 C++ Do Case

The Do Case syntax is

switch (integer variable or expression) {
case value1:
special processing when the variable contains value1
break;
case value2:
special processing when the variable contains value2
break;
...
case valuen:
special processing when the variable contains valuen
break;
default:
special processing when the variable is none of these
break;
}

Returning to the opening situation of five departments and their charge back totals,
assuming that deptno was defined to be an integer type, then the following represents the Do
Case solution. Notice it simplifies all the If-Then-Else statements.
switch (deptno) {
case 1:
total1 += cost;
Character Processing and Do Case 361

break;
case 2:
total2 += cost;
break;
case 3:
total3 += cost;
break;
case 4:
total4 += cost;
break;
case 5:
total5 += cost;
break;
default:
cerr << "Error: incorrect department number. It was: "
<< deptno << endl;
}

When the switch statement is encountered, the compiler evaluates the contents of
deptno. Depending upon the current contents of deptno, it then goes to the corresponding case.
So if deptno contains a 4, then case 4 is executed adding the cost to total4 and then the break
causes it to leave the entire Do Case statement.

Here is another example. Suppose that the program needed to display the day of the week.
Assume that variable day is an integer type. The following shows how this can easily be done
using a Do Case.
cin >> day;
switch (day) {
case 1:
cout << "Sunday" << endl;
break;
case 2:
cout << "Monday" << endl;
break;
case 3:
cout << "Tuesday" << endl;
break;
case 4:
cout << "Wednesday" << endl;
break;
case 5:
cout << "Thursday" << endl;
break;
case 6:
cout << "Friday" << endl;
break;
case 7:
Character Processing and Do Case 362

cout << "Saturday" << endl;

break;
default:
cerr << "Error: incorrect day: must be 1-7. It was: "
<< day << endl;
}

Now consider this version of the program. What is the output when the user enters a 4 for
the day?
cin >> day;
switch (day) {
case 1:
cout << "Sunday" << endl;
case 2:
cout << "Monday" << endl;
case 3:
cout << "Tuesday" << endl;
case 4:
cout << "Wednesday" << endl;
case 5:
cout << "Thursday" << endl;
case 6:
cout << "Friday" << endl;
case 7:
cout << "Saturday" << endl;
default:
cerr << "Error: incorrect day: must be 1-7. It was: "
<< day << endl;
}
It produces
Wednesday
Thursday
Friday
Saturday
Error: incorrect day: must be 1-7. It was 4
This is because the break; statements were left out. When it finishes one case, it falls on down
into the next case’s set of things to do.

The switch can be on any integer type of variable (char, short, int or long) and even an
expression that results in an integer value. It can also be used on ASCII characters since they are
stored in a char type. The values used in the case statements do not need to be in increasing or
even sequential order, but often are for readability. The values can also be ASCII characters.

Suppose that the tax on an item is based upon whether or not the item is classified as a
food item or a non-food item. Assume that the following variables have been defined and values
inputted for a specific purchase.
Character Processing and Do Case 363

char taxtype; // F or N
short quantity;
double cost;
double taxrate;
double tax;
The following switch statement obtains the correct tax.
switch (taxtype) {
case 'F':
taxrate = .01;
break;
case 'N':
taxrate = .07;
break;
default:
taxrate = 0;
}
tax = quantity * cost * taxrate;

When would the C++ fall through and do the next case’s statements ever be useful? That
is, are there any circumstances when break statements are not desired? Yes, there can be some
situations where the basic behavior is desirable. Suppose that we needed to write a
daysInMonth() function that returned the number of days in a month. Notice how streamlined
the coding can be by making use of switch statement and the default C++ behavior.
int daysInMonth (int month) {
int days;
switch (month) {
case 1:
case 3:
case 5:
case 7:
case 8:
case 10:
case 12:
days = 31;
break;
case 2:
days = 28;
break;
default:
days = 30;
}
return days;
}

One can nest additional switch statements within a case clause as long as that inner Do
Case structure is entirely contained in the outer case clause. However, no new variables can be
defined within a Do Case statement.
Character Processing and Do Case 364

The Do Case structure is entirely a matter of convenience. A series of If-Then-Elses can

accomplish the same task.

More on the break Statement and the continue Statement

The break statement has additional uses. It causes control to leave the current block of coding.
Suppose that we wish to write a “Guess the Letter” game in which the user tries to guess which
letter you have chosen. The main loop centers on the input of the next guess, repeating while the
guess is not correct. The user gets 10 tries. One way to write such a loop is as follows.
char theLetter; // the original letter to be guessed
char guess;
int tries = 0;
cout << "Guess my letter. Enter a letter: ";
cin >> guess;
while (cin && tries < 10) {
if (guess == theLetter)
break;
tries++;
cout << "Try again. Enter a letter: ";
cin >> guess;
}
if (cin) { // if not eof or user giving up
if (tries == 10)
cout << "You lose\n";
else
cout << "You win in " << tries << " attempts\n";
}

Here, if the user’s guess ever equals the letter, the break statement terminates the loop.
One could have just as easily written the while clause as
while (cin && tries < 10 && guess != theLetter) {
And this would have been a much clearer way to have written it. The break statement does have
valid uses in later chapters in which the alternate loop termination condition does not fit
particularly well in a while clause.

The continue statement causes control to go to the bottom of the loop. If it is a for loop,
then the bump expression(s) is(are) activated. The continue statement provides a means of
skipping the body of a loop on a specific iteration, yet not terminate the loop. Suppose that a
sumNumber() function is passed a number between one and one hundred. The function is to
return the sum of all integers between one and one hundred but not including the passed number.
Here is a good use of the continue statement.
long sumNumber (long number) {
long sum = 0;
long j;
Character Processing and Do Case 365

for (j=1; j<101; j++) {

if (j == number)
continue; // bypass adding this j to sum; go do next j
sum += j;
}
return sum;
}

Enumerated Data Types

Many integer variables hold key values that a program checks within while, If-Then-Else and Do
Case statements. Take the department number example in which the company has five that are
identified by values ranging from one to five. Throughout a program, one might see these
numbers appearing for various processing needs. However, these abstract numbers bear little
relation to what they represent. For example, it is not obvious to anyone that department number
1 represents the Appliance department; 2, the Automotive department and so on. To make the
program clearer, one can add comments such as these.
case 1: // appliance department
if (deptno == 2) { // the automotive department

A better approach would be to make a set of #define symbols or constant ints such as
these two versions.
#define Appliances 1
#define Automotive 2
...
case Appliances:
if (deptno == Automotive) {
or
const int Appliances = 1;
const int Automotive = 2;
...
case Appliances:
if (deptno == Automotive) {

The #define creates a symbolic name that represents the value that comes after the name.
It is a preprocessor directive. The preprocessor handles all the lines that begin with a # sign. It
includes the various header files. In the case of #defines, it does a text substitution. That is, the
preprocessor looks for all occurrences of Appliances within the program and replaces that
symbol with its corresponding value, 1. Because the #define is just a simple text substitution, the
const int approach is preferred because the Appliances is actually an instance of the integer data
type.
Please note that although Appliances is a “variable” it really is a constant integer. The
values on the case statements cannot be variables. That is, the following is invalid because in this
case appliances is a variable, not a constant.
Character Processing and Do Case 366

int appliances = 1;
...
case appliances: // error not a constant value

For many uses, using either a const int or a #define to create a more meaningful name is
extremely important for readability. However, when there are more than just a few possibilities,
errors can occur with this approach. What would result if the user coded the following and then
used these as the values in a Do Case statement?
const int Appliances = 1;
const int Automotive = 2;
const int Housewares = 3;
const int Toys = 3;
const int Accounting = 5;
Because Toys was accidentally given the wrong value, incorrect results will occur that are
difficult to find. There is a better way.

An enumerated data type allows us to create a “new” data type and specify precisely
what possible values that variables of this new type can contain. The syntax is
enum new_data_type {enumerator-1, enumerator-2,
..., enumerator-n };
where the form of the enumerators is
identifier = integer value

Usually the new data type is capitalized as are the enumerator identifiers. If the integer
values are not coded, the first one is given the value 0, the next one is given a value of 1, and so
on. Specifically, to handle these department numbers, an enum is a terrific way to proceed.
Consider this version.
enum DeptNum {Invalid, Appliances, Automotive,
Housewares, Toys, Accounting};
This creates a new data type known as DeptNum. Variables of this type can only have the six
indicated values represented by Invalid, Appliances and so on.

Actually, an enum is implemented as an int, so you can look upon an enum as a

“disguised int.” By default, the first value the enum can hold is given the value 0. Thus, in the
above definition, Invalid represents a 0; Appliances, a 1; Automotive, a 2; and so on. That is,
the compiler from the point of definition of this enum now knows to replace all occurrences of
Appliances with its value 1.

Where are most all enum definitions placed within a program? They, like #defines and
const ints, must be defined before their first use. They are usually placed after the #includes and
before the prototypes of our functions.

How do we create an instance of a DeptNum? One is created precisely the same way we
create an instance of a double or a long. The following creates several instances of the enum
Character Processing and Do Case 367

DeptNum.
DeptNum deptnum;
DeptNum previousDeptNum;
DeptNum x;

Given these new variables, then the following are valid.

if (deptnum == Appliances)
if (x == Automotive)
case Toys:
But these are invalid because specific integer values are being used.
if (deptnum == 1)
if (x == 2)
case 3:
An enum can be passed to a function and one can be returned by a function. The
prototypes require a data type and that is simple to code.
DeptNum someFunction (DeptNum z);
In other words, we use our new enum just as if it were any other intrinsic or built-in data type.

Here is the total view of how the enum fits into the program. The function
getAlternativeDept() returns an alternative department number; perhaps the item desired is
located in that department, and so on.
#include <iostream>
using namespace std;

enum DeptNum {Invalid, Appliances, Automotive,

Housewares, Toys, Accounting};

DeptNum getAlternativeDept (DeptNum thisDept);

int main () {
DeptNum thisDept;
...
switch (thisDept) {
case Appliances:
...
break;
case Automotive:
...
break;
}
DeptNum altDept = getAlternativeDept (thisDept);
...
}

DeptNum getAlternativeDept (DeptNum thisDept) {

switch (thisDept) {
case Appliances:
Character Processing and Do Case 368

return Housewares;
case Automotive:
return Appliances;
...
}
...
}

As you look this rather contrived example over, notice one major feature. No where in it
are the actual numerical values that would have had to be there if enums were not used! Also
note that since the first enum symbol has the value 0 by default and since no department number
can have the value 0, I chose to place another identifier to represent the value 0, Invalid. Thus,
Appliances has the value 1 as required.

Here is another example of an enum. This time, I use convenient names for the month.
enum Month {Invalid, Jan, Feb, Mar, Apl, May, Jun,
Jul, Aug, Sep, Oct, Nov, Dec};
Then within main() an instance can be created and used.
Month month;
...
if (month == Jun || month == Jul || month == Aug)
cout << "Vacation Time!\n";

An enum can be very useful in the right circumstance. A store might use an enum to
represent its products:
enum ProductId {Coffee, Tea, Milk, Soda};
Or a pet shop program could define an enum to help identify all the animals it has for sale:
enum Pets {Cat, Dog, Hamster, Snake, Goldfish, Bird};

The bottom line is simple: an English identifier that is easily read replaces abstract
numerical values. The result is far fewer errors in such programs!

There must be a catch you say. Well, yes, there is a catch. enums are implemented as an
integer data type internally. No I/O operations use enums directly; instead, all I/O operations
read and write the integer value. Using the Pet enum above, if one coded
Pets myNewPet;
cin >> myNewPet; // error - cannot I/O an enum value
The user would have to key in the integer 1 if the new pet was a dog, for instance. Ugh.

Likewise, if you outputted

cout << myNewPet; // error - cannot I/O an enum value
and if the pet was a dog, then a 1 appears on the output, not “Dog.” Ugh. (Note that the newer
ostream implementation now gives a compiler error message on the above line.)
Character Processing and Do Case 369

For clear output, there is an easy way around the enum problem. Use a Do Case
statement and display the appropriate string.
switch (myNewPet) {
case Cat:
cout << "Cat"; break;
case Dog:
cout << "Dog"; break;
...
}
This is very commonly done for output. It makes the output very readable.

How can the input operation be accomplished and still be readable? There is nothing to
prevent the user from inputting 42 when asked to enter the new pet number! Instead of inputting
the enumerated value directly, input something more meaningful and use a Do Case to assign the
correct enum value. One way would be to use a single letter for the pet type, ‘D’ for Dog, ‘C’ for
Cat and so on. Here is a way it can be written.
#include <iostream>
#include <cctype>
using namespace std;

enum Pets {Cat, Dog, Hamster, Snake, Goldfish, Bird};

istream& getPet (istream& infile, Pets& newPet);

int main () {
Pets thePet;
while (getPet (cin, thePet)) {
... here we have a valid thePet
}
...
return 0;
}

istream& getPet (istream& infile, Pets& newPet) {

char c;
bool validEntry = false;
while (infile && !validEntry) {
cout << "Enter Pet type - C (for cat), D (for dog),\n"
<< "H (for hamster), S (for snake), G (for goldfish)\n"
<< "B (for bird)\n";
infile >> c;
if (!infile)
return infile;
c = toupper (c);
switch (c) {
case 'C':
newPet = Cat;
Character Processing and Do Case 370

validEntry = true;
break;
case 'D':
newPet = Dog;
validEntry = true;
break;
case 'H':
newPet = Hamster;
validEntry = true;
break;
case 'S':
newPet = Snake;
validEntry = true;
break;
case 'G':
newPet = GoldFish;
validEntry = true;
break;
case 'B':
newPet = Bird;
validEntry = true;
break;
default:
cout << "Invalid entry - try again\n";
}
}
return infile;
}

In the getPet() function, the boolean variable validEntry is initialized to false. If the user
ever enters a valid pet character, then the validEntry is set to true. The while loop repeats the
sequence as long as infile is still in the good state and a valid entry has not yet been made. After
prompting the user and getting their input, if the end of file has been reached, the function returns
at once. Otherwise, the character is converted to uppercase and a Do Case sorts out the
possibilities. If the user entry matches one of the valid characters, the reference parameter enum
newPet is assigned the corresponding value and validEntry is set to true. If no match occurs,
then after an error message is displayed, the loop repeats until the user enters a proper pet code.

Thus we have ways of inputting and outputting these enumerated values in a much more
readable, user-friendly manner than just a set of abstract numbers.

Suppose that a program must deal with coins at a financial institution. Here is another
way the enum can be setup in which each identifier is given its value instead of taking the
default.
enum CoinAmt {Penny = 1, Nickel = 5, Dime = 10,
Quarter = 25, HalfDollar = 50,
Character Processing and Do Case 371

SilverDollar = 100};
Given these values, one can now use them anywhere an integer data type could be used. For
example
long totalAmount = 124; // total pennies

// the number of quarters in the change

int numQuarters = totalAmount / Quarter;

// remaining change
totalAmount %= Quarter;

Once one stops providing unique values, the next enumerator’s values go up by one from
the previous ones. With the department number enum, if one did not wish to have Invalid be a
part of the enumerated values, then it could have been coded this way.
enum DeptNum {Appliances = 1, Automotive, Housewares, Toys,
Accounting};
The month enum could have been done this way.
enum Month {Jan = 1, Feb, Mar, Apl, May, Jun,
Jul, Aug, Sep, Oct, Nov, Dec};
In these two cases, Automotive and Feb are given the value of 2; Housewares and Mar, 3; and
so on.

Finally, if no instances of the enumerated data type are ever going to be created, the name
can be omitted. This is called an anonymous enum. Of course, the enumerated values can be
used wherever an integer type can be used. In the coin amount example above, I did not create
any instances of CoinAmt. I did, however, make use of the enumerated values. It could have
been defined this way.
enum {Penny = 1, Nickel = 5, Dime = 10, Quarter = 25,
HalfDollar = 50, SilverDollar = 100};
Anonymous enums are often used to create good symbolic names for numerical values.

When an application has need of a series of numerical values that are used for some form
of identification purposes, consider making those values as part of an enumerated data type to
increase readability of the program and lessen the chance for errors.

Aborting the Program

Sometimes a function detects that an error has arisen that it cannot handle. One action the
function could to is to return a special error value, such as TIME_ERROR back in chapter 7.
However, if a function that is to open a file discovers it cannot be opened, the function may wish
to abort the program, to terminate it, instead of returning back to its caller. This is done using the
exit() function whose sole parameter is the integer return code to give back to DOS.
exit (1); // aborts the program with a return code of 1
Character Processing and Do Case 372

Section B: Computer Science Examples

Cs08a — Inventory on Hand Program
To illustrate many of the principles found in this chapter, let’s examine the inventory on hand
problem at Acme Chocolates Company. They make chocolate candies that are coated with a thin
shell of these colors: yellow, red, blue, green, and brown. The master.file contains the current
quantity on hand of each of the different colors. Each line in the master file contains first a letter
that corresponds to a color (Y, R, U, G, B where U is blue) followed by the number of that color
on hand.

At the end of the work day, two additional files have been produced from other programs:
orders.txt and made.txt, representing the orders that have been taken throughout the day for
chocolates and the number of new ones made that day. Both of these files have the same form as
the master file; that is, first is a letter identifying the color and then the number of that color.

The program should begin by inputting the current master.txt file. Next, the program
inputs the made.txt file and adds the number of the specific colors made today to the
corresponding number on hand. Then the program inputs the orders.txt file and subtracts the
number sold of each color from the number on hand. Finally, the program then writes a new
master file of the quantity on hand, called newmaster.txt. Note, should the final quantity on
hand even become negative, a special line is written to the file makeAtOnce.txt, indicating the
deficit information so that the manufacturing group can make those kinds immediately the next
shift to catch up with the demand.

In this case, an enum for the colors simplifies things. Let’s make an enum like this.
enum Color {Yellow, Red, Blue, Green, Brown};

Each line in the three files has a character color abbreviation followed by an integer
number. Notice the commonality here. Rather than write independent coding for each of the three
files, since they all have a common line characteristic, let’s write one function, getLine(), that
can be used to input a line of data from any of the three files. Since for each file the character
color must be converted into a Color enum type, let’s have a function convertToColor() to do
that coding one time only. Let’s also make four other functions that parallel the work to be done:
loadMaster(), addMades(), subOrders() and makeNewMaster(). The Top-Down Design is
thus shown in Figure 8.4.

The main() function is really simple in this program. It must define the five totals
variables and then call each of the four functions in turn. The variables are: totalYellow,
totalRed, totalBlue, totalGreen and totalBrown. The instruction sequence is just:
loadMaster (totalYellow, totalRed, totalBlue, totalGreen, totalBrown);
addMades (totalYellow, totalRed, totalBlue, totalGreen, totalBrown);
Character Processing and Do Case 373

subOrders (totalYellow, totalRed, totalBlue, totalGreen, totalBrown);

makeNewMaster (totalYellow, totalRed, totalBlue, totalGreen, totalBrown);
The variables must be passed by reference to the first three functions since those functions are
altering their contents. However, makeNewMaster() can just be passed a copy.

Figure 8.4 Top-Down Design of Cs08a

The sequence for loadMaster() which is passed the five variables by reference is as
follows.
define a Color variable called color
define a long variable to hold the incoming quantity, thisCount.
open the master file, displaying an error message and aborting if it cannot be opened.
get a line of data filling color and thisCount.
while the input stream is still good do the following
switch (color) {
for case Yellow, do this
totalYellow = thisCount;
end do this
repeat for other 4 color cases
end switch
get another line of data
end while
close the file

The sequence for addMades() is exactly the same as loadMaster(), except that within
the switch statement, a calculation must be done like this.
totalYellow += thisCount.

The sequence for subOrders() is also the same except the calculation is
totalYellow -= thisCount.
Character Processing and Do Case 374

The sequence for makeNewMasterFile() is also simple. It simply writes a line for each
type consisting of the color letter and the amount on hand. Then, it checks to see if the amount
for a given color is less than zero and, if so, displays a message to the make now file.

The sequence for the getLine() function, which is passed a reference to the Color and
amount, is as follows.
define a variable to hold the character color called col
input col and the amount
if the file is not in the good state, return the file
assign to the passed color reference parameter the returned value from
convertToColor giving it the col character to convert.
return the file reference

The sequence for the convertToColor() function first converts the color character to
uppercase and then does a switch on the character color variable. For each color case, it returns
the correct enumerated value.

Figure 8.5 shows the main storage diagram for the program. An easy way to handle the
reference variables in the functions is to place “ref” within the reference variable boxes. Then
when you are desk checking a call to a function which has one or more reference variables, draw
a line from the reference variable box with the ref in it to the calling function’s variable to which
it is a reference.

Figure 8.5 Main Storage for Inventory Program

Character Processing and Do Case 375

Here is the completed program. Notice how easy the enums make reading the coding.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs08a - Inventory on Hand *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs08a: Inventory On Hand Program */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 #include <cctype> *
* 11 using namespace std; *
* 12 *
* 13 enum Color {Yellow, Red, Blue, Green, Brown}; *
* 14 *
* 15 istream& getLine (istream& infile, Color& color, long& amount); *
* 16 Color convertToColor (char color); *
* 17 *
* 18 void loadMaster (long& totalYellow, long& totalRed, *
* 19 long& totalBlue, long& totalGreen, long& totalBrown);*
* 20 void addMades (long& totalYellow, long& totalRed, *
* 21 long& totalBlue, long& totalGreen, long& totalBrown);*
* 22 void subOrders (long& totalYellow, long& totalRed, *
* 23 long& totalBlue, long& totalGreen, long& totalBrown);*
* 24 void makeNewMaster (long totalYellow, long totalRed, *
* 25 long totalBlue, long totalGreen, long totalBrown); *
* 26 *
* 27 int main () { *
* 28 *
* 29 // total quantity on hand of each color *
* 30 long totalYellow = 0; *
* 31 long totalRed = 0; *
* 32 long totalBlue = 0; *
* 33 long totalGreen = 0; *
* 34 long totalBrown = 0; *
* 35 *
* 36 // main processing steps *
* 37 loadMaster (totalYellow, totalRed, totalBlue, totalGreen, *
* 38 totalBrown); *
* 39 addMades (totalYellow, totalRed, totalBlue, totalGreen, *
* 40 totalBrown); *
* 41 subOrders (totalYellow, totalRed, totalBlue, totalGreen, *
* 42 totalBrown); *
* 43 makeNewMaster (totalYellow, totalRed, totalBlue, totalGreen, *
* 44 totalBrown); *
* 45 *
* 46 return 0; *
* 47 } *
* 48 *
Character Processing and Do Case 376

* 49 /***************************************************************/*
* 50 /* */*
* 51 /* getLine: input a line: Color and amount from a file */*
* 52 /* */*
* 53 /***************************************************************/*
* 54 *
* 55 istream& getLine (istream& infile, Color& color, long& amount) { *
* 56 char col; // color letter *
* 57 infile >> col >> amount; // input one line *
* 58 if (!infile) *
* 59 return infile; *
* 60 color = convertToColor (col); *
* 61 return infile; *
* 62 } *
* 63 *
* 64 /***************************************************************/*
* 65 /* */*
* 66 /* convertToColor: given a char color letter, return */*
* 67 /* corresponding Color enumeration value */*
* 68 /* */*
* 69 /***************************************************************/*
* 70 *
* 71 Color convertToColor (char color) { *
* 72 color = toupper (color); *
* 73 switch (color) { *
* 74 case 'Y': *
* 75 return Yellow; *
* 76 case 'R': *
* 77 return Red; *
* 78 case 'G': *
* 79 return Green; *
* 80 case 'U': *
* 81 return Blue; *
* 82 case 'B': *
* 83 return Brown; *
* 84 default: *
* 85 cerr << "Invalid color letter. It was: " << color << endl; *
* 86 exit (1); *
* 87 } *
* 88 } *
* 89 *
* 90 /***************************************************************/*
* 91 /* */*
* 92 /* loadMaster: load master file of qty on hand for each color */*
* 93 /* */*
* 94 /***************************************************************/*
* 95 *
* 96 void loadMaster (long& totalYellow, long& totalRed, *
* 97 long& totalBlue, long& totalGreen, long& totalBrown) {*
* 98 Color color; // color of current one *
* 99 long thisCount; // quantity of current one *
* 100 *
Character Processing and Do Case 377

* 101 //open the master file *

* 102 ifstream masterFile ("master.txt"); *
* 103 if (!masterFile) { *
* 104 cerr << "Error: unable to open master.txt file.\n"; *
* 105 exit (2); *
* 106 } *
* 107 *
* 108 // input the amount on hand for each color *
* 109 while (getLine (masterFile, color, thisCount)) { *
* 110 switch (color) { *
* 111 case Yellow: *
* 112 totalYellow = thisCount; break; *
* 113 case Red: *
* 114 totalRed = thisCount; break; *
* 115 case Green: *
* 116 totalGreen = thisCount; break; *
* 117 case Blue: *
* 118 totalBlue = thisCount; break; *
* 119 case Brown: *
* 120 totalBrown = thisCount; break; *
* 121 } *
* 122 } *
* 123 masterFile.close (); *
* 124 } *
* 125 *
* 126 /***************************************************************/*
* 127 /* */*
* 128 /* addMades: add in the totals that were made today */*
* 129 /* */*
* 130 /***************************************************************/*
* 131 *
* 132 void addMades (long& totalYellow, long& totalRed, *
* 133 long& totalBlue, long& totalGreen, long& totalBrown) {*
* 134 Color color; // color of current one *
* 135 long thisCount; // quantity of current one *
* 136 *
* 137 //open the made today file *
* 138 ifstream madeFile ("made.txt"); *
* 139 if (!madeFile) { *
* 140 cerr << "Error: unable to open made.txt file.\n"; *
* 141 exit (3); *
* 142 } *
* 143 *
* 144 //for each one made today, add amount made to total for color *
* 145 while (getLine (madeFile, color, thisCount)) { *
* 146 switch (color) { *
* 147 case Yellow: *
* 148 totalYellow += thisCount; break; *
* 149 case Red: *
* 150 totalRed += thisCount; break; *
* 151 case Green: *
* 152 totalGreen += thisCount; break; *
Character Processing and Do Case 378

* 153 case Blue: *

* 154 totalBlue += thisCount; break; *
* 155 case Brown: *
* 156 totalBrown += thisCount; break; *
* 157 } *
* 158 } *
* 159 madeFile.close (); *
* 160 } *
* 161 *
* 162 /***************************************************************/*
* 163 /* */*
* 164 /* subOrders: subtract out the current orders to be filled */*
* 165 /* */*
* 166 /***************************************************************/*
* 167 *
* 168 void subOrders (long& totalYellow, long& totalRed, *
* 169 long& totalBlue, long& totalGreen, long& totalBrown) {*
* 170 Color color; // color of current one *
* 171 long thisCount; // quantity of current one *
* 172 *
* 173 // open the orders file *
* 174 ifstream ordersFile ("orders.txt"); *
* 175 if (!ordersFile) { *
* 176 cerr << "Error: unable to open orders.txt file.\n"; *
* 177 exit (4); *
* 178 } *
* 179 *
* 180 // process each line *
* 181 while (getLine (ordersFile, color, thisCount)) { *
* 182 // subtract total that corresponds to each color *
* 183 switch (color) { *
* 184 case Yellow: *
* 185 totalYellow -= thisCount; break; *
* 186 case Red: *
* 187 totalRed -= thisCount; break; *
* 188 case Green: *
* 189 totalGreen -= thisCount; break; *
* 190 case Blue: *
* 191 totalBlue -= thisCount; break; *
* 192 case Brown: *
* 193 totalBrown -= thisCount; break; *
* 194 } *
* 195 } *
* 196 ordersFile.close (); *
* 197 } *
* 198 *
* 199 /***************************************************************/*
* 200 /* */*
* 201 /* makeNewMaster: makes new master file with errors reported */*
* 202 /* */*
* 203 /***************************************************************/*
* 204 *
Character Processing and Do Case 379

* 205 void makeNewMaster (long totalYellow, long totalRed, *

* 206 long totalBlue, long totalGreen, long totalBrown) {*
* 207 // open both files *
* 208 ofstream makeNowFile ("makeAtOnce.txt"); *
* 209 ofstream newMasterFile ("newmaster.txt"); *
* 210 *
* 211 // make a new master file *
* 212 newMasterFile << 'Y' << " " << totalYellow << endl; *
* 213 newMasterFile << 'R' << " " << totalRed << endl; *
* 214 newMasterFile << 'G' << " " << totalGreen << endl; *
* 215 newMasterFile << 'U' << " " << totalBlue << endl; *
* 216 newMasterFile << 'B' << " " << totalBrown << endl; *
* 217 newMasterFile.close (); *
* 218 *
* 219 // report on deficits that must be made at once *
* 220 if (totalYellow < 0) { *
* 221 makeNowFile << "Yellow is short: " << setw(6) << totalYellow *
* 222 << endl; *
* 223 } *
* 224 if (totalRed < 0) { *
* 225 makeNowFile << "Red is short: " << setw(6) << totalRed *
* 226 << endl; *
* 227 } *
* 228 if (totalGreen < 0) { *
* 229 makeNowFile << "Green is short: " << setw(6) << totalGreen *
* 230 << endl; *
* 231 } *
* 232 if (totalBlue < 0) { *
* 233 makeNowFile << "Blue is short: " << setw(6) << totalBlue *
* 234 << endl; *
* 235 } *
* 236 if (totalBrown < 0) { *
* 237 makeNowFile << "Brown is short: " << setw(6) << totalBrown *
* 238 << endl; *
* 239 } *
* 240 makeNowFile.close (); *
* 241 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Cs08b — Inventory on Hand Program — Using a Generic

processFile() Function
Note that there is still some redundant coding in Cs08a. Namely, loadMaster() and addMades()
are nearly identical. They could be reduced to a single function by some slight changes. If main()
initialized the five totals to zero and passed a controlling parameter in addition to the five
references to the long amounts, then one function could be used to handle both situations. Taking
it one step further, if that common function was also passed a way to determine if it was a
subtraction that was needed, then it could also be used for the subOrders() as well.
Character Processing and Do Case 380

Suppose we made another enum that identifies which file we want to process. Let’s call it
FileType and it is defined as
enum FileType {Master, Mades, Orders};
Now the prototype for the generic single function to read and process a file of these data could
then be
void processFile (FileType which, long& totalYellow,
long& totalRed, long& totalBlue,
long& totalGreen, long& totalBrown);
Then the correct file can be opened this way.
ifstream infile;
switch (which) {
case Master:
infile.open ("master.txt");
break;
case Mades:
infile.open ("made.txt");
break;
case Orders:
infile.open ("orders.txt");
break;
default:
cerr << "Error: incorrect parameter to processFile\n";
exit (3);
}
if (!infile) {
cerr << "Error cannot open the input files\n";
exit (2);
}
Within the switch on the Color enum, a slight change handles all three cases:
case Yellow:
if (which == Orders)
totalYellow -= thisCount;
else
totalYellow += thisCount;
break;
And so on with the other four cases. In other words, we can write only one input a file and
process its data function, streamlining the program significantly. Here is the revised program
Cs08b.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs08b - Inventory on Hand - Generic Process Function *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs08b: Inventory On Hand Program */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
Character Processing and Do Case 381

* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 #include <cctype> *
* 11 using namespace std; *
* 12 *
* 13 enum Color {Yellow, Red, Blue, Green, Brown}; *
* 14 enum FileType {Master, Mades, Orders}; *
* 15 *
* 16 istream& getLine (istream& infile, Color& color, long& amount); *
* 17 Color convertToColor (char color); *
* 18 *
* 19 void processFile (FileType which, long& totalYellow, *
* 20 long& totalRed, long& totalBlue, long& totalGreen,*
* 21 long& totalBrown); *
* 22 void makeNewMaster (long totalYellow, long totalRed, *
* 23 long totalBlue, long totalGreen, long totalBrown);*
* 24 *
* 25 int main () { *
* 27 // total quantity on hand of each color *
* 28 long totalYellow = 0; *
* 29 long totalRed = 0; *
* 30 long totalBlue = 0; *
* 31 long totalGreen = 0; *
* 32 long totalBrown = 0; *
* 33 *
* 34 // main processing steps *
* 35 processFile (Master, totalYellow, totalRed, totalBlue, *
* 36 totalGreen, totalBrown); *
* 37 processFile (Mades, totalYellow, totalRed, totalBlue, *
* 38 totalGreen, totalBrown); *
* 39 processFile (Orders, totalYellow, totalRed, totalBlue, *
* 40 totalGreen, totalBrown); *
* 41 makeNewMaster (totalYellow, totalRed, totalBlue, *
* 42 totalGreen, totalBrown); *
* 43 *
* 44 return 0; *
* 45 } *
* 46 *
* 47 /***************************************************************/*
* 48 /* */*
* 49 /* getLine: input a line: Color and amount from a file */*
* 50 /* */*
* 51 /***************************************************************/*
* 52 *
* 53 istream& getLine (istream& infile, Color& color, long& amount) { *
* 54 char col; // color letter *
* 55 infile >> col >> amount; // input one line *
* 56 if (!infile) *
* 57 return infile; *
* 58 color = convertToColor (col); *
* 59 return infile; *
* 60 } *
Character Processing and Do Case 382

* 61 *
* 62 /***************************************************************/*
* 63 /* */*
* 64 /* convertToColor: given a char color, return corresponding */*
* 65 /* Color enumeration value */*
* 66 /* */*
* 67 /***************************************************************/*
* 68 *
* 69 Color convertToColor (char color) { *
* 70 color = toupper (color); *
* 71 switch (color) { *
* 72 case 'Y': *
* 73 return Yellow; *
* 74 case 'R': *
* 75 return Red; *
* 76 case 'G': *
* 77 return Green; *
* 78 case 'U': *
* 79 return Blue; *
* 80 case 'B': *
* 81 return Brown; *
* 82 default: *
* 83 cerr << "Invalid color letter. It was: " << color << endl; *
* 84 exit (1); *
* 85 } *
* 86 } *
* 87 *
* 88 /***************************************************************/*
* 89 /* */*
* 90 /* processFile: process a file of colors and amounts */*
* 91 /* */*
* 92 /***************************************************************/*
* 93 *
* 94 void processFile (FileType which, long& totalYellow, *
* 95 long& totalRed, long& totalBlue, long& totalGreen,*
* 96 long& totalBrown) { *
* 97 Color color; // color of current one *
* 98 long thisCount; // quantity of current one *
* 99 *
* 100 //open the master file *
* 101 ifstream infile; *
* 102 switch (which) { *
* 103 case Master: *
* 104 infile.open ("master.txt"); *
* 105 if (!infile) { *
* 106 cerr << "Error: unable to open master.txt file.\n"; *
* 107 exit (2); *
* 108 } *
* 109 break; *
* 110 case Mades: *
* 111 infile.open ("made.txt"); *
* 112 if (!infile) { *
Character Processing and Do Case 383

* 113 cerr << "Error: unable to open made.txt file.\n"; *

* 114 exit (3); *
* 115 } *
* 116 break; *
* 117 case Orders: *
* 118 infile.open ("orders.txt"); *
* 119 if (!infile) { *
* 120 cerr << "Error: unable to open orders.txt file.\n"; *
* 121 exit (4); *
* 122 } *
* 123 break; *
* 124 default: *
* 125 cerr << "Error: incorrect parameter to processFile\n"; *
* 126 exit (5); *
* 127 } *
* 128 *
* 129 // input the amount on hand for each color *
* 130 while (getLine (infile, color, thisCount)) { *
* 131 switch (color) { *
* 132 case Yellow: *
* 133 if (which == Orders) *
* 134 totalYellow -= thisCount; *
* 135 else *
* 136 totalYellow += thisCount; *
* 137 break; *
* 138 case Red: *
* 139 if (which == Orders) *
* 140 totalRed -= thisCount; *
* 141 else *
* 142 totalRed += thisCount; *
* 142.5 break; *
* 143 case Green: *
* 144 if (which == Orders) *
* 145 totalGreen -= thisCount; *
* 146 else *
* 147 totalGreen += thisCount; *
* 147.5 break; *
* 148 case Blue: *
* 149 if (which == Orders) *
* 150 totalBlue -= thisCount; *
* 151 else *
* 152 totalBlue += thisCount; *
* 152.5 break; *
* 153 case Brown: *
* 154 if (which == Orders) *
* 155 totalBrown -= thisCount; *
* 156 else *
* 157 totalBrown += thisCount; *
* 158 } *
* 159 } *
* 160 infile.close (); *
* 161 } *
Character Processing and Do Case 384

* 162 *
* 163 /***************************************************************/*
* 164 /* */*
* 165 /* makeNewMaster: makes new master file with errors reported */*
* 166 /* */*
* 167 /***************************************************************/*
* 168 *
* 169 void makeNewMaster (long totalYellow, long totalRed, *
* 170 long totalBlue, long totalGreen, long totalBrown) {*
* 171 // open both files *
* 172 ofstream makeNowFile ("makeAtOnce.txt"); *
* 173 ofstream newMasterFile ("newmaster.txt"); *
* 173.1if (!makeNowFile || !newMasterFile) { *
* 173.2 cerr << "Cannot open either newMasterFile or newmaster.txt\n"; *
* 173.3 exit (6); *
* 174 } *
* 175 // make a new master file *
* 176 newMasterFile << 'Y' << " " << totalYellow << endl; *
* 177 newMasterFile << 'R' << " " << totalRed << endl; *
* 178 newMasterFile << 'G' << " " << totalGreen << endl; *
* 179 newMasterFile << 'U' << " " << totalBlue << endl; *
* 180 newMasterFile << 'B' << " " << totalBrown << endl; *
* 181 newMasterFile.close (); *
* 182 *
* 183 // report on deficits that must be made at once *
* 184 if (totalYellow < 0) { *
* 185 makeNowFile << "Yellow is short: " << setw(6) << totalYellow *
* 186 << endl; *
* 187 } *
* 188 if (totalRed < 0) { *
* 189 makeNowFile << "Red is short: " << setw(6) << totalRed *
* 190 << endl; *
* 191 } *
* 192 if (totalGreen < 0) { *
* 193 makeNowFile << "Green is short: " << setw(6) << totalGreen *
* 194 << endl; *
* 195 } *
* 196 if (totalBlue < 0) { *
* 197 makeNowFile << "Blue is short: " << setw(6) << totalBlue *
* 198 << endl; *
* 199 } *
* 200 if (totalBrown < 0) { *
* 201 makeNowFile << "Brown is short: " << setw(6) << totalBrown *
* 202 << endl; *
* 203 } *
* 204 makeNowFile.close (); *
* 205 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
The rule of thumb to follow in programming is this. If you find that you are coding
nearly the same exact coding in more than one function, see if there is a way that coding
can easily be factored out and done just once.
Character Processing and Do Case 385

Section C: Engineering Examples — Numerical Integration

The Trapezoid Method of Numerical Integration
Often a programmer is called upon to integrate a function f(x). While the computer cannot
integrate from zero to infinity or from minus infinity to plus infinity, if there are some limits, say
integrating from point A to point B, then the computer can easily perform the integration. The
first method of handling the integration is called the trapezoid method.

A way of looking at the integration of

is that the result is actually the area under the curve as shown in Figure 8.6.

Figure 8.6 Integration Using One Trapezoid

The trapezoid method divides the interval A to B into a number of trapezoids and
calculates the area of the trapezoids which becomes the area under the curve or the result of the
integration. In Figure 8.6, there is one trapezoid. The area of that trapezoid is given by

One trapezoid does not provide a very accurate value for the area under the curve. In
Figure 8.7, the interval A to B is divided into four equal portions. The sum of the areas of these
trapezoids is a closer approximation to the area under the curve.
Character Processing and Do Case 386

Figure 8.7 Integration Using Four Trapezoids

When there are four evenly spaced trapezoids, let w be the uniform width of each trapezoid. The
area becomes
.5 * w * (f(a) + f(1)) + .5 * w * (f(1) + f(2))
+ .5 * w * (f(2) + f(3)) + .5 * w * (f(3) + f(b))

If we group these, we have

I = .5 * w * ( f(a) + f(b) + 2 (f(1) + f(2) + f(3)) )
From this, we can generalize the solution using n total trapezoids as

This equation can be used to perform the integration. However, one major difficulty
remains. How do we know how many trapezoids are sufficient to yield an accurate solution?
Certainly, one could solve the equation using one trapezoid, and then do it again for two, and
then three and so on. Eventually, adding one more trapezoid should make no appreciable impact
upon the resulting answer, to the desired degree of accuracy. While this would work, it is going
to be a very slow method doing many redundant calculations. There is a better way to proceed. In
other words, if we have already calculated the area using one number of trapezoids, is there a
way we can use that result in the next iteration so as to avoid recalculating the entire summation
each time? If so, then when the next iteration’s result is less than the desired degree of accuracy
from the previous result, we have found the answer.
Character Processing and Do Case 387

One thing that can be done to more quickly zero in on the result is to increase the number
of trapezoids by a power of 2. Let’s let n, the number of trapezoids, be given by

where k begins at 0 and goes up by one each time. Thus, the number of trapezoids goes rapidly
up: 1, 2, 4, 8, 16, 32, 64, 128, 256, 1024,...

Also by doubling the number of trapezoids each time, we introduce some interesting
aspects. The width of the trapezoids, w, is given by

The implication is that the width of the kth trapezoids is ½ of the (k – 1)th trapezoids. But there is
also an even more important aspect that this doubling of the number of trapezoids has on the
calculation. If we go from two to four trapezoids, then two of the points have already had their
f(x)’s calculated and added into the result. Compare Figure 8.7 with four trapezoids to Figure 8.8
which uses two trapezoids. Notice that point 2 in Figure 8.7 is the same as point 1 in Figure 8.8.

Figure 8.8 Integration Using Two Trapezoids

We have then the two areas given by

T2= ½ w2 (f(a) + f(b) + 2 (f1) )
T4 = ½ w4 (f(a) + f(b) + 2 (f1 + f2 + f3) )
but w4 is ½ w2 and f2 of T4 is the same point as f1 of T2 . So substituting we get
T4 = ½ T2 + w4 (f1 + f3)

From this we can generalize based on k, the power of 2:

Character Processing and Do Case 388

This is the version that we can efficiently use to integrate a function on the computer.

Engr08a — Numerical Integration with the Trapezoid Rule

Let’s integrate a simple function that has a known answer so that we can compare the results. The
function is

This yields ln(x); ln(1) is 0, and ln(2) is .69314718. Here is the output from running the program.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Engr08a - Integration using Trapezoid Method *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Integration of 1/x *
* 2 Enter the range a and b, the desired degree of accuracy *
* 3 such as 1E-6, and the maximum number of iterations *
* 4 separated by blanks *
* 5 1 2 .000001 50 *
* 6 The result is: 0.6931464 found after 25 iterations *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

The main() function does nothing more than prompt the user to enter the four needed
values, calls the Trap() function and displays the results. Function F() is trivial. All of the work
is done in the Trap() function. The logic within Trap() follows very similarly to that used in root
solving. The main loop is executed as long as k is less than the maximum number of iterations
and the difference between this result and the previous result is greater than the desired accuracy.
Here is the complete program.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr08a - Integration using Trapezoid Method *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr08a Integrate f(x) using the Trapezoid Method */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 double Trap (double a, double b, int imax, double eps, int& i); *
Character Processing and Do Case 389

* 12 double F (double x); *

* 13 *
* 14 int main () { *
* 15 *
* 16 double a, b; *
* 17 int imax; *
* 18 double eps; *
* 19 cout << "Integration of 1/x\n" *
* 20 << "Enter the range a and b, the desired degree of accuracy\n"*
* 21 << "such as 1E-6, and the maximum number of iterations\n" *
* 22 << "separated by blanks\n"; *
* 23 cin >> a >> b >> eps >> imax; *
* 24 if (!cin) *
* 25 return 1; *
* 26 *
* 27 double answer; *
* 28 int numberIterations; *
* 29 answer = Trap (a, b, imax, eps, numberIterations); *
* 30 cout << "The result is: " << setw(10) << setprecision (7) *
* 31 << answer << " found after " << numberIterations *
* 32 << " iterations\n"; *
* 33 *
* 34 return 0; *
* 35 } *
* 36 *
* 37 /***************************************************************/*
* 38 /* */*
* 39 /* Trap: integrates F(x) using the trapezoid method */*
* 40 /* */*
* 41 /***************************************************************/*
* 42 *
* 43 double Trap (double a, double b, int imax, double eps, int& k) { *
* 44 *
* 45 const double wba = b-a; // original width of integration *
* 46 double prevT = 0; // the previous integration result *
* 47 double newT = .5 * wba * (F(a) + F(b)); // current result *
* 48 double sum; *
* 49 double w; *
* 50 int n; *
* 51 int i; *
* 52 *
* 53 k = 1; // number of iterations *
* 54 *
* 55 while (k < imax && fabs (newT - prevT) > eps) { *
* 56 prevT = newT; // save prev result for this iteration *
* 57 n = (int) (pow (2., (double) k)); // new number of trapezoids *
* 58 w = wba / n; // get new uniform width of trapezoids *
* 59 sum = 0; // find sum of all odd terms *
* 60 for (i=1; i<n-1; i+=2) { *
* 61 sum = sum + F(a + i * w); *
* 62 } *
* 63 newT = .5 * prevT + w * sum; // calc new area *
Character Processing and Do Case 390

* 64 // for debugging, display these intermediate results *

* 65 //cout << k << " " << setprecision (8) << newT << endl; *
* 66 k++; *
* 67 } *
* 68 // check for non-convergence *
* 69 if (k == imax) { *
* 70 cerr << "After " << imax *
* 71 << " iterations, Trap has not converged.\n" *
* 72 << "The result so far is " << setprecision (8) << newT *
* 73 << endl << "The accuracy is " << fabs (newT - prevT) *
* 74 << endl; *
* 75 } *
* 76 return newT; *
* 77 } *
* 78 *
* 79 /***************************************************************/*
* 80 /* */*
* 81 /* F - the function to integrate - here 1/x */*
* 82 /* */*
* 83 /***************************************************************/*
* 84 *
* 85 double F (double x) { *
* 86 return 1/x; *
* 87 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Integration Using Simpson’s Rule

Simpson’s Rule fits a parabola to three equally spaced points instead of the simple trapezoid. The
general form of the parabola is
y(x) = a (x – x0)2 + b (x – x0) + c
Again if we use the number of panels as given by

and the uniform width, w, as (b-a) / n, then Simpson’s Rule becomes

Since the derivation is commonly found in Calculus books, it is not shown here.

To use Simpson’s Rule to perform the integration, we must enter the k term which is the
power of 2 that determines the number of panels in which to divide the interval a to b. The
programming solution of this equation is totally straightforward. Two summations must be done;
both are extremely straightforward.
Character Processing and Do Case 391

Engr08b — Numerical Integration with Simpson’s Rule

The function Simp() performs the calculation or integration. It is passed the two limits a and b
along with the k term which is the power of two specifying the number of panels to use. The only
refinement from the original equation above is the calculation of the next x term to be used in the
call to F(x) within the summations. As the equation is written, for each term, the new x value is
given by a + i * w. Performing a multiplication each time is a slow operation. It is far faster to
obtain the next x value by adding deltaX to the previous x where deltaX is 2*w.

Here is the result of a program execution. So that we can compare the results, I have
entered a k value of 25, identical to what the Trap() function found for the integration result.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Engr08b - Integration using Simpson's Rule *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Integration of 1/x using Simpson's Rule *
* 2 Enter the range a and b and enter the number of iterations *
* 3 separated by blanks *
* 4 1 2 25 *
* 5 The result is: 0.6931472 found using k of 25 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Compare this result to that from the trapezoid method and to the precise mathematical calculated
result:
precise value: 0.69314718
Trapezoid: 0.6931464
Simpson’s: 0.6931472
Simpson’s rule provides a more accurate value than does the trapezoid method, which is one
reason that this method is a popular one for integration work.

Here is the complete program.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr08b - Integration using Simpson's Rule *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr08b Integrate f(x) using Simpson's Rule */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 using namespace std; *
* 11 double Simp (double a, double b, int k); *
* 12 double F (double x); *
* 13 *
* 14 int main () { *
* 15 *
* 16 double a, b; *
Character Processing and Do Case 392

* 17 int k; *
* 18 cout << "Integration of 1/x using Simpson's Rule\n" *
* 19 <<"Enter the range a and b and enter the number of iterations\n"*
* 20 << "separated by blanks\n"; *
* 21 cin >> a >> b >> k; *
* 22 if (!cin) *
* 23 return 1; *
* 24 *
* 25 double answer; *
* 26 answer = Simp (a, b, k); *
* 27 cout << "The result is: " << setw(10) << setprecision (7) *
* 28 << answer << " found using k of " << k << endl; *
* 29 return 0; *
* 30 } *
* 31 *
* 32 /***************************************************************/*
* 33 /* */*
* 34 /* Simp: integrates F(x) using Simpson's Rule */*
* 35 /* */*
* 36 /***************************************************************/*
* 37 *
* 38 double Simp (double a, double b, int k) { *
* 39 *
* 40 int n = (int) (pow (2., (double) k)); // new number of panels *
* 41 double w = (b-a) / n; // get the uniform width *
* 42 *
* 43 // calculate first sums of odd terms *
* 44 int i; *
* 45 double sum2 = 0; *
* 46 double deltaX = 2 * w; // amount to add to x to get next odd x*
* 47 double x = a + w; // initial x to use *
* 48 for (i=1; i<n-1; i+=2) { // do all odd terms *
* 49 sum2 += F(x); // add in this term *
* 50 x += deltaX; // find next odd x *
* 51 } *
* 52 sum2 *= 4; // multiply sum by 4 *
* 53 *
* 54 // calculate second sum of even terms *
* 55 double sum4 = 0; *
* 56 x = a + deltaX; // initial x to use *
* 57 for (i=2; i<n-2; i+=2) { // do all even terms *
* 58 sum4 += F(x); // add in this term *
* 59 x += deltaX; // find next even term *
* 60 } *
* 61 sum4 *= 2; // multiply sum by 2 *
* 62 *
* 63 // create final sum of all terms *
* 64 double sum = F(a) + F(b) + sum2 + sum4; *
* 65 return w * sum / 3; // return Simpson's Rule for the area *
* 66 } *
* 67 *
* 68 /***************************************************************/*
Character Processing and Do Case 393

* 69 /* */*
* 70 /* F - the function to integrate - here 1/x */*
* 71 /* */*
* 72 /***************************************************************/*
* 73 *
* 74 double F (double x) { *
* 75 return 1/x; *
* 76 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Engr08c — Using Menus to Control Program Operation

A major use of character data in scientific type programming involves menus and menu choices.
The following represents the main menu presented to the user.
Integration of F(x)

A. Perform the integration using the Trapezoid method

B. Perform the integration using Simpson’s method
C. Quit the program

Enter the letter of your choice: _

The user then enters their choice and the program performs the requisite action and then
redisplays the main menu. A character is input as the user’s choice.

To be compatible, the original Trapezoid method is used so that it can be directly

compared to the results from Simpson’s Rule.

The coding to display a menu is a straightforward series of text lines to be displayed.

They are encapsulated in a function called ShowMenu(). The function GetMenuChoice() must
obtain a valid choice, A, B or C. I initialized the character choice with a 0 so that within the loop,
if choice is ever anything else, an invalid entry error message can be displayed and a prompt to
reenter the letter can be done. If the user presses Ctrl-Z for end of file, GetMenuChoice() simply
sets choice to C; the main loop handles the termination request as if a ‘C’ had been pressed.

Here is a sample run of what the program produces.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from - Engr08c - Using a Menu to Control Program Actions *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 *
* 2 *
* 3 *
* 4 Integration of F(x) *
* 5 *
* 6 A. Perform the integration using the Trapezoid method *
* 7 B. Perform the integration using Simpson's method *
* 8 C. Quit the program *
Character Processing and Do Case 394

* 9 *
* 10 Enter the letter of your choice: a *
* 11 Enter k factor 10 *
* 12 *
* 13 Integration of 1/x using Trapezoid method with k of 10 = 0.69265872 *
* 14 *
* 15 *
* 16 Integration of F(x) *
* 17 *
* 18 A. Perform the integration using the Trapezoid method *
* 19 B. Perform the integration using Simpson's method *
* 20 C. Quit the program *
* 21 *
* 22 Enter the letter of your choice: b *
* 23 Enter k factor 10 *
* 24 *
* 25 Integration of 1/x using Simpson's method with k of 10 = 0.69216998 *
* 26 *
* 27 *
* 28 Integration of F(x) *
* 29 *
* 30 A. Perform the integration using the Trapezoid method *
* 31 B. Perform the integration using Simpson's method *
* 32 C. Quit the program *
* 33 *
* 34 Enter the letter of your choice: q *
* 35 Choice must be A, B or C. Try again *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Here is the completed menu driven program.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr08c - Using a Menu to Control Program Actions *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr08c Menu Choices for Integration */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <cmath> *
* 10 #include <cctype> *
* 11 using namespace std; *
* 12 double Simp (double a, double b, int k); *
* 13 double Trap (double a, double b, int k); *
* 14 double F (double x); *
* 15 *
* 16 void ShowMenu (); *
* 17 void GetMenuChoice (char& choice); *
* 18 *
* 19 int main () { *
Character Processing and Do Case 395

* 20 *
* 21 const double a = 1; // fixed limits on a and b *
* 22 const double b = 2; *
* 23 int k; *
* 24 cout << setprecision (8); *
* 25 char choice; *
* 26 *
* 27 ShowMenu (); // display menu *
* 28 GetMenuChoice (choice); // get initial choice *
* 29 while (choice != 'C') { *
* 30 cout << "\tEnter k factor "; // get k factor *
* 31 cin >> k; *
* 32 if (!cin) break; *
* 33 // perform the integration; display results *
* 34 switch (choice) { *
* 35 case 'A': *
* 36 cout << *
* 37 "\nIntegration of 1/x using Trapezoid method with k of "*
* 38 << k << " = " << Trap (a, b, k); *
* 39 break; *
* 40 case 'B': *
* 41 cout << *
* 42 "\nIntegration of 1/x using Simpson's method with k of "*
* 43 << k << " = " << Simp (a, b, k); *
* 44 break; *
* 45 } *
* 46 ShowMenu (); // redisplay menu *
* 47 GetMenuChoice (choice); // get next choice *
* 48 } *
* 49 return 0; *
* 50 } *
* 51 *
* 52 /***************************************************************/*
* 53 /* */*
* 54 /* ShowMenu: displays the menu of choices */*
* 55 /* */*
* 56 /***************************************************************/*
* 57 *
* 58 void ShowMenu () { *
* 59 cout << endl << endl << endl; *
* 60 cout << "\tIntegration of F(x)\n\n" *
* 61 << "\tA. Perform the integration using the Trapezoid method\n"*
* 62 << "\tB. Perform the integration using Simpson's method\n" *
* 63 << "\tC. Quit the program\n\n" *
* 64 << "\tEnter the letter of your choice: "; *
* 65 } *
* 66 *
* 67 /***************************************************************/*
* 68 /* */*
* 69 /* GetMenuChoice: get the user's menu choice */*
* 70 /* */*
* 71 /***************************************************************/*
Character Processing and Do Case 396

* 72 *
* 73 void GetMenuChoice (char &choice) { *
* 74 choice = 0; *
* 75 while (choice < 'A' || choice > 'C') { *
* 76 if (choice) *
* 77 cout << "Choice must be A, B or C. Try again "; *
* 78 cin >> choice; *
* 79 if (!cin) *
* 80 choice = 'C'; *
* 81 else *
* 82 choice = toupper (choice); *
* 83 } *
* 84 } *
* 85 *
* 86 /***************************************************************/*
* 87 /* */*
* 88 /* Simp: integrates F(x) using Simpson's Rule */*
* 89 /* */*
* 90 /***************************************************************/*
* 91 *
* 92 double Simp (double a, double b, int k) { *
* 93 *
* 94 int n = (int) (pow (2., (double) k)); // new number of panels *
* 95 double w = (b-a) / n; // get the uniform width *
* 96 *
* 97 // calculate first sums of odd terms *
* 98 int i; *
* 99 double sum2 = 0; *
* 100 double deltaX = 2 * w; // amount to add to x to get next odd x*
* 101 double x = a + w; // initial x to use *
* 102 for (i=1; i<n-1; i+=2) { // do all odd terms *
* 103 sum2 += F(x); // add in this term *
* 104 x += deltaX; // find next odd x *
* 105 } *
* 106 sum2 *= 4; // multiply sum by 4 *
* 107 *
* 108 // calculate second sum of even terms *
* 109 double sum4 = 0; *
* 110 x = a + deltaX; // initial x to use *
* 111 for (i=2; i<n-2; i+=2) { // do all even terms *
* 112 sum4 += F(x); // add in this term *
* 113 x += deltaX; // find next even term *
* 114 } *
* 115 sum4 *= 2; // multiply sum by 2 *
* 116 *
* 117 // create final sum of all terms *
* 118 double sum = F(a) + F(b) + sum2 + sum4; *
* 119 return w * sum / 3; // return Simpson's Rule for the area *
* 120 } *
* 121 *
* 122 /***************************************************************/*
* 123 /* */*
Character Processing and Do Case 397

* 124 /* Trap: integrates F(x) using the trapezoid method */*

* 125 /* */*
* 126 /***************************************************************/*
* 127 *
* 128 double Trap (double a, double b, int k) { *
* 129 *
* 130 int n = (int) (pow (2., (double) k)); // number of trapezoids *
* 131 double w = (b-a) / n; // get the uniform width of traps *
* 132 double x = a + w; // get the initial x to use *
* 133 double sum = 0; *
* 134 int i; *
* 135 for (i=1; i<n-1; i++) { // sum all terms *
* 136 sum += F(x); *
* 137 x += w; *
* 138 } *
* 139 sum *= 2; // create 2 times the sum *
* 140 sum = sum + F(a) + F(b); // add up all the other terms *
* 141 return .5 * w * sum; // return the integration result *
* 142 } *
* 143 *
* 144 /***************************************************************/*
* 145 /* */*
* 146 /* F - the function to integrate - here 1/x */*
* 147 /* */*
* 148 /***************************************************************/*
* 149 *
* 150 double F (double x) { *
* 151 return 1/x; *
* 152 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

New Syntax Summary

The get Function
It inputs the next character no matter what it is, that is, it can input whitespace. It is used
when the input of whitespace is significant, such as a student’s answer to a multiple choice test
question where a blank means the student failed to answer that question. Unlike the extraction
operator, it does not first skip over any whitespace.
char c;
cin.get(c);
infile.get(c);

The put Function

The function displays the next character.
cout.put(c);

With char data type, since it can hold a signed number, an unsigned number or an ASCII
character, the computer interprets the meaning of a char variable by its usage in C++ statements.
Character Processing and Do Case 398

If a char variable is inputted or outputted, it is assumed to be an ASCII character. If any math

operations are performed on the char variable, it is assumed to be a number.

The Do Case Structure

switch (variable) {
case value1:
series of statements to do when the variable
contains value1
break;
case value2:
series of statements to do when the variable
contains value2
break;
default:
series of statements to do when it does not
contains any of the values
break;
}
Without the break; statement, control falls on down into the next case or default statements.

Enumerated Data Types

Enums allow us to create a consistent set of identifiers and make code more readable and
less error prone. Under the hood, it is actually an integer.
enum newDataType {value1, value2, ... valuen};

enum PetType {Cat, Dog, Hamster, Goldfish};

int main () {
PetType p = Dog;
if (p == Goldfish) {

If the first enum value is not given a specific value, 0 is used. Each one after that, unless
given a specific value is one greater than the preceding value. So in the above, Cat is 0, Dog is 1
and so on.
enum PetType {Cat = 2, Dog = 42, Hamster = 18, Goldfish =
25};
Here each enum value has a unique and different value.

The enum definition occurs most often after the const ints and before the function
prototypes because instances of the enum can be passed or returned from functions.

An anonymous enum is used when no actual instances are needed. For example,
enum {Penny = 1, Nickel = 5, Dime = 10, Quarter = 25,
HalfDollar = 50, SilverDollar = 100};
Here, we intend to use only the enum values in coding.
Character Processing and Do Case 399

The exit Function

The exit function aborts a program returning the passed integer back to Windows/DOS. It
is commonly found in functions in which a runtime error is found, such as being unable to open a
file. It’s prototype is in the iostreams headers, but is in <stdlib.h> if using the older header files.
exit (1); // abort returning 1 to DOS

Design Exercises
1. Design a Weekly Sales Summary Program

Acme Music Company handles violins, cellos, drums, guitars and pianos. When the weekly sales
summary report program is run, the sales file is input and summarized.

Each line in the sales file represents sales in which the customer purchased an item that
was in the store. A line contains the quantity sold, its cost, its id number, a letter code indicating
which kind of instrument it is, and finally a nine-digit customer id number. This letter code
consists of one of these letters: V, C, D, G, and P — taken from the first letter of each musical
instrument type

The summary report should accumulate the total sales of each instrument type. When the
end of the file is reached, display a summary of the sales for each instrument type. The report
looks like this.
Acme Music Company
Weekly Sales Summary

Instrument Quantity Total

Sold Sales
Violin 999 $9,999.99
Cello 999 $9,999.99
Drum 999 $9,999.99
Guitar 999 $9,999.99
Piano 999 $9,999.99
---------
$99,999.99

Design the program. First, do a Top-Down Design of the problem to isolate what
functions are needed. Next, decide upon whether or not to use an enum — that is, how you wish
to handle the instrument codes. Then, layout main storage for your functions. Finally,
pseudocode the functions needed.
Character Processing and Do Case 400

2. Design a petTypeConversion() function

Acme Pet Shop handles many kinds of pets. Each is identified in many of their data files by a
single letter, either upper or lowercase. Having heard about enums, management has decided to
convert all of their programs over to using a Pet enum. A single function is to be written that all
the programs can use to handle this process. The Pet enum handles the following: dog, cat, bird,
rat, hamster, snake, and gerbil. The letter codes consist of the first letter of the animal types.

The petTypeConversion() function is passed four parameters. The first parameter is a

bool called which that indicates which way the conversion is to go. If which is true, convert the
second parameter, a char pet code letter, into the appropriate enum value and store it in the third
parameter which is a reference to a Pet enum called pet. If which is false, then display the
English word, such as dog, for the passed Pet enum, pet. The display is done using the fourth
parameter, a reference to an ostream, outfile.

Stop! Do These Exercises Before Programming

1. The following program segment does not produce the correct results. Why? How can it be
fixed?
char quantity;
double cost;
cin >> quantity >> cost;
cout << "Total is " << quantity * cost << endl;

2. Acme Department Store has new product information stored in a file whose lines consist of
product id, product type and cost. The product id is a long integer while the cost is a double. The
product type is a letter code indicating the category of merchandise, such as A (automotive), C
(clothing), S (sports) and so on. A typical line looks like this with one blank on either side of the
product type letter.
23455 A 4.99

However, since these are new items, sometimes the type of product has not yet been
determined and that field is blank in that line. The programmer wrote the following input
function. It does not work. Why? What can be done to fix it up so that it properly inputs the data
whether or not the product type is temporarily blank?
istream& getData (istream& infile, long& id, char& type,
double& cost) {
infile >> id >> type >> cost;
return infile;
}
Character Processing and Do Case 401

3. Another Acme programmer attempted to fix the program in Problem 2 above by coding the
following function. It does not work properly either. Why? How could it be fixed to work
correctly?
istream& getData (istream& infile, long& id, char& type,
double& cost) {
infile >> id;
infile.get (type);
infile >> cost;
return infile;
}

4. What is wrong with the following Do Case coding? How can it be fixed up so that it would
work?
double month;
switch (month) {
case 1:
case 2:
case 12:
// winter costs are 25% higher
double sum;
sum = qty * cost * 1.25;
break;
default:
double sum;
sum = qty * cost;
break;
}
cout << sum;

5. The programmer goofed while coding this Do Case to calculate the shift bonus for the
employee payroll. What is wrong and how can it be fixed?
char shift;
switch (shift) {
case '1':
pay = hours * rate;
case '2':
pay = hours * rate * 1.05;
case '3':
pay = hours * rate * 1.12;
}
cout << pay;
Character Processing and Do Case 402

6. The programmer wanted to setup an enumerated data type to handle the employee’s shift.
However, the following coding fails. Why? How can it be repaired?
Enum ShiftType = First, Second, and Third;

7. A programmer setup the following enum to handle the product types.

enum ProductTypes {Games, Auto, Clothing, Appliances};
In the input a set of data function, the user is instructed to enter a letter for the product type: G,
A, C or A. What is the design flaw and why does not the following input coding work? How can
it be repaired?
ProductTypes prodType;
infile >> prodType;

8. In Problem 7, the programmer got frustrated and then did the following which does not
compile. Why? Can this coding be repaired?
ProductTypes prodType;
char c;
infile >> c;
if (c == 'A')
prodType = 1;
if (c == 'G')
prodType = 0;
Character Processing and Do Case 403

Programming Problems
Problem Cs08-1 — Triangle Processing
A triangle has three sides. If all three sides are equal, it is called an equilateral triangle. If any two
sides are equal, it is an isosceles triangle. If none of the three sides are equal, it is called a scalene
triangle. For a triangle to even exist, the sum of any two sides must be longer than the remaining
side.

Write a program inputs a set of three potential sides of a triangle; use doubles for their
data type. After processing a set of sides, input another set of sides until end-of-file occurs. For
each set of data, echo print the original values, and then call two functions and display the results
accordingly. The first function is called IsTriangle() that is passed the three sides and returns a
bool: either true if they make a triangle or false if they do not make a triangle. If the three sides
do not make a triangle, display “Not a triangle” and get the next set of data.

If they form a triangle, then call a function GetTriangleType() that is passed the three
sides and returns an enumerated value. You need to setup an enumerated type whose values are
Scalene, Isosceles, Equilateral. Store the return value in an instance of this enum. Next, the main
program switches on the stored return value (a Do Case) and displays an appropriate message
based on the kind of triangle. For example, “This is an isosceles triangle.”

Thoroughly test your program.

Problem Cs08-2 — Tax Rate Processing

Acme Sales operates stores in a number of states. Each state has its own tax rate. When a set of
sales data is input, the state abbreviation is stored in two character variables, stateLetter1 and
stateLetter2. Each line of input contains the order id number (a long), the two letters of the state
code, and the total cost of the order before taxes are applied (a double). For example, an input
line might be
123456 IL 42.00
The following table shows the states in which Acme is currently selling and the corresponding
tax rates.
State Two Letter Tax Rate
Input Code
Illinois IL 7.5%
Iowa IA 7%
Idaho ID 6.5%
Character Processing and Do Case 404

Indiana IN 8%
Hawaii HI 8%
Georgia GA 6.5%

The program should input the file, Sales.txt, that is provided in the test data
accompanying the book, and produce a sales report that looks like this.
Acme Sales Report

Order State Total

Number Code Sales
12345 IL 42.33

The program must use a Do Case to sort out the state code abbreviations. Use a switch on
the first letter. If the letter is an I, then switch on the second letter to sort out those that begin
with the letter I. Realistically, Acme may end up selling in all fifty states and the switching logic
would then be expanded to include all fifty states. Make sure your solution would be easily
expandable to include additional states.

Note that the input may contain lowercase letters and may contain invalid state codes. If
the code is not valid (perhaps a new state not yet in the program tables), display an error message
to cerr including the order number and the offending state code letters and go on to get the next
input line.

Problem Cs08-3 — Fun with Phone Numbers

A telephone number can utilize letters. The conversion is
ABC = 2, DEF = 3, GHI = 4, JKL = 5, MNO = 6, PRS = 7, TUV = 8, WXY = 9

Write a function called digitToLetter() that is passed an ASCII digit, ‘0’ through ‘9’ and
returns the character letter that corresponds to the number. If there is no corresponding letter,
return the ASCII digit instead. Use a Do Case to sort out the possibilities.

Write a function called letterToDigit() that is passed a letter and returns the ASCII digit
that corresponds to that letter. If there is no corresponding digit for a letter, such as Q, then return
a ‘*’ character. Use a Do Case to sort out the possibilities.

Now write a main() function that inputs a phone number in ASCII format. The number
might be entered as 699-9999 or as CAT-PETS. On a new line, echo print the original number
and then display “ = ” and then display the reverse of each digit or character of the number. For
example, if the CAT-PETS was entered, you program should display
CAT-PETS = 228-7387
Character Processing and Do Case 405

Or if 555-4242 were entered

555-4242 = JJJ-GAGA
Repeat until the end of the file is reached. Thoroughly test the program.

Problem Cs08-4 — Areas of Shapes

The areas of some shapes are given below.
Triangle: A = ½ base * height
Rectangle: A = height * width
Circle: A = PI * radius2
Square: A = side2

Define an enum called Shapes that has the following values: Triangle, Rectangle,
Square, Circle. Then define five doubles: base, height, width, radius, and side. Next, input a
shape, calculate and display its area. The first item on each line is a letter indicating which shape
the line represents. The character is T, R, C or S. The rest of the line contains the relevant
information for that shape. If it is a triangle, then after the T character comes the base and
height. If it is a rectangle, after the R character comes the height and width. If it is a circle, after
the C character comes the radius. If it is a square, after the S comes the side.

The program should input the character and then based upon its value, input the
remaining items and calculate the area accordingly. A switch statement can be used. Assume that
all data lines contain correct data. To test the program, use the provided file, shapes.txt.

Problem Engr08-1 — Integrate F(x) = sin (x) / x

Modify the Trap() and Simp() functions to perform integrations of

Specifically, use first 10 panels or trapezoids and then use 20 for both Trap() and Simp().
Display the results so that they can be easily compared. Format the output similar to this
Integration Results of F(x) = sin (x) / x

From xxxx to yyyy

method 10 panels 20 panels

Trap nnnnnnnnn nnnnnnnnn
Simp nnnnnnnnn nnnnnnnnn

Test the program with a = 1 and b = 6.5.

Character Processing and Do Case 406

Then, test the program with a = –3 and b = 3. Caution: you must modify the F(x)
function; the second test will crash at x = 0. However, the function F (x = 0) does have a value in
this particular case. Use that special value at point x = 0.

Problem Engr08-2 — Find the Integrals of Functions

Write a program(s) to integrate the following equations:

Use the Trap() function with a desired accuracy of .00001.

Problem Engr08-3 — Determine the RMS Current

(Electrical Engineering)
Alternating currents or oscillating currents produce a positive current flow and then a negative
flow. Over the course of one period, which may appear as a sine wave if monitored on an
oscilloscope, the average current flow may be zero since the positive values cancel the negative
ones out. Despite the fact that the average flow is zero does not mean that no work can be done
by that current. It does create heat and can do work. However, when working with such circuits,
one utilizes not the average current but the root mean square current flow.

The RMS current flow is given by the following equation.

where T is the period.

By using the square of the current i, the problem of positive and negative values canceling
each other out is eliminated. However, an integration is always involved when calculating IRMS.
For this problem, the i(t) function is given as

for 0 <= t <= T/2 and I(t) = 0 for T/2 <= t <=T.
Character Processing and Do Case 407

Thus, we need to integrate from a=0 through b=T/2. Assume that the period is to be one
second, so the T = 1 sec for this analysis. Write a program to calculate the IRMS in this case. To
assist you in debugging, the resultant integral of

equals 15.41261. Use a desired accuracy of .00001.

Problem Engr08-3 — Find the Cross-sectional Area of a River

(Water Resource Engineering)
To predict flooding effects of a river, a water resource engineer measures the cross-sectional area
of a riverbed. Using a boat, they have measured the depth of the river at uniform intervals across
its width. These depths have been tabulated below. All measurements are in meters. Write a
program to compute the cross-sectional area of the river from these data. Use the Trap() function
with the required number of trapezoids to match the data.
distance depth
from of
left bank river
0 0
2 2
4 3
6 6
8 10
10 12
12 10
14 8
16 6
18 4
20 0

Problem Engr08-4 — Menu for Volume of Shapes

Write a program that displays the following menu, gets the user’s choice and performs the
indicated calculation. Note additional prompts are needed to input the dimensions shown in
parentheses.

Volume of Shapes Calculator

A. Find the volume of a cube (side)

B. Find the volume of a rectangular block (height, width, length)
C. Find the volume of a sphere (radius)
D. Quit the program

Enter a letter choice:

Arrays 408

Chapter 9 — Arrays

Section A: Basic Theory

Definitions and Need for Arrays
An array is a consecutive series of data objects of the same data type. It represents a collection
of similar objects. For example, your local post office has an array of postal boxes, each identical
in size and shape (usually). Each object in the array is known as an array element. An element is
accessed by providing its subscript or index, which is a numerical value beginning with the
value zero. Similarly at the post office, to access the contents of a specific postal box, one must
know its box number. All box numbers must be unique; there cannot be two boxes with the same
number at a given post office. Subscripts follow a similar pattern, they must be unique and they
always begin with value zero. This is sometimes called zero-based offsets.

Why are arrays needed? Consider this problem. The Weather Service monitors the
temperatures at a particular location at six-minute intervals, yielding ten temperature
measurements every hour. They have collected a 24-hour period worth of observations, or a total
of 240 values. Now suppose they wish the report to display the average, the highest and the
lowest temperatures on the first line of the report followed by all of the actual temperatures
measured. The program needs to input and store all 240 observations so that the results can be
calculated and displayed first before it displays all of the data. If the result’s line could be
displayed last instead of first, then we could have just one input field, and write a simple loop to
input a temperature, perform the needed calculations and comparisons and then display that value
on the report. Clearly, one would not want to define 240 separate variables, write 240 separate
input operations. Imagine what the calculations coding would be like! Instead, here is an ideal
application for an array. We simply define one array of doubles to hold the temperatures; input
all of the temperatures into the array; then do our calculations and make the report, using and
reusing the values stored in the array.

Defining Arrays
To define an array, follow the variable name with [maximum number of elements].
double temps[240];
This defines temps to be an array of 240 doubles. Similarly, consider these array definitions
Arrays 409

long total[10];
char tally[50];
short count[100];
Here, total is an array of 10 longs; tally is an array of 50 characters; count is an array of 100
shorts.

However, good programming practice dictates that the numerical values should be
replaced by a more maintainable method. The array bounds should be either a #define symbol or
a const int variable. For example,
#include <iostream>
using namespace std;

const int MAXTEMPS = 240;

int main () {
double temps[MAXTEMPS];

Why use the constant integer MAXTEMPS instead of hard coding 240? From a program
maintenance point of view, array bounds often change. Suppose that the weather service went to
5-minute intervals instead of the 6-minute ones. Now there are 12 observations per hour or a total
of 288. With the constant integer approach shown above, all that would need to be done is
change the initial value from 240 to 288 and rebuild the program. If one had hard coded the 240,
then a thorough search of the entire program for all numerical values of 240 would have to be
made; such action is tedious and error prone.

Accessing Array Elements

An array is accessed by an integer type subscript that can be a constant, variable or an expression.
The subscript is also enclosed inside of [] brackets. The following are valid array references
temps[0] temps[i] temps[j + 1]
where i and j are both integer types.

The following are valid.

temps[i] = 42; // stores 42 in the ith element
temps[i]++; // adds 1 to the ith element of temps
cin >> temps[i]; // input the ith element of temps
cout << temps[i]; // output the ith element of temps
The following are invalid.
cin >> temps;
cout << temps;
An array cannot be input as a group in a single input statement. All assignment and math
operations must be done on an individual element basis. So these are invalid as well.
temps += count;
temps++;
Arrays 410

One array cannot be assigned to another array; instead, a loop must be written to copy
each element.
double save_temps[MAXTEMPS];
save_temps = temps; // illegal - use a loop instead
Rather, if a copy of the array is needed, then the following loop accomplishes it.
for (i=0; i<MAXTEMPS; i++) {
save_temps[i] = temps[i];
}

When using subscripts, it is vital that the subscript be within the range defined by the
array. In the case of the 240 temperatures, valid subscripts are in the range from 0 to 239. Note a
common cause of errors is to attempt to access temps[240], which does not exist as it would be
the 241st element. This attempt to access an element that is beyond the bounds of the array is
called subscript out of bounds.

What exactly happens when one uses a subscript that is out of bounds? To see the effect,
let’s examine how memory is laid out for an array. Suppose our company had five departments
and we wished to store the total income for each department. We might have the following.
const int MAX_DEPTS = 5;
int main () {
long totals[MAX_DEPTS];

The array totals consists of 5 long integers in a row. Assume that each long occupies 4
bytes of memory and that the array begins at memory address 100.
Memory
Address
100 104 108 112 116 120
---------------------------------------------------------------
| | | | | | | |
| | | | | | | |
---------------------------------------------------------------
[0] [1] [2] [3] [4] x
The array ends at “x” marks the spot. Memory beginning at address 120 is beyond the end of the
array. What is there? It could be the memory of some other variable in the program; it may not
even be a part of your program.

Now if the subscript out of bounds is in retrieval mode, such as

grandtotal += totals[5];
Then the computer attempts to access what it thinks is element 6 at memory address 120 and
retrieves the four bytes there, assumes that it is a long and adds it into grandtotal. Even though it
is effectively a garbage value, any combination of bits turned on with the byte corresponds to an
integer value of some kind. So if the memory exists, something, usually quite bizarre, is added
into grandtotal. However, if totals were an array of doubles, then not every possible
combination of bits within an 8-byte memory area corresponds to a valid floating point number.
In such cases a runtime error can result with a message of floating point overflow or underflow.
Arrays 411

On the other hand, if the subscript out of bounds is in a storage mode, such as
total[5] = 42;
then things become more unpredictable. If the memory is occupied by another program variable,
then that variable is overlaid. On Windows 95/98 systems, if that memory is actually part of
some Windows system code (part of a runtime dll, for example), then the code is overlaid.
Further, if that Windows code is ever actually subsequently executed, wildly unpredictable things
can occur, from a simple page fault exception, to a system lock up and reboot, to a resetting of
the computer’s CMOS settings, to a reformatting of hard drives, and so on. Under Windows NT,
any attempt to overlay system code is automatically handled by terminating the offending
program with a system exception.

Thus, when using arrays, the programmer must always be certain that subscripts are
within the bounds of the array.

Methods of Inputting Data into an Array

Okay, so with that in mind, how can these 240 temperatures be input into the array temps?
Consider the following methods.

Method A: Inputting a Known Number of Elements

ifstream infile;
infile.open ("myfile.txt");
if (!infile) {
cerr << "Cannot open myfile.txt\n";
return 1;
}
int i;
for (i=0; i<MAXTEMPS && infile; i++) {
infile >> temps[i];
}
if (!infile) {
cerr << " Error: Could not input " << MAXTEMPS
<< " temperatures\n";
infile.close ();
return 1;
}
infile.close ();

This method can be used only when there are precisely MAXTEMPS, 240, long values
in the file. Memory cannot be overwritten because of the for loop’s test condition. To guard
against bad data, the loop’s test condition also checks on infile’s goodness.
Arrays 412

Unfortunately, there are not many circumstances in which the precise number of data
values to be input is known at compile time or when the program is written. An alternative that is
often used is to make the first line of the input file contain the number of items in the array that
follows in the file.

Method B: Inputting the Number of Array Elements To Be Input

Here, the number of temperatures in the file is inputted first. If that number exceeds the
maximum size of the array, then an error message is printed. One could abort the program or test
it using the maximum number of temperatures the array can hold. What is very important in this
example is the use of the variable numTemps that is to hold the actual number of temperatures
in the array on THIS run of the program.

ifstream infile;
infile.open ("myfile.txt");
if (!infile) {
cerr << "Cannot open myfile.txt\n";
return 1;
}
int i;
int numTemps;
infile >> numTemps;
if (numTemps > MAXTEMPS) {
cerr << "Error: too many temperatures\n"
<< " using the first " << MAXTEMPS << endl;
numTemps = MAXTEMPS;
}
for (i=0; i<numTemps && infile; i++) {
infile >> temps[i];
}
if (!infile) {
cerr << " Error: Could not input " << numTemps
<< " temperatures\n";
infile.close ();
return 1;
}
infile.close ();

The drawback of this approach is that it can place a terrible burden on the user to count
accurately the number of temperatures to be input. While this is fine when the numbers are small,
I would not want to count to 240! Instead, the most common approach is to have the program
input all values until it reaches the end of the file.
Arrays 413

Method C: Inputting an Unknown Number of Elements Until EOF Is

Reached
ifstream infile;
infile.open ("myfile.txt");
if (!infile) {
cerr << "Cannot open myfile.txt\n";
return 1;
}
int i = 0;
int numTemps;
while (i<MAXTEMPS && infile >> temps[i]) {
i++;
}
// guard against too many temperatures in the file
if (i == MAXTEMPS && infile >> ws && infile.good()) {
cerr << "Error: too many temperatures\n";
infile.close ();
return 2;
}
// guard against bad data in the input file
else if (!infile.eof() && infile.fail()) {
cerr << "Error: bad data in the file\n";
infile.close ();
return 3;
}
// set the number of temps in the array on this run
numTemps = i;
infile.close ();

In the while loop, notice that the first test condition guarantees that the subscript must be
within the bounds of the array. Recall the effect of the && operator — the second half of the test
condition is never executed when i becomes equal or exceeds MAXTEMPS. The second half of
the test condition inputs a single temperature and returns a good state if successful. The loop
ends when the end of file is reached or bad data occurs or the array bound is exceeded.

Note the infile >> ws && infile.good() action. The ws is another manipulator function
that skips over white space. It provides us a convenient way to find out if there are more data still
in the input stream. If it can successfully skip over white space, then there are more data to be
input and the stream is in the good state. If it is not successful, then it is most likely the end of
file. Remember that the current position in an input stream is located right after the last character
that was input. Often the current position in the input stream is pointing to the <cr> at the end of
the line that was just input or is pointing to the EOF code (the hidden ^Z byte). Consider the
following two circumstances.
Arrays 414

Assume that the program inputs pairs of quantity and cost values and has just finished
this instruction.
infile >> quantity >> cost;
Assume the following is in the file being read. After inputting the first line, the current position
in the input stream is pointing to the boldfaced <cr>.
123 42.50<cr>
345 99.99<cr>
^Z
Coding
infile >> ws;
successfully skips over white space and positions the current position to the 3 digit of the 345
quantity value on the next line. After inputting the second line, the current position is now
pointing to the second line’s <cr>. Now executing
infile >> ws;
is unsuccessful as the end of file is reached and infile goes into the EOF state.

Next, the two error circumstances are sorted out, bad data and too much data. If the
count equals MAXTEMPS then two possibilities exist. The program may have just inputted
the last value in the file and the current position in the input stream is on the < CRLF>
before the EOF marker. Or there actually are some more data in the file. To sort out which is
the case, attempt to skip over whitespace. If there are no more data, the stream goes into the
non-good state and all is well. On the other hand, if there are more values in the input file,
then the skip over whitespace succeeds and the stream is in the good state, triggering the error
message. The bad data circumstance is easily checked by calling the fail() function.

An alternative method to detecting too much data in the file is to define a character, say
c, and attempt to extract it. If it is successful, then there are more data in the file.
// guard against too many temperatures in the file
char c;
if (i == MAXTEMPS && infile >> c) {
cerr << "Error: too many temperatures\n";
infile.close ();
return 2;
}

Again, it is vital to save the number of values actually inputted on this run. In this
example, it is saved in the variable numTemps;

Working with Arrays —The Calculations

With the array loaded with either MAXTEMPS or numTemps worth of temperatures, a
program often does one or more calculations using the array element values. Quite often,every
element in the array must be used. In this example, the average temperature must be found. If
Arrays 415

the program adds each temperature to a sum and then divides that sum by the number of
temperatures, the average can be found. In the following coding, it is assumed that
numTemps holds the actual number of temperatures inputted in this execution of the program.
double sum = 0;
for (i=0; i< numTemps; i++) {
sum += temps[i];
}
double average = sum / numTemps;

How are the highest and lowest temperatures found? Assuming that high and low
variables are to contain the results, each element in the array, that is, each temperature, must be
compared to these two values. If the next temperature is greater than or less than the
corresponding high or low temperature, then that value replaces the current high or low value.
This can be done as follows.
double high = temps[0];
double low = temps[0];
for (i=1; i<numTemps; i++) {
if (temps[i] > high)
high = temps[i];
if (temps[i] < low)
low = temps[i];
}
// here high and low have the largest and smallest
// temperatures in the array

Here is a common coding error that yields a mostly-working program. Can you spot the
error?
double high = 0;
double low = 0;
for (i=0; i<numTemps; i++) {
if (temps[i] > high)
high = temps[i];
if (temps[i] < low)
low = temps[i];
}
If not, suppose the location is Florida in the summer. What will this code produce as the low
temperature for the day? A temperature of 0 degrees would mark a catastrophe for the citrus
crops! Now suppose that this location is Fairbanks, Alaska in the dead of winter. What will this
code produce for highest temperature? Hint: 0 degrees would be considered a heat wave.

Working with arrays: the Output Process

Frequently all the elements of the array must be displayed. Writing another loop, outputting each
element in turn does this.
Arrays 416

cout << setprecision (1);

for (i=0; i<numTemps; i++) {
cout << setw(12) << temps[i] << endl;
}

Initializing an Array
When an array is defined or declared, it may also be initialized if desired. Suppose that one
needed to define an array of totals to hold the total sales from each of ten departments within a
store. One could code
double totals[10];
However, in practice, the first usage of any element in the array is likely to be of the form
total[j] = total[j] + sales;
This requires that each element in the array be initialized to zero. The initialization syntax in
general is
data type variable-name[limit] = {value0, value1,
value2, … valuen};
Initialization of arrays begins with a begin brace { and ends with an end brace }. The syntax
provides for giving each element its unique initial value. In the example of store totals, one could
code
double totals[10] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0};

However, C++ provides a default value of 0 for all elements not specifically initialized. In
reality, one would define totals this way
double totals[10] = {0};
This tells the compiler to place a 0 in the first element and then to place a default value (that
happens to also be 0) in all remaining elements. Once you stop providing initial values, all
remain elements are defaulted to a 0 starting value. Values neither can be omitted nor can there
be more values than array elements. The following coding is illegal, since once you stop
providing values (after element zero in this case) you cannot later on resume providing explicit
values.
double totals[10] = {0, , , 0};// error - cannot omit values

Passing Arrays to Functions

Very commonly, arrays are passed to functions so that the workload can be broken down into
more manageable units. Often a program has a loadArray() function whose purpose is to
encapsulate the entire array’s input operation. The main() function is then left with just a single
function call to loadArray(), thereby streamlining main(). What would main() need to give
loadArray() and what should the function return? The loadArray() function must be given the
array to fill up and the maximum number of elements in the array; it must return the actual
number of elements inputted on this run. Thus, if we were to write a loadArray() function to
Arrays 417

load the array of temperatures, the only coding in main() would now be
int numTemps = loadArray (temps, MAXTEMPS);

This immediately raises the question how are arrays passed to functions. The answer lies
in asking what is the data type of the name of an array? The data type of the name of an array is
always a constant pointer or memory address to the first element. Thus, in the above
temperature’s example, the name temps is a const double*. This reads backwards “is a pointer
to a double that is constant.” A pointer is just the memory address of something. In this context,
when used without the [i] notation, temps is just a memory address where the 240 doubles
begin. So when main() calls loadArray(), it is not passing a copy of the entire array of 240
doubles, but rather just the memory location at which the array begins. This is highly desirable;
consider the overhead of actually trying to pass a copy of the entire array — that is, 240 doubles
or 1920 bytes, assuming a double occupies 8 bytes. When coding the prototype for loadArray(),
two different notations can be used.
int loadArray (double temps[], int limit);
int loadArray (double* temps, int limit);
In the first version, the [] tells the compiler that the preceding symbol temps is not just a double
but is also an array of doubles, which therefore means that this parameter is really just the
memory address where the array begins.

Notice one significant detail when setting up the prototype of a function that has an array
as a parameter. The actual array bound is not used as it was when defining the array. Compare the
two statements in main():
// defines storage for the array
double temps[MAXTEMPS];

// prototype for the loadArray function

int loadArray (double temps[], int limit);
The first defines the array and allocates space for 240 doubles. The second line says that within
loadArray(), the temps parameter is an array of doubles. The compiler never cares how many
elements are in this passed array, only that it is an array of doubles. One could also have coded
the prototype as
int loadArray (double temps[MAXTEMPS], int limit);
However, the compiler simply ignores the array bounds; only the [] is important, for the [] tells
the compiler that the symbol is an array. As usual, the programmer must be vigilant in making
sure that the array bound is not exceeded by passing in that maximum number of elements as the
second parameter and using it to avoid inputting too many values.

Next, let’s see how the loadArray() function could be implemented. The changes from
the previous version are in boldface.
int loadArray (double temps[], int limit) {
ifstream infile;
infile.open ("myfile.txt");
if (!infile) {
Arrays 418

cerr << "Cannot open myfile.txt\n";

exit (1);
}
int i = 0;
while (i<limit && infile >> temps[i]) {
i++;
}
// guard against too many temperatures in the file
if (i == limit && infile >> ws && infile.good()) {
cerr << "Error: too many temperatures\n";
infile.close ();
exit (2);
}
// guard against bad data in the input file
else if (!infile.eof() && infile.fail ()) {
cerr << "Error: bad data in the file\n";
infile.close ();
exit (3);
}
// set the number of temps in the array on this run
infile.close ();
return i;
}

The first change to notice is the reference to the constant integer MAXTEMPS is
replaced with the parameter limit, which should contain the value MAXTEMPS has in it.
However, one could have just used the global constant integer MAXTEMPS and not even
bothered passing it as the second parameter, limit. But doing so is not the optimum method. Very
often in larger applications, many of the functions are actually stored in their own separate cpp
files. If loadArray() were in its own file, LoadArray.cpp, then the compiler would not have
access to the constant integer MAXTEMPS. By passing its value of 240 as the parameter limit,
all dependencies upon the symbol MAXTEMPS defined in the main.cpp file have been
removed. This makes a better design and a more flexible program.

The other change arises from the detection of error situations, such as, being unable to
open the file, or having too much data. When the coding to load the array is found in the main()
program, after displaying the error message, the instruction
return 1;
causes the program to return back to DOS and terminate. However, in loadArray(), any return
statement causes the computer to return back into main() at the point where main() calls or
invokes the loadArray() function and then copies the returned integer into the variable
numTemps! Oops! This is not what is desired at all. A function can directly invoke program
termination by calling the exit() function, which takes one parameter, the integer return code to
give to DOS. The coding
exit (1);
Arrays 419

immediately terminates the program without ever returning to the calling function and gives a
return code of 1 to DOS.

The main() function could be streamlined even further by creating three more functions,
findAverage(), findRanges() and printResults(). The prototypes for findAverage() and
printResults() are straightforward.
double findAverage (double temps[], int num_temps);
void printResults(double temps[], int num_temps,
double avg, double high, double low);
The findRanges() function cannot return two values, that is, the highest and lowest temperatures,
so these two variables must be passed by reference.
void FindRanges (double temps[], int num_temps,
double& high, double& low);

With these changes, main() now becomes very streamlined.

int main () {
double temps[MAXTEMPS];
int numTemps;
double average;
double high;
double low;

numTemps = LoadArray (temps, MAXTEMPS);

average = FindAverage (temps, numTemps);
FindRanges (temps, numTemps, high, low);
PrintResults (temps, numTemps, average, high, low);

return 0;
}

Here is the complete program listing for Basic09a. Pay particular attention to how the
array of temperatures is passed to the functions and how it is then used within the functions.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic09a - Produce Temperature Statistics *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Basic09a Produce temperature statistics from a set of temps */*
* 4 /* */*
* 5 /* Inputs up to 240 temperatures and calculates the average, */*
* 6 /* highest and lowest temperatures and then displays the */*
* 7 /* temperatures as inputted */*
* 8 /* */*
* 9 /***************************************************************/*
* 10 *
* 11 #include <iostream> *
* 12 #include <iomanip> *
* 13 #include <fstream> *
Arrays 420

* 14 using namespace std; *

* 15 *
* 16 const int MAXTEMPS = 240; // the maximum number of temperatures *
* 17 *
* 18 int LoadArray (double temps[], int limit); *
* 19 double FindAverage (double temps[], int numTemps); *
* 20 void FindRanges (double temps[], int numTemps, *
* 21 double& high, double& low); *
* 22 void PrintResults (double temps[], int numTemps, *
* 23 double avg, double high, double low); *
* 24 *
* 25 int main () { *
* 26 double temps[MAXTEMPS]; // the array of temperatures *
* 27 int numTemps; // the actual number inputted *
* 28 double average; // the average of all temperatures *
* 29 double high; // the highest temperature input *
* 30 double low; // the lowest temperature input *
* 31 *
* 33 cout << fixed; // setup floating point output format *
* 35 *
* 36 numTemps = LoadArray (temps, MAXTEMPS); // load the array *
* 36.2 if (!numTemps) { // guard against no data *
* 36.4 cout << "No temperatures in the file\n"; *
* 36.6 return 1; *
* 36.8 } *
* 37 average = FindAverage (temps, numTemps); // calc the average *
* 38 FindRanges (temps, numTemps, high, low); // find high and low *
* 39 PrintResults (temps, numTemps, average, high, low); *
* 40 *
* 41 return 0; *
* 42 } *
* 43 *
* 44 /***************************************************************/*
* 45 /* */*
* 46 /* LoadArray: load an array of temperatures */*
* 47 /* */*
* 48 /***************************************************************/*
* 49 *
* 50 int LoadArray (double temps[], int limit) { *
* 51 // open the data file - abort the program if not found *
* 52 ifstream infile; *
* 53 infile.open ("temperatures.txt"); *
* 54 if (!infile) { *
* 55 cerr << "Cannot open temperatures.txt\n"; *
* 56 exit (1); *
* 57 } *
* 58 *
* 59 // input all the temperatures upto limit total *
* 60 int i = 0; *
* 61 while (i<limit && infile >> temps[i]) { *
* 62 i++; *
* 63 } *
Arrays 421

* 64 *
* 65 // guard against too many temperatures in the file *
* 66 if (i == limit && infile >> ws && infile.good()) { *
* 67 cerr << "Error: too many temperatures\n"; *
* 68 infile.close (); *
* 69 exit (2); *
* 70 } *
* 71 *
* 72 // guard against bad data in the input file *
* 73 else if (!infile.eof() && infile.fail()) { *
* 74 cerr << "Error: bad data in the file\n"; *
* 75 infile.close (); *
* 76 exit (3); *
* 77 } *
* 78 infile.close (); *
* 79 *
* 80 // return the number of temperatures in the array on this run *
* 81 return i; *
* 82 } *
* 83 *
* 84 /***************************************************************/*
* 85 /* */*
* 86 /* FindAverage: determines the average temperature */*
* 87 /* */*
* 88 /***************************************************************/*
* 89 *
* 90 double FindAverage (double temps[], int numTemps) { *
* 91 double sum = 0; *
* 92 for (int i=0; i< numTemps; i++) { *
* 93 sum += temps[i]; // accumulate all temperatures *
* 94 } *
* 95 return sum / numTemps; // return the average temperature *
* 96 } *
* 97 *
* 98 /***************************************************************/*
* 99 /* */*
* 100 /* FindRanges: determines the highest and lowest temperatures */*
* 101 /* */*
* 102 /***************************************************************/*
* 103 *
* 104 void FindRanges (double temps[], int numTemps, *
* 105 double& high, double& low) { *
* 106 high = temps[0]; // initialize high to the first temperature *
* 107 low = temps[0]; // initialize low to the first temperature *
* 108 *
* 109 // now check all other temperatures against these *
* 110 for (int i=1; i<numTemps; i++) { *
* 111 if (temps[i] > high) *
* 112 high = temps[i]; // replace high with the larger temperature*
* 113 if (temps[i] < low) *
* 114 low = temps[i]; // replace low with the lower temperature *
* 115 } *
Arrays 422

* 116 // high & low now have the largest & smallest temps in the array*
* 117 } *
* 118 *
* 119 /***************************************************************/*
* 120 /* */*
* 121 /* PrintResults: prints the complete report */*
* 122 /* */*
* 123 /***************************************************************/*
* 124 *
* 125 void PrintResults (double temps[], int numTemps, *
* 126 double avg, double high, double low) { *
* 127 cout << setprecision (1); *
* 128 *
* 129 // display heading of report *
* 130 cout << "Temperature Analysis\n\n"; *
* 131 *
* 132 // display the statistical results lines *
* 133 cout << "Average: " << setw (6) << avg << endl *
* 134 << "Highest: " << setw (6) << high << endl *
* 135 << "Lowest: " << setw (6) << low << endl << endl; *
* 136 *
* 137 // display column heading *
* 138 cout << "Observations\n"; *
* 139 *
* 140 // display all temperatures as entered *
* 141 for (int i=0; i<numTemps; i++) { *
* 142 cout << setw(7) << temps[i] << endl; *
* 143 } *
* 144 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Here is a sample test run showing the output of the program.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Basic09a - Sample Test Run *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Temperature Analysis *
* 2 *
* 3 Average: 95.0 *
* 4 Highest: 100.0 *
* 5 Lowest: 90.0 *
* 6 *
* 7 Observations *
* 8 91.0 *
* 9 92.0 *
* 10 93.0 *
* 11 94.0 *
* 12 95.0 *
* 13 96.0 *
* 14 97.0 *
* 15 98.0 *
* 16 99.0 *
Arrays 423

* 17 100.0 *
* 18 99.0 *
* 19 98.0 *
* 20 97.0 *
* 21 96.0 *
* 22 95.0 *
* 23 94.0 *
* 24 93.0 *
* 25 92.0 *
* 26 91.0 *
* 27 90.0 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Section B: Computer Science Examples

Cs09a — Sales Data Analysis
Acme Sales wishes a statistical analysis of their monthly sales data. The input file contains the
sales person’s id number and their sales for that transaction. The sales person ids range from 1 to
10 inclusively. Each person is paid a commission based on the total of those sales as follows:
if the sales are less than 100.00, commission is 5%
if the sales are 100.00 or above but less than 500.00, commission is 6%
if the sales are 500.00 or above, the commission is 7%
Accumulate the sales for each sales person and their commission. When all sales data have been
input, produce the sales analysis report which lists the results for each person on one line. Print
that sales person’s id number, their total sales, their average sales, their total commission, their
average commission and their percentage their average commission from the company average
commission. At the end of the report, show the grand totals for each of the four dollar amounts.
Note that any given sales person might not have sold anything in this particular month.

Since this report is rather complex, here is the output from the sample run.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs09a Sales Data Analysis - Sample Report *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Acme Monthly Sales Analysis Report *
* 2 *
* 3 ID Number Total Average Total Average Percentage *
* 4 Sales Sales Sales Commission Commission of Avg Comm *
* 5 *
* 6 *
* 7 1 4 1300.00 325.00 88.00 22.00 -46.43 *
* 8 2 6 2600.00 433.33 173.00 28.83 -29.79 *
* 9 3 6 3200.00 533.33 217.00 36.17 -11.93 *
* 10 4 5 3200.00 640.00 220.00 44.00 7.15 *
* 11 5 6 4400.00 733.33 308.00 51.33 25.00 *
* 12 *
Arrays 424

* 13 6 5 3800.00 760.00 266.00 53.20 29.55 *

* 14 7 6 3300.00 550.00 228.00 38.00 -7.46 *
* 15 8 0 0.00 0.00 0.00 0.00 0.00 *
* 16 9 4 2100.00 525.00 143.00 35.75 -12.94 *
* 17 10 4 3600.00 900.00 246.00 61.50 49.76 *
* 18 ----- ------- ------- ------- ------- *
* 19 46 27500.00 597.83 1889.00 41.07 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Let’s examine some of the needed processing details before making the Top-Down
Design. It’s clear that the input operation must bring in each set of sales data and add the sales to
the corresponding salesperson’s total and increment that sales person’s number of sales. At that
time, we can easily calculate the commission on that sale and also add it into that sales persons
total commission. Effectively, we are just accumulating the basic data, hence an Accumulate()
function.

The calculations pose a more interesting situation. Once all the data have been
accumulated, we can traverse each sales person in the array and determine their two averages,
average sales and commission. However, to obtain the percentage of average commission, the
grand totals must first be determined so that the company average commission, here $41.07 can
be found. Once that is known, we can go back through each sales person and calculate their
percentage. So a DoCalcs() function is definitely in order. With the values known, it is a simple
matter to print the contents of all the arrays and grand total fields in a PrintReport() function.
This yields our Top-Down Design shown in Figure 9.1.

Figure 9.1 Top-Down Design

Examine first the coding for the main() function given below in the Cs10a listing. The
main() function defines all of the arrays and totals, initializing all to 0. Notice that the array
bounds is set to 11. The element at subscript 0 is being ignored in this example. A variation
might be to instead use the subscript 0 elements in the arrays to contain the grand totals, thereby
reducing the number of variables that need to be passed to the functions. main() then simply calls
in turn Accumulate(), DoCalcs() and PrintReport(), very streamlined indeed.

What will main storage look like for function main()? How do we draw effective
graphical representations for an array? Remember, one of the purposes of drawing the memory
layouts is to have a convenient way to desk check the solution. If an array were defined to contain
100 elements, then theoretically, one would have to draw 100 adjacent and connected boxes to
represent that array. However, such is not really needed. I usually just include several
Arrays 425

representing the first few and the last one. The main() function needs a const int for the array
bounds, say MAXID. It needs six arrays for the accumulated results for each salesperson. Let’s
call these: totalSales, totalComm, numSales, avgSales, avgComm and percent. Additionally,
there are five grand total fields called grandNum, grandSales, grandComm, grandAvgSales
and grandAvgComm — all of which must be initialized to 0. Main storage for the main()
function appears in Figure 9.2.

Figure 9.2 Main Storage for Function main()

In Accumulate(), the input fields, id and sales, will hold the data actually input from the
input file. Then, if the id number is invalid, an error message is written to the cerr stream and is
visible onscreen, not in the report. Realistically, the error message should also appear in the
report. Once validated, the sales person’s id is used as the subscript into the arrays. Note that no
checking is done for the possibility of bad data being entered. In reality, one should check
infile.fail() and report on that circumstance when the loop has finished. If the id number is within
range, then accumulate the totals and calculate the commission and accumulate it as well.

The Accumulate() function’s parameters are the three arrays — totalsales, totalcomm,
and numsales — and a copy of the const int MAXID here called limit is as follows. Notice
how I have laid out main storage for the array parameters in Figure 9.3. Notice how I set up the
arrays in main() and in Accumulate().
Arrays 426

Figure 9.3 Main Storage for Accumulate()

Remember when an array is passed, only the memory location of that array is actually
passed, so Accumulate()’s numsales is really pointing to main()’s numSales as shown. Here is
the coding sequence for Accumulate().
open infile
if it fails to open, display an error message on cerr and exit the program
while (infile >> id >> sales) do the following
if (id < 0 || id > limit) then display an error message to cerr and skip this set
otherwise do these steps
numsales[id] = numsales[id] + 1
totalsales[id] = totalsales[id] + sales
if (sales < 100.00) then commission = .05 * sales
else if (sales < 500.00) then commission = .06 * sales
else commission = .07 * sales
totalcomm[id] = totalcomm[id] + commission;
end otherwise clause
end while
close infile

The other functions can similarly be sketched out. The function DoCalcs() is passed
arrays: totalSales, totalComm, numSales, avgSales, avgComm, and percent. It is passed the
limit, and a reference to the grand totals: grandNum, grandSales, grandComm,
grandAvgSales, and grandAvgComm. The sequence of steps are as follows.
for i=1 and continuing until i<limit each time through the loop increment i
if (numSales[i] != 0) then
avgSales[i] = totalSales[i] / numSales[i];
avgComm[i] = totalComm[i] / numSales[i];
end the if
Arrays 427

add numSales[i] to grandNum

add totalSales[i] to grandSales
add totalComm[i] to grandComm
end the for loop
// find the grand averages
grandAvgSales = grandSales / grandNum;
grandAvgComm = grandComm / grandNum;
// use the grand averages to calculate each sales person's percentage
for (i=1; i<limit; i++) do the following
if (numSales[i] != 0)
percent[i] = (avgComm[i] – grandAvgComm) * 100 / grandAvgComm;
end the for loop

The function PrintReport() is straightforward with one exception. We are to leave a

blank line every 5 detail lines. This is easily done by
if ((i-1) % 5 == 0) // insert blank line every five lines

This implementation of Cs09a shown below is an example of a “mostly working”

program. Specifically, one possible situation was NOT checked. If that condition should occur,
then the program would crash with a divide by zero fatal error at run time. Can you spot what
that situation is? Clue: the run time error would occur at line 114. Okay. If the input file contains
no valid data (is empty or otherwise goofed up with all wrong data), the Accumulate() function
does not check for this possibility. Neither does the main() function which then calls DoCalcs().
Since all arrays and variables are initialized to zero, the disaster does not occur until line 114.
Here, grandNum still contains 0 and is used as the divisor. How can the program be
bulletproofed against such occurrences? There are several ways. See if you can find one or more
ways to fix Cs09a.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs09a Produce the monthly sales analysis report *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs09a Produce the monthly sales analysis report */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 using namespace std; *
* 11 *
* 12 const int MAXID = 11; // the maximum number sales persons *
* 13 // their id's range from 1 to 10 *
* 14 // ignore element [0] *
* 15 *
Arrays 428

* 16 void Accumulate (double totalSales[], double totalComm[], *

* 17 int numSales[], int limit); *
* 18 void DoCalcs (double totalSales[], double totalComm[], *
* 19 int numSales[], double avgSales[], double avgComm[],*
* 20 int limit, double percent[], int& grandNum, *
* 21 double& grandSales, double& grandComm, *
* 22 double& grandAvgSales, *
* 23 double& grandAvgComm); *
* 24 void PrintReport (double totalSales[], double totalComm[], *
* 25 int numSales[], double avgSales[],double avgComm[],*
* 26 int limit, double percent[], int grandNum, *
* 27 double grandSales, double grandComm, *
* 28 double grandAvgSales, double grandAvgComm); *
* 29 *
* 30 int main () { *
* 31 double totalSales[MAXID] = {0}; // total sales of each person *
* 32 double totalComm[MAXID] = {0}; // total commission each person *
* 33 int numSales[MAXID] = {0}; // number of sales each person *
* 34 double avgSales[MAXID] = {0}; // average sales of each person *
* 35 double avgComm[MAXID] = {0}; // average commission per sale *
* 36 double percent[MAXID] = {0}; // percent of avg commission *
* 37 int grandNum = 0; // grand number of sales *
* 38 double grandSales = 0; // grand total sales of everyone*
* 39 double grandComm = 0; // grand commission paid *
* 40 double grandAvgSales = 0; // average amount of each sale *
* 41 double grandAvgComm = 0; // average comm paid per sale *
* 42 *
* 43 Accumulate (totalSales, totalComm, numSales, MAXID); *
* 44 DoCalcs (totalSales, totalComm, numSales, avgSales, avgComm, *
* 45 MAXID, percent, grandNum, grandSales, grandComm, *
* 46 grandAvgSales, grandAvgComm); *
* 47 PrintReport (totalSales, totalComm, numSales, avgSales, avgComm,*
* 48 MAXID, percent, grandNum, grandSales, grandComm, *
* 49 grandAvgSales, grandAvgComm); *
* 50 return 0; *
* 51 } *
* 52 *
* 53 /***************************************************************/*
* 54 /* */*
* 55 /* Accumulate: input the data, accumulating all data */*
* 56 /* */*
* 57 /***************************************************************/*
* 58 *
* 59 void Accumulate (double totalsales[], double totalcomm[], *
* 60 int numsales[], int limit) { *
* 61 // open the data file - abort the program if not found *
* 62 ifstream infile; *
* 63 infile.open ("SalesData.txt"); *
* 64 if (!infile) { *
* 65 cerr << "Cannot open salesdata.txt\n"; *
* 66 exit (1); *
* 67 } *
Arrays 429

* 68 *
* 69 int id; *
* 70 double sales; *
* 71 int lineNumber = 1; *
* 72 while (infile >> id >> sales) { *
* 73 if (id < 1 || id >= limit) { *
* 74 cerr << "Error: invalid sales person number. It was " << id *
* 75 << " on line " << lineNumber *
* 76 << "\nAction taken - skipping this person\n"; *
* 77 } *
* 78 else { *
* 79 numsales[id]++; // increment their number of sales *
* 80 totalsales[id] += sales; // accumulate their sales *
* 81 double commission; // calculate the comm on this sales *
* 82 if (sales < 100.00) *
* 83 commission = .05 * sales; *
* 84 else if (sales < 500.00) *
* 85 commission = .06 * sales; *
* 86 else *
* 87 commission = .07 * sales; *
* 88 totalcomm[id] += commission; // accumulate commission *
* 89 } *
* 90 lineNumber++; *
* 91 } *
* 92 if (!infile.eof()) { *
* 93 cerr << "Error - bad data in the input file on line " *
* 94 << lineNumber << endl; *
* 95 infile.close (); *
* 96 exit (1); *
* 97 } *
* 98 infile.close (); *
* 99 } *
* 100 *
* 101 /***************************************************************/*
* 102 /* */*
* 103 /* DoCalcs: find the averages, grand totals and percentage */*
* 104 /* */*
* 105 /***************************************************************/*
* 106 *
* 107 void DoCalcs (double totalSales[], double totalComm[], *
* 108 int numSales[], double avgSales[], double avgComm[],*
* 109 int limit, double percent[], int& grandNum, *
* 110 double& grandSales, double& grandComm, *
* 111 double& grandAvgSales, double& grandAvgComm) { *
* 112 int i; *
* 113 // calculate all sales person's average sales and commission *
* 114 for (i=1; i<limit; i++) { *
* 115 if (numSales[i] != 0) { *
* 116 avgSales[i] = totalSales[i]/numSales[i]; *
* 117 avgComm[i] = totalComm[i]/numSales[i]; *
* 118 } *
* 119 grandNum += numSales[i]; // accumulate the grand totals *
Arrays 430

* 120 grandSales += totalSales[i]; *

* 121 grandComm += totalComm[i]; *
* 122 } *
* 123 // find the grand averages *
* 124 grandAvgSales = grandSales / grandNum; *
* 125 grandAvgComm = grandComm / grandNum; *
* 126 // use the grand averages to calculate sales person's percentage*
* 127 for (i=1; i<limit; i++) { *
* 128 if (numSales[i] != 0) { *
* 129 percent[i] = (avgComm[i] - grandAvgComm) * 100 / grandAvgComm;*
* 130 } *
* 131 } *
* 132 } *
* 133 *
* 134 /***************************************************************/*
* 135 /* */*
* 136 /* PrintReport: prints the complete report */*
* 137 /* */*
* 138 /***************************************************************/*
* 139 *
* 140 void PrintReport (double totalSales[], double totalComm[], *
* 141 int numSales[], double avgSales[], double avgComm[],*
* 142 int limit, double percent[], int grandNum, *
* 143 double grandSales, double grandComm, *
* 144 double grandAvgSales, double grandAvgComm) { *
* 145 ofstream outfile ("SalesReport.txt"); *
* 146 // setup floating point output format *
* 147 outfile << fixed << setprecision (2); *
* 150 *
* 151 // display heading of report and column headings *
* 152 outfile << *
* 153 " Acme Monthly Sales Analysis Report\n\n";*
* 154 outfile << " ID Number Total Average Total Average" *
* 155 " Percentage\n" *
* 156 << " Sales Sales Sales Commission Commission"*
* 157 " of Avg Comm\n\n"; *
* 158 // print each sales person's results *
* 159 for (int i=1; i<limit; i++) { *
* 160 if ((i-1) % 5 == 0) // insert blank line every five lines *
* 161 outfile << endl; *
* 162 outfile << setw(3) << i << setw(5) << numSales[i] *
* 163 << setw(11) << totalSales[i] << setw(10) << avgSales[i]*
* 164 << setw(10) << totalComm[i] << setw(10) << avgComm[i] *
* 165 << setw(10) << percent[i] << endl; *
* 166 } *
* 167 outfile <<" ----- ------- ------- ------- -------\n";*
* 168 outfile << " " << setw (5) << grandNum *
* 169 << setw (11) << grandSales *
* 170 << setw (10) << grandAvgSales *
* 171 << setw (10) << grandComm *
* 172 << setw (10) << grandAvgComm << endl; *
* 173 outfile.close(); *
Arrays 431

* 174 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Section C: Engineering Examples

Engr09a — Vector Coordinate Conversions
A vector has both a magnitude and a direction. Unlike a scalar, a quantity that can be represented
by a single number, a vector requires two numbers. Vectors can be defined by either their
magnitude and direction (in polar coordinates) or the projection of the vector along the axis of a
rectangular coordinate system. The two representations are equivalent. Frequently, one needs to
convert a vector from one coordinate system to another. These quantities are conveniently stored
in an array. For a two-dimensional vector, the conversions are represented by the following
equations derived from Figure 9.4.

V = Vx i + Vy j where i and j are the unit vectors in the x and y directions

Vx = V cos theta
Vy = V sin theta

theta = atan (Vy / Vx)

Figure 9.4 Vector

The two conversion functions are passed the rectangular coordinate’s array and the polar
coordinate’s array. Element 0 represents Vx in rectangular coordinates or V in polar. Similarly,
element 1 represents the Vy or the angle.

Another use of arrays is to lower the total number of parameters, by collecting similar,
related items into one array. Here, we can consider the rect array to be a vector containing the x
and y axis components. Likewise, the array polar can be though of containing the vector in polar
coordinates.
Arrays 432

Examine the PolarConvert() function first which is passed the arrays rect and polar. The
magnitude is calculated; the angle is found and then converted into degrees.
polar[0] = sqrt (rect[0] * rect[0] + rect[1] * rect[1]);
radians = atan (rect[1] / rect[0]);
polar[1] = ToDegrees (radians);
The function RectConvert() passed polar and rect arrays first converts the angle into
radians and then calculates the Vx and Vy quantities as follows.
radians = ToRadians (polar[1]);
rect[0] = polar[0] * cos (radians);
rect[1] = polar[0] * sin (radians);
Notice that constant subscripts, 0 and 1, are used to access the specific elements.

Again in the radian conversion helper functions, the value of PI is found by acos(-1.)
which gives the most nearly accurate value of PI as a double.

Listing Engr09a shows the complete program including a small test main program. A
test run is also shown below.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Listing for Program Engr09a - Vector Conversions *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr09a Vector Conversions */*
* 4 /* */*
* 5 /* Convert a vector from rectangular coords to polar coords */*
* 6 /* Convert a vector from polar coords to rectangular coords */*
* 7 /* */*
* 8 /***************************************************************/*
* 9 *
* 10 #include <iostream> *
* 11 #include <iomanip> *
* 12 #include <cmath> *
* 13 using namespace std; *
* 14 void PolarConvert (double rect[], double polar[]); *
* 15 void RectConvert (double polar[], double rect[]); *
* 16 *
* 17 double ToRadians (double angle); *
* 18 double ToDegrees (double radians); *
* 19 *
* 20 int main () { *
* 21 *
* 22 double rect[2]; *
* 23 double polar[2]; *
* 24 double rect2[2]; *
* 25 *
* 26 // setup floating point output format *
* 27 cout << fixed << setprecision (3); *
* 30 *
Arrays 433

* 31 // input a vector in rectangular coordinates *

* 32 cout << "Enter the Vector in rectangular coordinates\n" *
* 33 << "X axis value first, then Y axis - separated by a blank\n";*
* 34 cin >> rect[0] >> rect[1]; *
* 35 *
* 36 // convert vector to polar and then use polar to convert back *
* 37 PolarConvert (rect, polar); *
* 38 RectConvert (polar, rect2); *
* 39 *
* 40 // display results *
* 41 cout << "Original rect " << setw (8) << rect[0] << " " *
* 42 << setw (8) << rect[1] << endl; *
* 43 cout << "Polar " << setw (8) << polar[0] << " " *
* 44 << setw (8) << polar[1] << endl; *
* 45 cout << "Rect from polar " << setw (8) << rect2[0] << " " *
* 46 << setw (8) << rect2[1] << endl; *
* 47 return 0; *
* 48 } *
* 49 *
* 50 /***************************************************************/*
* 51 /* */*
* 52 /* PolarConvert: convert vector in rectangular coords to polar */*
* 53 /* */*
* 54 /* polar[0] = V polar[1] = angle */*
* 55 /* rect[0] = Vx rect[1] = Vy */*
* 56 /* */*
* 57 /***************************************************************/*
* 58 *
* 59 void PolarConvert (double rect[], double polar[]) { *
* 60 polar[0] = sqrt (rect[0] * rect[0] + rect[1] * rect[1]); *
* 61 double radians = atan (rect[1] / rect[0]); *
* 62 polar[1] = ToDegrees (radians); *
* 63 } *
* 64 *
* 65 /***************************************************************/*
* 66 /* */*
* 67 /* ToRadians: convert an angle in degrees into radians */*
* 68 /* */*
* 69 /***************************************************************/*
* 70 *
* 71 double ToRadians (double angle) { *
* 72 double pi = acos (-1.); *
* 73 return angle / 180. * pi; *
* 74 } *
* 75 *
* 76 /***************************************************************/*
* 77 /* */*
* 78 /* ToDegrees: convert an angle in radians to degrees */*
* 79 /* */*
* 80 /***************************************************************/*
* 81 *
* 82 double ToDegrees (double radians) { *
Arrays 434

* 83 double pi = acos (-1.); *

* 84 return radians * 180. / pi; *
* 85 } *
* 86 *
* 87 /***************************************************************/*
* 88 /* */*
* 89 /* RectConvert: convert a vector in polar coords to rect coords*/*
* 90 /* */*
* 91 /* polar[0] = V polar[1] = angle */*
* 92 /* rect[0] = Vx rect[1] = Vy */*
* 93 /* */*
* 94 /***************************************************************/*
* 95 *
* 96 void RectConvert (double polar[], double rect[]) { *
* 97 double radians = ToRadians (polar[1]); *
* 98 rect[0] = polar[0] * cos (radians); *
* 99 rect[1] = polar[0] * sin (radians); *
* 100 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Sample Execution of Program Engr09a - Vector Conversions *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Enter the Vector in rectangular coordinates *
* 2 X axis value first, then Y axis - separated by a blank *
* 3 5 5 *
* 4 Original rect 5.000 5.000 *
* 5 Polar 7.071 45.000 *
* 6 Rect from polar 5.000 5.00 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Engr09b — Plotting Graphs

Plotting graphs uses arrays in other ways. Typically, the function to be plotted is evaluated at a
number of uniformly spaced points. The x and y values are stored in a pair of arrays and passed
to a plotting function along with the number of points in the arrays.

The plotting function must scale the range of x and y values so that the graph can be
displayed within the dimensions of the screen or printer. If you have access to a plotter device,
high quality plots can be made and there often is a library of graphics or plotting functions
available for your program to invoke. If not, one can make a crude plot using the * character to
represent a point. This sample does just that to illustrate the basic mechanics of plotting a graph
using arrays of x and y values. Here is the sample output from this program.
Arrays 435

Plot of the function

Ymin Ymax
0.00 1.00
X + + +
|--------------------------------------------------|
0.00 |* |
0.06 | * |
0.12 | * |
0.17 | * |
0.23 | * |
0.29 | * |
0.35 | * |
0.41 | * |
0.47 | * |
0.52 | * |
0.58 | * |
0.64 | * |
0.70 | * |
0.76 | * |
0.81 | * |
0.87 | * |
0.93 | * |
0.99 | * |
1.05 | * |
1.11 | * |
1.16 | * |
1.22 | * |
1.28 | * |
1.34 | * |
1.40 | *|
1.45 | *|
1.51 | *|
1.57 | *|
1.63 | *|
1.69 | *|
1.75 | *|
1.80 | * |
1.86 | * |
1.92 | * |
1.98 | * |
2.04 | * |
2.09 | * |
2.15 | * |
2.21 | * |
2.27 | * |
2.33 | * |
2.39 | * |
Arrays 436

2.44 | * |
2.50 | * |
2.56 | * |
2.62 | * |
2.68 | * |
2.73 | * |
2.79 | * |
2.85 | * |
2.91 | * |
2.97 | * |
3.03 | * |
3.08 | * |
3.14 |* |
|--------------------------------------------------|

Let's begin with what a user of a plotting function would need to do to be able to plot a
graph of some function. Assume that the prototype for the Plot() function is
void Plot (double x[], double y[], int numpts, ostream& out);

The Plot() function is given an array of x and y points along with the number of points in
these arrays. Plot() displays the graph on the passed output stream, which can be either a file on
disk or cout for screen displays. Clearly, all that the user of the Plot() function must do is define
and fill an array of points that represent the function's behavior over the desired interval. Assume
that we wish to plot the sin function from 0 to PI. Since the points along the x axis must be
uniformly spaced, first calculate a delta x value, that is the incremental x amount to add to the
previous point's x value to get to the next x point to plot. The magnitude of delta x is determined
by the total range of x values to be plotted and the number of points desired. Since in this simple
plot, one line represents one point's x value, the number of points should be fairly small.
const int MaxPts = 55;
int main () {
double x[MaxPts]; // contains the points to plot
double y[MaxPts];
double pi = acos (-1.);
double deltax = pi / (MaxPts - 1);

The basic algorithm to calculate the ith x,y pair is to add delta x to the previous point's x
value to give the new x value. Thus, the initial x value must be set by hand.
x[0] = 0;
y[0] = sin (x[0]);

Now a loop can be done to automatically determine the next point's x value from the
previous point’s x value and then the new y value.
int i = 1;
while (i<MaxPts) {
x[i] = x[i-1] + deltax;
y[i] = sin (x[i]);
Arrays 437

i++;
}
When the loop is complete, the x and y arrays are ready to be graphed.
Plot (x, y, MaxPts, out);
Thus, the user of a plotting function only needs to create the array of points.

Writing the Plot() function is more difficult, especially if it is to handle all possible
graphical situations. The fundamental principle is that the range of y values must be scaled to fit
within the available physical space. In this case, all y values must lie within 50 columns so that
the graph may be displayed on the screen or on a printed page. In this simple plotting function,
each line represents the next x value. In a more robust plotting function, the x values would also
be scaled to fit the available space.

The minimum y value would be in the first column and the maximum y value would be in
th
the 50 column. The Plot() function begins by finding the maximum and minimum values of the
array of y values using the two helper functions, FindMax() and FindMin().
double ymax = FindMax (y, numpts);
double ymin = FindMin (y, numpts);
double scale_denom = ymax - ymin;

The columns array is defined as follows

char columns[50];
It must hold 49 blanks plus one *, where the * represents the actual point being plotted. To
initialize all 50 characters to blanks, use a short loop assigning a blank to each element.

Let's examine the main processing loop of the Plot() function as it plots a single point, the
th
i one. First, the y[i] value must be scaled and converted into a subscript that lies between 0 and
49.
ratio = (y[i] - ymin) / scale_denom;
int j = (int) (50. * ratio);
However, it is always wise practice when calculating a subscript to guarantee that the final result
lies within the range of the array, columns in this case.
if (j < 0)
j = 0;
else if (j > 49)
j = 49;
Now it is a simple matter of inserting the * character in the jth element.
columns[j] = '*';

The next issue is how to display that array of characters. One cannot just do
out << columns;
because the ostream considers that this is a character string which ends with a byte of 0. Strings
are discussed in a later chapter. Thus, the columns array must be outputted one character at a
time in a short loop which is followed by outputting the newline code.
for (int k=0; k<50; k++)
Arrays 438

out.put (columns[k]);
out << "|\n";
Finally, the * character must be replaced by a blank so that the columns array is ready for the
next point.
columns[j] = ' ';

Listing Engr09b contains the complete Plot program.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Listing for Program Engr09b - Plot Function *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr09b Plot Function - display a simple text graph of a fun*/*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 #include <cmath> *
* 11 using namespace std; *
* 12 void Plot (double x[], double y[], int numpts, ostream& out); *
* 13 double FindMax (double x[], int numpts); *
* 14 double FindMin (double x[], int numpts); *
* 15 *
* 16 const int MaxPts = 55; *
* 17 *
* 18 int main () { *
* 19 *
* 20 ofstream out ("Engr09b.txt"); // send the plot to this file *
* 21 *
* 22 double x[MaxPts]; // contains the points to plot *
* 23 double y[MaxPts]; *
* 24 *
* 25 double pi = acos (-1.); *
* 26 double deltax = pi / (MaxPts - 1); // uniform x interval *
* 27 *
* 28 // plot sin(x) from 0 to pi - assign the starting point values *
* 29 x[0] = 0; *
* 30 y[0] = sin (x[0]); *
* 31 int i = 1; *
* 32 // calculate all other points in the range *
* 33 while (i<MaxPts) { *
* 34 x[i] = x[i-1] + deltax; *
* 35 y[i] = sin (x[i]); *
* 36 i++; *
* 37 } *
* 38 *
* 39 // plot the graph *
* 40 Plot (x, y, MaxPts, out); *
* 41 *
Arrays 439

* 42 return 0; *
* 43 } *
* 44 *
* 45 /***************************************************************/*
* 46 /* */*
* 47 /* Plot: display a text style graph of a set of x,y points */*
* 48 /* on the passed output stream */*
* 49 /* */*
* 50 /* */*
* 51 /***************************************************************/*
* 52 *
* 53 void Plot (double x[], double y[], int numpts, ostream& out) { *
* 54 // setup floating point output format *
* 55 out << fixed <<setprecision (2); *
* 58 *
* 59 // get max range so that we can scale each point to fit ourrange*
* 60 double ymax = FindMax (y, numpts); *
* 61 double ymin = FindMin (y, numpts); *
* 62 double scale_denom = ymax - ymin; *
* 63 double ratio; *
* 64 *
* 65 char columns[50]; // will hold 49 blanks and one * per point *
* 66 int i; *
* 67 for (i=0; i<50; i++) { *
* 68 columns[i] = ' '; *
* 69 } *
* 70 *
* 71 // display title and headings *
* 72 out << "Plot of the function\n\n"; *
* 73 out << *
* 74 " Ymin Ymax\n";*
* 75 out << setw (12) << ymin << setw (46) << ymax << endl; *
* 76 out << *
* 77 " X + + +\n";*
* 78 out << *
* 79 " |--------------------------------------------------|\n";*
* 80 // plot each point *
* 81 for (i=0; i<numpts; i++) { *
* 82 out << setw (5) << x[i] << " |"; *
* 83 // scale this point *
* 84 ratio = (y[i] - ymin) / scale_denom; *
* 85 int j = (int) (50. * ratio); *
* 86 // force j to not exceed the boundaries of the columns array *
* 87 if (j < 0) *
* 88 j = 0; *
* 89 else if (j > 49) *
* 90 j = 49; *
* 91 // insert * character for this point *
* 92 columns[j] = '*'; *
* 93 // display all 50 characters one at a time *
* 94 for (int k=0; k<50; k++) *
* 95 out.put (columns[k]); *
Arrays 440

* 96 out << "|\n"; *

* 97 // remove * character for this point *
* 98 columns[j] = ' '; *
* 99 } *
* 100 out << *
* 101 " |--------------------------------------------------|\n";*
* 102 } *
* 103 *
* 104 /***************************************************************/*
* 105 /* */*
* 106 /* FindMin: finds the minimum value of an array of numpts vals */*
* 107 /* */*
* 108 /***************************************************************/*
* 109 *
* 110 double FindMin (double x[], int numpts) { *
* 111 double min = x[0]; *
* 112 for (int i=1; i<numpts; i++) { *
* 113 if (x[i] < min) *
* 114 min = x[i]; *
* 115 } *
* 116 return min; *
* 117 } *
* 118 *
* 119 /***************************************************************/*
* 120 /* */*
* 121 /* FindMax: finds the maximum value of an array of numpts vals */*
* 122 /* */*
* 123 /***************************************************************/*
* 124 *
* 125 double FindMax (double x[], int numpts) { *
* 126 double max = x[0]; *
* 127 for (int i=1; i<numpts; i++) { *
* 128 if (x[i] > max) *
* 129 max = x[i]; *
* 130 } *
* 131 return max; *
* 132 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

New Syntax Summary

Arrays: the [ ] always means an array, what is inside the brackets has two meanings.

Defining:
data type variableName [maximum number of elements];
The maximum number must be of the integer family. Normally, it is a const int value.
const int MAX = 10;
float temps[MAX]; // defines array of 10 floating point numbers
Arrays 441

Defining and Initializing:

data type variableName [maximum number of elements] = {value0, value1, . . . valuen};
If the number of values is less than the maximum number of elements, a value of 0 in stored in
the remaining elements. If too many values are given, a compiler error occurs. Always, at least
one value must be coded, the rest may default to 0.
float temps[MAX] = {0};

Accessing Elements: use [element desired or subscript or index]

The first element in any array in C++ is always [0]. The index value must be in the integer family
and can be a constant, variable, or an expression.
temps[0] temps[i] temps[i+j]

Typical Input Loop:

int numTemps;
j = 0;
while (j < MAX && infile >> temps[j]) {
j++;
}
numTemps = j;

Typical Sequential Accessing:

for (j = 0; j < numTemps; j++) {
cout << temps[j];
}

Passing Arrays to Functions:

The name of an array is always a constant pointer or memory address of the first element. Copies
of arrays are never passed. The function prototype and header can be coded in three ways. The
left most dimension value can be omitted because in a prototype the [] means is an array which
means memory address of the first element. The * means is a pointer to.
main:
Fun (temps);
prototype:
void Fun (float temps[MAX]);
void Fun (float temps[]);
void Fun (float* temps);
Arrays 442

Design Exercises
1. Mail Sorting Statistics

Design a program that displays mail sorting statistics for the Post Office. At the end of each
worker’s shift at the mail sorting room, that mail sorter’s id and the number of pieces of mail
they have sorted are appended to the DailyStats.txt file. At the end of the day, this program is
run to produce the Mail Sorting Stats Report.

The Average Number Sorted Today: 999999

Sorter Number Percentage

Id Sorted of Average
9999 99999 999.99%
9999 99999 999.99%
9999 99999 999.99%

Allow for a maximum of 100 workers. The percentage of the average sorted is given by
that workers number sorted * 100 / average number sorted.

2. The Optimum Hours

A fast food restaurant has installed a traffic counter to count the number of cars driving by its
store. It logs a count of the number of cars in each ten-minute period from 8am to 10pm. Each
line contains a time of the form of hh:mm followed by the count of the cars passing during that
interval. The program should first load the file counts.txt into three arrays, hour, minute, count.
Find the average number of cars per ten-minute interval throughout the day. Now display that
average count and the three highest counts that were observed along with their time. The report
appears similar to this.
Summary of Passing Cars
Average count: 9999
hh:mm count
99:99 9999
99:99 9999
99:99 9999
Based upon these results, management will ensure that more employees are working during the
potential higher traffic times.
Arrays 443

Stop! Do These Exercises Before Programming

1. This attempt to input the five tax rates for the states in which ACME sells products won’t
compile. Why? How can it be repaired so that it works?
int main () {
double taxrates[5];
cin >> taxrates;
...

2. This attempt also does not work properly, though it compiles. Why? How can it be fixed?
int main () {
double taxrates[5];
for (int j=1; j<6; j++)
cin >> taxrates[0];
...

3. This attempt compiles fine, but at run time it crashes. Why? How can it be fixed so that it
works properly?
int main () {
double taxrates[5];
for (int j=1; j<6; j++)
cin >> taxrates[j];
...

4. Now that the tax rates have been correctly input, a major design flaw surfaced. If the main
program inputs a state code, such as 13 (for Illinois), how can the corresponding tax rate from the
array of five tax rates be found? A file was created in which each input line of the taxrates.txt
file contains the integer state code and the corresponding tax rate. This attempt at making a
function to load the arrays fails to work properly. Why? How can it be fixed?
int main () {
double taxrates[5];
int states[5];
loadArray (states[], taxrates[5]);
...
void loadArray (int states[], double taxrates) {
ifstream infile ("taxrates.txt");
int j=0;
while (infile >> states[j] >> taxrates) {
j++);
}
infile.close();
}
Arrays 444

5. Since the previous attempt to make the load function failed, the programmer threw the whole
thing away and started over. This is the next attempt, also doomed, though the programmer
finally got it to nearly compile. Why? How can it be fixed up?
const int MAX = 5
int main () {
double taxrates[MAX];
int states[MAX];
loadArray (states[MAX], taxrates[MAX], MAX);
...
void loadArray (int states, double taxrates, int MAX) {
ifstream infile ("taxrates.txt");
int j=0;
while (infile >> states >> taxrates && j < MAX) {
j++);
}
infile.close();
}

6. Undaunted by his previous difficulties, the programmer decided that a matchState() function
was needed. Here is the first attempt. It compiles but does not work. Why? How can you fix it so
that it does work properly?
int matchState (int states[], int state) {
int j = 0;
while (state != states[0])
j++;
}
return j;
}

7. With a working way to find the matching state, work began on the main function’s
calculation’s loop. This is what the programmer produced. It fails to work properly. Why? How
can it be repaired?
int statecodes[5];
double taxrates[5];
double cost;
int quantity;
int statecd;
double tax;
double total;
double grandTotal;
int matchingStateCodeSubscript;
while (infile2 >> quantity >> cost >> statecd) {
total = quantity * cost;
matchState (statecodes, statecd);
tax = total * taxrates[matchingStateCodeSubscript];
Arrays 445

grandTotal = total + tax;

cout << grandTotal << endl;
}

8. A programmer was asked to make a program to convert students’ final raw grades into a letter
grade. The specifications called for the use of arrays. One array holds the lowest possible score
for a particular letter grade and the other array holds the corresponding letter grade. The
programmer produced the following coding. While it works, his boss immediately ordered a
revision and told him to just initialize the arrays not assign them values at run time. How can this
be done?
int main () {
int rawScores[5];
char grades[5];
rawScores[0] = 90;
grades[0] = 'A';
rawScores[1] = 80;
grades[1] = 'B';
rawScores[2] = 70;
grades[3] = 'C';
rawScores[3] = 60;
grades[3] = 'D';
rawScores[4] = 0;
grades[4] = 'F';

9. Next, having gotten the two arrays loaded properly, the programmer proceeded to write the
code to convert students’ raw scores into letter grades. The students complained bitterly about the
results. Desk check with a grade of 94.78. What is wrong with this and how can it be fixed to
work properly?
int main () {
int rawScores[5]....
char grade[5]....
double rawscore;
long idNum;
char grade;
while (cin >> idNum >> rawscore) {
for (int j=4; j>=0; j++)
if (rawscore > rawScores[j]) break;
}
grade = grades[j];
cout << idNum << ' ' << grade << endl;
}
Arrays 446

10. Hastily the programmer spotted his errors and recoded the program as follows. This is now a
“mostly working” program. Far fewer students were complaining about their grades. Can you
spot the remaining error? One of the students complaining bitterly about their grade received a
raw score of 89.997. How can the program now be fixed?
int main () {
int rawScores[5]....
char grade[5]....
double rawscore;
long idNum;
char grade;
while (cin >> idNum >> rawscore) {
for (int j=0; j<5; j++)
if (rawscore >= rawScores[j]) break;
}
grade = grades[j];
cout << idNum << ' ' << grade << endl;
}
Arrays 447

Programming Problems
Problem CS09-1 — Write a Program to Grade a Multiple Choice
Test
All input lines except the first line contain a 9-digit student ID number followed by one blank
followed by their letter answers to the questions. The first line of input contains a dummy student
ID, a blank space and then the answer key. For example,
000000000 ABCDABCDABCD
123123123 ABDCABACDDAD
143434193 BACDACDABACD
323737347 B A D D AAAA

In this example, there were 12 test questions. Please allow for up to 100 test questions.
All answers will be in uppercase letters. Please note that students sometimes fail to answer a
question. Their corresponding letter is a blank. This means you must use the get() function to
retrieve each letter (or blank).

After inputting the dummy ID number and using get() to retrieve the blank, make a loop
to input each letter answer key and store it in an array called answer_key. You need only this
ONE array in this problem. The bounds of the answer_key array should be 100. However, you
must know how many answers there really are on this particular test. Thus, continue inputting
answer characters and storing them into the array until you input a \n, the new line code. At this
point, you know the number of questions and their answers.

Now write a loop that goes until the end of the file is reached. Input the student ID
number (nine digits long) and then write a loop to input their answers. Since you now know how
many questions there are, you can use a for loop. As an answer comes in, compare it to the
correct answer letter and accumulate totals of correct answers for that student. After inputting all
the student answers, print on line that contains the student ID number followed by the percent
correct answers. (Remember when converting into a percentage, to multiply by 100 before you
divide.)

Finally, when the end of the file is reached, print the average percentage correct found by
totaling the individual percent correct scores and dividing by the number of students. Your output
should look something like:
STUDENT PERCENT
ID CORRECT
xxxxxxxxx xxx.xx
xxxxxxxxx xxx.xx
xxxxxxxxx xxx.xx
Arrays 448

AVERAGE xxx.xx
Test your program with the 2 provided test files, TestGrading1.txt and TestGrading2.txt.

Problem CS09-2 — The ComputeAverage() Function

Write a function to find and return the average of the numbers in an array of double type
elements. Name the function ComputeAverage(). The function is passed two arguments: the
array, and the count of numbers in the array. Test your function using the main() function
provided on disk contained in file CS09-2.cpp. Here are the results you should get.
Average of numbers in array x = 51.11
Average of numbers in array y = 5.50
Average of numbers in array z = 429.50

Problem CS09-3 — FindHighest() and FindLowest() Functions

Write a function to find and return the largest number in an array of double type elements. Name
the function FindHighest(). The function will be passed two arguments: the
array, and the count of numbers in the array. Also write a function to find and return the smallest
number in an array of double type elements. Name the function FindLowest(). The function is
passed the same arguments as FindHighest(). Test your function using the main() function
provided on disk contained in the file CS09-3.cpp. Here are the results you should get.
Array x: high = 88.88 low = 11.11
Array y: high = 9.00 low = 2.00
Array z: high = 789.00 low = 102.00

Problem CS09-4 — FindIndexHighest() Function

Write a function to find and return the index of the largest number in an array of double type
elements. That is, if the largest number in the array is in the element with the subscript 7, then the
function should return 7. Name the function FindIndexHighest(). Decide what the return data
type ought to be. The function is passed two arguments: the array, and the count of numbers in
the array. Test your function using the main() function provided on disk called CS09-4.cpp.
Here are the results you should get.
Array x: element index = 2 element contents = 88.88
Array y: element index = 0 element contents = 9.00
Array z: element index = 7 element contents = 789.00
Arrays 449

Problem CS09-5 — Occurrences() Function

Write a function that may be called like this:

int n = Occurrences (array, count, target);
The first argument is an array of double type elements. The second argument is the count of
number of elements used in the array. The function counts how many times the third argument
appears in the array and then returns that number of occurrences. Test your function using the
main() function provided on disk in file CS09-5.cpp. Here are the results you should get.
11.11 occurs 1 time(s) in array x.
11.11 occurs 2 time(s) in array y.
11.11 occurs 0 time(s) in array z.

33.33 occurs 2 time(s) in array x.

33.33 occurs 0 time(s) in array y.
33.33 occurs 3 time(s) in array z.

88.88 occurs 5 time(s) in array x.

88.88 occurs 2 time(s) in array y.
88.88 occurs 1 time(s) in array z.

Problem Engr09-1 — Vector Cross Product

Given two vectors in three-dimensional space, the cross product is often needed. Assuming the
two vectors are V1 and V2 and are defined in rectangular coordinates as
V1 = Vx1i + Vy1j + Vz1k
V2 = Vx2i + Vy2j + Vz2k

The cross product is given by

V1 xV2 = (Vy1 Vz2 – Vy2 Vz1) i +
(Vz1 Vx2 – Vz2 Vx1) j +
(Vx1 Vy2 – Vx2 Vy1) k

Write a function, CrossProduct(), that is passed three arrays, each with three elements.
The first array is V1, the second is V2 and the third is the answer vector. Write a main program
to prompt the user to input the two vectors V1 and V2 and then output the results. The general
format to output a vector to follow is
V1 [–2.5, 4.3, 6.7]
Test your program with the two vectors: V1 [–2.5, 4.3, 6.7], V2 [.05, 3.1, 2.2]
Arrays 450

Problem Engr09-2 — Cross Product Application —Velocity of an

Orbiting Satellite
A satellite is in orbit around the earth at a radius r from the center of the earth and has a known
angular velocity. Determine the velocity of the satellite. The angular velocity omega of an object
moving with a velocity v at distance r from the origin of the coordinate system is
v = r x omega

Write a program that inputs the distance vector and the angular velocity vector into two
arrays of three elements each. Then calculate the velocity vector using the previous problem's
cross product function.

Test your program with a satellite that is at r = 350000i + 450000j + 55000k meters with
an angular velocity of –6.1E–3i + 2.3E–3j + –9.2E–4k radians per second. The resulting velocity
is in meters per second.

Problem Engr09-3 — Vector Dot Products

Given two vectors in three-dimensional space, the dot product is often needed. Assuming the two
vectors are V1 and V2 and are defined in rectangular coordinates as
V1 = Vx1i + Vy1j + Vz1k
V2 = Vx2i + Vy2j + Vz2k
The dot product scalar is given by
.
V1 V2 = Vx1 Vx2 + Vy1 Vy2 + Vz1 Vz2

Write a function, DotProduct(), that is passed the arrays, each with three elements. The
first array is V1, and the second is V2. The function returns the dot product as a double.
Write a main program to prompt the user to input the two vectors V1 and V2 and then output the
results. The general format to output a vector to follow is
V1 [–2.5, 4.3, 6.7]

Test your program with the two vectors: V1 [–2.5, 4.3, 6.7], V2 [.05, 3.1, 2.2]
Arrays 451

Problem Engr09-4 — Dot Product Application – Power Supplied to

an Object
From physics, if an object is being pushed by a force of F at a velocity of v, then the power being
supplied to the object by that force is
p=F.v

Write a program that inputs the force vector and the velocity vector into two arrays of
three elements each. Then calculate the power being supplied by using the previous problem's dot
product function, Engr09-3.

Test your program with an object that has a velocity of v = 7.5i + 4.56j + 5.5k meters per
second with a force of 6.1i + 2.3j + 9.2k newton. The resulting power is in watts.

Problem Engr09-5 — Plotting the Cosine Function

Modify the Plot() graph function to plot the cosine function across the range 0-2pi. Since the plot
function used an array of 50 columns and since this function is uniformly spaced across the
positive and negative y axis over this range, make the columns array contain 51 elements. Let 0
be the array element in the middle.

Problem Engr09-6 — Gaussian (Normal) Distribution Function

The Gaussian distribution is a random distribution with the classic bell-shape curve. If the
distribution has an average of zero and a standard deviation of 1.0, it is called the standardized
normal distribution. The probability of any given value occurring in this standardized distribution
is given by the formula

Use the Plot() function to graph this equation over the range from –4 to +4 in 51 intervals.
Using Arrays 452

Chapter 10 — Using Arrays

Section A: Basic Theory

Introduction
Arrays are widely used in programming. Sometimes a single-dimensioned array is called a list or
a table. Thinking in general about an abstract list of things, there are a number of actions that
might commonly be done to and with and a list. Lists may be unsorted (unordered) or they may
be sorted into some order, numerically low to high or perhaps alphabetically. An unordered list
may need to be sorted into some order. A new item may be added to either an ordered or
unordered list. A list may be searched for a matching item; this is sometimes called a table
lookup. Two lists may be related; these are called parallel arrays. For example, one array might
hold the student id number while the parallel array holds the student grades for a course. In such
a case, a specific element of one array corresponds to that same element in another array. That is,
element 0 of the id array contains the id of the student whose grade is in the corresponding
element 0 of the grades array. In this chapter, we explore these different uses of arrays.

Using an Array for Direct Lookup Operations

When working with dates, one often needs to know how many days there are in a given month.
Using an array can streamline such operations. Given the month number (1 through 12), the
program should be able to access directly the array element that contains the number of days in
that month. If the month number is 1 for January, then days_in_month[1] should contain 31
days. When setting up the days_in_month array, since all arrays begin with element 0 and since
0 is not normally a month number, it is permissible to make the array one element larger, placing
a dummy value in the never-to-be-used element 0. The array could be defined as follows
const int days_in_month[13] = {0, 31, 28, 31, 30, 31, 30,
31, 31, 30, 31, 30, 31}

Notice also the usage of the const keyword. Once the array elements are given their initial
values, they should never be changed. Making the array constant ensures that no accidental
changes to these values can be made.
Using Arrays 453

In this example, the month number is used as the subscript to directly access the correct
number of days in that month. The following illustrates this.
int month;
cout << "Enter a month number: ";
cin >> month;
while (month < 1 || month > 12) {
cout << "Invalid month number – please re-enter: ";
cin >> month;
}
cout << "Month " << month << " contains "
<< days_in_month[month] << " days\n"

Parallel Arrays and Sequential Searches — Inquiry Programs

Consider two single-dimensioned arrays, one contains the student id number and the other
contains their course grade. Clearly, the two arrays must be kept synchronized at all times. The
grade stored in element 1 of the grade array corresponds to the student whose id is stored in
element 1 of the id array. Once the information is loaded into the two arrays, then the inquiry
operations can begin. An inquiry program is one in which the user is prompted to enter a
specific id of some kind and the program then finds the corresponding data and displays it.
Inquiry programs are widespread in the modern world. Checking on your bank account balance,
credit card limit, and even the grade that you received in a course — all are inquiry type
programs.

Let's first examine how the inquiry array is loaded and then how it is used or searched.
Assume that each line of input contains a long student id number followed by the letter grade
they received. The following loads both arrays
long id[MaxStudents];
char grade[MaxStudents];
int numberStudents;
int j = 0;
while (j < MaxStudents && cin >> id[j] >> grade[j]) {
j++;
}
numberStudents = j;

Notice the while test condition checks first to see if there is still another available element
and if so, attempts the input operation and if successful, increments the subscript for the next
iteration. Assume that the following Illustration 10.1 represents the arrays after all the data have
been input. The variable numberStudents contains the number actually input into the arrays and
is 5.
Using Arrays 454

Illustration 10.1 The Id and Grades Arrays

subscript id array grade array
0 111111111 A
1 444444444 B
2 222222222 A
3 555555555 C
4 333333333 B

Next, the inquiry program prompts the user to enter the id of the student whose grade is to
be found.
long studentId;
char studentGrade;
cout << "Enter student id number: ";
cin >> studentId;

Now the objective is to search the id array looking for a match on studentId, obtain the
subscript of the matching id and use that subscript to get at that student's grade. Let's encapsulate
the matching process in a function, MatchId(), whose header begins
int MatchId (const long id[], int num, long findId) {
MatchId() must be passed the array of id numbers whose data are constant in this function and
the current number in the array along with the id number to find, findId. It should return the
subscript of that element of the id array that matched findId.

Look over Illustration 10.1 above; suppose that the user enters an id number of
555555555. Counting down the array elements, the MatchId() function should return 3.

But what would happen if the user asks MatchId() to find a student id of 666666666?
That id number is not in the list. Thus, when MatchId() ends, if there is no match on the findId,
MatchId() must have a way to notify the caller of that fact. Because no subscript can ever be a
negative integer, we can adopt some negative number to return to indicate no match found.
Commonly –1 is used for this purpose.

Following good programming practice, define a constant integer for to represent it and
place it in the global namespace above the main function.
const int NoMatch = -1;

The logic of the MatchId() function is

int MatchId (const long id[], int num, long findId) {
for (int j=0; j<num; j++) {
if (findId == id[j])
return j;
}
return NoMatch;
}
Using Arrays 455

The main() program then invokes MatchId() as follows.

int match = MatchId (id, numberStudents, studentId);
if (match != NoMatch) {
studentGrade = grade[match];
cout << studentID << "received a grade of "
<< studentGrade << endl;
}
else {
cout << "Error: invalid student id\n";
}

Inserting Another Element into an Unsorted Array

Suppose that a student with an id number of 666666666 takes a make-up exam and scores a
grade of B. One could alter the input file to add this sixth line and rerun the program. However,
in some applications, it is neither possible nor desirable to terminate the program and restart it
just to reload the arrays. Instead, the new information is additionally inserted into the array. In an
unsorted array, the new information added into the first empty element. Make sure that the total
number of elements in the array is incremented. The following InsertStudent() function
illustrates how this may be done. Notice this time, the arrays are not constant.
bool InsertStudent (long id[], char grade[],
int& num, int maxlimit,
long newid, char newgrade) {
if (num >= maxlimit) return false;
id[num] = newid;
grade[num] = newgrade;
num++;
return true;
}

Notice that the function returns false if there is no more room left in the array. Observe
that the number in the array, num, must be passed by reference so that the number in main() can
be incremented. The two arrays now appear as follows as shown in Illustration 10.2.

Illustration 10.2 Updated Id and Grade Arrays

numberStudents is 6 - main()’s variable

subscript id array grade array

0 111111111 A
1 444444444 B
2 222222222 A
3 555555555 C
4 333333333 B
5 666666666 B
Using Arrays 456

Ordered (Sorted) Lists

One problem of unsorted lists is the time that it takes to search through the array sequentially
looking for a matching value in the array. If there are only a few elements, the amount of time is
negligible. However, suppose that these arrays contained a store’s inventory numbers, quantity
on hand and unit cost. Further, suppose that the store handles 100,000 separate items. If the item
number desired was the last one in the list, a significant amount of time is needed to find that
match. The answer is not “Get a faster computer” but rather devise a better algorithm. If the list is
sorted into numerical or alphabetical order depending upon the type of data the array contains,
then far faster searches can be devised. Returning to the student id and grades arrays, let’s
assume that the arrays have been sorted into increasing numerical order on the ids. The arrays
appear as shown in Illustration 10.3.

Illustration 10.3 Sorted Id and Grade Arrays

numberStudents is 6 - main()’s variable

subscript id array grade array

0 111111111 A
1 222222222 A
2 333333333 B
3 444444444 B
4 555555555 C
5 666666666 B

The array of ids can still be matched sequentially. However, we can take advantage of the
ordered nature to detect no matching id number more quickly. Suppose that the findId this time
was 345678999. Notice that when we are at subscript 3 which contains id 444444444, we know
for certain that this id is not in the array and can return false at once without having to check any
further subscripts. The slight modification is in boldface
int MatchSortedId (const long id[], int num, long findId) {
for (int j=0; j<num && findId >= id[j]; j++) {
if (findId == id[j])
return j;
}
return NoMatch;
}

On the average, some increase in speed results. However, for items near the end of the
array are still going to take a large number of iterations through the loop to find them.

The binary search method uses a different searching algorithm, one that drastically
reduces the number of comparisons that need to be done to find the match. Before looking at the
coding for the search, let's examine in detail how the binary search works. Let N represent the
number of ids in the array. The first subscript to use in the search is N/2 — the midpoint. We
Using Arrays 457

compare the findId to the element in the middle. If we are lucky, we have an exact match and are
done. More likely it does not match, but if the findId is smaller than the one in the middle, we
can eliminate the entire higher half of the array from further consideration. Likewise if the findId
is greater than that in the middle, we can eliminate all those values in the lower half. Thus, on
one test, we have eliminated one-half of the array from further consideration! Now that same
process is repeated, halving the new interval and testing the one in the middle again, and so on
until we find the match or run out of array, indicating no match.

Let's do a concrete example using the student data above in Illustration 10.3. Say the
findId is 22222222. The first subscript to try is (0 + 5) / 2 or index 2 which stores id 333333333.
The findId is smaller so if this one is in the array it must lie in the lower half, between indexes 0
and 1. The new index to try is halfway between. At subscript 1, we have our match.

The binary search function should be designed so that it returns true if it finds a match;
the index of the match is stored in a reference parameter for use by the caller. However, if it does
not find a match, the index stored in the passed reference parameter should be the index of where
that value ought to have been if it was in the list. Why? Code reuse. True, for a simple query,
match this id, just a return value of false for not present is sufficient. But the next feature one
might need to implement is to add this id into the sorted list where it belongs. Thinking ahead,
when an id is not in the list, it is a simple matter to also provide the index of where this element
should be if it were in the list. Then only one BinarySearch() function need be written.
bool BinarySearch (const long id[], int num,
long findId, int& foundIndex) {
int firstidx = 0;
int lastidx = num - 1;
int middleidx = 0; // in case array is empty
bool foundMatch = false;
while (lastidx >= firstidx && !foundMatch) {
middleidx = (firstidx + lastidx) / 2;
if (findId < id[middleidx])
lastidx = middleidx – 1;
else if (findId > id[middleidx])
firstidx = middleidx + 1;
else foundMatch = true;
}
foundIndex = middleidx;
return foundMatch;
// note that if there is no match, then this new value
// goes either before or after the returned foundIndex
// and an insertion should check which
}
The main function would then do the following to obtain the corresponding subscript in match.
if (BinarySearch (id, numberStudents, studentId, match))
cout << studentID << "received a grade of "
<< grade[match]<< endl;
Using Arrays 458

Inserting New Data into a Sorted List

Inserting new elements into a sorted array is more difficult. Consider the above id and grade
arrays with the six elements currently in it, Illustration 10.3. Suppose that a student id of
255555555 with a grade of B needs to be inserted. What would have to be done to actually insert
this new student?

First, we would have to find the subscript where that id would be the proper sequence. In
this case, 255555555 belongs in the element with a subscript of 2, between the values 222222222
and 333333333. Since element 2 is already occupied, that element and all others must be moved
down one element. That is, the data at index 2 must be moved into subscript 3; 3 must be moved
into index 4; 4 into 5 and 5 into the unoccupied 6.

Caution. The movement of elements must be done in reverse order. If we move

33333333 into the id array at subscript 3, it replaces the data that is there, id 444444444. Thus,
the movement must be 5 into 6, 4 into 5, 3 into 4 and finally 2 into 3. Once the data in the
element of index 2 has been copied into element 3, we can then copy in the new id of 255555555
into the element at index 2.

Of course, nothing can be inserted if all the elements are used. To be robust, the insert
function should also make sure the requested id is not already in the list. Remember too when
parallel arrays are involved, what is done to one array must be echoed in the other parallel arrays.

For this example, assume the following const int definitions are available in the global
namespace.
const int InsertErrorBoundsExceeded = -1;
const int InsertErrorDuplicateId = -2;
const int InsertSuccess = 0;

Further assume that the new data to be inserted into the list are contained in the following
main() program's variables
long newId;
char newGrade;

The main() function invokes the InsertStudent() function as follows

int retcd = InsertStudent (id, grade, numberStudents,
MaxStudents, newId, newGrade);
if (retcd == InsertErrorBoundsExceeded) {
cout << "Error: Cannot add more students\n"
<< "The list is full\n";
}
else if (retcd == InsertErrorDuplicateId) {
cout << "Error: student id " << newId
<< " is already present in the list\n";
Using Arrays 459

}
The coding for the robust InsertStudent() function is as follows.
int InsertStudent (long id[], char grade[],
int& num, int maxlimit,
long newId, char newGrade) {
if (num >= maxlimit) // out of elements
return InsertErrorBoundsExceeded;
int index; // subscript where id belongs
if (BinarySearch (id, num, newId, index))
return InsertErrorDuplicateId; // found this id
if (num && newId > id[index]) // is insert after this one?
index++;// yes, so move index down one for move operations
if (index != num) { // do we need to move elements down?
// yes, so move all items down one index to make room
for (int j=num-1; j >= index; j--) {
id[j+1] = id[j];
grade[j+1] = grade[j];
}
}
// copy new data into lists
id[index] = newid;
grade[index] = newgrade;
num++;
return InsertSuccess;
}

Sorting an Array
The next question is how is an array sorted into numerical or alphabetical order? There are
numerous sorting algorithms; each offers benefits under proper circumstances. One method of
sorting may work well on a list that is in more or less random order, yet be dismal when perhaps
only one element in the list is actually out of order. Another method may work very well when
only one element in the list is out of sequence, but performs badly if the list is in random order.

The Straight Selection Sort, is the easiest to remember how to code. Begin with the first
element in the list. Compare the initial element with all the other elements in the list below it.
Any time one below it is smaller than that initial element, switch them, placing the smallest
valued element in that initial position in the list. When the end of the list is reached on that pass,
the smallest valued element is in the first element. Proceed to the second element and look
through all the remaining elements and place the next smallest remaining element in it. Once the
next-to-the-last row has been done, all the elements are now in low to high order, either
numerically or alphabetically.
Using Arrays 460

When performing an exchange because the next element is smaller than the one at hand,
the swap action must use a temporary variable. For example, supposing one wanted to swap the
contents of int variables x and y. You must also have a temporary int variable as shown.
int temp = x;
x = y;
y = temp;

If sorting parallel arrays such as the id and grade lists, if you swap a pair of elements in
the id array, you must also swap the same elements in the grade array to maintain the
corresponding values.

Here is a simple SortArrays() function that sorts the id and grade arrays into increasing
id number order.
void SortArrays (long id[], char grade[], int num) {
long tempid;
char tempgrade;
int i, j;
for (i=0; i<num-1; i++) {
// look through all elements below this one
// place smallest one in this the ith one
for (j=i+1; j<num; j++) {
if (id[j] < id[i]) {
// jth one is smaller, so swap them
tempid = id[i];
id[i] = id[j];
id[j] = tempid;
// also swap parallel grade array
tempgrade = grade[i];
grade[i] = grade[j];
grade[j] = tempgrade;
}
}
}
}

There are many other sorting algorithms. Each has specific benefits in certain
circumstances.

Arrays are very useful aggregates of data. Unsorted lists are often used to look up
information, to find matching items and can easily have new items added to them by appending
them onto the end of the list. Lists can be sorted into numerical order. Matching with sorted lists
can be highly optimized for speed of lookup but inserting a new item into the list requires more
work. Parallel arrays of data provide a convenient means of keeping related information
organized.
Using Arrays 461

Section B: A Computer Science Example

Cs10B — Account Processing using a Menu and Sorted Arrays
In this example, all of the techniques of array processing are combined into a working example
along with a menu. The objective is to store an array of account numbers and their account
balances for our credit card company. When the company first opens for business, there are no
accounts in the main file. However, soon, the accounts file will exist. Thus, when the program
first begins, it attempts to load the parallel arrays from the master file, if it exists.

A menu of choices is shown. If another customer is added, the two arrays must be kept in
increasing account number order. Hence, a binary search is used to find the correct insertion
point for the new account number. Provisions are made to display all accounts on the screen and
to look up the balance of a specific account. At the end of the run, a new master file can be
written.

Here is a sample run of the program.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs10b Sample Test Run *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 *
* 2 *
* 3 *
* 4 Acme Loan Services *
* 5 *
* 6 1 Add a New Account *
* 7 2 Find an Account Balance *
* 8 3 Display All Accounts *
* 9 4 Save the Accounts Data *
* 10 5 Quit the Program *
* 11 *
* 12 Enter the number of your choice: 1 *
* 13 Enter the new account number and balance: 10 *
* 14 1000 *
* 15 Data has been added successfully *
* 16 *
* 17 *
* 18 *
* 19 Acme Loan Services *
* 20 *
* 21 1 Add a New Account *
* 22 2 Find an Account Balance *
* 23 3 Display All Accounts *
* 24 4 Save the Accounts Data *
* 25 5 Quit the Program *
* 26 *
* 27 Enter the number of your choice: 1 *
* 28 Enter the new account number and balance: 20 *
Using Arrays 462

* 29 2000 *
* 30 Data has been added successfully *
* 31 *
* 32 *
* 33 *
* 34 Acme Loan Services *
* 35 *
* 36 1 Add a New Account *
* 37 2 Find an Account Balance *
* 38 3 Display All Accounts *
* 39 4 Save the Accounts Data *
* 40 5 Quit the Program *
* 41 *
* 42 Enter the number of your choice: 1 *
* 43 Enter the new account number and balance: 30 3000 *
* 44 Data has been added successfully *
* 45 *
* 46 *
* 47 *
* 48 Acme Loan Services *
* 49 *
* 50 1 Add a New Account *
* 51 2 Find an Account Balance *
* 52 3 Display All Accounts *
* 53 4 Save the Accounts Data *
* 54 5 Quit the Program *
* 55 *
* 56 Enter the number of your choice: 1 *
* 57 Enter the new account number and balance: 5 500 *
* 58 Data has been added successfully *
* 59 *
* 60 *
* 61 *
* 62 Acme Loan Services *
* 63 *
* 64 1 Add a New Account *
* 65 2 Find an Account Balance *
* 66 3 Display All Accounts *
* 67 4 Save the Accounts Data *
* 68 5 Quit the Program *
* 69 *
* 70 Enter the number of your choice: 1 *
* 71 Enter the new account number and balance: 25 2500 *
* 72 Data has been added successfully *
* 73 *
* 74 *
* 75 *
* 76 Acme Loan Services *
* 77 *
* 78 1 Add a New Account *
* 79 2 Find an Account Balance *
* 80 3 Display All Accounts *
Using Arrays 463

* 81 4 Save the Accounts Data *

* 82 5 Quit the Program *
* 83 *
* 84 Enter the number of your choice: 3 *
* 85 Account Number Balance *
* 86 *
* 87 5 $ 500.00 *
* 88 10 $ 1000.00 *
* 89 20 $ 2000.00 *
* 90 25 $ 2500.00 *
* 91 30 $ 3000.00 *
* 92 *
* 93 *
* 94 *
* 95 *
* 96 *
* 97 Acme Loan Services *
* 98 *
* 99 1 Add a New Account *
* 100 2 Find an Account Balance *
* 101 3 Display All Accounts *
* 102 4 Save the Accounts Data *
* 103 5 Quit the Program *
* 104 *
* 105 Enter the number of your choice: 2 *
* 106 Enter the account number to look up: 5 *
* 107 Account Number 5 has a balance of $500.00 *
* 108 *
* 109 *
* 110 *
* 111 Acme Loan Services *
* 112 *
* 113 1 Add a New Account *
* 114 2 Find an Account Balance *
* 115 3 Display All Accounts *
* 116 4 Save the Accounts Data *
* 117 5 Quit the Program *
* 118 *
* 119 Enter the number of your choice: 2 *
* 120 Enter the account number to look up: 30 *
* 121 Account Number 30 has a balance of $3000.00 *
* 122 *
* 123 *
* 124 *
* 125 Acme Loan Services *
* 126 *
* 127 1 Add a New Account *
* 128 2 Find an Account Balance *
* 129 3 Display All Accounts *
* 130 4 Save the Accounts Data *
* 131 5 Quit the Program *
* 132 *
Using Arrays 464

* 133 Enter the number of your choice: 2 *

* 134 Enter the account number to look up: 42 *
* 135 Error: account 42 is not in the database *
* 136 *
* 137 *
* 138 *
* 139 Acme Loan Services *
* 140 *
* 141 1 Add a New Account *
* 142 2 Find an Account Balance *
* 143 3 Display All Accounts *
* 144 4 Save the Accounts Data *
* 145 5 Quit the Program *
* 146 *
* 147 Enter the number of your choice: 4 *
* 148 Data saved in file newAccounts.txt *
* 149 *
* 150 *
* 151 *
* 152 Acme Loan Services *
* 153 *
* 154 1 Add a New Account *
* 155 2 Find an Account Balance *
* 156 3 Display All Accounts *
* 157 4 Save the Accounts Data *
* 158 5 Quit the Program *
* 159 *
* 160 Enter the number of your choice: 5 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Here is the Top-down design for Cs10b.

Cs10b Top-down Design

To better handle responding to menu choices, an enum is used:

enum Choice {AddNewAccount = 1, FindAccount, DisplayAccounts,
Save, Quit};
Using Arrays 465

The main function becomes very streamlined and is just this.

long accountNum[MAX];
double balance[MAX];

// load the initial accounts array

int numAccounts = LoadAccounts (accountNum, balance, MAX);

// process menu choices until Quit is chosen

Choice c = GetValidMenuChoice ();
while (cin && c != Quit) {
switch (c) {
case AddNewAccount:
AddAccount (accountNum, balance, numAccounts);
break;
case FindAccount:
FindTheAccount (accountNum, balance, numAccounts);
break;
case DisplayAccounts:
DisplayAllAccounts (accountNum, balance, numAccounts);
break;
case Save:
SaveAccounts (accountNum, balance, numAccounts);
break;
}
c = GetValidMenuChoice ();
}
Each of the functions is very simple and straightforward. Here is the complete Cs10b program.
Notice that by using lots of functions, the problem has become relatively easy to code.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs10b Account Processing with Menu *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 #include <iostream> *
* 2 #include <iomanip> *
* 3 #include <fstream> *
* 4 using namespace std; *
* 5 *
* 6 /***************************************************************/*
* 7 /* */*
* 8 /* Cs10b: Account Processing */*
* 9 /* */*
* 10 /***************************************************************/*
* 11 *
* 12 const int MAX = 100; *
* 13 *
* 14 enum Choice {AddNewAccount = 1, FindAccount, DisplayAccounts, *
* 15 Save, Quit}; *
* 16 *
* 17 int LoadAccounts (long accountNum[], double balance[], *
* 18 int limit); *
* 19 void DisplayMenu (); *
* 20 Choice GetValidMenuChoice (); *
* 21 bool AddAccount (long accountNum[], double balance[], *
* 22 int& limit); *
* 23 void FindTheAccount (long accountNum[], double balance[], *
Using Arrays 466

* 24 int limit); *
* 25 void DisplayAllAccounts (long accountNum[], double balance[], *
* 26 int limit); *
* 27 bool SaveAccounts (long accountNum[], double balance[], *
* 28 int limit); *
* 29 bool BinarySearch (long id[], int num, long findId, *
* 30 int& foundIndex); *
* 31 *
* 32 int main () { *
* 33 cout << fixed << setprecision (2); *
* 35 *
* 36 long accountNum[MAX]; *
* 37 double balance[MAX]; *
* 38 *
* 39 // load the initial accounts array *
* 40 int numAccounts = LoadAccounts (accountNum, balance, MAX); *
* 41 *
* 42 // process menu choices until Quit is chosen *
* 43 Choice c = GetValidMenuChoice (); *
* 44 while (cin && c != Quit) { *
* 45 switch (c) { *
* 46 case AddNewAccount: *
* 47 AddAccount (accountNum, balance, numAccounts); *
* 48 break; *
* 49 case FindAccount: *
* 50 FindTheAccount (accountNum, balance, numAccounts); *
* 51 break; *
* 52 case DisplayAccounts: *
* 53 DisplayAllAccounts (accountNum, balance, numAccounts); *
* 54 break; *
* 55 case Save: *
* 56 SaveAccounts (accountNum, balance, numAccounts); *
* 57 break; *
* 58 } *
* 59 c = GetValidMenuChoice (); *
* 60 } *
* 61 *
* 62 return 0; *
* 63 } *
* 64 *
* 65 /***************************************************************/*
* 66 /* */*
* 67 /* LoadAccounts: loads arrays from the file of accounts */*
* 68 /* returns the number in the array */*
* 69 /* */*
* 70 /***************************************************************/*
* 71 int LoadAccounts (long accountNum[], double balance[], *
* 72 int limit) { *
* 73 ifstream infile ("Accounts.txt"); *
* 74 if (!infile) *
* 75 return 0; // no accounts sold as yet *
* 76 int i = 0; *
Using Arrays 467

* 77 while (i<limit && infile >> accountNum[i] >> balance[i]) *

* 78 i++; *
* 79 if (i == limit && infile >> ws) { *
* 80 cerr << "Error: too many accounts for the program to handle\n";*
* 81 infile.close (); *
* 82 exit (1); *
* 83 } *
* 84 if (!infile.eof() && infile.fail()) { *
* 85 cerr << "Error: bad data in the accounts file\n"; *
* 86 infile.close (); *
* 87 exit (2); *
* 88 } *
* 89 infile.close (); *
* 90 return i; *
* 91 } *
* 92 *
* 93 /***************************************************************/*
* 94 /* */*
* 95 /* DisplayMenu: shows the menu on the screen */*
* 96 /* */*
* 97 /***************************************************************/*
* 98 void DisplayMenu () { *
* 99 cout << "\n\n\n Acme Loan Services\n\n" *
* 100 << "\t1 Add a New Account\n" *
* 101 << "\t2 Find an Account Balance\n" *
* 102 << "\t3 Display All Accounts\n" *
* 103 << "\t4 Save the Accounts Data\n" *
* 104 << "\t5 Quit the Program\n\n" *
* 105 << "Enter the number of your choice: "; *
* 106 } *
* 107 *
* 108 /***************************************************************/*
* 109 /* */*
* 110 /* GetValidMenuChoice: returns a valid menu choice */*
* 111 /* */*
* 112 /***************************************************************/*
* 113 Choice GetValidMenuChoice () { *
* 114 int num; *
* 115 do { *
* 116 DisplayMenu (); *
* 117 cin >> num; *
* 118 } while ((num < AddNewAccount || num > Quit) && cin); *
* 119 if (!cin) num = Quit; *
* 120 return (Choice) num; *
* 121 } *
* 122 *
* 123 /***************************************************************/*
* 124 /* */*
* 125 /* AddAccount: adds another account to the array - ret false */*
* 126 /* if it could not add another */*
* 127 /* */*
* 128 /***************************************************************/*
Using Arrays 468

* 129 bool AddAccount (long accountNum[], double balance[], *

* 130 int& limit) { *
* 131 if (limit == MAX) { *
* 132 cerr << "Error: accounts array size at maximum capacity\n"; *
* 133 return false; *
* 134 } *
* 135 long newAcctNum; *
* 136 double newBalance; *
* 137 cout << "Enter the new account number and balance: "; *
* 138 cin >> newAcctNum >> newBalance; *
* 139 if (!cin) { *
* 140 cout << "Error: bad data inputted\n"; *
* 141 return false; *
* 142 } *
* 143 int index; *
* 144 if (BinarySearch (accountNum, limit, newAcctNum, index)) { *
* 145 cerr << "Error: account " << newAcctNum *
* 146 << " is already in the database\n"; *
* 147 return false; *
* 148 } *
* 149 if (limit && newAcctNum > accountNum[index]) *
* 150 index++; *
* 151 if (index != limit) { // do we need to move elements down? *
* 152 // yes, so move all items down one index to make room *
* 153 for (int j=limit-1; j >= index; j--) { *
* 154 accountNum[j+1] = accountNum[j]; *
* 155 balance[j+1] = balance[j]; *
* 156 } *
* 157 } *
* 158 accountNum[index] = newAcctNum; *
* 159 balance[index] = newBalance; *
* 160 limit++; *
* 161 cout << "Data has been added successfully\n"; *
* 162 return true; *
* 163 } *
* 164 *
* 165 /***************************************************************/*
* 166 /* */*
* 167 /* FindTheAccount: displays account info for this account */*
* 168 /* */*
* 169 /***************************************************************/*
* 170 void FindTheAccount (long accountNum[], double balance[], *
* 171 int limit) { *
* 172 long findAcctNum; *
* 173 cout << "Enter the account number to look up: "; *
* 174 cin >> findAcctNum; *
* 175 if (!cin) { *
* 176 cout << "Error: bad data inputted\n"; *
* 177 return; *
* 178 } *
* 179 int index; *
* 180 if (BinarySearch (accountNum, limit, findAcctNum, index)) *
Using Arrays 469

* 181 cout << "Account Number " << findAcctNum *

* 182 << " has a balance of $" << balance[index] << endl; *
* 183 else *
* 184 cerr << "Error: account " << findAcctNum *
* 185 << " is not in the database\n"; *
* 186 } *
* 187 *
* 188 /***************************************************************/*
* 189 /* */*
* 190 /* DisplayAllAccounts: on-screen display of all accounts */*
* 191 /* */*
* 192 /***************************************************************/*
* 193 void DisplayAllAccounts (long accountNum[], double balance[], *
* 194 int limit) { *
* 195 cout << "Account Number Balance\n\n"; *
* 196 for (int i=0; i<limit; i++) *
* 197 cout << setw(10) << accountNum[i] << " $" << setw (10) *
* 198 << balance[i] << endl; *
* 199 cout << endl << endl; *
* 200 } *
* 201 *
* 202 /***************************************************************/*
* 203 /* */*
* 204 /* SaveAccounts: returns true if data is saved to a file */*
* 205 /* */*
* 206 /***************************************************************/*
* 207 bool SaveAccounts (long accountNum[], double balance[], *
* 208 int limit) { *
* 209 ofstream outfile ("newAccounts.txt"); *
* 210 if (!outfile) { *
* 211 cerr << "Error: cannot open the newAccounts.txt file\n"; *
* 212 return false; *
* 213 } *
* 214 for (int i=0; i<limit; i++) *
* 215 outfile << setw(10) << accountNum[i] << setw (10) *
* 216 << balance[i] << endl; *
* 217 outfile.close(); *
* 218 cout << "Data saved in file newAccounts.txt\n"; *
* 219 return true; *
* 220 } *
* 221 *
* 222 /***************************************************************/*
* 223 /* */*
* 224 /* BinarySearch: returns true if found; foundIndex has the idx */*
* 225 /* of the matching one or the insertion point if not fnd */*
* 226 /* */*
* 227 /***************************************************************/*
* 228 bool BinarySearch (const long id[], int num, long findId, *
* 229 int& foundIndex) { *
* 230 int firstidx = 0; *
* 231 int lastidx = num - 1; *
* 232 int middleidx = 0; *
Using Arrays 470

* 233 bool foundMatch = false; *

* 234 while (lastidx >= firstidx && !foundMatch) { *
* 235 middleidx = (firstidx + lastidx) / 2; *
* 236 if (findId < id[middleidx]) *
* 237 lastidx = middleidx - 1; *
* 238 else if (findId > id[middleidx]) *
* 239 firstidx = middleidx + 1; *
* 240 else foundMatch = true; *
* 241 } *
* 242 foundIndex = middleidx; *
* 243 return foundMatch; *
* 244 // note that if there is no match, then this new value *
* 245 // goes either before or after the returned foundIndex *
* 246 // and an insertion should check which *
* 247 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Cs10A — Merging Arrays

In this example, some methods of merging two arrays into one final array are examined. While
this problem could be solved a number of different ways, here, let’s use arrays to illustrate how
arrays can be merged.

Acme Salvage Company has just bought out one of their competitor’s companies, Jones
Salvage. Both are supplying parts and management wishes to merge their master file of parts into
one new file. Each line in each company’s master file of parts consists of a part id, quantity on
hand, manufacturer id, and cost. Originally, both companies’ master files are sorted into
increasing part id order. Write a program to merge the two separate master files into one new
master file, again in increasing part id order.

Two complexities arise. In a few cases, both companies carry a specific manufacturer’s
part. If a part in the Jones master file has a manufacturer id that matches one that is already in the
Acme master file, simply add the Jones quantity on hand to that of the Acme quantity. The other
potential problem is duplicate id numbers. When adding a Jones set of data to the merged arrays,
if the Jones id number is the same as an Acme id number, then go ahead and add the Jones data
using that duplicate id number. However, write out an error message to a log file so that
management can review the magnitude of the problem and recommend appropriate steps.

In this problem, while we could write a LoadAcmeArrays() and a LoadJonesArrays()

pair of functions, this is not good programming design. Since both master files contain the same
fields in the same order, one should write one LoadArrays() function that can input either
master file. The main() function first calls LoadArrays() twice, once for each company’s master
file. Then main() calls MergeArrays() and finally OutputArrays().
Using Arrays 471

In main(), notice that both files should be opened and tested before calling the
LoadArrays() function. It would be a waste of computer resources to open one file and input its
data before attempting to open the second file only to find that the file is not present.

How is the merge operation handled? If there were no duplicate manufacturer id numbers,
it would be simple. We would define three subscripts, each initialized to zero; one for the Acme
arrays, one for the Jones arrays and one for the merged arrays. Next, compare the current id
numbers and copy the set of data with the lower id numbers into the merged arrays. If the id
numbers were the same, copy both sets of data and write an error log message.

Figure 10.1 Top-Down Design for Merge Program

However, the possibility of duplicate manufacturer id numbers requires an additional

step. The simplest method is to traverse all of the Jones manufacturer ids looking for a
corresponding match in the Acme array. If a duplicate is found, add the Jones quantity to the
corresponding Acme quantity. Thus, MergeArrays() needs a MatchAcmeManuId() function to
check for matches and a RemoveJonesElement() function to remove one set of data from the
Jones arrays. Figure 10.1 shows the Top-Down Design.

Here are the Acme input file and the Jones input file that we can use for desk checking.
Notice I have underlined the two duplicate manufacturing id occurrences and bold faced the pair
of duplicate id numbers.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* The Input file: Cs10a-AcmeMaster.txt *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 111111 10 11223344 42.00 *
* 2 111112 100 11323355 445.55 *
* 3 111113 200 13223366 88.99 *
* 4 111222 300 14223377 123.45 *
* 5 111224 110 51223388 234.22 *
* 6 222221 33 51223399 19.99 *
* 7 222222 42 21224411 24.99 *
* 8 222223 55 14424444 255.55 *
Using Arrays 472

* 9 333331 84 31225544 35.99 *

* 10 444444 55 61226644 55.49 *
* 11 555555 444 21227744 56.31 *
* 12 777777 13 41228833 88.88 *
* 13 999995 15 61229944 84.84 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* The Input file: Cs10a-JonesMaster.txt *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 111112 100 11323355 445.55 *
* 2 111113 200 22323366 88.99 *
* 3 131111 10 11233444 42.00 *
* 4 141222 300 14453377 123.45 *
* 5 151224 110 51224444 234.22 *
* 6 252221 33 51223399 19.99 *
* 7 262222 42 25624411 24.99 *
* 8 272223 55 14656444 255.55 *
* 9 383331 84 31225666 35.99 *
* 10 484444 55 61226688 55.49 *
* 11 595555 444 21456454 56.31 *
* 12 787777 13 41223633 88.88 *
* 13 988995 15 61244944 84.84 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

The main() function defines the arrays as shown in Figure 10.2. Its coding steps are as
follows.

Figure 10.2 Main Storage for main()

In order to display the final summary of the results, main() must have the three counters
as shown in Figure 10.2, initialized to 0. Here is the sketch of the main() function.
attempt to open acmein and jonesin
if either fail to open, display an error message and quit
let acmeNum = LoadArrays (acmeId, acmeQty, acmeManuId, acmeCost, MAX,
acmein)
Using Arrays 473

close acmein
let jonesNum = LoadArrays (jonesId, jonesQty, jonesManuId, jonesCost, MAX,
jonesin)
close jonesin
open the logfile
let mergedNum = MergeArrays (
acmeId, acmeQty, acmeManuId, acmeCost, acmeNum,
jonesId, jonesQty, jonesManuId, jonesCost, jonesNum,
mergedId, mergedQty, mergedManuId, mergedCost, MAX,
mergedcount, dupids, addedFromJones, logfile)
open outfile
OutputArrays (mergedId, mergedQty, mergedManuId, mergedCost, mergedNum,
outfile)
close outfile
display on the logfile the mergedcount, the dupids and addedFromJones
close logfile

The LoadArrays() function is very simple. It is given: id[], qty[], manufid[], cost[],
limit and infile.
let i = 0;
while (i < limit && infile >> id[i] >> qty[i] >> manufid[i] >> cost[i])
i++;
if (i == limit && infile >> ws) then display an error and exit
return i;

The merge operation has four loops. Initially, we must look through all of the Jones array
for duplicate manufacturer id numbers. If any are found, the Jones quantity must be added to the
Acme quantity and that entry must be removed from the Jones arrays. The actual merging can
then take place. The second loop continues while there still exists at least one set of data in both
Acme and Jones arrays. When the first loop ends, there could still be additional items in the
Acme array or there could still be items left in the Jones arrays that have not yet been added.
Thus, two additional loops are needed to add these to the new arrays. Here is the sequence of
MergeArrays() which is passed:
aid[], aqty[], amanuid[], acost[], numAcme,
jid[], jqty[], jmanuid[], jcost[], numJones,
mid[], mqty[], mmanuid[], mcost[], limit,
mergedcount, dupids, addedFromJones and logfile
let ida and idj and idm = 0 - our subscripts for the three arrays
// look through all of the Jones arrays to find and remove all those with duplicate
// manufacturer id numbers
while (idj < numJones) do the following to look for duplicate manufacturer ids
let match = MatchAcmeManuId (amanuid, numAcme, jmanuid[idj])
if match >= 0, then we have found a duplicate
Using Arrays 474

aqty[match] += jqty[idj]; // add Jones’s qty to Acme’s qty

And remove this entry from the Jones arrays
RemoveJonesElement (idj, jid, jqty, jmanuid, jcost, numJones);
increment mergedcount
end if
else increment the Jones subscript idj
end while
reset the Jones array subscript idj back to 0

The second loop continues while there still exists at least one set of data in both Acme
and Jones arrays. We want to add the smaller of the two current id numbers (between Acme and
Jones) into the resultant merged arrays. Thus, we check to see which is smaller and move the
data accordingly.
while (ida < numAcme && idj < numJones) {
if the Acme product id smaller than Jones — aid[ida] < jid[idj]? If so then do
// copy this Acme item into the merged arrays
mid[idm] = aid[ida];
mqty[idm] = aqty[ida];
mmanuid[idm] = amanuid[ida];
mcost[idm] = acost[ida];
increment both the Acme and the merged array subscripts, ida and idm
end the then clause
otherwise see if the Jones id is smaller aid[ida] > jid[idj]? If so then do
// copy this Jones item into the merged arrays
mid[idm] = jid[idj];
mqty[idm] = jqty[idj];
mmanuid[idm] = jmanuid[idj];
mcost[idm] = jcost[idj];
increment both the subscripts, idj and idm
increment the count of those added from Jones, addedFromJones
end the otherwise clause
else here they are duplicate product id numbers

If they are duplicate numbers, then we must add both to the merged arrays and write an
informatory message to the log file.
mid[idm] = aid[ida]; // add in Acme's first
mqty[idm] = aqty[ida];
mmanuid[idm] = amanuid[ida];
mcost[idm] = acost[ida];
increment the merged array subscript for the next entry, idm
mid[idm] = jid[idj]; // add in Jones
mqty[idm] = jqty[idj];
mmanuid[idm] = jmanuid[idj];
Using Arrays 475

mcost[idm] = jcost[idj];
increment the merged array subscript for the next entry, idm
log an error message about these duplicate product ids to logfile
increment the duplicates count, dupids
increment the Jones and Acme subscripts, ida and idj
end the else
end the while loop

When the first loop ends, there could still be additional items in the Acme array or there
could still be items left in the Jones arrays that have not yet been added. Thus, two additional
loops are needed to add any remaining in these to the new arrays.
while (ida < numAcme) do the following
mid[idm] = aid[ida];
mqty[idm] = aqty[ida];
mmanuid[idm] = amanuid[ida];
mcost[idm] = acost[ida];
increment both ida and idm
end the while loop for extra Acme products
Now add in any left over Jones products
while (idj < numJones) do the following
mid[idm] = jid[idj];
mqty[idm] = jqty[idj];
mmanuid[idm] = jmanuid[idj];
mcost[idm] = jcost[idj];
increment both idj and idm
increment addedFromJones
end the while loop for extra Jones products
return the total number in the new arrays, idm

The MatchAcmeManuId() function is very straightforward. It is given manuid[], num,

matchmanuid. Its sequence is as follows.
for (int j=0; j<num; j++) do the following
if (manuid[j] == matchmanuid) then
simply return j which is the subscript of the matching manufacturer id
end the for loop
return NoMatch;

The RemoveJonesElement() function must remove the data at the passed thisone
subscript from the Jones arrays. It is passed thisone, jid[], jqty[], jmanuid[], jcost[] and
numJones. The method of removal is to move all the elements above this one down one row in
the arrays. Thus, it is a simple loop. The subscript of the last row in the arrays is numJones – 1.
for (int j=thisone; j<numJones–1; j++) {
jid[j] = jid[j+1];
Using Arrays 476

jqty[j] = jqty[j+1];
jmanuid[j] = jmanuid[j+1];
jcost[j] = jcost[j+1];
end the for loop
lower the numJones count by one element

The OutputArrays() function must write the new master file. It is given id[], qty[],
manuid[], cost[], limit and outfile. The sequence of steps is short.
setup floating point output format on outfile
for (int i=0; i<limit; i++) {
display to outfile id[i], qty[i], manuid[i] and cost[i]

Here are the log report and the merge error report produced when the program runs.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs10a - Merge Master Files - Output Log Report *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Error: duplicate product ids: 111113 *
* 2 Acme manufacturer id: 13223366 *
* 3 Jones manufacturer id: 22323366 *
* 4 *
* 5 2 Duplicate Jones manufacturer ids *
* 6 1 Duplicate product ids to be examined *
* 7 10 New products added from Jones *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* The Output file: MergeErrorLog.txt *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Error: duplicate product ids: 111113 *
* 2 Acme manufacturer id: 13223366 *
* 3 Jones manufacturer id: 22323366 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Here is the completed program, Cs10a. Notice how easy it is to convert the sketch into
the actual program coding.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs10a - Merge Master Files *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs10a Merge Master Files */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 using namespace std; *
* 11 *
* 12 const int MAX = 100; // the maximum number of parts *
Using Arrays 477

* 13 const int NoMatch = -1; // code for no match found *

* 14 *
* 15 int LoadArrays (long id[], int qty[], long manufid[], *
* 16 double cost[], int limit, istream& infile); *
* 17 int MergeArrays (long aid[], int aqty[], long amanuid[], *
* 18 double acost[], int numAcme, *
* 19 long jid[], int jqty[], long jmanuid[], *
* 20 double jcost[], int& numJones, *
* 21 long mid[], int mqty[], long mmanuid[], *
* 22 double mcost[], int limit, *
* 23 int& mergedcount, int& dupids, *
* 24 int& addedFromJones, ostream& logfile); *
* 25 int MatchAcmeManuId (long manuid[], int num, long matchmanuid); *
* 26 void RemoveJonesElement (int thisone, long jid[], int jqty[], *
* 27 long jmanuid[], double jcost[], *
* 28 int& numJones); *
* 29 void OutputArrays (long id[], int qty[], long manuid[], *
* 30 double cost[], int limit, ostream& outfile); *
* 31 *
* 32 int main () { *
* 33 long acmeId[MAX]; // the Acme master file data *
* 34 int acmeQty[MAX]; *
* 35 long acmeManuId[MAX]; *
* 36 double acmeCost[MAX]; *
* 37 int acmeNum; *
* 38 *
* 39 long jonesId[MAX]; // the Jones master file data *
* 40 int jonesQty[MAX]; *
* 41 long jonesManuId[MAX]; *
* 42 double jonesCost[MAX]; *
* 43 int jonesNum; *
* 44 *
* 45 long mergedId[MAX]; // the merged file data *
* 46 int mergedQty[MAX]; *
* 47 long mergedManuId[MAX]; *
* 48 double mergedCost[MAX]; *
* 49 int mergedNum; *
* 50 *
* 51 // attempt to open both master files *
* 52 ifstream acmein ("Cs10a-AcmeMaster.txt"); *
* 53 if (!acmein) { *
* 54 cerr << "Error: cannot open Cs10a-AcmeMaster.txt file\n"; *
* 55 return 1; *
* 56 } *
* 57 ifstream jonesin("Cs10a-JonesMaster.txt"); *
* 58 if (!jonesin) { *
* 59 cerr << "Error: cannot open Cs10a-JonesMaster.txt file\n"; *
* 60 acmein.close (); *
* 61 return 1; *
* 62 } *
* 63 *
* 64 // load both files of data *
Using Arrays 478

* 65 acmeNum = LoadArrays (acmeId, acmeQty, acmeManuId, acmeCost,MAX,*

* 66 acmein); *
* 67 acmein.close (); *
* 68 jonesNum = LoadArrays (jonesId, jonesQty, jonesManuId,jonesCost,*
* 69 MAX, jonesin); *
* 70 jonesin.close (); *
* 71 *
* 72 int mergedcount = 0; // count of duplicate Jones products *
* 73 int dupids = 0; // count of duplicate product ids *
* 74 int addedFromJones = 0;// count of new products added from Jones*
* 75 ofstream logfile ("results-log.txt"); *
* 76 *
* 77 // merge both sets together, updating results counts *
* 78 mergedNum = MergeArrays ( *
* 79 acmeId, acmeQty, acmeManuId, acmeCost, acmeNum, *
* 80 jonesId, jonesQty, jonesManuId, jonesCost, jonesNum,*
* 81 mergedId, mergedQty, mergedManuId, mergedCost, MAX, *
* 82 mergedcount, dupids, addedFromJones, logfile); *
* 83 ofstream outfile ("master-new.txt"); *
* 84 *
* 85 // write out new master file *
* 86 OutputArrays (mergedId, mergedQty, mergedManuId, mergedCost, *
* 87 mergedNum, outfile); *
* 88 outfile.close (); *
* 89 *
* 90 // print final results to log file *
* 91 logfile << endl << setw (4) << mergedcount *
* 92 << " Duplicate Jones manufacturer ids\n" *
* 93 << setw (4) << dupids *
* 94 << " Duplicate product ids to be examined\n" *
* 95 << setw (4) << addedFromJones *
* 96 << " New products added from Jones\n"; *
* 97 logfile.close (); *
* 98 return 0; *
* 99 } *
* 100 *
* 101 /***************************************************************/*
* 102 /* */*
* 103 /* LoadArrays: input the data */*
* 104 /* */*
* 105 /***************************************************************/*
* 106 *
* 107 int LoadArrays (long id[], int qty[], long manufid[], *
* 108 double cost[], int limit, istream& infile) { *
* 109 int i = 0; *
* 110 while (i < limit && *
* 111 infile >> id[i] >> qty[i] >> manufid[i] >> cost[i]) { *
* 112 i++; *
* 113 } *
* 114 if (i == limit && infile >> ws) { *
* 115 cerr << "Error: array bounds exceeded\n"; *
* 116 exit (1); *
Using Arrays 479

* 117 } *
* 118 // not checking for bad input data - these are production files *
* 119 return i; *
* 120 } *
* 121 *
* 122 /***************************************************************/*
* 123 /* */*
* 124 /* MergeArrays: merge the two arrays together but check for */*
* 125 /* duplicate manufacturer id values - if found, */*
* 126 /* add jones to acme */*
* 127 /* */*
* 128 /***************************************************************/*
* 129 *
* 130 int MergeArrays ( *
* 131 long aid[], int aqty[], long amanuid[], double acost[], *
* 132 int numAcme, *
* 133 long jid[], int jqty[], long jmanuid[], double jcost[], *
* 134 int& numJones, *
* 135 long mid[], int mqty[], long mmanuid[], double mcost[], *
* 136 int limit, *
* 137 int& mergedcount, int& dupids, int& addedFromJones, *
* 138 ostream& logfile) { *
* 139 int ida = 0, idj = 0, idm = 0; *
* 140 // find all duplicate manufacturer id numbers and add Jones *
* 141 // into corresponding Acme qty and remove the dupe from Jones *
* 142 while (idj < numJones) { *
* 143 int match = MatchAcmeManuId (amanuid, numAcme, jmanuid[idj]); *
* 144 if (match >= 0) { // found a duplicate *
* 145 aqty[match] += jqty[idj]; // add to Acme's qty *
* 146 RemoveJonesElement (idj, jid, jqty, jmanuid, jcost, numJones);*
* 147 mergedcount++; // increment merged into Acme count *
* 148 } *
* 149 else idj++; *
* 150 } *
* 151 idj = 0; *
* 152 // perform main merge of Jones and Acme into new arrays *
* 153 // preserving sort order on product id array *
* 154 while (ida < numAcme && idj < numJones) { *
* 155 if (aid[ida] < jid[idj]) { // Acme prod id is smaller *
* 156 mid[idm] = aid[ida]; // copy over this Acme item *
* 157 mqty[idm] = aqty[ida]; *
* 158 mmanuid[idm] = amanuid[ida]; *
* 159 mcost[idm] = acost[ida]; *
* 160 ida++; *
* 161 idm++; *
* 162 } *
* 163 else if (aid[ida] > jid[idj]) { // Jones prod id is smaller *
* 164 mid[idm] = jid[idj]; // copy over this Jones item *
* 165 mqty[idm] = jqty[idj]; *
* 166 mmanuid[idm] = jmanuid[idj]; *
* 167 mcost[idm] = jcost[idj]; *
* 168 idj++; *
Using Arrays 480

* 169 idm++; *
* 170 addedFromJones++; // increment count of added ones*
* 171 } *
* 172 else { // error - duplicate prod ids *
* 173 mid[idm] = aid[ida]; // add in Acme's first *
* 174 mqty[idm] = aqty[ida]; *
* 175 mmanuid[idm] = amanuid[ida]; *
* 176 mcost[idm] = acost[ida]; *
* 177 idm++; *
* 178 mid[idm] = jid[idj]; // add in Jones *
* 179 mqty[idm] = jqty[idj]; *
* 180 mmanuid[idm] = jmanuid[idj]; *
* 181 mcost[idm] = jcost[idj]; *
* 182 idm++; *
* 183 // log an error message about these duplicate product ids *
* 184 logfile << "Error: duplicate product ids: " << aid[ida] *
* 185 << endl; *
* 186 logfile << " Acme manufacturer id: " << amanuid[ida] *
* 187 << endl; *
* 188 logfile << " Jones manufacturer id: " << jmanuid[idj] *
* 189 << endl; *
* 190 dupids++; // increment the duplicates count *
* 191 ida++; *
* 192 idj++; *
* 193 } *
* 194 } *
* 195 while (ida < numAcme) { // add in any left over Acme products *
* 196 mid[idm] = aid[ida]; *
* 197 mqty[idm] = aqty[ida]; *
* 198 mmanuid[idm] = amanuid[ida]; *
* 199 mcost[idm] = acost[ida]; *
* 200 ida++; *
* 201 idm++; *
* 202 } *
* 203 while (idj < numJones) { // add in any left over Jones products *
* 204 mid[idm] = jid[idj]; *
* 205 mqty[idm] = jqty[idj]; *
* 206 mmanuid[idm] = jmanuid[idj]; *
* 207 mcost[idm] = jcost[idj]; *
* 208 idj++; *
* 209 idm++; *
* 210 addedFromJones++; *
* 211 } *
* 212 // return the total number in the new arrays *
* 213 return idm; *
* 214 } *
* 215 *
* 216 /***************************************************************/*
* 217 /* */*
* 218 /* MatchAcmeManuId: match this jones manufacturer id to acme's */*
* 219 /* */*
* 220 /***************************************************************/*
Using Arrays 481

* 221 *
* 222 int MatchAcmeManuId (long manuid[], int num, long matchmanuid) {*
* 223 for (int j=0; j<num; j++) { *
* 224 if (manuid[j] == matchmanuid) { *
* 225 return j; // return subscript of the matching manufacturer id *
* 226 } *
* 227 } *
* 228 return NoMatch; *
* 229 } *
* 230 *
* 231 /***************************************************************/*
* 232 /* */*
* 233 /* RemoveJonesElement: remove this one from jones arrays */*
* 234 /* */*
* 235 /***************************************************************/*
* 236 *
* 237 void RemoveJonesElement (int thisone, long jid[], int jqty[], *
* 238 long jmanuid[], double jcost[], *
* 239 int& numJones) { *
* 240 // move all elements down one *
* 241 for (int j=thisone; j<numJones-1; j++) { *
* 242 jid[j] = jid[j+1]; *
* 243 jqty[j] = jqty[j+1]; *
* 244 jmanuid[j] = jmanuid[j+1]; *
* 245 jcost[j] = jcost[j+1]; *
* 246 } *
* 247 numJones--; // reset the number of elements in the arrays *
* 248 } *
* 249 *
* 250 /***************************************************************/*
* 251 /* */*
* 252 /* OutputArrays: rewrite the new master file */*
* 253 /* */*
* 254 /***************************************************************/*
* 255 *
* 256 void OutputArrays (long id[], int qty[], long manuid[], *
* 257 double cost[], int limit, ostream& outfile) { *
* 258 // setup floating point output format *
* 259 outfile << fixed << setprecision (2); *
* 262 *
* 263 // write a new master file *
* 264 for (int i=0; i<limit; i++) { *
* 265 outfile << setw (10) << id[i] << setw (10) << qty[i] *
* 266 << setw (10) << manuid[i] << setw (10) << cost[i] *
* 267 << endl; *
* 268 } *
* 269 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Using Arrays 482

Section C: An Engineering Example

Engr10a — Statistical Computations
Single dimensioned arrays are often used in statistical applications. We have seen that
determining the average, the highest and lowest values of an array are simple to calculate. If the
array is subsequently sorted, the mean value, that is, the value in the middle, can be found.
Given the average of a distribution, one can then calculate the standard deviation from the
average, which represents the spread of the data about the average. If the data are tightly grouped
about the mean, the deviation is small. The variance is defined to be the square of the deviation.
The final statistic is the coefficient of variance, which is a ratio of the standard deviation to the
mean often in the form of a percentage.

To find the median or middle value of a distribution, the values must be sorted into
numerical order. If a distribution of values has N elements or values in it, then the median value
is located at element N/2 if N is an odd number. Assume N is 3, then 3/2 yields subscript 1 for
the median value. However, if N is an even number, say 4 for example, then N/2 yields a
subscript of 2. But there are effectively two elements in the middle. Thus, if N is even, the
median value must be the average of the two elements in the middle. For example, suppose that
N is 4. The median would be the average of the values of the elements located at index 1 and 2.
Thus, we have
median value is array[N/2] when N is odd
median value is (array[N/2] + array[(N–1)/2]) / 2. when N is even

The standard deviation from the mean or average is given by the formula

Here xi is the ith observation and xmean is the mean or average value. Notice that the divisor is
N–1. There is no such thing as a spread of values about a single point.

Our example this time comes from Electrical Engineering. The electrical resistance for a
circuit element has been measured in the laboratory. The file resist.txt contains an unknown
number of these measurements. The company wishes to publish its findings. The published value
of the resistance is the mean of these values, but the standard deviation and coefficient of
variance must be published as well, providing an estimate of the accuracy of this resistance
value.

Engr10a provides a solution to this problem. The main() function calls LoadArray()
first to input the values. Next, SortArray() is called to get them into order as required by the
Using Arrays 483

formula. Now the mean value can be found and the FindStdDev() function can be called to find
the standard deviation from the mean. Then the results are printed. Since the problem is so
similar to the coding samples presented in the Basic Section above, I have only reproduced the
final program and results.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr10a - Statistics on a set of resistance observations *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr10a: Statistics on a set of resistance observations */*
* 4 /* Finds the mean, std deviation, and */*
* 5 /* coefficient of variance */*
* 6 /* */*
* 7 /* */*
* 8 /***************************************************************/*
* 9 *
* 10 #include <iostream> *
* 11 #include <iomanip> *
* 12 #include <fstream> *
* 13 #include <cmath> *
* 14 using namespace std; *
* 15 *
* 16 int LoadArray (double resist[], int limit); *
* 17 double FindAverage (const double resist[], int numpts); *
* 18 void SortResist (double resist[], int numpts); *
* 19 double FindStdDev (const double resist[], int numpts,double avg);*
* 20 *
* 21 const int MaxPts = 1000; *
* 22 *
* 23 int main () { *
* 24 *
* 25 // setup floating point output format *
* 26 cout << fixed << setprecision (2); *
* 29 *
* 30 double resist[MaxPts]; // array of resistance values *
* 31 double mean; // final resistance, the mean value *
* 32 double stdDev; // standard deviation of the mean *
* 33 double cv; // percentage coefficient of variance *
* 34 int numpts; // number of resistance values in array *
* 35 *
* 36 numpts = LoadArray (resist, MaxPts); // load resistance values *
* 37 double average = FindAverage (resist, numpts); // find average *
* 38 SortResist (resist, numpts); // sort resistances low to high *
* 39 // find the median resistance *
* 40 if (numpts % 2) { // odd number of points *
* 41 mean = resist[numpts/2]; *
* 42 } *
* 43 else { // even number of points *
* 44 mean = (resist[numpts/2] + resist[(numpts-1)/2]) / 2.; *
* 45 } *
* 46 stdDev = FindStdDev (resist, numpts, average); *
Using Arrays 484

* 47 cv = (stdDev*stdDev) / mean * 100; // find cv in percentage *

* 48 *
* 49 //display results *
* 50 cout << "Mean Resistance: " << setw (10) << mean << " ohms\n";*
* 51 cout << "Std Deviation: " << setw (10) << stdDev << endl; *
* 52 cout << "CV: " << setw (10) << cv << '%' << endl; *
* 53 *
* 54 return 0; *
* 55 } *
* 56 *
* 57 /***************************************************************/*
* 58 /* */*
* 59 /* LoadArray: load an array of resistance values */*
* 60 /* aborts program is file not found, too much data, bad data*/*
* 61 /* */*
* 62 /***************************************************************/*
* 63 *
* 64 int LoadArray (double resist[], int limit) { *
* 65 ifstream infile; *
* 66 infile.open ("resist.txt", ios::in); *
* 67 if (!infile) { *
* 68 cerr << "Error: cannot open file resist.txt\n"; *
* 69 exit (1); *
* 70 } *
* 71 int i = 0; *
* 72 while (i<limit && infile >> resist[i]) { *
* 73 i++; *
* 74 } *
* 75 if (i == limit && infile >> ws) { *
* 76 cerr << "Error: too many resistance values - maximum is " *
* 77 << limit << endl; *
* 78 infile.close (); *
* 79 exit (2); *
* 80 } *
* 81 infile.close (); *
* 82 if (!infile.eof() && infile.fail()) { *
* 83 cerr << "Error: bad data on line " << i+1 << endl; *
* 84 infile.close (); *
* 85 exit (2); *
* 86 } *
* 87 infile.close (); *
* 88 return i; *
* 89 } *
* 90 *
* 91 /***************************************************************/*
* 92 /* */*
* 93 /* SortResist: sorts the array into low to high order */*
* 94 /* */*
* 95 /***************************************************************/*
* 96 *
* 97 void SortResist (double resist[], int numpts) { *
* 98 int i, j; *
Using Arrays 485

* 99 double temp; *
* 100 for (i=0; i<numpts-1; i++) { *
* 101 for (j=i+1; j<numpts; j++) { *
* 102 if (resist[j] < resist[i]) { *
* 103 temp = resist[i]; *
* 104 resist[i] = resist[j]; *
* 105 resist[j] = temp; *
* 106 } *
* 107 } *
* 108 } *
* 109 } *
* 110 *
* 111 /***************************************************************/*
* 112 /* */*
* 113 /* FindAverage: finds the average resistance r */*
* 114 /* */*
* 115 /***************************************************************/*
* 116 *
* 117 double FindAverage (const double resist[], int numpts) { *
* 118 double sum = 0; *
* 119 for (int i=0; i<numpts; i++) { *
* 120 sum += resist[i]; *
* 121 } *
* 122 return sum / numpts; *
* 123 } *
* 124 *
* 125 /***************************************************************/*
* 126 /* */*
* 127 /* FindStdDev: finds the standard deviation from the average */*
* 128 /* aborts the program if the number of points is 1 */*
* 129 /* */*
* 130 /***************************************************************/*
* 131 *
* 132 double FindStdDev (const double resist[], int numpts, *
* 133 double avg) { *
* 134 // guard against a no solution *
* 135 if (numpts == 1) { *
* 136 cerr << *
* 137 "Error: data has only one point, cannot calculate std dev\n";*
* 138 exit (3); *
* 139 } *
* 140 double dev; *
* 141 double factor; *
* 142 double sum = 0; *
* 143 // compute the sum of the square of the differences *
* 144 for (int i=0; i<numpts; i++) { *
* 145 factor = resist[i] - avg; *
* 146 sum += factor * factor; *
* 147 } *
* 148 // calculate the std dev *
* 149 dev = sqrt (sum / (numpts - 1)); *
* 150 return dev; *
Using Arrays 486

* 151 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
The resist.txt file contains these data:
128.35
130.44
129.33
128.75
129.22
130.21
128.56
129.45
129.88
129.12

The program produces these results:

Mean Resistance: 129.28 ohms
Std Deviation: 0.69
CV: 0.37%

Least Squares Curve Fitting

Perhaps one of the most important numerical analysis tools is least squares curve fitting, a
process that determines the best fitting straight line to a set of observational data. Plotting a "best
fit" curve to data is an important facet of research. It is done by regression analysis; one
postulates a mathematical model that will model the data and by regression find the set of
coefficients that will minimize some measurement of the deviation from that model - the best fit
for that model. Often the model is a linear one; straight lines are the easiest to utilize:
y = f(x) = ax + b

Linear regression is used to find the best set of a and b values such that the square of the
distances between the line and the observed data is minimized. The raw data consists of points
(x1,y1), (x2,y2), ... (xn,yn) — for a total of n sets. We wish to minimize the distances (d) of the
observed points from the theoretical line.
Using Arrays 487

Figure 10.3 Least Squares Derivation

The formula for the distance d is:

d = y – f(x) = y – (ax + b)
d = y – ax – b
We wish to minimize these distances. But if we just sum the distances, some are likely
positive and some negative. They would tend to cancel each other out leaving us with an
incorrect measurement. Instead, we minimize the squares of the distances. This is shown in
Figure 10.3.

The total error E is then the sum of all distances squared is given by

Substituting for d, we get

How do we minimize E? Use differential calculus to get the partial derivatives of E with
respect to the two unknown coefficients a and b. A minimum of a function occurs when the
derivative is 0.
Using Arrays 488

Set these two equations to 0 and reduce and rearrange to get

Thus, we have two equations in two unknowns. Using Cramer's Rule to solve for a and b,
we get the following values.

Thus, given a set of observations, we can find the best fitting curve rather easily. This
theory is used in several problems below.

New Syntax Summary

If a function is not going to change any data in the passed array, that array should be made
constant.
void PrintTemps (const float temps[]);
Using Arrays 489

Design Exercises
1. Reverse Order

Acme Corporation has a master.txt file of item numbers and their corresponding costs. The file
has been sorted into ascending order on the item numbers. However, in order to compare their
items with a competitor, the printout must be in high to low order based on the item numbers.
Management does not want to spend the computer time to actually resort the file because it is a
very large one. Write a program that inputs the long item numbers and float costs into two arrays
that can handle a maximum of 10,000 items. Then print the report listing the item number and
associated cost.

2. Statistical Study of a Car’s Mpg (miles per gallon)

For 36 months, a record has been kept of gas station fill-ups. Specifically, the driver logged the
date (mm-dd-yyyy), the odometer reading in miles and the number of gallons of gas it took to fill
the car. It is postulated that one can use the average monthly mile per gallons (mpg) figures to
determine when the car needs maintenance. That is, when the mpg decreases substantially, it is
indicative of at least a tuneup being required. Write a program to prove or disprove the theory.

Input the log.txt file whose lines consist of the date, the mileage and the number of
gallons. For each line, calculate the mpg. Accumulate the mpg figures until the month changes
and then find that month’s average mpg. Store that month’s average mpg into an array which can
hold up to 36 values. The very first line of the file is special. It represents the very first full tank
of gas, so no average can be calculated for this entry.

Now print a report listing each month’s average mpg and print a ** beside that mpg
which is more than 20% different from the previous month. The report might appear like this.
Avg Monthly Mpg
35.6
36.5
25.5 **
34.5 **
33.9
Using Arrays 490

Stop! Do These Exercises Before Programming

1. Acme National Sales expanded and their parallel arrays of state codes and corresponding tax
rates have grown larger. The programmer has coded the following to load the tables. Why does it
not work? What must be done to fix the coding so that the tables are correctly loaded?
const int MaxLimit = 50;
int LoadArrays (int stateCode[], taxRate[], int limit);
int main () {
double taxRate[50];
int stateCode[50];
int numValues;
LoadArrays (stateCode, taxRate, 50);
...
int LoadArrays (int stateCode[], taxRate[], int limit) {
...// gets infile opened
int j=0;
while (infile >> stateCode[limit] >> taxRate[limit])
j++;
infile.close();
return limit;
}

2. With the arrays loaded, the programmer embarked on the writing of a sort function to get the
state codes in numerical order for faster table look ups within the main program. The following
function is to sort the state codes into increasing numerical order. It does not work properly.
Why? How can it be fixed up?
void SortStateCodes (int state[], int limit) {
int j, k, temp;
for (j=0; j<limit; j++) {
for (k=j; k<limit; k++) {
state[j] = temp;
temp = state[k];
state[k] = state[j];
}
}
}

3. Now that the states are sorted successfully, the programmer realized that the tax rates needed
to be sorted as well. Tired at the end of a long day, he produced the following patch to his
program. Assume that the SortTaxRates() function now works perfectly.
const int MaxLimit = 50;
int LoadArrays (int stateCode[], taxRate[], int limit);
void SortStateCodes (int states[], int limit);
void SortTaxRates (double rates[], int limit);
Using Arrays 491

int main () {
double taxRate[MaxLimit];
int stateCode[MaxLimit];
...
SortStateCodes (stateCode, numValues);
SortTaxRates (taxrate, numValues);
...
When he tested the program, all of the taxes were completely wrong. What is wrong with the
design of the entire sorting method? How can it be fixed? Write the sorting code that would
enable the program, given a state code, find the correct tax rate. (However, do not write any table
look up coding.)

4. Grateful for your assistance in bailing him out of his coding mess yesterday, the programmer
today has embarked upon construction of a fast matching function that is given the two arrays
and the number of elements in it and returns the tax rate. This is what has been produced thus far.
double FindTaxRate (int states[], double rates[], int num,
int findState) {
int j = 0;
double taxRate;
while (j<num) {
if (findState == states[num]) {
taxRate = rates[j];
}
j++;
}
return taxRate;
}
The function has serious design flaws and of course does not work properly. Why? What must be
done to correct this coding so that it produces the correct taxes in all circumstances?

5. Looking over Problem 4 just above, the FindTaxRate() function does not in any way make
use of the fact that the state codes are sorted into numerical order. Using a binary search function
would provide the speed up desired. Assume that the BinarySearch() function presented earlier
in this chapter has been rewritten using an array of ints instead of longs. Rewrite the
FindTaxRate() function above to use the BinarySearch() in finding the tax rate to return.
Using Arrays 492

Programming Problems
Problem Cs10-1 — Array Manipulations
The included file Cs10-1.cpp provides the main() function and some helper functions that enable
it to thoroughly test your coding. In this problem, you are going to practice various array
manipulations. You are to write each of the six functions described below. Do NOT define any
additional arrays inside your functions; work only with the array(s) passed as the argument(s).

1. Write the LoadData() function whose prototype is

int LoadData (istream &infile, int array[], int arraySize,
int &count);

The LoadData() function reads integers from the passed stream and loads them into the
successive elements of the array passed as the second argument. The third argument specifies the
size of the array (number of elements). The function loads data into the array until the array is
full or the end of the file is reached, whichever occurs first. The function assigns the count of
integers stored into the array to the fourth argument. The function must not overflow the
boundary of the array or read any more data after the array is full. The function returns a status
code: 0 if the whole input file is successfully loaded into the array, 1 if the input file is too large
to fully load into the array (the array is full but there are more data in the file), or –1 if there is
invalid (non-integer) data in the input file. In all cases, the count parameter must be assigned the
correct count of numbers stored in the array.

2. Write the List() function whose prototype is

void List (const int array[], int count);
The List() function prints the numbers in the array passed as the first argument; the second
argument is the count of numbers in the array. Print each number in a three-column field and
print a newline character after each set of twenty numbers. After all the numbers are printed,
print one blank line.

3. Write the CopyArray() function whose prototype is

void CopyArray (int destArray[], const int srcArray[], int count);
The Copy() function copies the contents of one array to another. The first argument is the
destination array and the second argument is the source array. The third argument specifies the
number of elements to copy.

4. Write the RotateLeft() function whose prototype is

void RotateLeft (int array[], int count);
The RotateLeft() function shifts each number in the array, except the first, one element to the
“left” (the element with the next lower subscript). The first number in the array is shifted to the
last element in the array.
Using Arrays 493

5. Write the RotateRight() function whose prototype is

void RotateRight (int array[], int count);
The RotateRight() function shifts each number in the array, except the last, one element to the
“right” (the element with the next higher subscript). The last number in the array is shifted to the
first element. (Note: you can RotateRight() by repeating RotateLeft() for count – 1 times.
However, this is extremely inefficient. Use an efficient method for rotating to the right; do not
repeatedly rotate the array to left.)

6. Write the Reverse() function whose prototype is

void Reverse (int array[], int count);
The Reverse() function reverses the sequence of the numbers in the array. (Note: you can reverse
an array by repeating RotateLeft() or RotateRight() for count – 1 times, while at the same time
decreasing the count after each repetition. However, this is extremely inefficient. Use an efficient
method for reversing; do not repeatedly rotate the array.)

Test your program on the provided test files Cs10-1a.txt, Cs10-1b.txt, Cs10-1c.txt,
Cs10-1d.txt and Cs10-1e.txt. Here are the outputs you should get (I have single-spaced the
output to reduce lines here in the book.)

Enter input file name: CS10-1a.txt

status = 0 count = 9

load: 99 88 77 66 55 44 33 22 11
copy: 99 88 77 66 55 44 33 22 11
left: 88 77 66 55 44 33 22 11 99
right: 11 99 88 77 66 55 44 33 22
reverse: 11 22 33 44 55 66 77 88 99

Enter input file name: CS10-1b.txt

status = 0 count = 20

load: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
copy: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
left: 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1
right: 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
reverse: 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1

Enter input file name: CS10-1c.txt

status = 1 count = 20

load: 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
copy: 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
left: 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 2
right: 40 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38
reverse: 40 38 36 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2

Enter input file name:CS10-1d.txt

status = -1 count = 5
Using Arrays 494
load: 90 80 70 60 50
copy: 90 80 70 60 50
left: 80 70 60 50 90
right: 50 90 80 70 60
reverse: 50 60 70 80 90

Problem Cs10-2 — Acme Sales Summary

The Acme has a master file, Cs10-2-master.txt, that contains the information on all the products
that they sell. Each line represents a specific product and includes an item number and the cost.
Item numbers can be up to six digits long. The file is in increasing item number order. Allow for
a maximum of 500 products.

Each day, Acme stores all of their daily sales in the Cs10-2-transactions.txt file. It
includes the invoice number (up to nine digits), the product id (the item number) and the quantity
ordered.

After loading the master file of data, input each transaction line and calculate the total
cost of that order and print it nicely formatted. When you reach the end of the transaction file,
display the grand total sales figure beneath the total cost column. The report should look
something like this.
Acme Sales Summary

Invoice Product Id Quantity Cost

999999999 999999 999 $999999.99

999999999 999999 999 $999999.99
----------
$9999999.99

Note that you need parallel arrays for the item number and cost arrays. Also the match
function should take advantage of the fact that the file is sorted into item number order.

If a given product id cannot be found in the master file, print that invoice, id and quantity
as usual in the report, but instead of printing the cost, display “error” in the cost column.
Using Arrays 495

Problem Cs10-3 — Acme Sales Inventory Update

Continuing work with the Acme master file described in Problem Cs10-2, each week Acme adds
new products to their line. Occasionally, they drop a line of products. The file Cs10-3-update.txt
contains the week’s purchasing transactions.

The first character on each update transaction line contains an uppercase letter that
describes this transaction. If the character is ‘A’, then this is an ‘add new product’ transaction and
the next two numbers represent the product id and its cost. If the character is a ‘D’, then this is a
deletion request and the next number is the product id.

Write a program to perform this master file update process. When processing an add
request, insert that product id and its cost into the correct position in the sorted arrays. When
processing a deletion request, remove that product id and its cost from the arrays.

Your program should produce a log.txt file of the requested transaction and its results,
similar to the following.
Acme Weekly Update Master File Log

Action Product Id Cost Result

A 123434 444.44 Success
A 123434 444.44 Error: duplicate product id
D 234444 Removed

Finally, when the update transactions have all been processed, output the new modified
master file (call it master-new.txt). For debugging purposes, also print the contents of the two
arrays in columnar format.

Problem Cs10-4 — Grade Statistics

Write a program that calculates the results of a test. Sample output is shown below. Each line of
input contains a student identification number and a test score. The identification number is a one
to nine digit whole number; the test score may contain a decimal point. The input file may
contain from zero to fifty students. Write separate functions as follows:
1. The ReadInput() function reads the input file, counts the number of students and
stores all the identification numbers and test scores in two arrays.

2. The OutputReport() function prints the report. Notice that the program prints one
asterisk for students who made a B on the test, and prints two asterisks for students who made an
A on the test.

3. The AverageScore() function finds and returns the average test score in the array of
test scores; the function does not read any input or write any output.
Using Arrays 496

4. The HighScore() function finds and returns the highest test score in the array of test
scores; the function does not read any input or write any output.

5. The LetterGrade() function is passed two arguments: an individual test score (not an
array) and the group average. The function returns the student's letter grade according to this
scale:
A = 20 points or more above average
B = 10 or more but less than 20 points above average
C = less than 10 points above or below the average
D = 10 or more but less than 20 points below average
F = 20 points or more below average

6. The main() function simply calls the other functions; main() does not perform any
input, calculations, or output. Test your program using each of the provided data files:
Cs10-4a.txt, Cs10-4b.txt and Cs10-4c.txt. Write the grade report to a disk output file. Input the
file names from the keyboard at run-time.
Print both the screen output and the file output.

Average score = 75.00

Highest score = 100.00

ID SCORE GRADE

100000000 54.00 F
100000001 70.00 C
111111111 65.00 D
*200000000 89.00 B
200000001 55.00 F
222222222 72.00 C
300000000 66.00 C
**300000003 96.00 A
333333333 50.00 F
**400000000 95.00 A
400000001 76.00 C
*400000004 85.00 B
*444444444 94.00 B
500000000 84.00 C
500000001 80.00 C
500000005 59.00 D
555555555 74.00 C
600000000 61.00 D
600000001 56.00 D
**600000006 100.00 A
*666666666 94.00 B
Using Arrays 497

Problem Engr10-1 — Other Statistical Means

There are three other statistical means that are sometimes used. These include the geometric
mean, the harmonic mean and the root-mean-square (RMS) average. They are given by the
following formulas.

Write a program to calculate the average, mean, standard deviation from the mean,
geometric mean, harmonic mean, and RMS average. Use the resist.txt data file from program
Engr10a. Format the results similar to Engr10a.

Problem Engr10-2 — Random Distributions

C++ has a function that generates random numbers. However, the random number generator
required an initial seed value. For any given seed value, the sequence of numbers generated are
always the same. If a program desires truly random integer numbers each time the program
executes, the seed value must be different for each run of the program. Programmers often use
the computer's time of day clock to create the initial seed value because it would be very unlikely
to launch the program at precisely the same second on two successive runs. The random function
is called rand() and the seed function is srand(). Both functions’ prototypes are in <iostream>.
Calling the time() function with a parameter of 0 results in the current time of day being
returned; its prototype is in <ctime>. The seed value is normally given only once per run of the
program. The following sets the seed to the current time and then generates 10 random numbers.
#include <iostream>
#include <ctime>
using namespace std;
int main () {
srand ((unsigned) time (0)); // current time as seed
for (int j=0; j<10; j++)
cout << rand () << endl;
return 0;
}
Using Arrays 498

Game designers often require a random die roll. Engineering applications sometimes need
a set of random values within a specified range. To create a random die roll on a 6-sided die, use
(rand() % 6) + 1 to generate numbers between 1 and 6, for example.

Write a program that inputs a specific unsigned seed value, generates the first 1,000
random numbers between 0 and 10 based on that seed. The question to explore is “for a specific
seed, are these numbers truly random?” Create a frequency count array of 11 elements initialized
to 0. For each of the 1,000 random numbers, increment the corresponding frequency count as the
random number is found. For example, if the random number is 3, then add one to the frequency
count of subscript 3.

If the distribution were truly random, one might expect to see all frequency counts
approximately the same value, if the sample size is sufficiently large. Calculate the average and
standard deviation of the frequency counts. The theoretical average for random numbers in the
range (0, 10) is 5. The theoretical standard deviation is .289.

The program should output the seed used, the average, the standard deviation and then the
frequency counts for all 11 values. Caution, the standard deviation requires that the array be
sorted. Thus, before calling the sort function, copy the frequency array into a new temporary
array and use the temporary array to sort and to find the standard deviation. Keep the original
frequency array so that the original values may be printed.

Run the program using 5 different seed values. Does the seed value affect the results?

Problem Engr10-3 — Least Squares Curve Fitting

The resistance characteristics of a given thermistor can be found by placing it on an oven, setting
the oven at various known temperatures and measuring the resistance of the thermistor at each of
the known temperatures. This was done for a series of temperatures. The data, pairs of
temperature and resistance, are
100 25580
120 14430
140 8576
160 5328
180 3446
200 2308
220 1578
240 1132
260 826
280 616
300 469
Using Arrays 499

Assuming that temperature is the independent variable, write a program to find the best
linear fit. Print the results, slope and intercept, nicely formatted. Then, modify the Plot() function
from Chapter 10 to plot two sets of data. Use ‘*’ characters to represent the observed data and
use the ‘x’ character to represent the least squares best-fit curve. Using your best fit solution,
calculate the resistance at the same temperatures as the original data and pass these as the second
array pair to Plot().

Problem Engr10-4—Exponential Curves and Least Squares Best

Fitting Lines
In the previous problem Engr10-3, the best fitting straight line is not a good fit because the
thermistor resistance versus temperature is not linear. Rather, it is exponential in nature. Thus,
fitting an exponential curve to the observed data yields a better result. The general exponential
curve to use is given by

However, an examination of the data shows that the simple slope intercept method will
fail because the curve never reaches the x axis; it is asymptotic at around 100 ohms. In such
cases, one can employ the trick of converting to x versus log y space — that is, plot the graph
using x and log y. If the data were asymptotic to both axises, one could try a plot in log x versus
log y space. In other words, we plot x as before, but use log y for each y value.

Rewrite the least squares best-fit program to find the best fitting straight line in x versus
log y space. As you input the Y data points, take the log of them. When plotting the two graphs,
plot log (y) instead of y values for the vertical axis.
Strings 500

Chapter 11 — Strings

Section A: Basic Theory

Defining Character Strings
A character string in C++, such as “Hello World,” is stored in an array of characters with an extra
byte on the end marking the end of the string. This extra end of string marker is called the null
terminator and consists of a byte whose value is zero, that is, all the bits are zero. In C++, nearly
all strings ever used are null-terminated. However, the language provides for non-null terminated
strings as well. But, in C++, unlike C, these are seldom used.

Suppose for example you used the string literal value “Sam.” We know that if we had
written
cout << "Sam";
then the output stream displays
Sam
But how does the computer know how long the literal string is and where it ends? The answer is
that the literal “Sam” is a null-terminated string consisting of four bytes containing 'S', 'a', 'm',
and 0. This null-terminator can be represented by a numerical 0 or by the escape sequence \0.
Most all of the C++ functions that take a character string as an argument expect that string to be
null-terminated.

The null-terminator marks the end of the characters in the variable. For example, suppose
that a variable is defined to hold a person's name as follows
char name[21];
This definition is saying that the maximum number of characters that can be stored is twenty plus
one for the null-terminator. This maximum length is different from the number of characters
actually stored when a person's name is entered. For example, assume that the program has
inputted the name as follows
cin >> name;
Assume that the user has entered “Sam” from the keyboard. In this instance, only four of the
possible twenty-one are in use with the null terminator in the 4th character. Make sure you
understand the distinction between the maximum size of a string and the actual size in a specific
instance.
Strings 501

When defining a string variable, it makes good programming sense not to hard-code the
array bounds but to use a const int, just as is done with other kinds of arrays. Thus, the name
variable ought to have been coded like this
const int NAMELENGTH = 21;
int main () {
char name[NAMELENGTH];
If at a later date you decide that twenty character names are too short, it is a simple matter of
changing the constant value and recompiling.

When character string variables are defined, they can also be initialized. However, two
forms are possible. Assume that when the name variable is defined, it should be given the value
of “Sam.” Following the initialization syntax for any other kind of array, one could code
char name[NAMELENGTH] = {'S', 'a', 'm', '\0'};
Here each specific character is assigned its starting value; do not fail to include the null
terminator. However, the compiler allows a string to be initialized with another literal string as
follows
char name[NAMELENGTH] = "Sam";
Clearly this second form is much more convenient.

With all forms of arrays, when defining and initializing an array, it is permissible to omit
the array bounds and let the compiler determine how many elements the array must have based
on the number of initial values you provide. Thus, the following is valid.
char name[] = "Sam";
However, in general, this approach is lousy programming style and error prone. Why? In the
above case, the compiler allocates an array just large enough to hold the literal “Sam.” That is,
the name array is only four characters long. What would happen if later on one attempted to
input another name that needed more characters? Disaster. Always provide the array bounds
whenever possible.

Inputting Character Strings

Using the Extraction Operator

The extraction operator can be used to input character strings. The specific rules of string
extraction follow those for the other data types we have covered. It skips over whitespace to the
first non-whitespace character, inputs successive characters storing them into successive bytes in
the array until the extraction operator encounters whitespace or the end of file. Lastly, it stores
the null terminator. There are two aspects of this input operation that frequently make the use of
the extraction operator useless.

Notice that the extraction operator does not permit a blank to be in the string. Suppose
that you prompted the user to input their name and age and then used cin to input them as
Strings 502

follows
cin >> name >> age;
What results if the user enters the following data?
Sam Spade 25
The input stream goes into the bad or fail state. It inputs the characters “Sam” and stores them
along with the trailing null-terminator into the name field. It skips over the blank and attempts to
input the character S of Spade into the age integer and goes immediately into the bad state. If you
reflect upon all the different kinds of strings that you might encounter in the real world of
programming (names, product descriptions, addresses, cities), the vast majority may have
embedded blanks in them. This rules out the extraction operation as a method of inputting them.

The other part of the extraction operator rules is quite destructive, especially if you are
running on the Windows 95/98 platform. It inputs all characters until it finds whitespace or EOF.
Now suppose that the field name is defined to be an array of 21 characters. What happens if in
response to the prompt to enter a name, the user enters the following name.
Rumplestillskinchevskikov
The computer attempts to store 26 characters into an array that is only 21 characters long. Four
bytes of memory are now overlaid. What happens next is unpredictable. If another variable in
your program occupies that overlaid memory, its contents are trashed. If that memory is not even
part of your program, but is part of some other program, such as a Windows system dll, it is
overlaid; even wilder things can happen! Under Windows NT/2000, if you attempt to overlay
memory that is not part of your data segment, the program is aborted instead. This is one reason
for so many system crashes under Windows 95/98.

One way to get around the extraction operator's disadvantages is to use either the get() or
getline() function. The get() function can be used in one of two ways. Note: while I am using cin
in these examples, any ifstream instance can be used as well.
cin.get (string variable, sizeof (string variable));
cin.get (string variable, sizeof (string variable),
delimiter character);
These input all characters from the current position in the stream until either the maximum
number of characters including the null terminator has been read or EOF or the delimiter is
found. By default the delimiter is a new line code. The delimiter is not extracted but remains in
the input stream.
cin.getline (string variable, sizeof (string variable));
cin.getline (string variable, sizeof (string variable),
delimiter character);
This function works the same way except the delimiter is removed from the input stream but
never stored in the string variable. It also defaults to the new line code.
Strings 503

Method A — All Strings Have the Same Length

This is a common situation. In the input set of data or file, all character strings are the same
length, the maximum. Shorter strings have blanks added onto the end of the character series to
fill out the maximum length. Assume that a cost record input set of data contains the item
number, quantity, description and cost fields. The program defines the input fields as follows.
const int DescrLimit = 21;
long itemnumber;
long quantity;
char description[DescrLimit];
double cost;

The description field can hold up to twenty characters plus one for the null terminator.
The input set of data would appear as
12345 10 Pots and Pans 14.99
34567 101 Cups 5.99
45667 3 Silverware, Finished 10.42
Notice how the shorter strings are padded with blanks so that in all circumstances the
description field is 20 characters long.

The data is then input this way.

infile >> itemnumber >> quantity >> ws;
infile.get (description, sizeof (description));
infile >> cost;
Observe that the first line ends by skipping over whitespace to position the input stream to the
first character of the description field. sizeof() always returns the number of bytes the variable
occupies. In the case of the description field, it yields twenty-one. If one used sizeof(quantity),
it would return four bytes, since longs occupy four bytes. One could also use the constant integer
DescrLimit instead of the sizeof(); this subtle difference will be important shortly.

Many company input data files are set up in this manner. What is input and stored in the
description field when the second line of data above is input? The description contains “Cups
" – that is, the characters C-u-p-s followed by sixteen blanks and then the null terminator.

There is one drawback to this method. The blanks are stored. Shortly we will see how
character strings can be compared to see if two contain the same values. Clearly, if we compared
this description to the literal “Cups,” the two would not be equal. Can you spot why? The
inputted description contains sixteen blanks that the literal does not contain! Thus, if the trailing
blanks are going to present a problem to the processing logic of the program, they need to be
removed. On the other hand, if the description field is only going to be displayed, the presence of
the blanks is harmless.

With a few lines of coding, the blanks can be removed. The idea is to begin at the end of
the string and if that byte contains a blank, back up another byte until a byte that is non-blank is
Strings 504

found. Then place a null terminator in the last blank position. Since the length of all strings must
be twenty characters (after the get() function is done, the null terminator is in the twenty-first
position), the location of the last byte that contains real data must be subscript 19. The null
terminator must be at subscript 20. The following coding can be used to remove the blanks at the
end, if any.

int index = DescrLimit - 2; // or 19

while (index >= 0 && description[index] == ' ')
index--;
// here index = subscript of the first non-blank char
index++;
description[index] = 0; // insert a null-terminator
// over last blank
If the description contains all blanks or if the string contains a non-blank character in the 20th
position, this coding still works well.

The main problem to consider when inputting strings with the get() function is handling
the detection of the end of file properly. We are used to seeing coding such as
while (cin >> itemnumber >> quantity) {
But in this case, the input operation cannot be done with one chained series of extraction
operators. Rather, it is broken into three separate statements. Consider replacing the three lines of
coding with a new user helper function.
while (GetData (infile, itemnumber, quantity,
description, cost, DescrLimit)) {
The function would be
istream& GetData (istream& infile, long& itemnumber,
long& quantity, char description[],
double& cost, int descrLimit) {
infile >> itemnumber >> quantity >> ws;
if (!infile) return infile;
infile.get (description, descrLimit);
if (!infile) return infile;
infile >> cost;
if (!infile) return infile;
int index = descrLimit - 2;
while (index >= 0 && description[index] == ' ')
index--;
index++;
description[index] = 0;
return infile;
}
Vitally important is that the number of bytes to use in the get() function this time is not
sizeof(description). Why? Within the function, the description is the memory address of where
the first element of the array of characters is located. Memory addresses are always four bytes in
size on a 32-bit platform. Thus, had we used sizeof(description), then 4 bytes would have been
the limit!
Strings 505

Method A, where all strings are the same length, also applies to data files that have more
than one string in a line of data. Consider a customer data line, which contains the customer
number, name, address, city, state and zip code. Here three strings potentially contain blanks,
assuming the state is a two-digit abbreviation. Thus, Method A is commonly used.

Method B – String Contains Only the Needed Characters, But Is the Last
Field on a Line

In certain circumstances, the string data field is the last item on the input data line. If so, it can
contain just the number of characters it needs. Assume that the cost record data were reorganized
as shown (<CRLF> indicates the enter key).
12345 10 14.99 Pots and Pans<CRLF>
34567 101 5.99 Cups<CRLF>
45667 3 10.42 Silverware, Finished<CRLF>
This data can be input more easily as follows.
infile >> itemnumber >> quantity >> cost >> ws;
infile.get (description, sizeof (description));
Alternately, the getline() function could also be used. There are no excess blanks on the end of
the descriptions to be removed. It is simpler. However, its use is limited because many data entry
lines contain more than one string and it is often impossible to reorganize a company's data files
just to put the string at the end of the data entry lines.

Method B works well when prompting the user to enter a single string. Consider the
action of asking the user to enter a filename for the program to use for input. Note on the open
function call for input, we can use the ios::in flag and for output we use the ios::out flag.
char filename[_MAX_PATH];
cin.getline (filename, sizeof(filename));
ifstream infile;
infile.open (filename, ios::in);

When dealing with filenames, one common problem to face is just how many characters
long should the filename array actually be? The compiler provides a #define of _MAX_PATH
(in the header file <iostream>) that contains the platform specific maximum length a complete
path could be. For Windows 95, that number is 256 bytes.
Strings 506

Method C — All strings Are Delimited

The problem that we are facing is knowing where a string actually ends because a blank is not
usually a good delimiter. Sometimes quote marks are used to surround the string data. Here a "
mark begins and ends a string. Suppose that the input data appeared as follows.
12345 10 "Pots and Pans" 14.99
34567 101 "Cups" 5.99
45667 3 "Silverware, Finished" 10.42

When a string is delimited, the data can be input rather easily if we use the alternate form
of the get() function, supplying the delimiter ‘\"’.
char junk;
infile >> itemnumber >> quantity >> junk;
infile.get (description, sizeof (description), '\"');
infile >> junk >> cost;
Notice that we must input the beginning quote mark. The get() function leaves the delimiter in
the input stream, so we must extract it before continuing on with the next field, cost.

On the other hand, the getline() function removes the delimiter. Coding becomes simpler.
char junk;
infile >> itemnumber >> quantity >> junk;
infile.getline (description, DescrLimit, '\"');
infile >> cost;

Outputting Character Strings

Outputting strings presents a different set of problems, ones of spacing and alignment. In most all
cases, the insertion operator handles the output of strings quite well. In the most basic form one
might output a line of the cost record as follows
cout << setw (10) << itemnumber
<< setw (10) << quantity
<< description
<< setw (10) << cost << endl;
If the entire program output consisted of one line, the above is fine. Usually, the output consists
of many lines, columnarly aligned. If so, the above fails utterly.

With a string, the insertion operator outputs all of the characters up to the null terminator.
It does not output the null terminator. With strings of varying length, there is going to be an
unacceptable jagged right edge in the description column. On the other hand, if Method A was
used to input the strings and all strings are of the same length, all is well until the setw() function
is used to define the total field width. Suppose that the description field should be displayed
within a width of thirty columns. One might be tempted to code
cout << setw (10) << itemnumber
Strings 507

<< setw (10) << quantity

<< setw (30) << description
<< setw (10) << cost << endl;

The default field alignment of an ostream is right alignment. All of our numeric fields
display perfectly this way. But when right alignment is used on character strings, the results are
usually not acceptable as shown below
12345 10 Pots and Pans 14.99
34567 101 Cups 5.99
45667 3 Silverware, Finished 10.42

Left alignment must be used when displaying strings. Right alignment must be used when
displaying numerical data. The alignment is easily changed by using the setf() function.
cout << setw (10) << itemnumber
<< setw (10) << quantity;
cout.setf (ios::left, ios::adjustfield);
cout << setw (30) << description;
cout.setf (ios::right, ios::adjustfield);
cout << setw (10) << cost << endl;

In the call to setf(), the second parameter ios::adjustfield clears all the justification flags
— that is, turns them off. Then left justification is turned on. Once the string is output, the
second call to setf() turns right justification back on for the other numerical data. It is vital to use
the ios::adjustfield second parameter. The Microsoft implementation of the ostream contains
two flags, one for left and one for right justification. If the left justification flag is on, then left
justification occurs. Since there are two separate flags, when setting justification, failure to clear
all the flags can lead to the weird circumstance in which both left and right justification flags are
on. Now you have left-right justification (a joke) — from now on, the output is hopelessly
messed up justification-wise.

Alternatively, one can use the much more convenient manipulator functions: left and
right.
cout << setw (10) << itemnumber << setw (10) << quantity
<< left << setw (30) << description << right
<< setw (10) << cost << endl;

Finally, the insertion operator displays all characters in a string until it encounters the null
terminator. What happens if by accident a string is missing its null terminator? Simple, the
insertion operator displays all bytes until it finds a null terminator. I often refer to this action as a
“light show.” Yes, one sees the contents of the string appear, but “garbage” characters follow
that. If a line gets full, DOS line wraps and continues on the next line. If the screen fills, DOS
scrolls. All of this occurs at a blazing speed. Sit back and relax; don't panic if this happens to
you. It is harmless. Enjoy the show. It will stop eventually when it finds a byte with a zero in it.
Strings 508

Passing a String to a Function

When passing a string to a function, the prototype of the string is just like that of any other array.
Suppose that we have a PrintRecord() function whose purpose was to display one cost record.
The description string must be passed. The prototype of the PrintRecord() function is
void PrintRecord (const char description[],...
and the main() function could invoke it as
PrintRecord (description,...

Recall that the name of an array is always the memory address of the first element, or a
pointer. Sometimes you may see the prototype for a string using pointer notation instead of array
notation.
void PrintRecord (const char* description, ...
These are entirely equivalent notations when passing a string to a function.

Remember, if a function is not going to alter the caller’s character string, it should have
the const qualifier.

Working with Strings

Working with character string fields presents some new problems that we have not encountered
before. Suppose that we have the following fields defined and have inputted some data into them.
const int NameLen = 21;
char previousName[NameLen];
char currentName[NameLen];
Suppose that we needed to compare the two names to see if they were equal or not — that
is, they contain the same series of characters. Further, suppose that if they are not the same, we
needed to copy the current name into the previous name field. One might be tempted to code the
following.
if (previousName != currentName) {
previousName = currentName;

Coding the above cannot possibly work. Why? Remember that the name of an array is the
memory address where that array begins in memory. For the sake of illustration, assume that the
previousName array begins at memory address 5000 and that the currentName array begins at
memory location 8000. If you substitute these values for the variable array names in the above
coding as the compiler does, you end up with this
if (5000 != 8000) {
5000 = 8000;
In all cases, the test condition is always true, for 5000 is not 8000, ever. But look at the
assignment, it is ludicrous. Although the test condition compiles with no errors, the assignment
line generates an error message.
Strings 509

To our rescue comes the library of string functions. The prototypes of all of these string
functions are in the header file <string>.

Comparing Strings

Here is where the new changes Microsoft has made in .NET 2005 come to the forefront. Older
code now recompiled using .NET 2005 will produce a large number of warning message about
function calls now being depricated, that is obsolete. First, let’s examine the older versions and
then see why Microsoft has made unilateral, not yet in the C++ Standard, changes.

The Old Way: To compare two strings, use either strcmp() or stricmp(). strcmp() is a
case sensitive string compare function. stricmp() is a case insensitive string compare function.
Both functions return an integer indicating the result of the comparison operation. The prototype
of the string comparison function is this.
int strcmp (const char* string1, const char* string2);
It is showing that we pass it the two strings to be compared. However, the notation, const char*
also indicates that the string’s contents are constant. That is, the comparison function cannot alter
the contents of either string. If the parameters were just char* string1, then potentially the
contents of the string we passed could be altered in some way. The const char* notation is our
guarantee that the function cannot alter the contents of the string we pass. It is rather like making
the string “read-only.”

The integer return code indicates the result:

0 => the two strings are the same
positive => the first string is larger
negative => the first string is smaller

The New Way: To compare two strings, use either strcmp() or _stricmp(). strcmp() is a
case sensitive string compare function. _stricmp() is a case insensitive string compare function.
Both functions return an integer indicating the result of the comparison operation. The prototype
of the string comparison function is this.
int strcmp (const char* string1, const char* string2);
int _stricmp (const char* string1, const char* string2);
It is showing that we pass it the two strings to be compared. Both functions abort the program if
either of the two passed memory addresses is zero or NULL.

While the meaning of the result’s phrase, “the two strings are the same,” is obvious, the
other two results might not be so clear. Character data is stored in an encoding scheme, often
ASCII, American Standard Code for Information Interchange. In this scheme, the decimal
number 65 represents the letter ‘A’. The letter ‘B’ is a 66; ‘C’, a 67, and so on. If the first string
begins with the letter ‘A’ and the second string begins with the letter ‘B’, then the first string is
said to be smaller than the second string because the 65 is smaller than the 66. The comparison
Strings 510

function returns the value given by ‘A’ – ‘B’ or (65 – 66) or a negative number indicating that the
first string is smaller than the second string.

When comparing strings, one is more often testing for the equal or not equal situation.
Applications that involve sorting or merging two sets of strings would make use of the
smaller/larger possibilities. To fix up the previous example in which we wanted to find out if the
previousName was not equal to the currentName, we should code the following assuming that
case was important.
if (strcmp (previousName, currentName) != 0) {
If we wanted to ignore case sensitivity issues, then code this.
if (_stricmp (previousName, currentName) != 0) {

Copying Strings

The older function to copy a string is strcpy(). Its prototype is

char* strcpy (char* destination, const char* source);
It copies all characters including the null terminator of the source string, placing them in the
destination string. In the previous example where we wanted to copy the currentName into the
previousName field, we code
strcpy (previousName, currentName);

Of course, the destination string should have sufficient characters in its array to store all
the characters contained in the source string. If not, a memory overlay occurs. For example, if
one has defined the following two strings
char source[20] = "Hello World";
char dest[5];
If one copies the source string to the destination string, memory is overlain.
strcpy (dest, source);
Seven bytes of memory are clobbered in this case and contain a blank, the characters “World”
and the null terminator.

This clobbering of memory, the core overlay, or more politically correct, buffer overrun,
has taken its toll on not only Microsoft coding but many other applications. Hackers and virus
writers often take advantage of this inherently insecure function to overwrite memory with
malicious machine instructions. Hence, Microsoft has unilaterally decided to rewrite the standard
C Libraries to prevent such from occurring. As of this publication, Microsoft’s changes are not in
the ANSII C++ standard.

The new string copy function looks like this.

char* strcpy_s (char* destination, size_t destSize,
const char* source);
It copies all characters including the null terminator of the source string, placing them in the
destination string, subject to not exceeding the maximum number of bytes of the destination
Strings 511

string. In all cases, the destination string will be null terminated. However, if the source or
destination memory address is 0 or if the destination string is too small to hold the result, the
program is basically terminated at run time. In a later course, a program can prevent this
abnormal termination and do something about the problem.

In the previous example where we wanted to copy the currentName into the
previousName field, we now code
strcpy_s (previousName, sizeof (previousName), currentName);
This text will consistently use these new Microsoft changes. If you are using another
compiler, either use the samples provided in the 2002-3 samples folder or remove the sizeof
parameter along with the _s in the function names.

Getting the Actual Number of Characters Currently in a String

The next most frequently used string function is strlen(), which returns the number of bytes that
the string currently contains. Suppose that we had defined
char name[21] = "Sam";
If we code the following
int len = strlen (name); // returns 3 bytes
int size = sizeof (name); // returns 21 bytes
then the strlen(name) function would return 3. Notice that strlen() does NOT count the null
terminator.

The sizeof(name) gives the defined number of bytes that the variable contains or 21 in
this case. Notice the significant difference. Between these two operations.

Concatenating or Joining Two Strings into One Larger String

Again, there is a new version of this function in .NET 2005. The older function is the strcat()
function which appends one string onto the end of another string forming a concatenation of the
two strings. Suppose that we had defined
char drive[3] = "C:";
char path[_MAX_PATH] = "\\Programming\\Samples";
char name[_MAX_PATH] = "test.txt";
char fullfilename[_MAX_PATH];
In reality, when users install an application, they can place it on nearly any drive and nearly any
path. However, the application does know the filename and then has to join the pieces together.
The objective here is to join the filename components into a complete file specification so that
the fullfilename field can then be passed to the ifstream open() function. The sequence would
be
strcpy (fullfilename, drive); // copy the drive string
strcat (fullfilename, path); // append the path
Strings 512

strcat (fullfilename, "\\"); // append the \

strcat (fullfilename, name); // append filename
infile.open (fullfilename, ios::in); // open the file

The new version is strcat_s() which now takes the destination maximum number of bytes
as the second parameter before the source string. The above sequence using the newer functions
is this.
strcpy_s (fullfilename, _MAX_PATH, drive); // copy the drive
strcat_s (fullfilename, _MAX_PATH, path); // append the path
strcat_s (fullfilename, _MAX_PATH, "\\"); // append the \
strcat_s (fullfilename, _MAX_PATH, name); // append filename
infile.open (fullfilename, ios::in); // open the file

The String Functions

There are a number of other string functions that are available. The next table lists some of these
and their use. The prototypes of all of these are in <string>. The data type size_t is really an
unsigned integer.

Name: strlen
Meaning: string length function
Prototype: size_t strlen (const char* string);
Action done: returns the current length of the string. size_t is
another name for an unsigned int.
Example: char s1[10] = "Sam";
char s2[10] = "";
strlen (s1) yields 3
strlen (s2) yields 0

Name-old: strcmp and stricmp

Meaning: string compare, case sensitive and case insensitive
Prototype: int strcmp (const char* string1, const char* string2);
int stricmp (const char* string1, const char* string2);
Action done: strcmp does a case sensitive comparison of the two
strings, beginning with the first character of each
string. It returns 0 if all characters in both
strings are the same. It returns a negative value if
the different character in string1 is less than that
in string2. It returns a positive value if it is
larger.
Example: char s1[10] = "Bcd";
char s2[10] = "Bcd";
char s3[10] = "Abc";
char s4[10] = "Cde";
Strings 513

char s5[10] = "bcd";

strcmp (s1, s2) yields 0 - stings are equal
stricmp (s1, s5) yields 0 - strings are equal
strcmp (s1, s3) yields a + value — s1 > s3
strcmp (s1, s4) yields a – value — s1 < s4

Name-new: strcmp and _stricmp

Meaning: string compare, case sensitive and case insensitive
Prototype: int strcmp (const char* string1, const char* string2);
int _stricmp (const char* string1, const char* string2);
Action done: strcmp does a case sensitive comparison of the two
strings, beginning with the first character of each
string. It returns 0 if all characters in both
strings are the same. It returns a negative value if
the different character in string1 is less than that
in string2. It returns a positive value if it is
larger. Both functions abort the program if the
memory address is null or 0.
Example: char s1[10] = "Bcd";
char s2[10] = "Bcd";
char s3[10] = "Abc";
char s4[10] = "Cde";
char s5[10] = "bcd";
strcmp (s1, s2) yields 0 - stings are equal
_stricmp (s1, s5) yields 0 - strings are equal
strcmp (s1, s3) yields a + value — s1 > s3
strcmp (s1, s4) yields a – value — s1 < s4

Name-old: strcat
Meaning: string concatenation
Prototype: char* strcat (char* desString, const char* srcString);
Action done: The srcString is appended onto the end of the
desString. Returns the desString address
Example: char s1[20] = "Hello";
char s2[10] = " World";
strcat (s1, s2); yields "Hello World" in s1.

Name-new: strcat_s
Meaning: string concatenation
Prototype: strcat (char* desString, size_t maxDestSize,
const char* srcString);
Action done: The srcString is appended onto the end of the
desString. Aborts the program if dest is too small.
Example: char s1[20] = "Hello";
char s2[10] = " World";
strcat_s (s1, sizeof(s1), s2); yields "Hello World" in s1.
Strings 514

Name-old: strcpy
Meaning: string copy
Prototype: char* strcpy (char* desString, const char* srcString);
Action done: All bytes of the srcString are copied into the
destination string, including the null terminator.
The function returns the desString memory address.
Example: char s1[10];
char s2[10] = "Sam";
strcpy (s1, s2); When done, s1 now contains "Sam".

Name-new: strcpy_s
Meaning: string copy
Prototype: char* strcpy (char* desString, size_t maxDestSize,
const char* srcString);
Action done: All bytes of the srcString are copied into the
destination string, including the null terminator.
The function returns the desString memory address.
It aborts the program if destination is too small.
Example: char s1[10];
char s2[10] = "Sam";
strcpy_s (s1, sizeof (s1), s2);
When done, s1 now contains "Sam".

Name: strchr
Meaning: search string for first occurrence of the character
Prototype: char* strchr (const char* srcString, int findChar);
Action done: returns the memory address or char* of the first
occurrence of the findChar in the srcString. If
findChar is not in the srcString, it returns NULL
or 0.
Example: char s1[10] = "Burr";
char* found = strchr (s1, 'r');
returns the memory address of the first letter r
character, so that found[0] would give you that 'r'.

Name: strstr
Meaning: search string1 for the first occurrence of find string
Prototype: char* strstr (const char* string1,
const char* findThisString);
Action done: returns the memory address (char*) of the first
occurrence of findThisString in string1 or NULL (0)
if it is not present.
Example: char s1[10] = "abcabc";
char s2[10] = "abcdef";
char* firstOccurrence = strstr (s1, "abc");
Strings 515

It finds the first abc in s1 and firstOccurrence has the

same memory address as s1, so that s1[0] and
firstOccurrence[0] both contain the first letter 'a' of
the string
char* where = strstr (s2, "def");
Here where contains the memory address of the 'd' in the
s2

Name-old: strlwr
Meaning: string to lowercase
Prototype: char* strlwr (char* string);
Action done: All uppercase letters in the string are converted
to lowercase letters. All others are left untouched.
Example: char s1[10] = "Hello 123";
strlwr (s1);
Yields "hello 123" in s1 when done.

Name-new: strlwr_s
Meaning: string to lowercase
Prototype: char* strlwr (char* string, size_t maxSizeOfString);
Action done: All uppercase letters in the string are converted
to lowercase letters. All others are left untouched.
Example: char s1[10] = "Hello 123";
strlwr_s (s1, sizeof (s1));
Yields "hello 123" in s1 when done.

Name-old: strupr
Meaning: convert a string to uppercase
Prototype: char* strupr (char* string);
Action done: Any lowercase letters in the string are converted
to uppercase; all others are untouched.
Example: char s1[10] = "Hello 123";
strupr (s1);
When done, s1 contains "HELLO 123"

Name-new: strupr_s
Meaning: convert a string to uppercase
Prototype: char* strupr_s (char* string, size_t maxSizeOfString);
Action done: Any lowercase letters in the string are converted
to uppercase; all others are untouched.
Example: char s1[10] = "Hello 123";
strupr_s (s1, sizeof(s1));
When done, s1 contains "HELLO 123"
Strings 516

Name-old: strrev
Meaning: string reverse
Prototype: char* strrev (char* string);
Action done: Reverses the characters in a string.
Example: char s1[10] = "Hello";
strrev (s1);
When done, string contains "olleH"

Name-new: _strrev
Meaning: string reverse
Prototype: char* _strrev (char* string);
Action done: Reverses the characters in a string. It aborts the
program if the memory address passed is null or 0;
Example: char s1[10] = "Hello";
_strrev (s1);
When done, string contains "olleH"

How Could String Functions Be Implemented?

Next, let’s examine how the strcpy() and strcmp() functions could be implemented using array
notation. The strcpy() function must copy all bytes from the srcString into the desString,
including the null-terminator. It could be done as follows.
char* strcpy (char* desString, const char* srcString) {
int i = 0;
while (desString[i] = srcString[i])
i++;
return desString;
}
The while clause first copies a character from the source into the destination string. Then it
compares the character it just copied. If that character was not equal to zero, the body of the loop
is executed; i is incremented for the next character. If the character just copied was the null
terminator, the test condition is false and the loop ends.

Here is how the strcmp() function might be implemented using array notation.
int strcmp (const char* string1, const char* string2) {
int i=0;
while (string1[i] && string1[i] == string2[i])
i++;
return string1[i] - string2[i];
}
The first test in the while clause is checking to see if we are at the null terminator of string1. If
so, the loop ends. If not, then the corresponding characters of string1 and string2 are compared.
If those two characters are equal, the loop body is executed and i is incremented. If the two
Strings 517

characters are different, the loop also ends. To create the return integer, the current characters are
subtracted. If the two strings are indeed equal, then both bytes must be the null terminators of the
respective strings; the return value is then 0. Otherwise, the return value depends on the ASCII
numerical values of the corresponding characters.

Section B: A Computer Science Example

Cs11a — Character String Manipulation — Customer Names
One task frequently encountered when applications work with people’s names is that of
conversion from “firstname lastname” into “lastname, firstname.” This problem explores some
techniques to handle the conversion. There are many ways to accomplish splitting a name apart.
Since the use of pointer variables (variables that contain the memory addresses of things), have
not yet been discussed, the approach here is to use subscripting to accomplish it. Indeed, a
programmer does need to be able to manipulate the contents of a string as well as utilize the
higher lever string functions. This example illustrates low-level character manipulation within
strings as well as utilizing some commonly used string functions.

The problem is to take a customer name, such as “John Jones” and extract the first and
last names (“John” and “Jones”) and then to turn it into the alternate comma delimited form,
“Jones, John.” Alternatively, take the comma form and extract the first and last names. At first
glance, the approach to take seems simple enough. When extracting the first and last names from
the full name, look for a blank delimiter and take what’s to the left of it as the first name and
what’s to the right as the last name.

But what about names like “Mr. and Mrs. John J. Jones?” To find the last name portion,
begin on the right or at the end of the string and move through the string in reverse direction
looking for the first blank. That works fine until one encounters “John J. Jones, Jr.”. So we must
make a further qualification on that first blank, and that is, there must not be a comma
immediately in front of it. If there is, ignore that blank and keep moving toward the beginning of
the string.

When extracting the first and last names from the comma version (such as “Jones, John
J.”), we can look for the comma followed by a blank pair. However, what about this one, “Jones,
Jr., John J.?” Clearly we need to start at the end of the string and work toward the beginning of
the string in the search for the comma-blank pair.

Once we know the subscript of the blank or the comma-blank pair, how can the pieces be
copied into the first and last name strings? This is done by copying byte by byte from some
starting subscript through some ending subscript, appending a null terminator when finished. In
this problem, I have made a helper function, CopyPartialString() to do just that.
Strings 518

The function NameToParts() takes a full name and breaks it into first and last name
strings. The original passed full name string is not altered and is declared constant.

The function NameToCommaForm() takes the first and last names and converts them
into the comma-formatted name, last, first. Since the first and last names are not altered, those
parameters are also declared constant.

The function CommaFormToNames() converts a comma-formatted name into first and

last names. Since the comma-formatted name is not altered, it is also declared constant.

Let’s begin by examining the output of the program to see what is needed. Here is the test
run of Cs11a.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs11a Character String Manipulation - Sample Execution *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Original Name: |John J. Jones| *
* 2 First Name: |John J.| *
* 3 Last Name: |Jones| *
* 4 Comma Form: |Jones, John J.| *
* 5 First and Last from comma form test ok *
* 6 *
* 7 Original Name: |Betsy Smith| *
* 8 First Name: |Betsy| *
* 9 Last Name: |Smith| *
* 10 Comma Form: |Smith, Betsy| *
* 11 First and Last from comma form test ok *
* 12 *
* 13 Original Name: |Mr. and Mrs. R. J. Smith| *
* 14 First Name: |Mr. and Mrs. R. J.| *
* 15 Last Name: |Smith| *
* 16 Comma Form: |Smith, Mr. and Mrs. R. J.| *
* 17 First and Last from comma form test ok *
* 18 *
* 19 Original Name: |Prof. William Q. Jones| *
* 20 First Name: |Prof. William Q.| *
* 21 Last Name: |Jones| *
* 22 Comma Form: |Jones, Prof. William Q.| *
* 23 First and Last from comma form test ok *
* 24 *
* 25 Original Name: |J. J. Jones| *
* 26 First Name: |J. J.| *
* 27 Last Name: |Jones| *
* 28 Comma Form: |Jones, J. J.| *
* 29 First and Last from comma form test ok *
* 30 *
* 31 Original Name: |Jones| *
* 32 First Name: || *
* 33 Last Name: |Jones| *
* 34 Comma Form: |Jones| *
Strings 519

* 35 First and Last from comma form test ok *

* 36 *
* 37 Original Name: |Mr. John J. Jones, Jr.| *
* 38 First Name: |Mr. John J.| *
* 39 Last Name: |Jones, Jr.| *
* 40 Comma Form: |Jones, Jr., Mr. John J.| *
* 41 First and Last from comma form test ok *
* 42 *
* 43 Original Name: |Mr. John J. Jones, II| *
* 44 First Name: |Mr. John J.| *
* 45 Last Name: |Jones, II| *
* 46 Comma Form: |Jones, II, Mr. John J.| *
* 47 First and Last from comma form test ok *
* 48 *
* 49 Original Name: |Mr. John J. Jones, MD.| *
* 50 First Name: |Mr. John J.| *
* 51 Last Name: |Jones, MD.| *
* 52 Comma Form: |Jones, MD., Mr. John J.| *
* 53 First and Last from comma form test ok *
* 54 *
* 55 Original Name: |The Honorable Betsy Smith| *
* 56 First Name: |The Honorable Betsy| *
* 57 Last Name: |Smith| *
* 58 Comma Form: |Smith, The Honorable Betsy| *
* 59 First and Last from comma form test ok *
* 60 *
* 61 Original Name: |Betsy O'Neill| *
* 62 First Name: |Betsy| *
* 63 Last Name: |O'Neill| *
* 64 Comma Form: |O'Neill, Betsy| *
* 65 First and Last from comma form test ok *
* 66 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Program Cs11a is going to input a file of customer names. For each name, it first
converts that full name into first and last names. Then, it forms the comma-blank version of the
full name. And finally, it extracts the first and last names from the comma-blank string. If the
first and last names from the two approaches do not agree, an error is written. If they agree, an
“Okay” message is displayed. Since blanks are important to this problem and since a blank is
hard to spot, the | character is printed before and after each string, making errant blanks quite
visible. The Top-Down Design is shown in Figure 11.1.
Strings 520

Figure 11.1 Top-Down Design of Name Program

The main() function defines the arrays as shown in Figure 11.2. The sequence of
processing steps for main() is as follows.
open the input file, if it fails, display an error message and quit
while we have successfully inputted a line into fullName do the following
call NameToParts (fullName, firstName, lastName, MaxNameLen);
call NameToCommaForm (commaName, firstName, lastName);
call CommaFormToNames (commaName, firstFromCommaForm,
lastFromCommaForm);
output the results which include fullName, firstName, lastName and commaName
if firstName and firstFromCommaForm are the same as well as
lastName and lastFromCommaForm then output an Ok message
else display an error message and the firstFromCommaForm and
lastFromCommaForm
end the while clause
close the input file

Figure 11.2 Main Storage for main()

Strings 521

NameToParts() must break a full name into first and last names and is passed four
parameters: fullName, firstName, lastName, and limit. As we work out the sequence of coding,
let’s work with a specific example. Suppose that fullName contains the following, where the 0
indicates the null terminator. I have written the subscripts below the corresponding characters.
Mr. John J. Jones, MD.0
00000000001111111111222
01234567890123456789012
The strlen(fullName) yields 22 characters as the current length and the subscript for the last
character in the string is thus 21. So working from the end of the string, look for a blank that does
not have a comma immediately in front of it.
i = strlen (fullName) – 1;
while (i >= 0) do the following
does fullName[i] == ‘ ’? If so do the following
if there is a previous character — that is, is i>0 and
if that previous character is not a comma, fullName[i – 1] != ‘,’ then
// we have found the spot — so we need to break out of the loop
break; with i on the blank
end the if test
end the does clause
back up to the previous character, i--;
end the while clause
Now split out the two names. Notice we pass i+1, which is the first non-blank character in the
last name.
CopyParitalString (lastName, fullName, i+1, strlen (fullName));
CopyParitalString (firstName, fullName, 0, i);

The CopyParitalString() function’s purpose is to copy a series of characters in a source

string from some beginning subscript through an ending subscript and then insert a null
terminator. It is passed the dest string, the src string, startAt and endAt.
is startAt >= endAt meaning we are starting at the ending point, there is nothing
to copy, so just make the dest string a properly null-terminated string.
dest[0] = 0
and return
end is
To copy the characters, we need a subscript variable for each string, isrc and ides.
let isrc = startAt
let ides = 0;
Now copy all characters from startAt to endAt
while isrc < endAt do the following
dest[ides] = src[isrc];
increment both isrc and ides
end while
Finally, insert the null terminator
Strings 522

dest[ides] = 0;

The NameToCommaForm() function is comparatively simple. From two strings

containing the first and last names, make one combined new string of the form last name, first
name. However, in some cases, there might not be any first name. In that case, the result should
just be a copy of the last name string. NameToCommaForm() is passed three strings: the answer
string to fill up — commaName — and the two source strings — firstName and lastName. The
sequence is as follows.
strcpy (commaName, lastName);
if a first name exists — that is, does strlen (firstName) != 0, if so do
append a comma and a blank — strcat (commaName, “, ”)
append the first name — strcat (commaName, firstName)
end if

The CommaFormToNames() function must convert a single string with the form of
“last name, first name” into first and last name strings. It is passed commaName to convert and
the two strings to fill up - firstName and lastName. This time, we again begin at the end of the
string looking for the first comma followed by a blank. Consider these two cases.
Jones, Jr., Mr. John J.0
Jones, Prof. William Q.0
Clearly, we want to stop at the first “, ” occurrence to avoid problems with “Jr.”.
let len = strlen (commaName)
let commaAt = len – 2
while commaAt > 0 do the following
if the current character at commaAt is a ‘,’ and
the character at commaAt + 1 is a blank, then break out of the loop
back up commaAt
end the while clause
However, this could be compacted a bit more by using ! (not) logic in the while test condition.
while commaAt > 0 and
!(commaName[commaAt] == ‘,’ && commaName[commaAt+1] == ‘ ’)) {

When the loop ends, we must guard against no comma and blank found.
if (commaAt <= 0) then there is no comma so do the following
strcpy (lastName, commaName)
firstName[0] = 0
and return
end the if
Finally, at this point, we have found the “, ” portion; copy the two portions as follows.
CopyParitalString (lastName, commaName, 0, commaAt)
CopyParitalString (firstName, commaName, commaAt+2, len)

As you study the coding, draw some pictures of some test data and trace what is occurring
Strings 523

if you have any doubts about what is going on. Here is the complete program.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs11a Character String Manipulation *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs11a Character String Manipulation - Customer Names */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 #include <string> *
* 11 using namespace std; *
* 12 *
* 13 const int MaxNameLen = 51; // the maximum length of names *
* 14 *
* 15 void NameToParts (const char fullName[],// converts a full name *
* 16 char firstName[], // to first & last names *
* 17 char lastName[], *
* 18 int limit); *
* 19 *
* 20 void NameToCommaForm (char commaName[], // converts a first and *
* 21 const char firstName[], // last name into a *
* 22 const char lastName[]); // full name string *
* 23 *
* 24 void CommaFormToNames (const char commaName[],// converts a comma*
* 25 char firstFromCommaForm[], // form of name into *
* 26 char lastFromCommaForm[]); // first & last names*
* 27 *
* 28 void CopyParitalString (char dest[], // copies a part of the src *
* 29 const char src[], // string into the dest *
* 30 int startAt, // beginning at startAt and *
* 31 int endAt); // ending at endAt *
* 32 *
* 33 int main () { *
* 34 char fullName[MaxNameLen]; // original full name as input*
* 35 char firstName[MaxNameLen]; // first name from full name *
* 36 char lastName[MaxNameLen]; // last name from full name *
* 37 char commaName[MaxNameLen]; // full name in comma form *
* 38 char firstFromCommaForm[MaxNameLen];// first name from commaform*
* 39 char lastFromCommaForm[MaxNameLen]; // last name from comma form*
* 40 *
* 41 ifstream infile ("Cs11a-Names.txt"); *
* 42 if (!infile) { *
* 43 cerr << "Error: cannot find the names file\n"; *
* 44 return 1; *
* 45 } *
* 46 ofstream out ("results.txt"); *
* 47 *
* 48 while (infile.getline (fullName, sizeof (fullName))) { *
Strings 524

* 49 // break full name inputted into first and last names *

* 50 NameToParts (fullName, firstName, lastName, MaxNameLen); *
* 51 *
* 52 // turn first and last names into a comma form of full name *
* 53 NameToCommaForm (commaName, firstName, lastName); *
* 54 *
* 55 // break comma form of full name into first and last names *
* 56 CommaFormToNames (commaName, firstFromCommaForm, *
* 57 lastFromCommaForm); *
* 58 *
* 59 // output results *
* 60 out << "Original Name: |" << fullName << '|' << endl; *
* 61 out << " First Name: |" << firstName << '|' << endl; *
* 62 out << " Last Name: |" << lastName << '|' << endl; *
* 63 out << " Comma Form: |" << commaName << '|' << endl; *
* 64 *
* 65 // test that first and last names agree from both forms *
* 66 // of extraction *
* 67 if (strcmp (firstName, firstFromCommaForm) == 0 && *
* 68 strcmp (lastName, lastFromCommaForm) == 0) *
* 69 out << " First and Last from comma form test ok" << endl; *
* 70 else { *
* 71 out << " Error from comma form - does not match\n"; *
* 72 out << " First Name: |" << firstFromCommaForm << '|' <<endl;*
* 73 out << " Last Name: |" << lastFromCommaForm << '|' <<endl;*
* 74 } *
* 75 out << endl; *
* 76 } *
* 77 infile.close (); *
* 78 out.close (); *
* 79 return 0; *
* 80 } *
* 81 *
* 82 /***************************************************************/*
* 83 /* */*
* 84 /* CopyParitalString: copies src from startAt through endAt */*
* 85 /* */*
* 86 /***************************************************************/*
* 87 *
* 88 void CopyParitalString (char dest[], const char src[], *
* 89 int startAt, int endAt) { *
* 90 if (startAt >= endAt) { // avoid starting after ending *
* 91 dest[0] = 0; // just set dest string to a null string*
* 92 return; *
* 93 } *
* 94 *
* 95 int isrc = startAt; *
* 96 int ides = 0; *
* 97 // copy all needed chars from startAt to endAt *
* 98 for (; isrc<endAt; isrc++, ides++) { *
* 99 dest[ides] = src[isrc]; *
* 100 } *
Strings 525

* 101 dest[ides] = 0; // insert null terminator *

* 102 } *
* 103 *
* 104 /***************************************************************/*
* 105 /* */*
* 106 /* NameToParts: break a full name into first and last name */*
* 107 /* */*
* 108 /***************************************************************/*
* 109 *
* 110 void NameToParts (const char fullName[], char firstName[], *
* 111 char lastName[], int limit) { *
* 112 // working from the end of the string, look for blank separator *
* 113 // that does not have a , immediately in front of it *
* 114 int i = (int) strlen (fullName) - 1; *
* 115 while (i >= 0) { *
* 116 if (fullName[i] == ' ') { // found a blank and *
* 117 if (i>0 && fullName[i-1] != ',') { // earlier char is not a, *
* 118 break; // end with i on the blank *
* 119 } *
* 120 } *
* 121 i--; *
* 122 } *
* 123 CopyParitalString (lastName,fullName i+1,(int)strlen(fullName));*
* 124 CopyParitalString (firstName, fullName, 0, i); *
* 125 } *
* 126 *
* 127 /***************************************************************/*
* 128 /* */*
* 129 /* NameToCommaForm: from first & last names, make last, first */*
* 130 /* */*
* 131 /***************************************************************/*
* 132 *
* 133 void NameToCommaForm (char commaName[], const char firstName[], *
* 134 const char lastName[]) { *
* 135 strcpy_s (commaName, MaxNameLen, lastName); *
* 136 if (strlen (firstName)) { // if a first name exists, *
* 137 strcat_s (commaName, MaxNameLen, ", "); // add a , and blank *
* 138 strcat_s (commaName, MaxNameLen, firstName); // add first name *
* 139 } *
* 140 } *
* 141 *
* 142 /***************************************************************/*
* 143 /* */*
* 144 /* CommaFormToNames: convert a last, first name to first & last*/*
* 145 /* */*
* 146 /***************************************************************/*
* 147 *
* 148 void CommaFormToNames (const char commaName[], char firstName[], *
* 149 char lastName[]) { *
* 150 // begin at the end and look for a ,blank *
* 151 int len = (int) strlen (commaName); *
* 152 int commaAt = len - 2; *
Strings 526

* 153 while (commaAt > 0 && *

* 154 !(commaName[commaAt] == ',' && commaName[commaAt+1] == ' ')) {*
* 155 commaAt--; *
* 156 } *
* 157 if (commaAt <= 0) { // here there is no comma so *
* 158 strcpy_s (lastName, MaxNameLen, commaName); *
* 159 firstName[0] = 0; // set first name to null string *
* 160 return; *
* 161 } *
* 162 CopyParitalString (lastName, commaName, 0, commaAt); *
* 163 CopyParitalString (firstName, commaName, commaAt+2, len); *
* 164 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Section C: An Engineering Example

Engineering problems primarily make use of strings as labels or identifiers associated with a set
of numerical values.

Engr11a — Weather Statistics Revisited

On a daily basis, weather statistics for cities scattered around the state are collected, summarized
and forwarded to our center for processing. Our company maintains an Internet web page that
lists the unusual weather occurrences within the last 24-hour period. Write a program that inputs
the daily weather file and displays those cities with unusual weather in a nicely formatted report.

An input line consists of the city surrounded by double quote marks, such as “Peoria.”
Next, come the high and low temperatures, the rainfall amount, the snowfall amount, and wind
speed. Unusual weather is defined to be a high temperature above 95, a low temperature below 0,
a rainfall amount in excess of two inches, snowfall accumulations in excess of six inches or a
wind speed greater than 45 mph.

Since each day’s data is stored in a different file, the program first should prompt the user
to enter the filename to be used for the input. Also prompt the user for the output file to which
the report is to be written.

An output line might appear as

City Hi Low Rain Snow Winds

Peoria 85 55 0 0 55*
Washington 99* 75 0 0 10
A * character is placed after the weather statistic that is unusual.
Strings 527

Since this problem is quite basic, I have not included the coding sketch. By now, the logic
should be obvious. Here are the program listing and the sample output. Make sure you examine
the instructions that process the new string variables.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Listing for Program Engr11a - Unusual Weather Statistics *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr11a: Unusual Weather Statistics report */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 #include <string> *
* 11 using namespace std; *
* 12 *
* 13 const int MaxCityLen = 21; // city name length is 20 chars *
* 14 *
* 15 int main () { *
* 16 *
* 17 char infilename[_MAX_PATH]; *
* 18 char reportname[_MAX_PATH]; *
* 19 cout << "Enter the filename with today's weather data\n"; *
* 20 cin.getline (infilename, sizeof (infilename)); *
* 21 cout << "\nEnter the report filename\n"; *
* 22 cin.getline (reportname, sizeof(reportname)); *
* 23 *
* 24 ifstream infile; *
* 25 infile.open (infilename); *
* 26 if (!infile) { *
* 27 cerr << "Error: cannot open file: " << infilename << endl; *
* 28 return 1; *
* 29 } *
* 30 *
* 31 ofstream outfile; *
* 32 outfile.open (reportname, ios::out); *
* 33 if (!outfile) { *
* 34 cerr << "Error: cannot open file: " << reportname << endl; *
* 35 return 1; *
* 36 } *
* 37 // setup floating point output format *
* 38 outfile << fixed << setprecision (1); *
* 41 *
* 42 outfile << "Unusual Weather Report\n\n"; *
* 43 outfile<<"City High Low Rain Snow"*
* 44 " Wind\n"; *
* 45 outfile<<" Fall Fall"*
* 46 " Speed\n\n"; *
* 47 *
Strings 528

* 48 char city [MaxCityLen]; // string to hold city name *

* 49 float high; // high temperature of the day - F *
* 50 float low; // low temperature of the day - F *
* 51 float rainfall; // rainfall in inches *
* 52 float snowfall; // snowfall in inches *
* 53 float windspeed; // wind speed in mph *
* 54 *
* 55 char junk; // to hold the " around city names *
* 56 int line = 0; // line count for error processing *
* 57 *
* 58 while (infile >> junk) { // input the leading " of city *
* 59 infile.get (city, sizeof (city), '\"'); *
* 60 infile.get (junk); *
* 61 infile >> high >> low >> rainfall >> snowfall >> windspeed; *
* 62 // abort if there is incomplete or bad data *
* 63 if (!infile) { *
* 64 cerr << "Error: incomplete city data on line " << line <<endl;*
* 65 infile.close (); *
* 66 outfile.close (); *
* 67 return 2; *
* 68 } *
* 69 if (high > 95 || low < 0 || rainfall > 2 || snowfall > 6 || *
* 70 windspeed > 45) { *
* 71 // unusual weather - display this city data *
* 72 outfile << left << setw (22) << city << right *
* << setw (7) << high; *
* 76 if (high > 95) *
* 77 outfile << '*'; *
* 78 else *
* 79 outfile << ' '; *
* 80 outfile << setw (7) << low; *
* 81 if (low < 0) *
* 82 outfile << '*'; *
* 83 else *
* 84 outfile << ' '; *
* 85 outfile << setw (7) << rainfall; *
* 86 if (rainfall > 2) *
* 87 outfile << '*'; *
* 88 else *
* 89 outfile << ' '; *
* 90 outfile << setw (7) << snowfall; *
* 91 if (snowfall > 6) *
* 92 outfile << '*'; *
* 93 else *
* 94 outfile << ' '; *
* 95 outfile << setw (7) << windspeed; *
* 96 if (windspeed > 45) *
* 97 outfile << '*'; *
* 98 else *
* 99 outfile << ' '; *
* 100 outfile << endl; *
* 101 } *
Strings 529

* 102 } *
* 103 infile.close (); *
* 104 outfile.close (); *
* 105 return 0; *
* 106 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr11a - Unusual Weather Report Output *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Unusual Weather Report *
* 2 *
* 3 City High Low Rain Snow Wind *
* 4 Fall Fall Speed *
* 5 *
* 6 Washington 99.0* 70.0 0.0 0.0 20.0 *
* 7 Morton 85.0 65.0 5.0* 0.0 40.0 *
* 8 Chicago 32.0 -5.0* 0.0 8.0* 25.0 *
* 9 Joliet 88.0 70.0 2.0 0.0 60.0* *
* 10 Springfield 99.0* 75.0 3.0* 0.0 55.0* *
* 11 New Salem 0.0 -3.0* 0.0 9.0* 55.0* *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

New Syntax Summary

A string is an array of char, so it is defined the same way as any other array.
const int MAX = 10;
char string [MAX];

Inputting a String:
Extraction: cin >> string;
Blanks end the extraction; thus the data cannot contain any imbedded blanks. Further, if more
than 9 characters are entered, memory is over-written or the program is aborted, depending upon
the operating system platform and what is being clobbered. Extraction of a string can only be
done in totally controlled situations.

All strings are the same length, padded with blanks to the max length:
Typically, any leading whitespace must be skipped so that the current position in the input stream
is the first character of the string to be input.
cin.get (string, sizeof(string));
cin.getline (string, sizeof(string));

istream& get (char* string, size_t maxlength,

char delimeterCharacter);
istream& getline (char* string, size_t maxlength,
char delimeterCharacter);
both functions input and store successive characters until
Strings 530

a. eof is reached
b. the maximum number of characters minus one for the null is input
c. the delimiter code is found
By default, if not coded, the delimiter character is a new line code, ‘\n.’ The only difference
between the two functions is if the delimiter code is found, getline removes it while get does not.

The string is the last item on an input line:

Typically, any leading whitespace must be skipped so that the current position in the input stream
is the first character of the string to be input.
cin.get (string, sizeof(string));
cin.getline (string, sizeof(string));
Here each string is only as long as it needs to be; they are not padded with blanks.

The string ends with a delimiter code:

Typical delimiter codes are a " and a , (comma). Again, any leading whitespace must be skipped
so that the current position in the input stream is the first character of the string to be input and
the leading " must be inputted first.
cin.get (string, sizeof(string), '\"');
cin.getline (string, sizeof(string), '\"');
Here each string is only as long as it needs to be. If get is used, remember to next extract the
trailing delimiter byte. If a comma ended the string, then use
cin.get (string, sizeof(string), ',');
cin.getline (string, sizeof(string), ',');

Output of a String:
Strings are left justified not the default right justification. Hence, use the left and right
manipulator functions.
cout << left << setw (sizeof(string)+2) << string << right <<...

To work with strings, use the built-in string functions. Be alert for the version of the compiler
you are using. .NET2005 changed the string functions significantly.
Strings 531

Design Exercises
1. Design a Grade Book Program

The Grade Book Program inputs a set of students grades for a semester. First, design the layout
of the data file to be used for input and then design the program to produce the Grade Report
shown below.

The data consists of a student id number which is their social security number, their name
which can be up to 20 characters long, the course name which can be up to 10 characters in
length, the course number and finally the letter grade earned. Design how the input lines must be
entered. Include in what order they are entered; pay particular attention to specifically how the
student names are going to be entered on your lines.

The Grade Report produced by the program that is to input your data file appears as
follows.
Student Grade Report

Student Student ----Course-----

Id Name Name Number Grade
111111111 Sam J. Jones Cmpsc 125 A
...

2. Design the Merge Conference Roster Program

Two sections of a conference course have been merged into one larger section. Each original
section has a file of the attendee names. You are to write a program that merges the two into one
new file. Each original file contains, in alphabetical order, the attendee names which can be up to
30 characters long, one name per line. The new file this program creates must also be in
alphabetical order.
Strings 532

Stop! Do These Exercises Before Programming

1. A programmer needs to input the day of the week as a character string. The following coding
failed to run properly. Why? What must be done to fix it up?
char dayName[9];
cin >> dayName;

2. A program needs to input the chemical compound names of two substances and then compare
to see if the names are the same. The following was coded and compiles without errors but when
run always produces the wrong results. Why? How can it be fixed?
char compound1[40];
char compound2[40];
infile1.get (compound1, sizeof (compound1));
infile2.get (compound2, sizeof (compound2));
if (compound1 == compound2) {
cout << "These compounds match\n";
else
cout << "These compounds do not match\n";

3. The programmer inputted a compound name and its cost and then wanted to check to see if it
was equal to “Sodium Chloride.” The following coding compiles with no errors but when it runs,
it fails to find Sodium Chloride when that is input. The input line is
Sodium Chloride 4.99
What is wrong and how can it be fixed?
char compound[20];
double cost;
cin.get (compound, sizeof (compound));
cin >> cost;
if (stricmp (compound, "Sodium Chloride") == 0) {
cout << "Found\n";
}

4. The input file consists of a long student id number followed by a blank and then the student’s
name. The following coding does not input the data properly. Why? What specifically is input
when the user enters a line like this?
1234567 Sam Spade<cr>
How can it be fixed so that it correctly inputs the data?
long id;
char name[20];
while (cin >> id) {
cin.get (name, sizeof (name));
...
}
Strings 533

5. A file of student names and their grades is to be input. The programmer wrote a
GetNextStudent() function. It does not work. How can it be fixed so that it does work properly?
char name[20];
char grade;
while (GetNextStudent (infile, name, grade, 20)) {
...
istream GetNextStudent (istream infile, char name[],
char grade, int maxLen) {
infile.get (name, sizeof (name));
infile.get (grade);
return infile;
}

6. The proposed Acme Data Records consist of the following.

12345 Pots and Pans 42 10.99
23455 Coffee #10 can 18 5.99
32453 Peanuts 20 1.25
The first entry is the item number, the second is the product description, the third is the quantity
on hand, and the fourth is the unit cost. Assume that no description can exceed 20 characters. The
programmer wrote the following code to input the data.
int main () {
long id;
char description[21];
int quantity;
double cost;
ifstream infile ("master.txt", ios::in | ios::nocreate);
while (infile >> id >> description >> quantity >> cost) {
...
However, it did not run at all right. What is wrong with it? Is it possible to fix the program so
that it would read in that data file? What would you recommend?
Strings 534

Programming Problems
Problem Cs11-1 — Life Insurance Problem
Acme Life Insurance has asked you to write a program to produce their Customer’s Premium
Paid Report. The report lists the person’s name, age and yearly premium paid. Yearly premiums
are based upon the age when the person first became a customer.

The table of rates is stored in the file Cs11-1-rates.txt on disk. The file contains the age
and the corresponding premium on a line. Since these rates are subject to change, your program
should read these values from the file. In other words, do not hard code them in the program.
Currently, the data appears as follows (column headings have been added by for clarity).
Age Premium
Limit Dollars
25 277.00
35 287.50
45 307.75
55 327.25
65 357.00
70 455.00
The ages listed are the upper limits for the corresponding premium. In other words, if a
person took out a policy at any age up to and including 25, the premium would be $277.00. If
they were 26 through 35, then their premium would be $287.50. If they were above 70, use the
age 70 rate of $455.00.

Your program should begin by inputting the two parallel arrays, age and premium.
Allow for a maximum of 20 in each array. Load these arrays from a function called
LoadArrays() that is passed the two arrays and the limit of 20. It returns the number of elements
in the parallel arrays.

After calling the LoadArrays(), the main() function, inputs the customers’ data from the
Cs11-1-policy.txt file. Each line in this file contains the policy number, name and age fields. The
policy number should be a long and the name can be up to 20 characters long. The customer
names contain the last name only with no imbedded blanks. For each customer, print out their
name, their age and their premium. The report should have an appropriate title and column
headings.
Strings 535

Problem Cs11-2 — Acme Personnel Report

Write a program to produce the Acme Personnel Report from the Cs11-2-personnel.txt file. In
the file are the following fields in this order: employee name (20 characters maximum), integer
years employed, the department (15 characters maximum) and the year-to-date pay. The report
should look like this.
Acme Personnel Report

Employee Years Department Year to

Name Emp. Date Pay

xxxxxxxxxxxxxxxxxxxx 99 xxxxxxxxxxxxxxx $99999.99

xxxxxxxxxxxxxxxxxxxx 99 xxxxxxxxxxxxxxx $99999.99
The employee name and the department should be left aligned while the numeric fields should be
right aligned.

Problem Cs11-3 — Palindrome Analysis

A palindrome is a string that is the same whether read forward or backwards. For example,
“level” and “Nod Don” and “123454321” are all palindromes. For this problem, case is not
important. Write a function IsPalindrome() that takes a constant string as its only argument and
returns a bool, true if the word is a palindrome or false if it is not.

Then write a main() function that inputs file Cs11-3-words.txt. A line in this file cannot
exceed 80 characters. For each line input, print out a single line as follows
Yes--Nod Don
No---Nod Jim

Problem Cs11-4 — Merging Customer Files

Write a Merge Files program to merge two separate customer data into one file. Each file
contains the following fields: the customer’s number (up to 7 digits), the customer’s last name
(20 characters maximum), the customer’s first name (15 characters maximum), the address (20
characters maximum), the city (15 characters maximum), the state code (2 characters) and the zip
code (5 digits).

The resulting file should be in order by customer last names (a through z). If there are two
identical last names, then check the first names to decide which to insert into the new master file
first. Names should be case insensitive.
Strings 536

Normally, the only output of the merge program is the new master file called
newMaster.txt. However, for debugging purposes, also echo print to the screen the customer last
and first names as they are written to the new master file.

The two input files are called Cs11-4-mast1.txt and Cs11-4-mast2.txt.

Problem Engr11-1—Liquids and Gases in Coexistence

(Chemical Engineering)
The chemical and physical interactions between gases and liquids are commonly encountered in
chemical engineering. For a specific substance, the mathematical description of the transition
from gas to liquid is vital. The basic ideal gas equation for one mole of gas is
P = RT / V
where P = pressure in N/m2
V = volume of one mole in m3
T = temperature in degrees K
R = ideal gas constant of 8.314 J/mol-K

This ideal gas equation assumes low pressures and high temperatures such that the liquid
state is not present at all. However, this assumption often is not a valid one; many situations exist
where there is a combination of a substance in both its gaseous and liquid state present. This
situation is called an imperfect gas. Empirical formulas have been discovered that model this
behavior. One of these is Van der Waal's equation of state for an imperfect gas. If the formula is
simplified, it is

where p, v and t are scaled versions of the pressure, volume and temperature. The scaling is done
by dividing the measurement by a known, published critical value of that measurement. These
scaled equations are
p = P/Pc v = V/Vc t = T/Tc
These critical measurements correspond to that point where equal masses of the gas and liquid
phase have the same density. The critical values are tabulated for many substances. See for
example the Handbook of Chemistry and Physics — “Critical Constants for Gases” section.

Since there are actually three variables, v, p and t, the objective for this problem is to see
how this equation behaves at that boundary where gas is turning into a liquid. To do so, plot p
versus v versus t. An easy way that this can be accomplished is to choose a specific t value and
Strings 537

calculate a set of p versus v values. Then change t and make another set of p versus v values. All
told, there are to be three sets of p versus v values.

The three t values to use are 1.1, 1.0, and 0.9. For all three cases, the v values range from
0.4 through 3.0; divide this range into 100 uniformly spaced intervals. Then for each of the 100 v
values, calculate the corresponding p value. This means that you should define a v array that
holds 100 elements. Define three p arrays, one for each of the three t values, each p array to hold
100 elements. One of the p arrays represents the t = 1.1 results; another, the t = 1.0 results; the
third, the t = 0.9 results. Create one for loop that calculates all of these values. It is most
convenient to define also a function p (v, t) to handle the actual calculation of one specific
pressure at a specific volume and temperature.

Since these results are scaled values, they can then be applied to any specific substance.
Prepare an input data file for the substances listed below. Enter the four fields in this order,
substance, Tc, Pc, Vc. Your program should input each of these lines. For each line, in other
words each substance, the four arrays are printed in a columnar format, with the scaled t, v, p
values converted into T, V and P. In the table below, Tc is in degrees Kelvin; Pc is in
atmospheres; Vc is in cubic meters per mole.
Substance Tc Pc Vc
Water 647.56 217.72 0.00000721
Nitrogen 126.06 33.5 0.00000436
Carbon dioxide 304.26 73.0 0.0000202
The report for a specific substance should appear similar to the following
Substance: Carbon dioxide

Critical Volume Critical Pressures for 3 temps

cubic meters/mole T = 334.69 T = 304.26 T = 273.83

0.00000808 1551.24 1843.23 1259.24

...

If you have access to a plotter, for each substance, plot all three sets of p versus v curves on the
same graph.

Problem Engr11-2 — Chemical Formula

Each line of the E11-2-formula.txt file contains the chemical formula for a compound. A blank
separates the formula from the compound name. For example, one line could be
NaClO3 Sodium Chlorate. In the formula, there can be no blanks; allow for a maximum of 40
characters in the formula and another 40 in the compound name. Further, in the formula, case is
significant. The atom identification is one or two characters long, the first of which must be
uppercase and the second, if any, must be lowercase. That is, the atom is identified by an
uppercase letter. Any trailing numbers represent the number of those atoms at that point in the
Strings 538

formula. In the above example, there is one Na (Sodium), one Cl (Chlorine) and three O
(Oxygen) atoms in the compound.

For each compound, print a line detailing its component atoms such as this.
Sodium Chlorate
1 Na
1 Cl
3 O
Sum all like atoms into a single total. For example, if we had Methanol — CH3OH, the totals
would be
1 C
4 H
1 O
Multidimensional Arrays 539

Chapter 12 — Multidimensional Arrays

Section A: Basic Theory

Introduction
Multidimensional arrays, that is, arrays with two or more dimensions, are extremely valuable in
many applications. A two-dimensional array can be thought of as having a number of rows each
with the same number of columns.

One common application of two-dimensional arrays is a spreadsheet. In a monthly budget

spreadsheet, for example, the rows represent the income and expenses while the columns
represent the monthly expenses. Consider the following budget.
item June July August
income 1500.00 1550.00 1500.00
rent 500.00 500.00 500.00
utilities 200.00 200.00 200.00
phone 40.00 40.00 40.00
movies 20.00 30.00 25.00

All of the above numerical values are doubles. While one could create five single-
dimensioned arrays, each containing three elements to hold the sets of monthly values, a single
two-dimensional array of five rows each with three columns greatly simplifies the programming
logic.

Defining Multidimensional Arrays

The above budget two-dimensional array is defined as
double budget[5][3];
In general, the syntax is
datatype name[limit1][limit2][limit3]…[limitn];
The number of dimensions is unlimited; however, for practical purposes, the amount of memory
available for a program to use on a specific platform becomes the limiting factor. How much
memory does the above budget array occupy? Assuming that a double occupies 8 bytes, then
budget takes 5 x 3 x 8 bytes, or 120 bytes.
Multidimensional Arrays 540

When defining multidimensional arrays, each array bound or limit should be either a
#define value or a const int. These limit values are likely to be used throughout the program. If a
symbolic limit is used, it is easier to later modify the limits to allow for more data. The above
budget array can be defined as follows.
const int NumberItems = 5;
const int NumberMonths = 3;
…
double budget[NumberItems][NumberMonths];

Consider another example. Suppose that we needed to accumulate the total sales from
various cash registers located in three different stores and that each store had four departments
each. We could define three separate total arrays, one for each store; each array would have four
elements, one for each department. However, defining one two-dimensional array of three stores
each with four departments greatly simplifies programming. The totals array could be defined as
follows.
#define STORES 3
#define DEPTS 4
…
double totals[STORES][DEPTS];

Suppose further that, within each department, there are always two cash registers. Now
the array would contain a third dimension.
#define REGS 2
…
double regtotals[STORES][DEPTS][REGS];

How are the individual elements within a multidimensional array accessed? By providing
all the needed subscripts. Remember that all subscripts begin with element 0. The following are
valid.
totals[0][1] = 5.; // store 0, dept 1
totals[1][3] = 10.; // store 1, dept 3
totals[0][0] = 0; // the first element in the array
x = totals[2][3]; // the last element in the array
regtotals[1][2][0] = 42; // store 1, dept 2, reg 0
The following are invalid.
totals[0,1] = 5; // each subscript must be within []
totals[1] = 1; // this only specifies row 1 –
// which has 4 columns
totals = 0; // unfortunately not allowed either

Normally, the subscripts are variables and not constants. The subscripts may also be
integer expressions. The following are valid.
totals[i][j]++; // increments this specific total
totals[k][0] = 5;
totals[k+j][j/2] = 5;
The following are invalid.
Multidimensional Arrays 541

totals[k++][0]; // incs k not the element in total

totals++[k][0]; // ++ op comes after the subscripts
totals[.5][.3] = 5; // subscripts must be integers

Physical Memory Layout Versus Logical Layout

The physical memory layout always follows the same sequence. In a two-dimensional array, all
of the columns of row 0 come first followed by all the columns for row 1 and so on. This is
called row-major order. Figure 12.1 shows how memory is laid out for the totals array.

Figure 12.2 shows how the regtotals array is stored in memory.

Figure 12.1 Memory Layout for totals

Figure 12.2 Memory Layout for regtotals

Programmatically, two-dimensional arrays are often thought of as having rows and

columns, rather like a table. It is more useful to take the following logical viewpoint of the totals
array, where the x-axis represents the columns and the y-axis represent the rows. This is shown in
Figure 12.3.

In a similar manner, a three-dimensional array has the first or leftmost dimension on the
z-axis (coming into or out of the page) and its second dimension is along the y-axis and the
rightmost dimension is along the x-axis.
Multidimensional Arrays 542

Figure 12.3 The Logical Rows and Columns View

Initialization of Multidimensional Arrays

The basic concept of single-dimensioned array initialization is extended in a similar fashion to
multidimensional arrays. Consider the totals array of 3 rows by 4 columns. First, all of row 0 is
initialized. However, since row 0 is an array of 4 columns, the array notation is used.
double totals[3][4] = { {1, 2, 3, 4}, {11. 12. 13. 14},
row 0 row 1

{21, 22, 23, 24} };

row 2

If all of the elements of a single-dimensioned array are not initialized, the default value of
0 is used for the remaining unspecified elements. Thus, if totals were to be initialized to 0, it
could be done as follows.
double totals[3][4] = { {0}, {0}, {0} };
To initialize all the elements of regtotals to 0, one could do the following.
double regtotals[3][4][2] = { { {0}, {0}, {0}, {0} },
// dept 0 1 2 3
// –– store 0 ---------
{ {0}, {0}, {0}, {0} }, { {0}, {0}, {0}, {0} } };
// ----- store 1 ----- ----- store 2 -----

As the number of dimensions increases, the initialization syntax becomes awful!

Frequently, it is much simpler just to write some loops to initialize the arrays at run time. The
regtotals array can be initialized as follows.
for (int i=0; i< STORES; i++) {
for (int j=0; j<DEPTS; j++) {
for (int k=0; k<REGS; k++) {
regtotals[i][j][k] = 0;
}
}
}
Multidimensional Arrays 543

Passing Multidimensional Arrays to Functions

With single-dimensioned arrays, the name of the array is a constant pointer or memory address of
the first element of the array. The same is true of a multidimensional array; its name is a constant
pointer or memory address of the first element. If the totals array were to be passed to a function,
say calcs(), the prototype is
void calcs (double totals[][DEPTS]);
It would not be wrong to provide all dimensions though, as in
void calcs (double totals[STORES][DEPTS]);
However, the compiler always ignores the leftmost dimension's value. However, all other
dimensions must be specified. The main() function would then invoke the calcs() function by
calcs (totals);

The following is in error — why?

void calcs (double totals[][]);
To see why, suppose that within calcs() one coded the following.
totals[1][0] = qty * sales;
How does the compiler locate the specific element to assign the calculation result? The compiler
finds the start of the 2nd row (subscript 1) by multiplying the number of elements in any row
times the size of the data type of the array. This is called the row offset from the start of the
array. It then finds the column offset by multiplying the column number by the size of the data
type of the array. Finally, the compiler adds the starting address of the array with the row offset
and then the column offset to yield the memory address of the requested element. Thus, to find
the row offset, the compiler must know how many elements are in a row, that is, the second
dimension.

Suppose that the array regtotals is to be passed to the calcs2() function. The prototype is
void calcs2 (double regtotals[][DEPTS][REGS]);
In all cases, it is permissible to omit only the leftmost dimension. However, it is always
permissible to provide all the dimension limits; this is also less error prone.
void calcs2 (double regtotals[STORES][DEPTS][REGS]);
The main() function would then invoke calcs2() by the following.
calcs2 (regtotals);

Loading a Multidimensional Array from an Input File

Consider again the budget array with which the chapter began. It was defined as
double budget[NumberItems][ NumberMonths];
Suppose that the data were stored in a file called budget.txt and that a LoadArray() function is
to read this data filling up the budget array. Recall with a single-dimensioned array, typically,
not all potential elements were used in a given execution of a program. We commonly track the
Multidimensional Arrays 544

number of elements actually input with an integer, numElements, for example. The input of a
multidimensional array presents some additional complications.

Suppose in true generality that not every column of every row was present. That is, for
row 0, we might only have the first two columns; for row 1, all three columns are present; for
row 2, only the first column's data is present; and so on. How could this be represented in the
input data and worse still, how would the program know when accessing a specific row how
many columns of data were actually in that row? True, we could input on a line by line basis and
say all columns for a given row were on one line so that line breaks ended that row's input, but
there is no easy way to “remember” how many elements each row has. If this needed to be done,
a parallel second array of integer column counts would have to be constructed in which the
number of elements actually in row 0 of the budget array was stored in element 0 of the counts
array. Notice how fast the complexity is rising!

In reality, very often the input of multidimensional arrays is simplified to one of two
approaches:
All elements of every row and all rows are entered
All elements of every row are entered, but not all rows are input
In other words, only the leftmost dimension can have a variable number input. For instance, with
the budget array, we could store the number of rows of budget items actually input in the integer
numItems. However, every row entered must have all three monthly values present.

The main() function calls LoadBudget() as follows.

int numItems = LoadBudget (budget, NumberItems,
NumberMonths);
Here is the LoadBudget() function that returns the number of items or rows actually input.
int LoadBudget (double budget[NumberItems][NumberMonths],
int itemLimit, int monthLimit) {
ifstream infile ("budget.txt");
if (!infile) {
cerr << "Error: cannot open budget.txt\n";
exit (1);
}
int j = 0;
int k;
while (j<itemLimit && infile >> ws && infile.good()) {
for (k=0; k<monthLimit && infile; k++) {
infile >> budget[j][k];
}
if (!infile) {
cerr << "Error: unexpected end of a budget row "
<< j+1 << endl;
exit (2);
}
j++;
Multidimensional Arrays 545

}
infile.close ();
return j;
}

Working with Multidimensional Arrays

When working with two-dimensional arrays, a programmer is frequently called upon to sum the
contents of an entire row (summing all the columns of that row) or to sum the contents of a
specific column (summing that column in all the rows). Let's examine the straightforward
approach to these two problems and then see what can be done to improve execution speed of the
operations. The main() function defines the array to use and calls the simple sumrow1() function
to sum all of the values in a row and display that sum.
#include <iostream>
#include <iomanip>
using namespace std;

const int NUMROWS = 3;

const int NUMCOLS = 4;

int sumrow1 (int x[][NUMCOLS], int whichrow);

int sumrow2 (int x[NUMCOLS]);

int sumcol1 (int x[][NUMCOLS], int whichcol);

int sumcol2 (int x[NUMCOLS], int whichcol);

int main() {
int n;
int array[NUMROWS][NUMCOLS] = { {1,2,3,4}, {11,12,13,14},
{21,22,23,24} };
// Method 1: normal sumrow function
for (n=0; n<NUMROWS; n++) {
cout << "sumrow1 = " << sumrow1(array,n) << endl;
}

The function sumrow1() is straightforward. For the given row, simply add up each
column.
int sumrow1 (int x[][NUMCOLS], int whichrow) {
int i, sum = 0;
for (i=0; i<NUMCOLS; i++) {
sum += x[whichrow][i];
}
return sum;
}
Multidimensional Arrays 546

How can this coding be speeded up at execution time? First of all, if the bounds are small
and the sum is invoked only one time, there is no need to try to improve its efficiency. However,
if the bounds are large and this function is to be invoked many times, then a speed up is in order.
What is slowing down the sumrow1() function is the need for two subscripts. Remember that to
find the current element to add to the sum, the compiler must first calculate the row offset and
then the column offset and add those to values to the beginning memory address of the array in
order to find the requested element. Both offset calculations involve multiplying by the size of
the data type, the sizeof a double or 8 in this case. The multiplication machine instruction is
fairly slow, though on the Pentium class chips, it has been drastically speeded up. If we can
reduce the number of multiplies, the function executes more quickly.

When summing a row, all the columns of that row are in consecutive memory locations,
that is, it can be thought of as a single-dimensioned array of four columns in this case. Thus, we
pass only the current row n of four columns. The notation is
array[n]
Here I have provided only the first subscript of the two-dimensional array. The compiler assumes
that I am specifying only the nth row, which is a single-dimensioned array of four doubles.

The main() function now does the following.

// results in 25% less code, 4% faster execution
for (n=0; n<NUMROWS; n++) {
cout << "sumrow2 = " << sumrow2(array[n]) << endl;
}
The sumrow2() function is now very simple indeed.
int sumrow2 (int x[NUMCOLS]) {
int i, sum = 0;
for (i=0; i<NUMCOLS; i++) {
sum += x[i];
}
return sum;
}

The other common operation is summing columns. For a specific column, find the sum of
that column by accumulating the sum of that column in every row. The main() function calls
sumcol1() illustrating the straightforward approach. Note that no matter which vertical column
whose sum is desired, the entire array is passed.
for (n=0; n<NUMCOLS; n++) {
cout << "sumcol1 = " << sumcol1(array,n) << endl;
}
The basic sumcol1() function is as follows. Here we simply go through each row adding up the
desired column.
int sumcol1 (int x[][NUMCOLS], int whichcol) {
int i, sum = 0;
for (i=0; i<NUMROWS; i++) {
sum += x[i][whichcol];
Multidimensional Arrays 547

}
return sum;
}

If performance requires the sumcol() function to be more efficient, how can it be

improved? The objective is once more to reduce the number of multiplication machine
instructions it takes to find the desired element. In this case, we cannot just pass the needed row
as an array of four column values; we need to go vertical through the rows summing the specific
column in each row. However, there is a trick that we can use. Recall that the compiler never
checks for the “subscript out of range” condition. Assume that we have passed only the very first
row of the two-dimensional array, so that the sumcol2() function sees only a single-dimension
array of four elements, those of row 0. Set the initial subscript to the desired column that we are
to sum — for example column 2. Then, to get to the next row’s corresponding column, to our
subscript add the number of columns in a row. In this case there are four columns per row. The
next subscript is 2 + 4 or 6 which is, in fact, really in column 2 of the next row.
// The results: it has 2% faster execution
for (n=0;n<NUMCOLS;n++) {
cout << "sumcol2 = " << sumcol2(array[0],n) << endl;
}
The improved sumcol2() function is shown below.
int i, j = whichcol, sum = 0;
for (i=0; i<NUMROWS; i++) {
sum += x[j];
j = j + NUMCOLS;
}

Here are the complete program listing and sample output for Basic12a that implements
all these functions. Multidimensional arrays have many uses. Some of these are explored in the
next sections.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Listing for Basic12a - Working with two-dimensional arrays *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Basic12a Working with 2-dimensional arrays */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 #include <iostream> *
* 7 #include <iomanip> *
* 8 using namespace std; *
* 9 const int NUMROWS = 3; *
* 10 const int NUMCOLS = 4; *
* 11 *
* 12 int sumrow1 (int x[][NUMCOLS], int whichrow); *
* 13 int sumrow2 (int x[NUMCOLS]); *
* 14 *
* 15 int sumcol1 (int x[][NUMCOLS], int whichcol); *
* 16 int sumcol2 (int x[NUMCOLS], int whichcol); *
Multidimensional Arrays 548

* 17 *
* 18 int main() { *
* 19 int n; *
* 20 int array[NUMROWS][NUMCOLS] = { {1,2,3,4}, {11,12,13,14}, *
* 21 {21,22,23,24} }; *
* 22 *
* 23 // Method 1: normal sum the rows function *
* 24 for (n=0; n<NUMROWS; n++) { *
* 25 cout << "sumrow1 = " << sumrow1(array,n) << endl; *
* 26 } *
* 27 cout << endl; *
* 28 *
* 29 // To get it faster, pass a single dimensioned array of 4 cols *
* 30 // that is the row to sum up. Access is now by a single *
* 31 // subscript array[n] is an array of 4 ints *
* 32 // The results: takes 25% less code and has 4% faster execution *
* 33 for (n=0; n<NUMROWS; n++) { *
* 34 cout << "sumrow2 = " << sumrow2(array[n]) << endl; *
* 35 } *
* 36 cout << endl; *
* 37 *
* 38 // Normal sum the columns approach *
* 39 for (n=0; n<NUMCOLS; n++) { *
* 40 cout << "sumcol1 = " << sumcol1(array,n) << endl; *
* 41 } *
* 42 cout << endl; *
* 43 *
* 44 // To get it faster, pass only the address of start of the first*
* 45 // row as a single dimension array of 4 columns *
* 46 // The results: it has 2% faster execution *
* 47 for (n=0;n<NUMCOLS;n++) { *
* 48 cout << "sumcol2 = " << sumcol2(array[0],n) << endl; *
* 49 } *
* 50 cout << endl; *
* 51 *
* 52 return 0; *
* 53 } *
* 54 *
* 55 /***************************************************************/*
* 56 /* */*
* 57 /* sumrow1: basic method to sum the contents of a specific row */*
* 58 /* */*
* 59 /***************************************************************/*
* 60 *
* 61 int sumrow1 (int x[][NUMCOLS], int whichrow) { *
* 62 int i, sum = 0; *
* 63 for (i=0; i<NUMCOLS; i++) { *
* 64 sum += x[whichrow][i]; *
* 65 } *
* 66 return sum; *
* 67 } *
* 68 *
Multidimensional Arrays 549

* 69 /***************************************************************/*
* 70 /* */*
* 71 /* sumrow2: faster method to sum the contents of a specific row*/*
* 72 /* */*
* 73 /***************************************************************/*
* 74 *
* 75 int sumrow2 (int x[NUMCOLS]) { *
* 76 int i, sum = 0; *
* 77 for (i=0; i<NUMCOLS; i++) { *
* 78 sum += x[i]; *
* 79 } *
* 80 return sum; *
* 81 } *
* 82 *
* 83 /***************************************************************/*
* 84 /* */*
* 85 /* sumcol1: basic method to sum the contents of a specific col */*
* 86 /* */*
* 87 /***************************************************************/*
* 88 *
* 89 int sumcol1 (int x[][NUMCOLS], int whichcol) { *
* 90 int i, sum = 0; *
* 91 for (i=0; i<NUMROWS; i++) { *
* 92 sum += x[i][whichcol]; *
* 93 } *
* 94 return sum; *
* 95 } *
* 96 *
* 97 /***************************************************************/*
* 98 /* */*
* 99 /* sumcol2: fater method to sum the contents of a specific col */*
* 100 /* */*
* 101 /***************************************************************/*
* 102 *
* 103 int sumcol2 (int x[], int whichcol) { *
* 104 int i, j = whichcol, sum = 0; *
* 105 for (i=0; i<NUMROWS; i++) { *
* 106 sum += x[j]; *
* 107 j = j + NUMCOLS; *
* 108 } *
* 109 return sum; *
* 110 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Basic12a - Working with two-dimensional arrays *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 sumrow1 = 10 *
* 2 sumrow1 = 50 *
* 3 sumrow1 = 90 *
* 4 *
* 5 sumrow2 = 10 *
* 6 sumrow2 = 50 *
Multidimensional Arrays 550

* 7 sumrow2 = 90 *
* 8 *
* 9 sumcol1 = 33 *
* 10 sumcol1 = 36 *
* 11 sumcol1 = 39 *
* 12 sumcol1 = 42 *
* 13 *
* 14 sumcol2 = 33 *
* 15 sumcol2 = 36 *
* 16 sumcol2 = 39 *
* 17 sumcol2 = 42 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Some More Examples of Array Processing

Let’s assume that we have defined a two-dimensional array this way.
const int MAX = 20;
int array[MAX][MAX];
Also, assume that there is a value in all twenty rows and all twenty columns of each row. And
let’s perform some actions on the array. First, set all elements of the 3rd row to 0.
for (col=0; col<MAX; col++) {
array[2][col] = 0;
}

Next, let’s set all the 2nd column values to 1.

for (row=0; row<MAX; row++) {
array[row][1] = 1;
}

Let’s set each element of the main diagonal from (0,0) to (19,19) to 42.
for (row=0; row<MAX; row++) {
array[row][row] = 42;
}

Let’s set each element of the back diagonal from (0,19) to (19,0) to 5.
for (row=0; row<MAX; row++) {
array[row][MAX-row-1] = 5;
}
Multidimensional Arrays 551

Section B: A Computer Science Example

Cs12a — Arrays of Strings
Suppose that we wished to store an array of character strings, such as the month names? Notice
that since a string is a single-dimensioned array of char, then the array of strings must therefore
be really a two-dimensioned array of characters. Here is how the array of month names could be
defined and initialized.
const char monthNames[13][10] = { "",
"January", "February", "March", "April", "May",
"June", "July", "August", "September", "October",
"November", "December" };
When initializing an array of strings, it is permissible to use “strings” as element values. Since
month numbers normally range from one through twelve, I have “wasted” element 0, storing a
null string in it. Also notice that I made the array constant. Thus, these values cannot be
accidentally altered by the program.

For this example, let’s write a word counter program. Our input can be a sample chapter
of someone’s writing. We must construct a table of the words that we find (ignoring case) and a
parallel array of frequencies of those words.

Initially, numWords, which holds the number of words in the table, is zero. The two-
dimensional array words holds all of the found words. The array freq holds the corresponding
frequency counts. The array word holds the current word that has been input.

What constitutes a “word?” My definition is overly simplified. It is any consecutive series

of letters only. Hyphenated words become two separate words. The main() program then defines
these as follows.
const int MaxWords = 1000;
const int WordLen = 20;
char words[MaxWords][WordLen]; // the table of found words
int freq[MaxWords] = {0};
char word[WordLen]; // place to input the next word
int numWords = 0; // number of words in the table

Figure 12.4 shows main storage for the main() function. The basic operation of the
program is to input a word and find out if it is in the table. If it is in the table already, then
increment that word’s corresponding frequency. If it is not yet in the table of words, add it to the
table and set its frequency count to one. This process is repeated until the end of the text file
occurs. The results are sorted into frequency of use order and printed. This then suggests the
following Top-Down Design shown in Figure 12.5.
Multidimensional Arrays 552

Figure 12.4 Main Storage for Word Program

Figure 12.5 Top-Down Design of Word Counter Program

From a design point of view, the GetNextWord() function should return the istream
reference from which it is inputting the new word. That way, we can test for the end of file
condition in the main() function’s while clause. SearchTableForWord() should return the
subscript of the matching word or the current number of words in the table if the new word is not
in the table. This allows the main() function to become very simple indeed. Here is the sequence.
open infile; if it fails to open, display an error message and quit
while (GetNextWord (infile, word, WordLen)) do the following
i = SearchTableForWord (words, numWords, word);
if i == numWords, then the word is not in the table, so do these steps
add this word by doing strcpy (words[i], word);
let freq[i] = 1;
increment the numWords in the table
end then clause
otherwise it is in the table, so increment its count: freq[i]++
end the while clause
sort the words array into order by calling into frequency by calling SortArray passing
it words, freq and numWords
call PrintArray passing it words, freq, numWords, and a string containing the
filename to use for the display.
Multidimensional Arrays 553

The GetNextWord() function retrieves the next word. It must skip over non letters to
next letter; then get and store each letter until it either encounters a non letter or reaches the
maximum word length – 1. Finally, it must insert the null terminator.

GetNextWord() is passed infile, word[] and the wordlen. Its sequence is as follows.
To skip over everything but a letter:
while (infile.get (c)) do the following
if c < ‘A’ || (c > ‘Z’ && c < ‘a’) || c > ‘z’ then it is not a letter, so continue
else break out of the loop with c containing the letter
end while
if !infile then it is EOF and no word was found, so do these steps
word[0] = 0 to null terminate the string
return infile
end the if
Now insert successive letters into the word array until a non letter or wordlen – 1 is reached or
EOF occurs:
word[count++] = c;
while count < wordlen -1 && infile.get (c) do the following
is c < ‘A’ || (c > ‘Z’ && c < ‘a’) || c > ‘z’ then it is not a letter so break
word[count++] = c to insert the letter into the array
end while loop
word[count] = 0 to null terminate the string
return infile;

The SearchTableForWord() function is passed table[][WordLen], count and the new

word. It must return the subscript of the matching word in the table or count if word is not in the
table yet. It is case insensitive.
for (int i=0; i<count; i++) do the following
if (stricmp (word, table[i]) == 0) then it is a match, so return i;
end for loop
no match, so return count

The SortArray() function sorts the table of words and frequencies into frequency order
from highest to lowest. It is passed the table[][WordLen], freq[] and count. Because table and
freq are parallel arrays, swapping must be done together. Assuming that temp and tfreq are the
temporary variables to hold a string and a frequency, the following represents the sorting
sequence.
for (int i=0; i<count-1; i++) do the following
for (int j=i+1; j<count; j++) do the following
is freq[j] > freq[i]? If so, then we have found a larger frequency, so swap
tfreq = freq[i]; // exchange frequencies
freq[i] = freq[j];
freq[j] = tfreq;
Multidimensional Arrays 554

strcpy (temp, table[i]); // exchange words in the table

strcpy (table[i], table[j]);
strcpy (table[j], temp);
end if
end inner loop
end outer loop

The PrintArray() function prints out a nicely formatted set of results and is passed the
table[][WordLen], freq[], count and the filename string to use for output.
attempt to open out; if it fails, display an error message and quit
display a title
for (int i=0; i<count; i++)
output the table[i] word and its frequency freq[i]
close out

Here are the completed program Cs12a and the abbreviated output for the first few page
of this chapter used as input.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Cs12a - Word Usage Frequency Analysis *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs12a Word frequency of use in the first part of Chapter 12 */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 #include <string> *
* 11 #include <cctype> *
* 12 using namespace std; *
* 13 *
* 14 const int MaxWords = 1000; *
* 15 const int WordLen = 20; *
* 16 *
* 17 istream& GetNextWord (istream& infile, char word[], int wordlen);*
* 18 int SearchTableForWord (char table[][WordLen], int count, *
* 19 char word[]); *
* 20 void SortArray(char table[][WordLen], int freq[], int count);*
* 21 void PrintArray(char table[][WordLen], int freq[], int count,*
* 22 const char filename[]); *
* 23 *
* 24 int main () { *
* 25 char words[MaxWords][WordLen]; // the table of found words *
* 26 int freq[MaxWords] = {0}; // corresponding frequencies *
* 27 int numWords = 0; // number of words in the table *
* 28 *
Multidimensional Arrays 555

* 29 char word[WordLen]; // an inputted word *

* 30 *
* 31 const char filename[] = "Cs12a-text.txt"; *
* 32 ifstream infile (filename); *
* 33 if (!infile) { *
* 34 cerr << "Error: cannot open input file " << filename << endl; *
* 35 return 1; *
* 36 } *
* 37 *
* 38 // input words and accumulate frequency of occurrence *
* 39 while (GetNextWord (infile, word, WordLen)) { *
* 40 int i = SearchTableForWord (words, numWords, word); *
* 41 if (i == numWords) { // word not in the table yet *
* 42 strcpy_s (words[i], sizeof(words[i]), word); // add this word *
* 43 freq[i] = 1; // set count to 1 time *
* 44 numWords++; // increment total word count *
* 45 } *
* 46 else freq[i]++; // is in the table, so inc frequency *
* 47 } *
* 48 infile.close (); *
* 49 *
* 50 // sort into frequency of use order *
* 51 SortArray (words, freq, numWords); *
* 52 *
* 53 // print the results *
* 54 PrintArray (words, freq, numWords, filename); *
* 55 *
* 56 return 0; *
* 57 } *
* 58 *
* 59 /***************************************************************/*
* 60 /* */*
* 61 /* GetNextWord: retrieves the next word */*
* 62 /* skip over non-letters to next letter */*
* 63 /* get and store each letter until either it encounters a */*
* 64 /* non-letter or reaches max word length -1 */*
* 65 /* null terminates the string */*
* 66 /* */*
* 67 /***************************************************************/*
* 68 *
* 69 istream& GetNextWord (istream& infile, char word[], *
* 70 int wordlen) { *
* 71 char c; *
* 72 int count = 0; *
* 73 while (infile.get (c)) { // skip over everything but a letter *
* 74 if (c < 'A' || (c > 'Z' && c < 'a') || c > 'z') *
* 75 continue; *
* 76 else *
* 77 break; *
* 78 } *
* 79 *
* 80 if (!infile) { // if eof & no word, at least return null string *
Multidimensional Arrays 556

* 81 word[0] = 0; *
* 82 return infile; *
* 83 } *
* 84 *
* 85 // insert successive letters until a non-letter is found or *
* 86 // it reaches wordlen - 1 or hits eof *
* 87 word[count++] = c; *
* 88 while (count < wordlen -1 && infile.get (c)) { *
* 89 if (c < 'A' || (c > 'Z' && c < 'a') || c > 'z') *
* 90 break; *
* 91 word[count++] = c; *
* 92 } *
* 93 *
* 94 word[count] = 0; // null terminate string *
* 95 return infile; *
* 96 } *
* 97 *
* 98 /***************************************************************/*
* 99 /* */*
* 100 /* SearchTableForWord: returns subscript of matching word */*
* 101 /* in table or count if it is not in the table yet */*
* 102 /* case insensitive comparison */*
* 103 /* */*
* 104 /***************************************************************/*
* 105 *
* 106 int SearchTableForWord (char table[][WordLen], int count, *
* 107 char word[]) { *
* 108 for (int i=0; i<count; i++) { *
* 109 if (_stricmp (word, table[i]) == 0) *
* 110 return i; *
* 111 } *
* 112 return count; *
* 113 } *
* 114 *
* 115 /***************************************************************/*
* 116 /* */*
* 117 /* SortArray: sorts the table of words & frequencies into freq */*
* 118 /* order from highest to lowest */*
* 119 /* */*
* 120 /***************************************************************/*
* 121 *
* 122 void SortArray (char table[][WordLen], int freq[], int count) { *
* 123 char temp[WordLen]; *
* 124 int tfreq; *
* 125 for (int i=0; i<count-1; i++) { *
* 126 for (int j=i+1; j<count; j++) { *
* 127 if (freq[j] > freq[i]) { // found a larger freq, so swap *
* 128 tfreq = freq[i]; // exchange frequencies *
* 129 freq[i] = freq[j]; *
* 130 freq[j] = tfreq; *
* 131 strcpy (temp, WordLen, table[i]); // exchange words in table *
* 132 strcpy (table[i], WordLen, table[j]); *
Multidimensional Arrays 557

* 133 strcpy (table[j], WordLen, temp); *

* 134 } *
* 135 } *
* 136 } *
* 137 } *
* 138 *
* 139 /***************************************************************/*
* 140 /* */*
* 141 /* PrintArray: print out a nicely formatted set of results */*
* 142 /* */*
* 143 /***************************************************************/*
* 144 *
* 145 void PrintArray (char table[][WordLen], int freq[], int count, *
* 146 const char filename[]) { *
* 147 ofstream out ("Cs12a-results.txt"); *
* 148 if (!out) { *
* 149 cerr<<"Error: cannot open the output file Cs12a-results.txt\n";*
* 150 exit (1); *
* 151 } *
* 152 out << "Word frequencies found in " << filename << endl << endl;*
* 153 for (int i=0; i<count; i++) { *
* 154 out << setw (4) << i+1 << ". " << left << setw (WordLen+5) *
* 156 << table[i] << right << setw (6) << freq[i] << endl; *
* 159 } *
* 160 out.close (); *
* 161 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Abbreviated output of Cs12a - Word Usage Frequency Analysis *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Word frequencies found in Cs12a-text.txt *
* 2 *
* 3 1. the 40 *
* 4 2. array 15 *
* 5 3. A 15 *
* 6 4. totals 15 *
* 7 5. of 12 *
* 8 6. each 12 *
* 9 7. are 10 *
* 10 8. is 9 *
* 11 9. be 8 *
* 12 10. Arrays 8 *
* 13 11. budget 8 *
* 14 12. for 8 *
* 15 13. two 8 *
* 16 14. to 7 *
* 17 15. In 7 *
* 18 16. dimensional 6 *
* 19 17. following 6 *
* 20 18. dimensioned 6 *
* 21 19. limit 6 *
* 22 20. columns 6 *
Multidimensional Arrays 558

* 23 21. subscripts 6 *
* 24 22. three 5 *
* 25 23. One 5 *
* 26 24. double 5 *
* 27 25. with 5 *
* 28 26. memory 5 *
* 29 27. that 5 *
* 30 28. define 5 *
* 31 29. total 5 *
* 32 30. stores 5 *
* 33 31. store 5 *
* 34 32. Multi 5 *
* 35 33. All 5 *
* 36 34. k 5 *
* 37 ... *
* 38 227. out 1 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Section C: Engineering Examples

Matrix Algebra
Multiple dimensioned arrays open new vistas in the types of problems that can be solved.
Specifically, matrices can be stored in two-dimensional arrays. Matrices can be used to solve
linear simultaneous equations, such as n equations in n unknowns. The starting point is a brief
review of the rules of Matrix Algebra.

Suppose that we had the following simultaneous equations.

5x + 4y + 3z = 40
9y + 3z + 8x = 10
4z + 3x + 6y = 20
They must be rearranged into the proper format.
5x + 4y + 3z = 40
8x + 9y + 3z = 10
3x + 6y + 4z = 20
In matrix notation, this becomes the following.

Or A X = B; so the solution is X = B/A

Multidimensional Arrays 559

The normal matrix notation for this case of 3 equations in 3 unknowns is show below.

Notice that the math matrix notation parallels C++ subscripts, but begins with subscript 1
not 0. Always remember to subtract 1 from the matrix math indices to get a C++ array subscript.

In this example, the a matrix is composed of 3 rows or row vectors, and 3 columns or
column vectors. In general a matrix is said to be an m by n matrix, m rows and n columns. When
m = n, it is called a square matrix. A matrix with only one row is a row matrix; one with only
one column is a column matrix. The x and b matrices are both column matrices.

Matrix Math Operations Summary

1. Two matrices are said to be equal if and only if they have the same dimensions and all
corresponding elements are equal.
aij = bij for all i=1,m and j=1,n

2. Addition and Subtraction operations require that the matrices involved have the same
number of rows and columns. To compute C = A + B or C = A – B, simply add or subtract all
corresponding elements. This can be implemented in C++ as follows.
for (int I=0; I<M; I++) {
for (int J=0; J<N; J++) {
C(I,J) = A(I,J) + B(I,J);
}
}

3. Multiplication of a matrix by a number is commutative. That is, rA is the same as Ar.

The result is given by r times each element.
for (int I=0; I<M; I++) {
for (int J=0; J<N; J++) {
A(I,J) = A(I,J) * r;
}
}

For example, assume A is defined to be the following.

Then 2A would be
Multidimensional Arrays 560

and 10/9A would be

4. A diagonal matrix is one whose elements above and below the principal diagonal are
0: namely aij=0 for all i!=j

5. An identity matrix is a diagonal matrix whose principal diagonal elements are all 1.

6. Matrix multiplication says that the product of a square matrix times a column matrix is
another column matrix. It is computed as follows: for each row in the square matrix, sum the
products of each element in the square matrix's row by the corresponding element in the column
matrix's column.

For a square matrix times a square matrix, the result is a square matrix of the same
dimensions, each element of the result is the sum of the products of each element of the
corresponding row of one matrix times each element of the corresponding column of the other
matrix
C=A*B
where Cij = ith row of A * jth column of B or in coding
for (int I=0; I<3; I++) {
for (int J=0; J<3; J++) {
C(I,J) = 0;
for (int K=0; K<3; K++) {
C(I,J) = C(I,J) + A(I,K)*B(K,J);
}
}
}

7. Determinants form a crucial aspect in solving systems of equations. What we are after
is the ability to solve: A*X = B so that we can solve it as X = B/A. However, matrix division is a
Multidimensional Arrays 561

real problem and really is not needed because there is a simpler method. A determinant can be
pictorially thought of as rather like a “greatest common denominator.”

If we had this simple two equations in two unknowns problem

a11x1 + a12x2 = b1
a21x1 + a22x2 = b2
then the long hand solution would be
x1 = (b1a22 - b2a12) x2 = (b2a11 - b1a12)
--------------- ---------------
(a11a22 - a21a12) (a11a22 - a21a12)
assuming the denominator, called the determinant, is not zero; Note that the determinant is a
single number. Notice that the determinant can be considered as the sum of the right slanted
diagonals — sum of the left slanted diagonals.

It is notated as |a|, the determinant of a. For a 3x3 matrix, the determinant is given by the
following number.
a11a22a33 + a12a23a31 + a13a32a21
- a11a23a32 - a21a12a33 - a31a22a13
For a larger matrix, that is, the general case, the Cofactor Matrix concept is used. Consult a
matrix math text for details. Normally another approach is used when the number of dimensions
becomes four or more.

Mathematical Theorems of Determinants

The following summarize the rules that apply to working with determinants. We will apply these
to the problem of solving simultaneous equations shortly.

1. The value of a determinant is not altered if its rows are written as columns in the same
order.

2. If all the elements of one row (or one column) of a determinant are multiplied by the
same factor k, the value of the determinant is k times the value of the determinant. Notice the
difference between k|D| and kA, where |D| is a determinant and A is a matrix. The operation
k|D| multiplies just one row or column by k but kA multiplies all elements by k.

3. If all elements of a row or column of a determinant are zero, the value of the
determinant is zero.
Multidimensional Arrays 562

4. If any one row is proportional to another row (or one column is proportional to another
column), then the determinant is zero.

5. If the elements of any one row are identical (in the same order) to another row, the
determinant is zero. Likewise for columns.

6. Any two rows or columns may be interchanged, and the determinant just changes sign.

7. The value of a determinant is unchanged if the elements of a row (or column) are
altered by adding to them any constant multiple of the corresponding elements in any other row
(or column).

Given these basic principles, we can examine the two common methods used to solve
linear equations. The Gauss Method is perhaps the simplest procedure used to solve linear
equations. The Gauss-Jordan Method, which is a refinement of the Gauss Method, makes the
process even easier.

The Gauss Method for Solving a System of Linear Equations

Suppose we had to solve:
2x + 3y - z = 1
3x + 5y + 2z = 8
x - 2y - 3z = -1
Gauss elimination begins by using the first equation to eliminate the first variable x in the second
and other equations. By the theorems above, we can add, subtract, and multiply without altering
the results. So we choose to subtract some multiple of the first equation to the remaining
equations to bring about 0x in the remaining equations:
3x +5.0y + 2.0z = 8.0 second equation
-(3x +4.5y - 1.5z = 1.5) subtract the first equation * 1.5
-----------------------
0x +0.5y +3.5z = 6.5 the replacement second equation

So we now have the following.

2x + 3y - z = 1
0x + .5y + 3.5z = 6.5
x - 2y - 3z = -1

Next we do the same for the third equation.

x - 2.0y - 3.0z = -1 third equation
-(x + 1.5y - 0.5z = .5) subtract first equation * .5
-----------------------
0x -3.5y - 2.5z = -1.5
Multidimensional Arrays 563

So now we have these results.

2x + 3y - z = 1
0x + .5y + 3.5z = 6.5
0x - 3.5y - 2.5z = -1.5

Now we move down to row 2 and repeat the process for all those equations that lie below
row 2 - here, only row 3. Notice that column 1, the x column, is 0x for all rows from 2 on down.
Hence, we only need to consider column 2 downward.
- 3.5y - 2.5z = -1.5 third equation
-(- 3.5y -24.5z =-45.5) subtract second equation *-7
-----------------------
0y +22.0z = 44

So now we have the following.

2x + 3y - z = 1
0x + .5y + 3.5z = 6.5
0x + 0y + 22z = 44

We can back solve for the answers at this point. Beginning at the bottom, that is, row 3,
we get the following.
z = 44/22 = 2
y = (6.5 - 3.5*2)/.5 = -1
x = (1 + 2 - 3*(-1))/2 = 3

Before we implement this procedure, let’s study the effects. In the Gauss method, the
“pivot row” is defined to be that row at which we are currently using to eliminate elements in the
equations below. The pivot row begins at row 1 and runs to the n–1 row. The last row is
complete — here 22z = 44. The “pivot coefficient” is the multiplicative factor we need to force
zero in the “pivot column” of the lower rows. How did we determine what the pivot coefficient
should be? Look again at the first step above. We took the second equation’s pivot column’s
number (3x) and divided it by the first equation's pivot column’s number (2x) to get 1.5.

Now you can see the problem. What happens if the current pivot row’s pivot column
contains zero? A divide by zero problem! The second problem is round off. If we divide by a tiny
number, we will get a large result that is then used to multiply all other elements. Clearly the
errors introduced will mushroom.

A better approach is known as “pivoting.” We know that we can reorder the equations
without altering the solution. So the idea is that, when reaching a specific pivot row and
corresponding pivot column, we scan on down to the nth row and replace the current pivot row
with that lower row that has the largest absolute magnitude pivot column number. Not only will
this reduce round off errors, it will help avoid dividing by zero as well.
Multidimensional Arrays 564

Gauss-Jordan Method of Solving Simultaneous Linear Equations

The Gauss-Jordan method is very similar to Gauss except that the elimination is carried further.
The objective is to turn matrix A into an identity matrix, in which case, the coefficient matrix B
actually holds the results.

For each pivot row r from 1 to n do the following two steps.

1. The pivot element is a(r,r); normalized by dividing all elements of
pivot row r and the b(r) by the pivot element a(r,r). This gets the pivot element set to 1.

2. Next add multiples of the pivot row r to every other row in such a manner so
that the pivot column r (the same number as the pivot row) has all zeros in it.

When the process is done, the matrix A will have become the identity matrix with the
answers in the B column matrix. Usually, the B matrix is copied into the X answer matrix for
consistency.

Listing Engr12a shows both the Gauss() and GaussJordan() functions. The main()
function is discussed next.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Engr12a Gauss and GaussJordan Solutions of Equations *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr12a Gauss and GaussJordan Solutions of Equations */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <fstream> *
* 9 #include <iomanip> *
* 10 #include <cmath> *
* 11 using namespace std; *
* 12 *
* 13 const int MaxDim = 25; *
* 14 bool Gauss (double A[][MaxDim], double B[], double X[], int N); *
* 15 bool Reorder (double A[][MaxDim], double B[], int N, int pivot); *
* 16 void Elim (double A[][MaxDim], double B[], int N, int pivot); *
* 17 void Backsub (double A[][MaxDim], double B[], int N, double X[]);*
* 18 bool GaussJordan (double A[][MaxDim], double B[], double X[], *
* 19 int N); *
* 20 void ElimJordan (double A[][MaxDim], double B[], int N, *
* 21 int pivot); *
* 22 *
* 23 int main () { *
* 24 // the arrays - one for Gauss and one for GaussJordan *
* 25 double A[MaxDim][MaxDim], B[MaxDim], X[MaxDim]; *
* 26 double AJ[MaxDim][MaxDim], BJ[MaxDim], XJ[MaxDim]; *
Multidimensional Arrays 565

* 27 int i, j, N; *
* 28 *
* 29 // setup floating point output format *
* 30 ofstream out ("GaussResults.txt"); *
* 31 out << fixed << setprecision (5); *
* 33 *
* 34 // open input file *
* 35 ifstream infile ("mirror.dat"); *
* 36 if (!infile) { *
* 37 cerr << "Error: cannot open file mirror.dat\n"; *
* 38 out.close (); *
* 39 return 1; *
* 40 } *
* 41 // input the number of equations in n unknowns *
* 42 infile >> N; *
* 43 if (N > MaxDim) { *
* 44 cerr << "Error: array dimensions exceeded - can handle 25\n"; *
* 45 out.close (); *
* 46 infile.close (); *
* 47 return 2; *
* 48 } *
* 49 // input the arrays and copy into arrays for GaussJordan *
* 50 for (i=0; i<N; i++) { *
* 51 for (j=0; j<N; j++) { *
* 52 infile >> A[i][j]; *
* 53 AJ[i][j] = A[i][j]; *
* 54 } *
* 55 infile >> B[i]; *
* 56 BJ[i] = B[i]; *
* 57 } *
* 58 infile.close (); *
* 59 *
* 60 // do both the Gauss and Gauss Jordan solutions *
* 61 if (Gauss (A, B, X, N)) { *
* 62 if (GaussJordan (AJ, BJ, XJ, N)) { *
* 63 // print the results *
* 64 out << "The solutions are\n\n I Gauss Gauss Jordan" *
* 65 << endl << endl; *
* 66 for (i=0; i<N; i++) { *
* 67 out << setw (3) << i+1 << setw (13) << X[i] << setw (15) *
* 68 << XJ[i] << endl; *
* 69 } *
* 70 } *
* 71 } *
* 72 out.close (); *
* 73 return 0; *
* 74 } *
* 75 *
* 76 /***************************************************************/*
* 77 /* */*
* 78 /* Gauss function */*
* 79 /* */*
Multidimensional Arrays 566

* 80 /* Problem Definition: Perform a Gauss Elimination to find */*

* 81 /* the solution of a set of n linear equations. */*
* 82 /* */*
* 83 /* Pass array of coefficients A, column array of answers B */*
* 84 /* and a column array X to hold the solution */*
* 85 /* Pass N number dims in use this run */*
* 86 /* */*
* 87 /* The answer is placed in the array X, with original arrays */*
* 88 /* A and B being destroyed in the process */*
* 89 /* */*
* 90 /* The solution utilizes pivoting and reordering to bring the */*
* 91 /* largest coefficient to the current pivot row on each */*
* 92 /* iteration reducing round off errors and */*
* 93 /* avoiding division by 0 */*
* 94 /* */*
* 95 /* However, should the pivot divisor become too small 1.E-5 or*/*
* 96 /* actually be 0, then no solution is possible and Gauss */*
* 97 /* returns false. */*
* 98 /* If all is well, Gauss returns true, solution found */*
* 99 /* */*
* 100 /***************************************************************/*
* 101 *
* 102 bool Gauss (double A[][MaxDim], double B[], double X[], int N) { *
* 103 int pivot = 0; // begin on the first row and go til N-1 row*
* 104 while (pivot < (N-1)) { *
* 105 if (!Reorder (A, B, N, pivot)) {; // bring largest to pivot row*
* 106 cerr << "Error - no solution is possible\n"; *
* 107 return false; *
* 108 } *
* 109 Elim (A, B, N, pivot); // perform the elimination *
* 110 pivot++; // and go on to the next row *
* 111 } *
* 112 Backsub (A, B, N, X); // now calculate the solution *
* 113 return true; *
* 114 } *
* 115 *
* 116 /***************************************************************/*
* 117 /* */*
* 118 /* Reorder: find the largest pivot element in this pivot row */*
* 119 /* if it is not in the current pivot row, */*
* 120 /* replace the pivot rowand this larger one's row */*
* 121 /* */*
* 122 /***************************************************************/*
* 123 *
* 124 bool Reorder (double A[][MaxDim], double B[], int N, int pivot) {*
* 125 int maxrow = pivot; // assume this one is the largest *
* 126 int row, col; *
* 127 double temp; *
* 128 // find if any other row has a larger one *
* 129 for (row=pivot+1; row<N; row++) { *
* 130 if (fabs (A[row][pivot]) > fabs (A[maxrow][pivot])) *
* 131 maxrow = row; *
Multidimensional Arrays 567

* 132 } *
* 133 // error check - is the largest one 0 or too small to continue? *
* 134 if (fabs (A[maxrow][pivot]) < 1.E-5) return false; *
* 135 // no, so check to see if rows will need to be swapped *
* 136 if (maxrow != pivot) { // yes, swap rows in A and also in B *
* 137 for (col=0; col<N; col++) { *
* 138 temp = A[maxrow][col]; *
* 139 A[maxrow][col] = A[pivot][col]; *
* 140 A[pivot][col] = temp; *
* 141 } *
* 142 temp = B[maxrow]; *
* 143 B[maxrow] = B[pivot]; *
* 144 B[pivot] = temp; *
* 145 } *
* 146 return true; // all is ok *
* 147 } *
* 148 *
* 149 /***************************************************************/*
* 150 /* */*
* 151 /* Elim: replace all rows below pivot row, */*
* 152 /* forcing 0 into the pivot col */*
* 153 /* */*
* 154 /***************************************************************/*
* 155 *
* 156 void Elim (double A[][MaxDim], double B[], int N, int pivot) { *
* 157 int row, col; *
* 158 double factor; *
* 159 for (row=pivot+1; row<N; row++) { *
* 160 factor = A[row][pivot] / A[pivot][pivot]; *
* 161 A[row][pivot] = 0; *
* 162 for (col=pivot+1; col<N; col++) { *
* 163 A[row][col] -= A[pivot][col] * factor; *
* 164 } *
* 165 B[row] -= B[pivot] * factor; *
* 166 } *
* 167 } *
* 168 *
* 169 /***************************************************************/*
* 170 /* */*
* 171 /* Backsub: perform back substitution to create answer array */*
* 172 /* */*
* 173 /***************************************************************/*
* 174 *
* 175 void Backsub (double A[][MaxDim], double B[], int N, double X[]){*
* 176 int row, col; *
* 177 // calculate directly the last one *
* 178 X[N-1] = B[N-1] / A[N-1][N-1]; *
* 179 // now repetatively substitute found X's to create next one *
* 180 for (row=N-2; row>=0; row--) { *
* 181 for (col=N-1; col>row; col--) { *
* 182 B[row] -= X[col] * A[row][col]; *
* 183 } *
Multidimensional Arrays 568

* 184 X[row] = B[row] / A[row][row]; *

* 185 } *
* 186 } *
* 187 *
* 188 /***************************************************************/*
* 189 /* */*
* 190 /* GaussJordan function */*
* 191 /* */*
* 192 /* Problem Definition: Perform a Gauss Jordan Elimination */*
* 193 /* to determine the solution of a set of n linear equations. */*
* 194 /* */*
* 195 /* Pass array of coefficients A, column array of answers B */*
* 196 /* and a column array X to hold the solution */*
* 197 /* Pass N number dims in use this run */*
* 198 /* */*
* 199 /* The answer is placed in the array X, with the original */*
* 200 /* A and B being destroyed in the process */*
* 201 /* */*
* 202 /* The solution utilizes pivoting and reordering to bring the */*
* 203 /* largest coefficient to current pivot row on each iteration */*
* 204 /* thereby reducing round off errors and avoiding dividing by0*/*
* 205 /* */*
* 206 /* However, should the pivot divisor become too small 1.E-5 or*/*
* 207 /* actually be 0, then no solution is possible and it returns */*
* 208 /* false. If all is well, returns true, solution found */*
* 209 /* */*
* 210 /***************************************************************/*
* 211 *
* 212 bool GaussJordan (double A[][MaxDim], double B[], double X[], *
* 213 int N) { *
* 214 int pivot = 0; // begin on the first row and go til N-1 row*
* 215 while (pivot < N) { *
* 216 if (!Reorder (A, B, N, pivot)) { // bring largest to pivot row *
* 217 cerr << "Error - no solution is possible\n"; *
* 218 return false; *
* 219 } *
* 220 ElimJordan (A, B, N, pivot); // perform the elimination *
* 221 pivot++; // and go on to the next row *
* 222 } *
* 223 for (int i=0; i<N; i++) { // copy results back into X array*
* 224 X[i] = B[i]; *
* 225 } *
* 226 return true; *
* 227 } *
* 228 *
* 229 /***************************************************************/*
* 230 /* */*
* 231 /* ElimJordan: replace all rows above and below pivot row */*
* 232 /* forcing 1 into main diagonal and 0 in all others*/*
* 233 /* */*
* 234 /***************************************************************/*
* 235 *
Multidimensional Arrays 569

* 236 void ElimJordan (double A[][MaxDim], double B[], int N, *

* 237 int pivot) { *
* 238 int row, col; *
* 239 // normalize pivot row, so A(pivot,pivot) = 1 *
* 240 for (col=pivot+1; col<N; col++) { *
* 241 A[pivot][col] = A[pivot][col] / A[pivot][pivot]; *
* 242 } *
* 243 B[pivot] = B[pivot] / A[pivot][pivot]; *
* 244 A[pivot][pivot] = 1.; *
* 245 // eliminate pivot column in all other rows, except pivot row *
* 246 for (row=0; row<N; row++) { *
* 247 if (row == pivot || A[row][pivot] == 0) continue; *
* 248 for (col=pivot+1; col<N; col++) { *
* 249 A[row][col] -= A[row][pivot] * A[pivot][col]; *
* 250 } *
* 251 B[row] -= B[pivot] * A[row][pivot]; *
* 252 A[row][pivot] = 0; *
* 253 } *
* 254 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Engr12a — Aligning the Mirrors of a Telescope (Astronomy)

Finally, we need a set of equations to solve. In astronomy, the larger the diameter of the mirror,
the better the performance. However, using glass mirrors beyond about 200 inches in diameter is
not feasible due to overall weight, cost and other factors. One way around this is to make an array
of inexpensive 12-inch mirrors and align them to focus at a single point, as if they were sections
of one large diameter mirror. Small servomotors can be used to keep all of the smaller mirrors in
precise alignment. Periodically, a set of optical sensors detect the current alignment. In this
example, there are 25 mirrors hooked together. Thus, one sample time period yields 25
coefficients of current alignment. When inserted into the corresponding 25 equations, and solved,
the resulting values are sent to the servomotors to bring each mirror into correct alignment. This
process is repeated periodically to keep the scope properly aligned.

The only difficulty in writing a main() function to do Gauss elimination is knowing in

advance how many equations and unknowns there are going to be. When we define the A, B, and
x matrices, we must provide the maximum number. In this example, I used an arbitrarily high
number, const int MaxDim = 25. Then, the program expects the very first number to be input to
contain the real number of unknowns on this particular test run. The real number of unknowns
must be less than or equal to 25. This MaxDim could have been set far higher on a Win32
platform because up to 2G of memory are potentially available. If MaxDim had been set to 100
unknowns, then the amount of memory used by the arrays would be 160k all told.

In this case, the first line in the test file mirror.dat contains the number of equations, 25.
Next comes the data for row 0’s 25 column coefficients followed by the corresponding B value.
Multidimensional Arrays 570

Next come row 1’s 25 coefficients and its B value and so on. The main() function inputs the A
and B matrices and does both a Gauss and Gauss-Jordan elimination, printing the solutions side
by side. Examine the above Engr12a.cpp file above for the coding in the main() function. Here
is the output of the program.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* The output from Engr12a - Gauss and Gauss-Jordan Methods *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 The solutions are *
* 2 *
* 3 I Gauss Gauss Jordan *
* 4 *
* 5 1 -0.91555 -0.91555 *
* 6 2 -0.40311 -0.40311 *
* 7 3 0.00197 0.00197 *
* 8 4 0.39354 0.39354 *
* 9 5 0.92317 0.92317 *
* 10 6 -0.81407 -0.81407 *
* 11 7 -0.16746 -0.16746 *
* 12 8 -0.00308 -0.00308 *
* 13 9 0.17736 0.17736 *
* 14 10 0.75808 0.75808 *
* 15 11 -0.85436 -0.85436 *
* 16 12 -0.26967 -0.26967 *
* 17 13 -0.00471 -0.00471 *
* 18 14 0.28572 0.28572 *
* 19 15 0.86823 0.86823 *
* 20 16 -0.81815 -0.81815 *
* 21 17 -0.17053 -0.17053 *
* 22 18 -0.00129 -0.00129 *
* 23 19 0.15691 0.15691 *
* 24 20 0.80501 0.80501 *
* 25 21 -0.91745 -0.91745 *
* 26 22 -0.39403 -0.39403 *
* 27 23 0.00120 0.00120 *
* 28 24 0.39109 0.39109 *
* 29 25 0.91720 0.91720 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Multidimensional Arrays 571

Design Exercises
1. Spreadsheet Design

You want to track your budget for a year’s time on a monthly basis. You have categorized your
expenses into ten categories and have only one job producing income. The input file consists of
11 lines. The first is the income line. It contains a character string description, “Monthly
Income,” which is then followed by 12 monthly income figures. The next 10 lines contain the
expense description, such as “House Rent” followed by the 12 monthly figures for that expense.
Sketch out the Top-Down Design and pseudo coding to produce your budget report shown
below.
My Budget Report

Item Jan Feb Mar ... Dec

Monthly Income 999.99 999.99 999.99 999.99
House Rent 999.99 999.99 999.99 999.99
...
------ ------ ------ ------
Total Expenses 9999.99 9999.99 9999.99 9999.99
Net Profit 9999.99 9999.99 9999.99 9999.99

You should make effective use of functions to eliminate as many repetitive actions as
possible.
Multidimensional Arrays 572

Stop! Do These Exercises Before Programming

1. The programmer wanted to define a two-dimensional array of grades. There are five sections
of twenty-four students each. What must be done to the following to get it defined correctly?
const int Sections = 5;
const int Students = 24;
int main () {
char grades[Students, Sections];
Which array bounds should come first, assuming that the normal processing handles all of the
students within a given section at one time? Why?

2. Since not every section has 24 students in it, the programmer decided to have another array
called numberStudentsInThisSection which is an array of five integers, one for each section.
Thus, numberStudentsInThisSection[0] contains the number of students in that section. With
this defined, a LoadStudentArrays() function was written but does not compile or work. Why?
What must be done to make this work properly?
const int Sections = 5;
const int Students = 24;
int LoadStudentArrays (char grades[][Students],
numberStudentsInThisSection[],
int maxSections, int maxStudents);
int main () {
char grades[Sections, Students];
int numberStudentsInThisSection[Sections];
int numSections = LoadStudentArrays (grades[][Students],
numberStudentsInThisSection[], Sections, Students);
...
int LoadStudentArrays (char grades[][Students],
numberStudentsInThisSection[],
int maxSections, int maxStudents){
int j = 0; // section subscript
while (cin >> ws) {
int k = 0; // student subscript
while (cin >> grades[k][j]) {
k++;
}
numberStudentsInThisSection[j] = j;
j++;
}
return j;
}
Multidimensional Arrays 573

3. Next the main() function attempted to printout all of the grades to see if they had been input
properly. The following coding does not work properly. Why? What must be done to get it to
properly print out the grades as entered?
const int Sections = 5;
const int Students = 24;
int LoadStudentArrays (char grades[][Students],
numberStudentsInThisSection[],
int maxSections, int maxStudents);
int main () {
char grades[Sections, Students];
int numberStudentsInThisSection[Sections];
int numSections = LoadStudentArrays (grades[][Students],
numberStudentsInThisSection[], Sections, Students);
for (int j=0; j<numSections; j++) {
cout << "\n\nSection: " << j << endl;
for (int k=0; k<numberStudentsInThisSection[k]; k++) {
cout << grades[k][j] << endl;
}
}

4. With the data properly input and printed, the next step is to calculate the average grade for
each section. Since the grades are letter grades, assume that a 4.0 system is in use. That is, an A
is worth 4 points, B is 3 and so on. The FindAvgGrades() function does not compile. Why?
How can it be made to work properly?
const int Sections = 5;
const int Students = 24;
void FindAvgGrades (char grades[][Students],
numberStudentsInThisSection[],
int numSections, double avgs[]);
int main () {
char grades[Sections, Students];
int numberStudentsInThisSection[Sections];
int numSections;
double averages;
...
FindAvgGrades (grades, numberStudentsInThisSection[],
int numSections, averages);
...
void FindAvgGrades (char grades[][Students],
int numberStudentsInThisSection[],
int numSections, double avgs[]){
double sum;
for (j=0; j<numberStudentsInThisSection[j]; j++) {
sum = 0;
for (k=0; k<numberStudentsInThisSection[j]; k++) {
switch (grades[j[k]) {
Multidimensional Arrays 574

case 'A':
sum += 4;
case 'B':
sum += 3;
case 'C':
sum += 2;
case 'D':
sum += 1;
case 'F':
sum += 0;
}
}
avgs[k] = sum / numberStudentsInThisSection[j];
}
}

5. Sorting of a two-dimensional array usually means sorting each row’s worth of column values
into order. Assume that there are 10 rows of raw scores and each row has 20 columns and that all
elements are present. That is, there are 200 values in the array. The following coding to sort the
array fails. Why? How can it be fixed so that the data are sorted properly?
const int Rows = 10;
const int Cols = 20;
double rawScores[Rows][Cols];
for (int j=0; j<Rows; j++) {
double temp;
for (int k=0; k<Cols; k++) {
for (int m=k; m<Cols; m++) {
if (rawScores[j][k] < rawScores[j][m]) {
temp = rawScores[k][j];
rawScores[j][k] = rawScores[j][m];
rawScores[j][m] = temp;
}
}
}
Multidimensional Arrays 575

Programming Problems
Problem Cs12-1 — Two-Dimensioned Array — Star Graph
The Jet Propulsion Laboratory wishes you to write a Star Graph program to quickly verify that
their digital images coming back from a space craft are operational. The spacecraft’s digital
images are focused onto a two-dimensional array of photo sensors, 30 rows of 30 columns of
sensors. Each sensor records the intensity of light hitting its surface during the exposure. The
intensity ranges from 0 to 20. Due to imperfections in the device, some stray light will scatter
about. Thus, a star is likely to be found at a given location only when the average of that sensor
and the surrounding 4 sensors intensity is greater than 6.0.

Create an array of integers that is 30 x 30 in size. The 4 surrounding neighbors would be

the same row but in the previous and in the next column as well as in the row above and below in
the same column as the reference intensity. In other words, no diagonals are used. Do not try to
find the intensity of the 4 surrounding neighbors on the outer rows and outer columns. There are
not 4 neighbors!

Call a function known as LoadArray() to actually input the basic intensities into the
array. Assume that there are 30 rows of 30 columns of data. However, guard against too many
rows and columns in the input data. The test data file is called Cs12-1-starplot.txt. The
LoadArray() prototype should be something like this.
bool LoadArray (int intensity[30][30]);
However, you should use a constant integer for the bounds of 30. If the function fails (cannot
find the file, for example), return false. If the array is loaded successfully, return true.

Next, the main() program should call a Plot() function whose prototype should be
something like this.
void Plot (int intensity[30][30]);

For each intensity in the array, print a ‘*’ character if there is a star at that location. Print a
blank if there is no star in that location. Thus, each row of intensities uses 30 columns and there
is 30 rows. Additionally, draw a frame around the image using the ‘-’ character for the horizontal
lines and a ‘|’ character for the vertical lines. Print your usual student id information first before
the graph is printed. The output, for example, might begin something like this:
------------------------------
| |
| * * |

Caution: you must consider the special cases of the top and bottom rows and the left and
right columns in the intensity averaging process. In these cases, there are not four nearest
neighbors. Also note that the 4 corners are different still.
Multidimensional Arrays 576

Problem Cs12-2 — Store Statistical Comparison

Acme Incorporated owns five chain stores located in five different states. Each of their stores has
five departments. To avoid check out lines, each department has two cash registers. Acme
Incorporated has collected together all of the monthly sales figures from all stores, all
departments and all registers within each department. The data is stored in master file Cs12-2-
sales.txt. The individual data items are total sales figures.

Define an array of doubles to hold the data. Each line of the file contains all the data for
one store and consists of ten sales amounts. The first pair represents the total sales from the two
cash registers within the first department within that store. The second pair represents the register
totals for the second department, and so on.

The desired report concerns a comparison of department sales from store to store. The
report appears as follows.

Acme Sales Analysis - Sales Figures

Department Store Grand
Alabama Illinois New York Virginia Texas Total
Automotive 99999.99 99999.99 99999.99 99999.99 99999.99 999999.99
Appliances 99999.99 99999.99 99999.99 99999.99 99999.99 999999.99
Clothing 99999.99 99999.99 99999.99 99999.99 99999.99 999999.99
Sporting Goods 99999.99 99999.99 99999.99 99999.99 99999.99 999999.99
Toys 99999.99 99999.99 99999.99 99999.99 99999.99 999999.99
Grand Totals 99999.99 99999.99 99999.99 99999.99 99999.99 999999.99

Acme Sales Analysis - Sales Percentages

Department Store
Alabama Illinois New York Virginia Texas
Automotive 99.9 99.9 99.9 99.9 99.9
Appliances 99.9 99.9 99.9 99.9 99.9
Clothing 99.9 99.9 99.9 99.9 99.9
Sporting Goods 99.9 99.9 99.9 99.9 99.9
Toys 99.9 99.9 99.9 99.9 99.9

The store that corresponds to subscript 0 is Alabama; subscript 4, Texas. The automotive
department is subscript 0; Toys is subscript 4, and so on. On the detail lines, the “Grand Total”
column represents the sum of all of the store’s automotive departments, for example. The last
line of totals represents each store’s total sales in all departments. The appearance of the first
report is similar to a spreadsheet. Once the grand totals are known, then the relative percentages
of each department can be calculated, store by store. For Alabama’s Automotive department,
multiply its total sales by 100 and divide by the grand total sales for the Automotive department.
The others are done similarly.
Multidimensional Arrays 577

Problem Cs12-3 — Tic-Tac-Toe Game

Write a program to play a tic-tac-toe game. The playing board consists of boxes laid out in three
rows of three columns. One player places X’s and the other player places O’s in the boxes. The
first player to have three in a row — horizontally, diagonally or vertically — wins the game. If
neither player wins, it is a draw. Store the “boxes” as a 3x3 array of char values. Initialize all
elements to a blank. When the X player places an X in a box, store the character ‘X’ in that box.
Use an ‘O’ character for the other player. Begin the game with the X player and take turns until
someone wins or it becomes a draw.

The interactive dialog should be something like this:

X player — enter row and column numbers (1 1 through 3 3): 2 2
Now place an X in array[1][1]. Notice that the array is zero-based, while the coordinates that the
players enter are one-based. A box is empty and available for a player to choose if it contains a
blank. After getting a valid choice from player X and placing the X in the correct box, determine
if that player has won. If so, the game ends with a winning message of your design. If not, then
repeat the sequence using the O player. Also, print out the final box representation showing the
placement of the X’s and O’s.

After a player has won or the game is a draw and the appropriate messages have been
shown, then ask the users if they want to play another game. If so, clear the array and repeat the
game.

Design your solution first. Notice that there are a lot of common operations going on. Try
to design your game to minimize duplicate coding, that is, doing a complete set of operations for
player X and then coding that same set for player O. Instead, write generalized functions that can
take a player’s mark, the X or O as a parameter and code the sequence one time as a function.
Multidimensional Arrays 578

Problem Engr12-1 — Weather Statistics using Arrays

Do some research on the weather. Pick a city in the northern hemisphere and obtain a series of
daily temperatures covering a 30-day period. Create a file containing these temperatures. Enter
one day’s set of temperature observations per line. Use a uniform number of temperatures for
each day. For example, you might have 10 daily temperature recordings each day for 30 days.

1. Read the temperatures into a two-dimensioned array, defined for example as:
double temps[NumDays][NumObservations];
where NumDays might be 30 and NumObservations might be 10. Use values that are
appropriate to your data.

2. Print a nicely formatted report with column heading headings listing the temperatures
for each day. Begin each day’s series on a new page. Remember printing a ‘\f’ code causes the
printer to eject to a new page.

3. Create single-dimensioned arrays to hold the daily high temperature, low temperature,
average temperature, the median temperature (the one in the middle), and the standard deviation.
These arrays should have NumDays elements, one for each day. Now, calculate these values.

4. Print a nicely formatted report; each line on the report represents the above five values
for that day. Begin this section of the report on a new page.

5. Also turn in a printout of the original data file used; for example:
C>COPY TEMPS.DAT PRN

6. Cite your source for the temperatures that you find. The following is an example of
some data you might find.
50 54 56 60 67 69 74 78 73 74 60 55
52 55 58 62 66 69 73 75 74 73 65 60
55 59 60 66 71 75 75 72 70 65 61 50
45 44 43 49 53 52 51 49 45 42 42 41
35 36 40 45 55 57 55 53 45 47 40 37
33 35 39 42 44 48 50 53 49 45 42 38
29 28 29 33 37 38 37 36 39 39 37 33
32 30 32 35 39 40 40 39 39 38 38 37
33 30 29 30 30 33 35 34 33 32 31 30
30 29 28 28 27 27 26 25 23 22 21 20
19 20 22 24 28 28 29 27 24 22 20 19
18 17 19 20 21 22 24 25 23 22 20 15
15 17 20 21 23 23 23 24 22 20 19 10
5 10 11 13 15 15 15 14 13 13 12 11
4 9 10 11 13 13 13 12 11 10 9 7
-1 0 2 1 0 1 2 4 3 2 1 -2
-4 -3 -2 1 1 3 4 2 -1 -1 -3 -5
Multidimensional Arrays 579

-8 -5 -4 -5 -3 -2 -1 -2 -3 -3 -4 -9
-11-3 -1 1 2 3 4 5 5 6 7 7
9 10 15 20 16 11 10 7 2 -1 0 1

Here are some URLs for the Internet to try out. There is no guarantee that they still exist.
Internet sites come and go.

For New Mexico, try entering this in the Location box:

http://weather.nmsu.edu/stations.html
This is the New Mexico State University data base. There is a “click here” to get the layout of the
hourly data. Pick a location; there’s lots of data.

For Utah, try entering this in the Location box:

http://climate.usu.edu/
Then pick “hourly data.” It is not as fancy an interface as New Mexico’s.

For Fairbanks, Alaska, try

http://newnet.jdola.lanl.gov/sel1705.html
Then choose Listing.

Problem Engr12-2 — Analysis of Static Truss (Civil Engineering)

In structural engineering, the calculations of the forces and reactions on a static truss are
important. In the following figure of a section of a bridge, Point 3 is constrained only to transmit
forces vertically to the pier, while Point 2 also can transmit in the horizontal direction. If an
external load of 1000 pounds downward is applied at Point 1, the load is distributed among the
various beams of the truss.

One can construct free-body-force diagrams for each of the three points. Since the system
is at rest, the sum of all the forces at each of the three node points must be 0. This leads then to
six equations that describe the situation.
-F1 cos 30 + F3 cos 60 + F1h = 0
-F1 sin 30 – F3 sin 60 + F1v = 0
F2 + F1 cos 30 + F2h + H2 = 0
F1 sin 30 + F2v + V2 = 0
-F2 –F3 cos 60 + F3h = 0
F3 sin 60 + F3v + V3 = 0
In this problem, F1v is –1000 pounds. F1h, F2h, F2v, F3h, F3v are 0. If we insert these numerical
values into the six equations and represent them in matrix format we have the following system
of equations to solve.
Multidimensional Arrays 580

Figure 12.6 The Force Diagram

Figure 12.7 The System of Equations

Use the data in the above matrix notation to make a data file to be used as input to the
main program. Write a program to input these data into arrays and perform both a Gauss and a
Gauss-Jordan elimination to solve for the six unknowns. For each variable, such as F1, print side
by side the formatted results from the Gauss and Gauss Jordan methods.

Problem Engr12-3 — A Matrix Math Package

When dealing with problems that involve matrices, it is convenient to have a set of canned
functions to handle the common matrix math operations. For simplicity, assume that all two-
dimensional matrices are square and that any corresponding column or row vectors have the
same dimension as their corresponding square matrix. Thus, one constant integer can provide
dimensions for the arrays, MaxDim in this case.

1. Write an IsEqual() function that returns true if the two square matrices are equal. Its
prototype should be
bool IsEqual (double A[MaxDim][MaxDim],
double B[MaxDim][MaxDim], int limit);
Multidimensional Arrays 581

2. Write an Add() function to add two square matrices A and B, placing the result in
matrix C. Its prototype is
void Add (double A[MaxDim][MaxDim],
double B[MaxDim][MaxDim],
double C[MaxDim][MaxDim], int limit);

3. Write a Subtract() function to subtract two square matrices (C = A – B) placing the

result in matrix C. Its prototype is
void Subtract (double A[MaxDim][MaxDim],
double B[MaxDim][MaxDim],
double C[MaxDim][MaxDim], int limit);

4. Write a MultiplyByConstant() function to multiply the matrix A by the value b and

place the result in matrix C. The prototype is
void MultiplyByConstant (double A[MaxDim][MaxDim],
double b,
double C[MaxDim][MaxDim],
int limit);

5. Write a function IsDiagonal() that returns true if the matrix is a diagonal matrix. Its
prototype is
bool IsDiagonal (double A[MaxDim][MaxDim], int limit);

6. Write a function IsIdentity() that returns true if the matrix is an identity matrix. Its prototype
is
bool IsIdentity (double A[MaxDim][MaxDim], int limit);

7. Write a function MatrixBySquare() that multiplies the square matrix A by the column
matrix B yielding the column matrix C. Its prototype is
void MatrixBySquare (double A[MaxDim][MaxDim],
double B[MaxDim],
double C[MaxDim], int limit);

8. Finally, write a function MatrixByMatrix() that multiplies the square matrix A by

square matrix B, yielding square matrix C. Its prototype is
void MatrixByMatrix (double A[MaxDim][MaxDim],
double B[MaxDim][MaxDim],
double C[MaxDim][MaxDim], int limit);

Now write a main function to test all of these. Thoroughly test all functions.
Structures 582

Chapter 13 — Structures

Section A: Basic Theory

Introduction
Suppose that a program needed to process car insurance premiums. Each line of input
represented one policy holder’s information about their insurance. Take a minute and reflect on
what fields of information might be needed. Certainly the line would contain the policy number,
the policy holder’s first and last name, their address, city, state, zip code, phone number, car
make, car model, VIN number, color, year and so on. There could easily be more than two dozen
individual fields. Following the principles of good programming design, the problem would
undoubtedly be functionally decomposed into several functions, for example input a set of data.
Can you imagine the function prototypes and headers required to pass two dozen fields? It would
be a nightmare.

Structures

What is needed is a way to group all these related fields into one large aggregate and then to be
able to use that aggregate to simplify the programming. This is exactly what a structure is and
does for us. A structure is a grouping of related fields of information that is often called a record
of data. Many convenient operations can be performed using this structure aggregate. If we had a
car insurance structure, we can pass this one instance of the large group to the functions instead
of the dozens of individual fields. The structure provides means for us to access the individual
member fields as needed.

Defining Structures

The starting point is to define the model or blueprint that the compiler uses when it needs to
create an actual instance of the structure in memory. This model is called the structure template
or definition. The template includes the keyword struct followed by the structure tag which is
the name that is used to identify this structure from all others. This is followed by all of the
member field definitions surrounded by braces {…} and ends with a semicolon.
Structures 583

Suppose that the program is to process cost records. Each cost record includes the item
number, quantity on hand, product description and its cost. Here is what the structure template
looks like.
const int DescrLen = 21; // max length of description

struct COSTREC {
long itemNum; // item number
short qty; // quantity on hand
char descr[DescrLen]; // item description
double cost; // item cost
};

The structure tag, COSTREC in this case, is used to identify this particular structure. By
convention, all structure tags either are wholly uppercase names (usually) or are capitalized.

The four data items contained between the braces { } are called the structure members.
Each structure member is a normal variable data definition. Notice that constant integers or
#defines can be used for array bounds as usual, but those definitions must precede the structure
template, following the “defined before first use” rule.

When any instance of COSTREC is created or used, the member fields are always
created and stored in the order shown in the template. The order of the structure members can
sometimes be important. If this program is part of a collection of programs, all sharing the same
files, such as a payroll system of programs, or if the data file to be input is in binary format, then
the structure members must be in the same order that the data is in the binary file. A binary file
is one in which all data is stored in internal format; binary files cannot be viewed with text
editors such as Notepad. They are discussed in detail near the end of this chapter. For most
problems, the fields can be in any order you choose.

Suppose that when recording weather statistics, data is measured and recorded every
hour. A daily weather record might be defined as follows.
const int NumObs = 24;
const int StationLen = 21;

struct WEATHER {
char stationName[StationLen]; // reporting location
float temps[NumObs]; // degrees Centigrade
float humidity[NumObs]; // such as 50%
float rainfall[NumObs]; // in millimeters
float windspeed[NumObs]; // in m/s
};
Notice that a structure can contain arrays.

Where are the structure templates located in a program? As with all data definitions, a
structure template must be defined before its first use. Where is its first usage? In a modular
Structures 584

program, structures or references to structures are commonly passed to other functions. Thus, the
structure templates must come before function prototypes that use them. The sequence is often
#includes
const ints or #defines
structure templates
int main () {

Creating Instances of a Structure

With the structure template defined, how are instances of it created? It is done in a manner
similar to any other intrinsic data type. For example, how would one define an instance of a
double called cost?
double cost;
The data type precedes the desired name of the variable. Structure instances follow the same
pattern. The data type is the structure tag in C++. The following creates a structure variable
called costRec and a structure variable called weaRec. A structure variable is just an instance of
a structure in memory.
COSTREC costRec;
WEATHER weaRec;

What does the structure variable costRec look like in memory when it is created by the
compiler? Figure 13.1 shows the memory layout of costRec and its member fields. Notice that
the fields are in the same order as in the COSTREC template.

Figure 13.1 The costRec Memory Layout

One can have arrays of structures as well. Suppose that the program needed to store a
maximum of 1000 cost records and a maximum of 500 weather records. The following defines
these two arrays and also shows the location of all the parts of the structure definitions.
#define MAXRECS 1000
const int DescrLen = 21; // max length of description

struct COSTREC {
long itemNum; // item number
short qty; // quantity on hand
char descr[DescrLen]; // item description
double cost; // item cost
};

int main () {
COSTREC arec[MAXRECS]; // array of 1000 cost records
Structures 585

...
or
#define LIMIT 500
const int NumObs = 24; // number observations per day
const int StationLen = 21; // max len of station

struct WEATHER {
char stationName[StationLen]; // reporting location
float temps[NumObs]; // degrees Centigrade
float humidity[NumObs]; // such as 50%
float rainfall[NumObs]; // in millimeters
float windspeed[NumObs]; // in m/s
};

int main () {
WEATHER weaArray[LIMIT];

A structure can also contain instances of other structures and arrays of other structures.
For example, consider a DATE structure which represents a calendar date. Using instances of a
DATE structure would make passing dates very convenient. Further, consider an employee
record that contained the employee’s id number, their salary and the date that they were hired.
The EMPLOYEE structure contains an instance of the DATE structure as shown below.
struct DATE {
char month;
char day;
short year;
};

struct EMPLOYEE {
long id;
double salary;
DATE hireDate;
};

Suppose that a CARMAINT structure must be defined to represent the periodic

maintenance requirements for a new car. Here the CARMAINT structure contains an array of
DATE structures.
const int numMaint = 10;
struct CARMAINT {
bool maintenanceDone[numMaint]; // true if the work was done
int maintenanceCode[numMaint]; // manufacturer’s maint. codes
DATE maintenanceDueDate[numMaint];// date maintenance is due
};
Structures 586

How are Structures Initialized?

An instance of a structure can be initialized when it is defined, just as any other variable.
However, since a structure typically has a number of data members, the values are surrounded by
braces {} as are single dimensioned arrays. The following structure represents a quarter coin
initialized as it is defined within main().
const int MaxLen = 10;
struct COIN {
int denomination;
char singular[MaxLen];
char multiple[MaxLen];
};
int main () {
COIN quarter = {25, "Quarter", "Quarters"};

How are Structure Members Accessed?

Having defined the structure template and created instance(s) of it, the next action is to utilize the
members within the structure. This is done by using the dot (.) operator. To the left of the dot
operator must be a structure variable and to the right must be a member variable of that
structure.

To access the qty member of the costRec instance, one codes

costRec.qty
To calculate the totalCost using the cost and qty members of the costRec instance, do the
following.
double totalCost = costRec.qty * costRec.cost;
To display the description, use
cout << costRec.descr;
To increment the costRec's quantity member or add another variable to it, one can code
costRec.qty++;
costRec.qty += orderedQty;

To input a set of data into the costRec variable, there are a number of ways. Here is one.
cin >> costRec.itemNum >> costRec.qty >> ws;
cin.get (costRec.descr, DescrLen);
cin >> costRec.cost;
The above assumes that no description field in the input data contains all blanks. It also assumes
that all descriptions contain DescrLen – 1 number of characters.

As you look these over, notice that there are no differences at all on input or output of
structure members, other than the requisite dot operator qualification with the structure variable.
Structures 587

Rules of Use for Structure Variables

Structure variables can be used for only five actions. These are the following.
A structure variable can be used to access structure members.
A structure variable or reference to one can be passed to a function.
A function can return a structure variable.
The address operator & returns the memory address of a structure variable
A structure variable can be assigned to another structure variable as long as they both
have the same structure tag.

We have already examined the first one, using the structure variable to access the
individual members, as in costRec.qty. The address operator & returns the address of the
structure variable. If one codes
&costRec
then the compiler provides the memory location where the instance begins. Normally, the
compiler does this automatically for us when we use reference variables.

Assume that the program also had defined another instance of the COSTREC.
COSTREC previousRec;
The fifth rule says that a complete copy of a structure variable can be done as follows.
previousRec = costRec;
This is very powerful indeed. Consider the alternative if this were not allowed. One would have
to write an assignment for each of the three numeric fields and then use strcpy() to copy the
string as shown below.
previousRec.itemNum = costRec.itemNum;
previousRec.qty = costRec.qty;
previousRec.cost = costRec.cost;
strcpy (previousRec.descr, costRec.descr);
Clearly, the ability to assign one structure variable to another instance can be a terrific operation
when it is needed.

A structure variable can be passed to a function or a reference to one can be passed.

Passing by reference is the best approach to take. Likewise, a function can return a copy of a
structure. However, in reality, returning a structure and passing a structure and not using
reference to a structure instance is generally avoided. Let's examine these two issues in detail.

Suppose that the main() program defined the cost record structure as we have been using
it thus far. Suppose further that the main() function then wanted to call a PrintRec() function
whose task is to print the data nicely formatted. The main() function does the following.
int main () {
COSTREC crec;
…
PrintRec (outfile, crec);
Structures 588

The PrintRec() function begins as follows.

void PrintRec (ostream& outfile, COSTREC crec) {
outfile << crec.itemNum…

When the compiler generates the instructions to make the call to PrintRec(), it must
make a new parameter instance of the COSTREC structure and then spend execution time to
copy all the data from the main()'s costRec instance into PrintRec()'s crec parameter instance.
For structures that contain a large number of members, this is wasteful of both memory (the
parameter copy) and execution speed (making the copy every time the function is called).

A far better approach is to pass the structure variable by reference. A simple change to
PrintRec() vastly improves both memory utilization and execution speed.
void PrintRec (ostream& outfile, COSTREC& crec) {
outfile << crec.itemNum…
Here the compiler actually passes only the memory address of main()'s costRec. PrintRec()'s
crec parameter is now a reference variable (usually occupying 4 bytes of memory). No copy of
the data is made.

Okay. But what about good programming design? PrintRec() should not be allowed to
modify the contents of main()'s costRec in any way. It should have read-only access, for
example. This can be enforced by using the const keyword as follows.
void PrintRec (ostream& outfile, const COSTREC& crec) {
outfile << crec.itemNum…
Here, the data being referenced is constant and cannot be changed within PrintRec(). If
PrintRec() were to attempt to assign 42 to the qty field as in
crec.qty = 42;
the compiler generates a compile time error. Thus, you should always use constant references in
functions that should not be allowed to alter the data. A ReadRec() function would certainly not
be passed a constant COSTREC reference. It should be filling up the structure instance with
input data.

This brings up the ReadRec() function whose job it is to input the data and somehow fill
up the main()'s costRec with that data. One way that the ReadRec() function can be defined is to
have it return a COSTREC structure. This is not a good way to do it, but let’s see how a function
can return a structure instance. Then, we will see how to better design the ReadRec() function. If
ReadRec() returns a structure, then main() would have to assign it to main()'s costRec variable.
From a design point of view, since main() is passing ReadRec() a reference to the input stream,
ReadRec() lets the main() function decide on what to for I/O errors, bad data and EOF detection.
The coding for main() is as follows.
int main () {
COSTREC costRec;
costRec = ReadRec (infile);
// now check on infile’s state
Structures 589

Now in ReadRec(), the coding can be done this way.

COSTREC ReadRec (istream& infile) {
COSTREC temp = {0};
if (infile >> ws && !infile.good()) {
return temp;
}
infile >> temp.itemNum >> and so on
return temp;
}
Here the structure variable temp is filled with the input file’s next set of data and then a complete
copy of temp is returned to main(). However, since EOF can occur as well as bad data and since
we have to return an instance of the COSTREC structure, temp is initialized to zeros. Back in
main(), when the function call to ReadRec() is completed, the compiler then must copy that
returned copy of temp into main()'s costRec variable. If the structure contained a large number
of member fields, memory is being wasted. In all cases, execution speed is going to suffer
because of all the copying operations needed to move the data from temp into costRec.

While there can be times when this overhead cannot be avoided, usually the answer is to
pass a reference to the ReadRec() function and have the function fill up main()’s costRec
directly. This then frees up the return value for other uses. And by now returning a reference to
the input stream being used for input, the caller of the ReadRec() function can make more
effective use of the language.

Suppose that ReadRec() was rewritten to be passed a reference to the caller's COSTREC
structure variable to be filled with the input data. The improved function is shown below.
istream& ReadRec (istream& infile, COSTREC& crec) {
if (infile >> ws && !infile.good()) {
return infile;
}
infile >> crec.itemNum >> and so on
return infile;
}

Now the main() function has more ways that it can utilize the ReadRec() function. Here
is the improved main() function.
int main () {
COSTREC costRec;
…
while (ReadRec (infile, costRec)) {
Certainly main() benefits from the change. The while clause is basically testing the goodness of
the input stream after the input operations are complete. Also, ReadRec() now avoids both the
extra memory overhead of returning a structure instance and the execution time needed to make
the copies.
Structures 590

User-Written Header Files

A programming application may involve a series of programs. A payroll application typically
consists of several individual programs to accomplish the total task of managing a company’s
payroll needs. One program might input the weekly time sheets and store the data into a master
file on disk. The next program in the series might calculate the wages of all the employees from
that master file on disk. Another program might input that master file and produce IRS reports,
and so on. All of these programs must use an EMPLOYEE structure to group the fields into a
manageable aggregate.

If each program defined its own version of the EMPLOYEE structure, chaos arises.
Suppose a simple change was requested — increase the size of the employee’s name field from
20 characters to 30. The programmers would have to go into every program in the system and
make that same change in every version of the EMPLOYEE structure. This is simply not
practical nor feasible. Instead, the EMPLOYEE structure definition or template and related
items such as the const ints or #defines are stored in a separate file, a header file with the
extension .h. Each program that wishes to use the EMPLOYEE structure only has to code a
#include for it.

The syntax for user header files differs slightly from system header files. The <> on the
#include compiler directives notify the compiler that the filename within the angle brackets is in
the system include folder(s). To include user header files, use double quote marks instead of the
angle brackets — “filename” for example.
#include "Employee.h"
This instructs the compiler to copy the contents of the file Employee.h into the program at this
point. The compiler assumes that the Employee.h file is located in the same folder as the project
cpp file(s).

If the header file is not in the project folder, a partial path, a relative path or a full path
can be included. The following are all valid, assuming the file is located in the indicated place.
#include "Employee.h"
#include "Include\\Employee.h"
#include "..\\IncludeFiles\\Employee.h"
#include "D:\\Programming\\Include\\Employee.h"
The first one implies the header is in the project folder. The second one implies that the header
file is in a subfolder of the project called Include. The third one implies that the header is in the
IncludeFiles folder located in the parent folder of the project folder. The fourth one provides an
absolute path.

In general, do not use absolute paths because the project is not portable. One cannot move
the project to other drives, other computers or other folders and still compile properly. User
headers are often found in the project folder.
Structures 591

The header file is built just like a cpp file. If you are working with the Microsoft Visual
C++ compiler, choose File-New and pick header file and enter the name desired.

If we wanted to place the COSTREC structure template into a header file, the contents of
the header file CostRec.h would be as follows.
const int DescrLen = 21; // max length of description

struct COSTREC {
long itemNum; // item number
short qty; // quantity on hand
char descr[DescrLen]; // item description
double cost; // item cost
};

In the main() function, you should include first all of the needed C++ standard library
header files that the program requires and then include all of the user header files.
#include <iostream>
#include <iomanip>
#include <cmath>
using namespace std;
#include "Costrec.h"

Why? Accidentally, you might have used a name that is already in use by the system
functions and system coding contained in the system header files. If you include the system
headers first, the compiler points to your accidental redefinition in your header file. However, if
you include the user headers first and the system headers second, then the compiler’s error
messages point to a redefinition error within the system headers!

Binary Files and Structures

Up to this point, every file of data that has been used is a text file. That is, the encoding scheme
is that of pure ASCII text. Such files are easily viewed and created with Notepad or any other text
editor. For example, suppose that the file contained the quantity of 1234. If you viewed this file
and the line with this number on it, four characters are shown: ‘1’, ‘2’, ‘3’ and ‘4’. The digits are
stored using their ASCII decimal values of 49, 50, 51 and 52. If the quantity were defined to be
a short and then inputted as usual with
cin >> quantity;
we know that quantity is stored in a binary format. In this case, since a short has 16 bits and
since the sign occupies the first or leftmost position (0 meaning it is a positive number), then 15
bits remain to store the value.

In memory the quantity variable appears this way.

0 0 0 0 0 1 0 0 1 1 0 1 0 0 1 0
Structures 592

This means the number stored is given by:

or
2 + 16 + 64 + 128 + 1024 = 1234
This is what is meant by internal numeric format. Floating point types have a different format.

When the input stream inputs the quantity, it brings in the ASCII text digits and must
then convert those digits into the binary internal representation. This is a process known as data
conversion. In general, data conversion is a slow, tedious process. It is very slow indeed for the
floating point types.

A binary file is one in which all data are stored in internal format. If the short quantity
were stored in a binary file, the two bytes would contain precisely the bit pattern that they would
have if it were in memory, namely
0 0 0 0 0 1 0 0 1 1 0 1 0 0 1 0
It is indistinguishable from the short quantity in memory. If the ifstream was to input a
quantity field from a binary file, no conversion is needed ever. It only has to input the two bytes
that comprise a short and place then into the two bytes of the quantity field.

Binary file I/O is significantly faster than text file I/O. The larger the file, the more
significant the speed difference. Consider some other examples: a double occupies eight bytes in
a binary file; a long, four bytes. How about a character string? Suppose that the string was
defined to contain a maximum of 20 bytes but that this particular instance currently is holding the
string “Sam”. Binary output of this string outputs all 20 bytes, not just the characters that are in
use at the moment. Similarly binary input of the string must input all 20 bytes. Contrast this with
what would be output using a cout operation.

If a company is storing its data in a data base or file, then it makes sense to store that data
in a binary file. Consider for a moment the data base or master file of a national car insurance
company. One expects that the company to have a significant number of programs accessing that
data for some purpose, such as billing and reporting. By making the master file be a binary file
instead of a text file, their entire system of programs runs significantly faster.

Mechanics of Binary Files

The iostreams are capable of reading and writing binary files. When opening a file for binary
operations, another flag must be set, ios::binary. If this flag is not set, the iostream assumes that
it is a text type of file.

Two new functions actually handle the binary input and output operations, read() and
write(). Let’s begin by writing a short program that inputs a text file of temperatures and writes
out a binary file of those same temperatures. When opening a binary file, normally we have used
Structures 593

the flags ios::in or ios::out. Now we must OR in the ios::binary flag.

ofstream outfile;
outfile.open ("temps.dat", ios::out | ios::binary);
float temp;
while (cin >> temp) {
outfile.write ((char*)&temp, sizeof (temp));
}
outfile.close ();
The new coding is in bold. First notice how ios::binary is ORed into the open flags.

The write() function looks complicated at first, but it really is simple. Its prototype is
ostream& write (char* buffer, unsigned int length);
The first parameter is the memory address from which the data is to be written and the second
parameter is how many bytes are to be written starting at that location. The memory address it
assumes is an array of characters or bytes. We use the address operator & to obtain the memory
location of temp. The (char*) typecast is used to convince the compiler that this memory address
of a float is what is desired. The sizeof() macro is used to obtain the implementation defined
number of bytes of the float variable temp.

Next, let’s input the temperatures from the binary file we have just written. The short
sequence is as follows.
ifstream infile;
infile.open ("temps.dat", ios::in | ios::binary);
float temp;
while (infile.read ((char*)&temp, sizeof (temp))) {
do something with temp
}
infile.close();
Again, the new coding is in bold face. The ios::binary flag is ORed with ios::in.

The syntax of the read() function is

istream& read (char* buffer, unsigned int numBytes);
The read() function inputs the number of bytes indicated by the second parameter into
consecutive bytes beginning with the memory address provided by the first parameter. It returns a
reference to the original input stream.

When arrays are involved, binary I/O is even more powerful. Suppose that we had an
array of up to 1000 temperatures with numTemps containing the number actually present on this
execution. We can output the entire array with a single write function call!
float temps[1000];
int numTemps;
outfile.write ((char*) temps, numTemps * sizeof (float));
Or if all 1000 entries were filled, just
outfile.write ((char*) temps, sizeof (temps));
Structures 594

It gets even better when structures are involved. Suppose that we wanted to write our cost
record instance, costRec, to a binary output file.
COSTREC costRec;
outfile.write ((char*) &costRec, sizeof (costRec));
One line and the entire structure with all of its member fields are written to disk. To input a
single COSTREC structure instance into costRec from a binary file of data, it is as simple as
infile.read ((char*) &costRec, sizeof (costRec));

Binary data files using structures opens up new processing avenues and leads directly to
inquiry and file update type applications. In an inquiry program, the user enters a key identifier,
such as social security number, and the program finds that specific record in the binary master
file, reads it into memory and typically displays the relevant information. In an update program,
after inputting the data for the specific item, various fields are altered and the new modified data
is rewritten to the master file, overlaying the previous contents. These type applications are
covered in my sequel book, Beginning Data Structures in C++. Structures represent the key to
most all advanced processing circumstances. Structures led historically to the development of
classes in object oriented programming.

Section B: Computer Science Examples

Cs13-1 — Credit Card Application with Sorting
Acme Credit Company has a master file of its card holders. Each record in the file
CreditMaster.txt represents the data for one customer. The data consists of the account number
(social security number), the customer name of up to 30 characters, a two-character state code
abbreviation, the credit limit to be extended and finally the current balance. The customer name
is enclosed in quote marks, for example, “John Doe.” The state codes are all uppercase letters.
Design a credit structure and create an array that can hold a maximum of 500 card holders. Note
that the file is in no particular order.

The program is to input the master file and create the Credit Extended by State report.
The report looks like this.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output Report of Cs13a Credit Extended by State Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Acme Credit Company - Credit Extended by State Report *
* 2 *
* 3 State Number of Credit Totals Extended *
* 4 Code Accounts Limit Balance *
* 5 AL 4 $ 22100.00 $ 20500.00 *
* 6 IA 5 $ 28300.00 $ 22642.42 *
* 7 IL 7 $ 36500.00 $ 33500.00 *
* 8 MO 4 $ 20200.00 $ 18500.00 *
* 9 -- --- --------- --------- *
Structures 595

* 10 4 20 $107100.00 $ 95142.42 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

The program should have a LoadArray() function and a SortArray() function to

subsequently sort the array of credit records into state code order. Once this has been done, the
data can easily be accumulated and the report printed from a PrintReport() function.

The main() program should prompt the user for the master file name to be used and for
the name of the report file to write. The screen display of the program should appear as follows;
note that the user enters the file names to use.
Enter the name of the credit master file: CreditMaster.txt
Enter the name of the output report file: CreditReport.txt

Array loaded: nnn records input

Sorting the array
Printing the report
Finished

In this program, the report is sent to an output file. However, there is lots of processing
going on during the execution of the program. Hence, to make the program more user-friendly, it
must periodically display the last four lines, indicating the action that is occurring at that point.
Otherwise, if the input file was large and the processing time therefore lengthy, the user would be
staring at a blinking cursor wondering what was going on and perhaps panicking a bit.

Figure 13.2 shows the Top-Down Design for the program.

Figure 13.2 Top-Down Design for the Program

The data for a single card holder is stored in a CREDITREC structure. It is defined as
follows.
struct CREDITREC {
long acctno; // account number =social security num
char custName[MaxNameLen]; // customer name up to 30 characters
char stateCode[MaxStCodeLen];// upper case state code
double creditLimit; // credit limit of customer
double balance; // customer's current balance due
};
Structures 596

The main() function defines an array of CREDITREC structures called creditRec and
numRecs contains the current number of elements actually in the array. Since the user must input
the two file names, let’s call them inputFilename and reportFilename. Main storage for main()
is shown in Figure 13.3. Notice how I have drawn the array of CREDITREC structures. At the
bottom of the array, I have included the member names.

Figure 13.3 Main Storage for main()

The instruction sequence for main() is very simple, IPO as usual. First, it must obtain the
input and report filenames from the user.
prompt for and input inputFilename
prompt for and input reportFilename
load the array of credit records by calling
numRecs = LoadArray (creditRec, MaxNum, inputFilename)
display a message to the user that numRecs records were input
display a message that we are sorting the array into state code order
call SortArray (creditRec, numRecs)
display a message that we are printing the report
call PrintReport (creditRec, numRecs, reportFilename);
display a finished message

LoadArray() loads the passed array of credit records from the input file. It is given
crec[], limit and filename. Its sequence is as follows. Notice how the ith element of the structure
is referenced in the input operations. When the custName field is to be input, the leading double
quotes mark is input into a character leaving the input stream pointing to the first character of the
name. The getline() function is used to extract all characters of the name up to the trailing double
quotes mark; getline() also removes this trailing character from the input stream.
define and open infile (filename), if it fails to open, display an error message and abort
let i = 0;
while i<limit and an attempt to partially input the data by
Structures 597

infile >> crec[i].acctno >> c is successful then do the following

input the customer name by infile.getline (crec[i].custName, MaxNameLen, ‘\"’)
input the next byte which can be the " if the name is max length infile.get (c)
then skip over whitespace to the start of the state code infile >> ws
input the stateCode by calling infile.get (crec[i].stateCode, MaxStCodeLen)
input the limit and balance by infile >> crec[i].creditLimit >> crec[i].balance
increment i
end while loop
guard against bad input data by checking infile.fail() — if bad, then do
display to cerr an error message stating what was the last good id
close the infile
abort the program
end the if bad block
guard against too many elements by checking if i == limit and
if infile >> ws is successful, then do the following
display an error message to cerr that the array size has been exceeded
close the infile
abort the program
end the if
close the infile
return i which is the number of records input

The SortArray() sorts the credit records into state code order. It is passed crec[] and
numRecs. In order to swap two elements of the array, a temporary CREDITREC called temp is
needed. Since the field to be sorted is a character string, the strcmp() string function is used.
for (i=0; i<numRecs-1; i++) {
for (j=i+1; j<numRecs; j++) {
if (strcmp (crec[j].stateCode, crec[i].stateCode) < 0) {
temp = crec[i];
crec[i] = crec[j];
crec[j] = temp;
end if
end inner for loop
end outer for loop

The PrintReport() function must summarize the data while printing the report. It is
passed crec[], numRecs and the filename to use. This is essentially a control break problem. We
must sum the number of accounts, their limits and balances for each state. Since the array is now
in state code order, the main loop just sums the values. When there is a change in state code, then
we display the totals for the state we just summed and reset the totals and begin summing the
next state.
Structures 598

Begin by identifying the key variables to hold our needed values. First, prevState should
contain the previous state code; we check it against the current state code of the ith element in the
array. The three state totals are numAcctsInState, totalLimit and totalBalance. The grand
nationwide totals are grandNumStates which is the total number of states serviced,
grandNumAccts, grandTotalLimit and grandTotalBalance. All of these seven totals must be
initialized to zero. Figure 13.4 shows the main storage layout for PrintReport(). Note for ease of
debugging, I have recopied the actual array defined in the main() function here in the diagram for
crec; in fact, crec is really just the memory address of main()’s creditRec array.

Figure 13.4 Main Storage for PrintReport()

The coding sequence for PrintReport() is as follows.

open outfile, but if it fails, display an error message to cerr and abort
output headings and column headings
if there are any elements in crec, then do the summation
copy the first state code into prevState — strcpy (prevState, crec[0].stateCode)
let totalLimit = crec[0].creditLimit
let totalBalance = crec[0].balance
and increment numAcctsInState
loop through all the remaining credit records: for i=1; i<numRecs; i++ do
has the state code changed — is strcmp (prevState, crec[i].stateCode) != 0 — yes, do
output the prevState, numAcctsInState, totalLimit and totalBalance
// add this state set of totals to the grand totals
grandNumAccts += numAcctsInState
increment grandNumStates
grandTotalLimit += totalLimit
grandTotalBalance += totalBalance
reset the previous state code — strcpy (prevState, crec[i].stateCode)
// set totals back to zero
Structures 599

numAcctsInState = 0;
totalLimit = 0;
totalBalance = 0;
end if state changed
increment numAcctsInState
totalLimit += crec[i].creditLimit
totalBalance += crec[i].balance
end for loop
// output last state’s totals - prevState, numAcctsInState, totalLimit totalBalance
grandNumAccts += numAcctsInState
increment grandNumStates
grandTotalLimit += totalLimit
grandTotalBalance += totalBalance
end if there are any elements
output the dash line
output grandNumStates, grandNumAccts, grandTotalLimit and grandTotalBalance
close outfile

As usual, one should thoroughly desk check the solution before converting it into actual
program coding. The complete program Cs13a is shown below.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Listing Cs13a - Credit Extended by State Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* CS13a - Credit Extended by State report program */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <fstream> *
* 9 #include <iomanip> *
* 10 #include <string> *
* 11 using namespace std; *
* 12 *
* 13 const int MaxNameLen = 31; // maximum length of customer name *
* 14 const int MaxStCodeLen = 3; // maximum length of state code *
* 15 const int MaxNum = 500; // maximum number of credit records*
* 16 *
* 17 struct CREDITREC { *
* 18 long acctno; // account number = ss number *
* 19 char custName[MaxNameLen]; // customer name up to 30 chars *
* 20 char stateCode[MaxStCodeLen];// upper case state code *
* 21 double creditLimit; // credit limit of customer *
* 22 double balance; // customer's current balance due*
* 23 }; *
* 24 *
* 25 int LoadArray (CREDITREC crec[], int limit, char filename[]); *
Structures 600

* 26 void SortArray (CREDITREC crec[], int numRecs); *

* 27 void PrintReport(CREDITREC crec[], int numRecs, char filename[]);*
* 28 *
* 29 int main () { *
* 30 CREDITREC creditRec[MaxNum]; // array of customer credit records*
* 31 int numRecs; // current number of customers *
* 32 *
* 33 // obtain the input and report filenames from the user *
* 34 char inputFilename[_MAX_PATH]; *
* 35 char reportFilename[_MAX_PATH]; *
* 36 cout << "Enter the name of the credit master file: "; *
* 37 cin.getline (inputFilename, _MAX_PATH); *
* 38 cout << "Enter the name of the output report file: "; *
* 39 cin.getline (reportFilename, _MAX_PATH); *
* 40 *
* 41 // load the array of credit records *
* 42 numRecs = LoadArray (creditRec, MaxNum, inputFilename); *
* 43 cout << "\nArray loaded: " << numRecs << " records input\n"; *
* 44 *
* 45 // sort the array into state code order *
* 46 cout << "Sorting the array\n"; *
* 47 SortArray (creditRec, numRecs); *
* 48 *
* 49 // produce the report *
* 50 cout << "Printing the report\n"; *
* 51 PrintReport (creditRec, numRecs, reportFilename); *
* 52 cout << "Finished\n"; *
* 53 return 0; *
* 54 } *
* 55 *
* 56 /***************************************************************/*
* 57 /* */*
* 58 /* LoadArray: loads an array of credit records from the file */*
* 59 /* */*
* 60 /***************************************************************/*
* 61 *
* 62 int LoadArray (CREDITREC crec[], int limit, char filename[]) { *
* 63 ifstream infile (filename); *
* 64 if (!infile) { *
* 65 cerr << "Error: cannot open " << filename << endl; *
* 66 exit (1); *
* 67 } *
* 68 int i = 0; *
* 69 char c; *
* 70 while (i<limit && infile >> crec[i].acctno >> c) { *
* 71 // input the customer name " delimited *
* 72 infile.getline (crec[i].custName, MaxNameLen, '\"'); *
* 73 // input next byte which can be the " if the name is max length*
* 74 infile.get (c); *
* 75 // if it was the " then skip ws *
* 76 infile >> ws; *
* 77 infile.get (crec[i].stateCode, MaxStCodeLen); *
Structures 601

* 78 infile >> crec[i].creditLimit >> crec[i].balance; *

* 79 i++; *
* 80 } *
* 81 // guard against bad input data *
* 82 if (!infile.eof() && infile.fail()) { *
* 83 cerr << "Error: bad data in input file.\nLast good id was "; *
* 84 if ((i-1)==0) *
* 85 cerr << "the first line\n"; *
* 86 else *
* 87 cerr << crec[i-1].acctno << endl; *
* 88 infile.close (); *
* 89 exit (2); *
* 90 } *
* 91 // guard against too much data for array size *
* 92 if (i == limit && infile >> ws && infile.good ()) { *
* 93 cerr << "Error: array size exceeded\n"; *
* 94 infile.close (); *
* 95 exit (3); *
* 96 } *
* 97 infile.close (); *
* 98 return i; // return the number of records inputted *
* 99 } *
* 100 *
* 101 /***************************************************************/*
* 102 /* */*
* 103 /* SortArray: sorts the credit records into state code order */*
* 104 /* */*
* 105 /***************************************************************/*
* 106 *
* 107 void SortArray (CREDITREC crec[], int numRecs) { *
* 108 CREDITREC temp; *
* 109 int i, j; *
* 110 for (i=0; i<numRecs-1; i++) { *
* 111 for (j=i+1; j<numRecs; j++) { *
* 112 if (strcmp (crec[j].stateCode, crec[i].stateCode) < 0) { *
* 113 temp = crec[i]; *
* 114 crec[i] = crec[j]; *
* 115 crec[j] = temp; *
* 116 } *
* 117 } *
* 118 } *
* 119 } *
* 120 *
* 121 /***************************************************************/*
* 122 /* */*
* 123 /* PrintReport: summarize data while printing the report */*
* 124 /* */*
* 125 /***************************************************************/*
* 126 *
* 127 void PrintReport (CREDITREC crec[], int numRecs, *
* 128 char filename[]) { *
* 129 char prevState[MaxStCodeLen]; // previous state code for break*
Structures 602

* 130 int numAcctsInState = 0; // num of accounts in this state*

* 131 double totalLimit = 0; // tot creditlimit in this state*
* 132 double totalBalance = 0; // tot credit extended in state*
* 133 int grandNumStates = 0; // tot number of states serviced*
* 134 int grandNumAccts = 0; // company total num of accounts*
* 135 double grandTotalLimit = 0; // company total credit limit *
* 136 double grandTotalBalance = 0; // company total credit extended*
* 137 *
* 138 // open report file and print headings and column headings *
* 139 ofstream outfile (filename); *
* 140 if (!outfile) { *
* 141 cerr << "Error: cannot open the output file: " << filename *
* 142 << endl; *
* 143 exit (2); *
* 144 } *
* 145 outfile << *
* 146 "Acme Credit Company - Credit Extended by State Report\n\n"*
* 147 << "State Number of Credit Totals Extended\n" *
* 148 << "Code Accounts Limit Balance\n"; *
* 149 outfile << fixed <<setprecision (2); *
* 152 *
* 153 // guard against no records in the array *
* 154 if (numRecs) { *
* 155 // initialize the previous state code to the first record *
* 156 // and initialize the initial state totals *
* 157 strcpy_s (prevState, sizeof(prevState), crec[0].stateCode); *
* 158 totalLimit = crec[0].creditLimit; *
* 159 totalBalance = crec[0].balance; *
* 160 numAcctsInState++; *
* 161 *
* 162 // loop through all the credit records *
* 163 for (int i=1; i<numRecs; i++) { *
* 164 // check for a change in state code *
* 165 if (strcmp (prevState, crec[i].stateCode) != 0) { *
* 166 // this state is the next state, print totals for prev state *
* 167 outfile << setw (4) << prevState *
* 168 << setw (9) << numAcctsInState *
* 169 << setw (7) << "$" << setw (9) << totalLimit *
* 170 << setw (6) << "$" << setw (9) << totalBalance *
* 171 << endl; *
* 172 // roll previous state totals into grand company totals *
* 173 grandNumAccts += numAcctsInState; *
* 174 grandNumStates++; *
* 175 grandTotalLimit += totalLimit; *
* 176 grandTotalBalance += totalBalance; *
* 177 // reset previous state code to the new one *
* 178 strcpy_s (prevState, sizeof(prevState), crec[i].stateCode); *
* 179 // reset this state totals to 0 *
* 180 numAcctsInState = 0; *
* 181 totalLimit = 0; *
* 182 totalBalance = 0; *
* 183 } *
Structures 603

* 184 // accumulate this state values *

* 185 numAcctsInState++; *
* 186 totalLimit += crec[i].creditLimit; *
* 187 totalBalance += crec[i].balance; *
* 188 } *
* 189 // print last state in the set of data *
* 190 outfile << setw (4) << prevState *
* 191 << setw (9) << numAcctsInState *
* 192 << setw (7) << "$" << setw (9) << totalLimit *
* 193 << setw (6) << "$" << setw (9) << totalBalance << endl;*
* 194 grandNumAccts += numAcctsInState; *
* 195 grandNumStates++; *
* 196 grandTotalLimit += totalLimit; *
* 197 grandTotalBalance += totalBalance; *
* 198 } *
* 199 // print the grand company totals *
* 200 outfile << " -- --- --------- ---------\n" *
* 201 << setw (4) << grandNumStates *
* 202 << setw (9) << grandNumAccts *
* 203 << setw (7) << "$" << setw (9) << grandTotalLimit *
* 204 << setw (6) << "$" << setw (9) << grandTotalBalance *
* 205 << endl; *
* 206 outfile.close (); *
* 207 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Cs13-2 — Writing a Binary File

Let’s examine how a binary file is created in the first place and then in the next example see how
it can be read back into a program. For this program, we deal with Student Course Records. At
the end of each semester, student grades are submitted. Each line of the input file contains the
student’s id number (social security number), the course (up to five characters), course number
(three characters), section (one or two characters) and the letter grade earned. The binary master
file that is output must contain these fields plus a float for the grade points earned (based on the
4.0 scale). An A earns 4.0 points, a B earns 3, and so on. So first, let’s create the structure to
contain these fields. The structure STUDREC that meets these requirements is defined as
follows.
const int CourseLen = 5;
const int CourseNumLen = 4;
const int SectionLen = 3;

struct STUDREC {
long ssno;
char course[CourseLen];
char courseNum[CourseNumLen];
char section[SectionLen];
char grade;
Structures 604

float gp;
};

The design is very simple. Whenever you need to input a set of data into a structure
instance, it is usually very convenient to create a GetData() function that is passed a reference to
the structure instance to fill up and a reference to the input stream to use, returning a reference to
that stream back to the caller. Since this is the only function needed in this program, I have
omitted the Top-Down Design drawing. The main storage diagram is also very tiny, containing
just a single instance of the structure to be filled. I have omitted the main storage diagram as
well.

The main processing loop is quite small. As long as the GetData() function is able to
input another student record, the grade points earned for that course are figured and the binary
record written to disk. Since the actual processing steps are so simple, I have omitted the pseudo
coding also. Here are the key lines to look at. The items dealing with the binary file are shown in
boldface.
STUDREC srec;
ofstream outfile ("Master.dat", ios::out | ios::binary);
while (GetData (infile, srec)) {
count++;
switch (srec.grade) { // figure grade points earned
...
outfile.write ((char*) &srec, sizeof (srec));
}

Here is the complete listing of Cs13b.

+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Listing of Cs13b Student Records - build a binary master file *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs13b Student Records - build a binary master file */*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <fstream> *
* 9 using namespace std; *
* 10 const int CourseLen = 6; *
* 11 const int CourseNumLen = 4; *
* 12 const int SectionLen = 3; *
* 13 *
* 14 struct STUDREC { *
* 15 long ssno; // student id number *
* 16 char course[CourseLen]; // course taken: CSMPS *
* 17 char courseNum[CourseNumLen]; // course number: 125 *
* 18 char section[SectionLen]; // course section: AA *
* 19 char grade; // letter grade received *
Structures 605

* 20 float gp; // grade point received-4.0 system*

* 21 }; *
* 22 *
* 23 istream& GetData (istream& infile, STUDREC& srec); *
* 24 *
* 25 int main () { *
* 26 STUDREC srec; *
* 27 *
* 28 ifstream infile ("Cs13b-student-records.txt"); *
* 30 if (!infile) { *
* 31 cerr << "Error: cannot open file Cs13b-student-records.txt\n"; *
* 32 return 1; *
* 33 } *
* 34 ofstream outfile ("Master.dat", ios::out | ios::binary); *
* 35 *
* 36 int count = 0; *
* 37 while (GetData (infile, srec)) { *
* 38 count++; *
* 39 switch (srec.grade) { // figure grade points earned *
* 40 case 'A': *
* 41 srec.gp = 4; *
* 42 break; *
* 43 case 'B': *
* 44 srec.gp = 3; *
* 45 break; *
* 46 case 'C': *
* 47 srec.gp = 2; *
* 48 break; *
* 49 case 'D': *
* 50 srec.gp = 1; *
* 51 break; *
* 52 default: *
* 53 srec.gp = 0; *
* 54 }; *
* 55 outfile.write ((char*) &srec, sizeof (srec)); // binary write *
* 56 } *
* 57 infile.close (); *
* 58 outfile.close (); *
* 59 cout << count << " Student records written to the masterfile\n";*
* 60 *
* 61 return 0; *
* 62 } *
* 63 *
* 64 /***************************************************************/*
* 65 /* */*
* 66 /* GetData: input a student grade record */*
* 67 /* */*
* 68 /***************************************************************/*
* 69 *
* 70 istream& GetData (istream& infile, STUDREC& srec) { *
* 71 infile >> srec.ssno >> ws; *
* 72 if (!infile) return infile; *
Structures 606

* 73 infile.get (srec.course, sizeof (srec.course)); *

* 74 infile >> ws; *
* 75 infile.get (srec.courseNum, sizeof (srec.courseNum)); *
* 76 infile >> ws; *
* 77 infile.get (srec.section, sizeof (srec.section)); *
* 78 infile >> srec.grade; *
* 79 return infile; *
* 80 } *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Cs13-3 — Reading a Binary File —

Building a Consolidated Student GPA Binary File
Next, let’s input the binary file of student course records just built with Program Cs13-2,
accumulate the data for each student and write a new binary file of GPA records. The student
course records from the previous program are, in fact, in order by student id (social security
number). For each student, accumulate their grade points and then calculate their GPA. The GPA
record has just two members, the student id (social security number) and the overall GPA.

Actually, this problem is a control break problem. Input and accumulate a student’s grade
points until there is a change in social security number. The ControlBreak() function is called
when there is a change in social security numbers. The break function fills in the GPA record,
writes that record to disk, formats a brief log message, and resets the counter and sum fields to
zero, ready for the next student. Don’t forget to call the break function one more time when the
main loop ends. Remember the program has been accumulating the grade points earned for the
last student; that last student’s GPA data must be written to disk as well.

The GPA structure is defined as

struct GPAREC {
long ssno; // student id number
float gpa; // grade point received - 4.0 system
};
As with all control break problems, the previous value of the control field, here the social
security number, must be initialized with the value from the first student course record. Thus, we
must use a primed loop approach and input the first record and give the previous field its value.

Let’s begin by examining the main() function’s processing. The key variables are
STUDREC srec; // student course record
int count; // count of courses of a student
float sum; // sum of gp of a student
long previousSsno; // the previous ssno

Next, main() reads in the first binary record and sets previousSsno variable and the
accumulators.
Structures 607

infile.read ((char*) &srec, sizeof (srec));

previousSsno = srec.ssno;
sum = srec.gp;
count = 1;

After the while clause successfully reads in the next student record, the first action is to
check and see if we are still working with the same student. If so, accumulate grade point
information. If not, then the ControlBreak() function is called to generate and output the student
GPA record. The previousSsno is reset to this new student.
while (infile.read ((char*) &srec, sizeof (srec))) {
if (previousSsno != srec.ssno) {
ControlBreak (outfile, previousSsno, sum, count, logfile);
previousSsno = srec.ssno;
}
sum += srec.gp;
count++;
}
ControlBreak (outfile, previousSsno, sum, count, logfile);
When the main loop ends, the ControlBreak() function is invoked one last time for that last
student.

Control break logic always follows a quite standard sequence. First, roll (or add) any
totals into any grand totals (there are none in this problem). Next, output this set of totals; here
we calculate the GPA and fill in the gpaRec structure instance with this student’s data and write
it to the new binary master file. Next, the break function would zero the totals and counters.
ControlBreak() is given outfile, ssno, sum and count. Its sequence is just this.
GPAREC gpaRec;
gpaRec.ssno = ssno;
gpaRec.gpa = sum / count;
outfile.write ((char*) &gpaRec, sizeof (gpaRec));
count = 0;
sum = 0;

Now we have a new binary file of student GPA values. The next step is to find out how
an inquiry program can directly access a specific student’s GPA record on disk. (Alas, you will
have to get my sequel book to see this.) Here are the complete listing for Cs13c and a sample test
run.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Listing of Cs13c Student GPA Builder Program - reads a binary file *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Cs13c Student GPA Builder Program- reads binary file of data*/*
* 4 /* writes another binary file of student gpa records */*
* 5 /* */*
* 6 /***************************************************************/*
* 7 *
Structures 608

* 8 #include <iostream> *
* 9 #include <iomanip> *
* 10 #include <fstream> *
* 11 using namespace std; *
* 12 const int CourseLen = 6; *
* 13 const int CourseNumLen = 4; *
* 14 const int SectionLen = 3; *
* 15 *
* 16 struct STUDREC { *
* 17 long ssno; // student id number *
* 18 char course[CourseLen]; // course taken: CSMPS *
* 19 char courseNum[CourseNumLen]; // course number: 125 *
* 20 char section[SectionLen]; // course section: AA *
* 21 char grade; // letter grade received *
* 22 float gp; // grade point received-4.0 system*
* 23 }; *
* 24 *
* 25 struct GPAREC { *
* 26 long ssno; // student id number *
* 27 float gpa; // grade point received-4.0 system*
* 28 }; *
* 29 *
* 30 void ControlBreak (ostream& outfile, long ssno, float& sum, *
* 31 int& count, ostream& logfile); *
* 32 *
* 33 int main () { *
* 34 ifstream infile ("Cs13c-StudentRecords.dat", *
* 35 ios::in | ios::binary); *
* 36 if (!infile) { *
* 37 cerr << "Error: cannot open file Cs13c-StudentRecords.dat\n"; *
* 38 return 1; *
* 39 } *
* 40 *
* 41 STUDREC srec; // student course record *
* 42 int count; // count of courses of a student *
* 43 float sum; // sum of gp of a student *
* 44 long previousSsno; // the previous ssno *
* 45 int totalStudents = 0; // the total number of students *
* 46 *
* 47 // read first record and initialize counters and save areas *
* 48 infile.read ((char*) &srec, sizeof (srec)); *
* 49 if (!infile) { *
* 50 cerr << "Masterfile is empty.\n"; *
* 51 infile.close (); *
* 52 return 1; *
* 53 } *
* 54 previousSsno = srec.ssno; *
* 55 sum = srec.gp; *
* 56 count = 1; *
* 57 *
* 58 // open output files *
* 59 ofstream outfile ("GPA-Master.dat", ios::out | ios::binary); *
Structures 609

* 60 ofstream logfile ("LogResults.txt"); *

* 61 // setup floating point output format *
* 62 logfile << fixed << setprecision (2); *
* 65 // display headings on the log file *
* 66 logfile << " GPA Log File\n\n Student ID GPA\n\n"; *
* 67 *
* 68 // process all student course records *
* 69 while (infile.read ((char*) &srec, sizeof (srec))) { *
* 70 // check for a change in student id (ssno) *
* 71 if (previousSsno != srec.ssno) { *
* 72 // finished with this student, so output their GPA record *
* 73 ControlBreak (outfile, previousSsno, sum, count, logfile); *
* 74 previousSsno = srec.ssno; *
* 75 totalStudents++; *
* 76 } *
* 77 // accumulate student grade points *
* 78 sum += srec.gp; *
* 79 count++; *
* 80 } *
* 81 // process last student being accumulated *
* 82 ControlBreak (outfile, previousSsno, sum, count, logfile); *
* 83 totalStudents++; *
* 84 *
* 85 infile.close (); *
* 86 outfile.close (); *
* 87 logfile << endl << totalStudents *
* 88 << " Student gpa records written\n"; *
* 89 logfile.close (); *
* 90 return 0; *
* 91 } *
* 92 *
* 93 /***************************************************************/*
* 94 /* */*
* 95 /* ControlBreak: fill in the GPA record and write it to master */*
* 96 /* and log files */*
* 97 /* */*
* 98 /***************************************************************/*
* 99 *
* 100 void ControlBreak (ostream& outfile, long ssno, float& sum, *
* 101 int& count, ostream& logfile) { *
* 102 GPAREC gpaRec; *
* 103 *
* 104 // fillup gpaRec *
* 105 gpaRec.ssno = ssno; *
* 106 gpaRec.gpa = sum / count; *
* 107 *
* 108 // write the binary master file *
* 109 outfile.write ((char*) &gpaRec, sizeof (gpaRec)); *
* 110 *
* 111 // write log information *
* 112 logfile << setw (15) << ssno << setw (6) << gpaRec.gpa << endl; *
* 113 *
Structures 610

* 114 // reset counters for next student *

* 115 count = 0; *
* 116 sum = 0; *
* 117 } *
* 118 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Logfile for Cs13c Student GPA Builder Program *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 GPA Log File *
* 2 *
* 3 Student ID GPA *
* 4 *
* 5 111111111 4.00 *
* 6 222222222 3.00 *
* 7 333333333 2.00 *
* 8 444444444 3.00 *
* 9 555555555 2.50 *
* 10 666666666 2.00 *
* 11 *
* 12 6 Student gpa records written *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-

Section C: An Engineering Example

Engr13a — Weather Statistics Revisited
The major impact that structures have on engineering problems is primarily that of consolidating
related fields. Weather statistics are far more involved than just merely recording hourly
temperatures. A structure allows us to group all the related fields into one simple package and
then to be able to pass a reference to the package to various statistical functions.

Assume that the weather stations are logging weather events on an hourly basis. For each
feature measured, there are going to be twenty-four such observations per record. The following
defines an advanced daily weather record for just one observing site within a state.
const int StationLen = 21;
const int NumObs = 24;
struct WEATHER_EVENTS {
char station[StationLen];
float temps[NumObs];
float rainfall[NumObs];
float pressure[NumObs];
float relHumidity[NumObs];
float windSpeed[NumObs];
char windDir[NumObs][3];
};
Structures 611

Notice that the WEATHER_EVENTS record contains one string, five arrays of twenty-
four elements each and one two-dimensional array of characters (in other words, a single-
dimensioned array of strings). This WEATHER_EVENTS record represents 145 separate
fields, occupying at least 573 bytes! Clearly, structures offer engineering programming a great
value. In fact, if the data was stored in binary format, one read statement is all that is needed to
input the entire set of fields!

This sample program is designed to show you how convenient arrays of structures can be.
Let’s modify Program Engr11a that produced a report of unusual weather conditions. If you look
back at Program Engr11a, the main() function is a large wall of coding. The objective is to easily
break this problem down into smaller units by using an array of weather structures. The
WEATHER structure is defined as
const int MaxCityLen = 21; // city name length is 20 chars
const int MaxRecs = 100; // max number of weather records

struct WEATHER {
char city [MaxCityLen]; // string to hold city name
float high; // high temperature of the day - F
float low; // low temperature of the day - F
float rainfall; // rainfall in inches
float snowfall; // snowfall in inches
float windspeed; // wind speed in mph
};

The main() function defines an array of these weather records. After prompting the user
to input the filenames to use both for input and for output, main() calls the LoadArray()
function to input all the weather records into the array. Then, main() calls the PrintReport()
function to print the actual report. Now main() is very streamlined and we have encapsulated the
input and output operations in a pair of functions.

In the LoadArray() function, the key loop to input all the data is this.
while (i < limit && infile >> junk) {
// input of junk retrieved the leading " of city string
infile.get (wrec[i].city, sizeof (wrec[i].city), '\"');
infile.get (junk);
infile >> wrec[i].high >> wrec[i].low >> wrec[i].rainfall
>> wrec[i].snowfall >> wrec[i].windspeed;
i++;
}

Notice the syntax to access the individual members of the ith element of the wrec array is
wrec[i].city, for example.

Here are the complete coding for Engr13a and the test run.
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
Structures 612

* Engr13a: Unusual Weather Statistics report using a structure *

/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 /***************************************************************/*
* 2 /* */*
* 3 /* Engr13a: Unusual Weather Statistics report using a structure*/*
* 4 /* */*
* 5 /***************************************************************/*
* 6 *
* 7 #include <iostream> *
* 8 #include <iomanip> *
* 9 #include <fstream> *
* 10 #include <string> *
* 11 using namespace std; *
* 12 *
* 13 const int MaxCityLen = 21; // city name length is 20 chars *
* 14 const int MaxRecs = 100; // maximum number of weather records *
* 15 *
* 16 struct WEATHER { *
* 17 char city [MaxCityLen]; // string to hold city name *
* 18 float high; // high temperature of the day - F *
* 19 float low; // low temperature of the day - F *
* 20 float rainfall; // rainfall in inches *
* 21 float snowfall; // snowfall in inches *
* 22 float windspeed; // wind speed in mph *
* 23 }; *
* 24 *
* 25 int LoadArray (char filename[], WEATHER wrec[], int limit); *
* 26 void PrintReport (char filename[], WEATHER wrec[], int limit); *
* 27 *
* 28 int main () { *
* 29 *
* 30 // prompt user for filenames for input and output *
* 31 char infilename[_MAX_PATH]; *
* 32 char reportname[_MAX_PATH]; *
* 33 cout << "Enter the filename with today's weather data\n"; *
* 34 cin.getline (infilename, sizeof (infilename)); *
* 35 cout << "\nEnter the report filename\n"; *
* 36 cin.getline (reportname, sizeof(reportname)); *
* 37 *
* 38 WEATHER wrec[MaxRecs]; *
* 39 int numRecs = LoadArray (infilename, wrec, MaxRecs); *
* 40 *
* 41 PrintReport (reportname, wrec, numRecs); *
* 42 return 0; *
* 43 } *
* 44 *
* 45 /***************************************************************/*
* 46 /* */*
* 47 /* LoadArray: loads an array of weather records */*
* 48 /* */*
* 49 /***************************************************************/*
* 50 *
Structures 613

* 51 int LoadArray (char filename[], WEATHER wrec[], int limit) { *

* 52 char junk; // to hold the " around city names *
* 53 int i = 0; *
* 54 ifstream infile; *
* 55 infile.open (filename); *
* 56 if (!infile) { *
* 57 cerr << "Error: cannot open file: " << filename << endl; *
* 58 exit (1); *
* 59 } *
* 60 *
* 61 while (i < limit && infile >> junk) { // input leading " of city*
* 62 infile.get (wrec[i].city, sizeof (wrec[i].city), '\"'); *
* 63 infile.get (junk); *
* 64 infile >> wrec[i].high >> wrec[i].low >> wrec[i].rainfall *
* 65 >> wrec[i].snowfall >> wrec[i].windspeed; *
* 66 *
* 67 // abort if there is incomplete or bad data *
* 68 if (!infile) { *
* 69 cerr << "Error: incomplete city data on line " << i+1 << endl;*
* 70 infile.close (); *
* 71 exit (2); *
* 72 } *
* 73 *
* 74 i++; *
* 75 } *
* 76 infile.close (); *
* 77 return i; *
* 78 } *
* 79 /***************************************************************/*
* 80 /* */*
* 81 /* PrintReport: prints a report of unusual weather */*
* 82 /* */*
* 83 /***************************************************************/*
* 84 *
* 85 void PrintReport (char filename[], WEATHER wrec[], int limit) { *
* 86 ofstream outfile; *
* 87 outfile.open (filename); *
* 88 if (!outfile) { *
* 89 cerr << "Error: cannot open file: " << filename << endl; *
* 90 exit (3); *
* 91 } *
* 92 // setup floating point output format *
* 93 outfile << fixed << setprecision (1); *
* 96 *
* 97 outfile << "Unusual Weather Report\n\n"; *
* 98 outfile<<"City High Low Rain Snow"*
* 99 " Wind\n"; *
* 100 outfile << *
* 101 " Fall Fall"*
* 102 " Speed\n\n"; *
* 103 *
* 104 for (int i=0; i<limit; i++) { *
Structures 614

* 105 if (wrec[i].high > 95 || wrec[i].low < 0 || *

* 106 wrec[i].rainfall > 2 || wrec[i].snowfall > 6 || *
* 107 wrec[i].windspeed > 45) { *
* 108 // unusual weather - display this city data *
* 109 outfile << left << setw (22) << wrec[i].city << right *
* 112 << setw (7) << wrec[i].high; *
* 113 if (wrec[i].high > 95) *
* 114 outfile << '*'; *
* 115 else *
* 116 outfile << ' '; *
* 117 outfile << setw (7) << wrec[i].low; *
* 118 if (wrec[i].low < 0) *
* 119 outfile << '*'; *
* 120 else *
* 121 outfile << ' '; *
* 122 outfile << setw (7) << wrec[i].rainfall; *
* 123 if (wrec[i].rainfall > 2) *
* 124 outfile << '*'; *
* 125 else *
* 126 outfile << ' '; *
* 127 outfile << setw (7) << wrec[i].snowfall; *
* 128 if (wrec[i].snowfall > 6) *
* 129 outfile << '*'; *
* 130 else *
* 131 outfile << ' '; *
* 132 outfile << setw (7) << wrec[i].windspeed; *
* 133 if (wrec[i].windspeed > 45) *
* 134 outfile << '*'; *
* 135 else *
* 136 outfile << ' '; *
* 137 outfile << endl; *
* 138 } *
* 139 } *
* 140 outfile.close (); *
* 141 *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
+))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))),
* Output from Engr13a: Unusual Weather Statistics report using struct *
/)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))1
* 1 Unusual Weather Report *
* 2 *
* 3 City High Low Rain Snow Wind *
* 4 Fall Fall Speed *
* 5 *
* 6 Washington 99.0* 70.0 0.0 0.0 20.0 *
* 7 Morton 85.0 65.0 5.0* 0.0 40.0 *
* 8 Chicago 32.0 -5.0* 0.0 8.0* 25.0 *
* 9 Joliet 88.0 70.0 2.0 0.0 60.0* *
* 10 Springfield 99.0* 75.0 3.0* 0.0 55.0* *
* 11 New Salem 0.0 -3.0* 0.0 9.0* 55.0* *
.)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))-
Structures 615

Design Exercises
1. Airline Scheduling Program

Acme Airline wants a new program to track their arrival and departure schedules. Create a
DATE structure to contain a date that consists of three numbers: month, day and year. Create a
TIME structure to contain a time that consists of two numbers: hours and minutes. All times are
on a 24-hour basis; that is, 10:00 p.m. would be entered as 22:00. Next, create a FLIGHT
structure that contains the departure date and time, the arrival date and time, the character string
flight number, the maximum number of passengers and the current number of passengers. The
arrival and departure dates and times should be instances of your DATE and TIME structures.
The flight number is a string of 10 characters maximum including the null-terminator.

Now write the sketch for the program that inputs a set of data on flights, stores them in an
array of FLIGHT structures. Then, print a report of the flights as entered. Each line of the input
file consists of the flight number, departure date and time, arrival date and time, maximum
passengers and current number of passengers.

2. Sports Event Reporting

For the time trails at a major sporting event, a series of programs needs to be designed and made
to track the results of the contestants. The judges desire to store the contestant’s three digit id
number, the contestant name of up to 30 characters, and the start and stop times. A time is input
as hh:mm:ss.ssss. First design a TIME structure to store the three portions of a time. Then,
design an EVENTREC structure to store the basic information for one contestant. There is one
additional field in the EVENTREC, the total elapsed time in seconds.

Now sketch the Data Entry program that inputs EVENTREC data and writes a binary
master file of the contestant data. It must calculate the total elapsed time in seconds.

Now sketch an Event Summary Program that inputs the binary master file into an array of
EVENTREC structures. The array should then be sorted into increasing total elapsed time order.
The printed report displays the contestant id number, name and total elapsed time.
Structures 616

Stop! Do These Exercises Before Programming

1. Acme wishes to make a database of all of its clients’ TV preferences. A TV preference
consists of the short int channel number, two short ints for the hour and minute the show is
broadcast but are stored in 24-hour format (that is, 8pm is stored as 20 hours), a short int for the
day of the week, and the name of the show which can hold thirty characters. The programmer
coded the following which does not compile among other goofs. Correct this sequence so that it
compiles and meets the specifications above.
Structure PREFERENCE {
short channel
short day
short hr
short min
char name[MAXLEN]
}
const int MAXLEN = 30;

2. Just to test the structure coding, the programmer decided to see if he could input one set of
preferences and then display it. However, his attempt met with complete failure and does not
compile. Why? What would have to be done to make this work?
int main () {
cin >> PREFERENCE.channel >> PREFERENCE.day
>> PREFERENCE.hr >> PREFERENCE.min
>> PREFERENCE.name;
cout << PREFERENCE.channel << PREFERENCE.day
<< PREFERENCE.hr << PREFERENCE.min
<< PREFERENCE.name << endl;

3. The problem specifications called for creating a LoadPreferences() function. Here the
programmer ran into a great deal of difficulty. What is wrong with this function? How can it be
fixed?
void LoadPreferences (PREFERENCE rec, int limit,
istream& infile) {
for (int j=0; j<limit && infile >> ws; j++) {
infile >> rec[j].channel >> rec[j].day >> rec[j].hr
>> rec[j].min >>rec[j].name;
}
}
Here are two typical data lines.
11 6 08:00pm Babylon 5
12 5 07:30pm Doctor Who
Structures 617

4. Having gotten the array loaded, the programmer decided to print what was in the array to be
sure the data had been entered correctly. However, it does not compile. How can it be made to
work properly?
int main () {
PREFERENCE recs[1000];
int numRecs;
for (int j=0; j<numRecs; j++) {
cout << recs.channel << setw (5) << recs.day << " "
<< setfill ('0') << setw (2) << recs.hr << ':'
<< recs.min << setfill (' ') << setw (40)
<< recs.name << endl;
}

5. Next, management wanted the entire array sorted into order based on the TV show’s name so
that the program could then calculate frequencies of each TV show. The sortArray() function
does not work properly, though it compiles. What is wrong and how can it be fixed?
void sortArray (PREFERENCE recs[], int numRecs) {
PREFERENCES temp;
for (int j=0; j<numRecs; j++) {
for (int k=j; k<numRecs; k++) {
if (recs[k].name < recs[j].name) {
temp = recs[k];
recs[k] = recs[j];
recs[j] = temp;
}
}
}
}
Structures 618

Programming Problems
Problem Cs13-1 — Acme Payroll Program

The Acme Company has a master file of employee data that is already sorted by employee id
number. Their employees work on various job sites around the city. At the end of the pay period,
a transactions file is created that contains their hours worked. Our task is to calculate their pay.

Some employees are classified as time-card workers and get paid an hourly rate. Some are
classified as salaried or management and are paid a fixed rate independent of hours worked.
Define an enumerated data type, PayType, to reflect these two kinds, Hourly or Salaried.

Next, define an EMPLOYEE structure to hold the following fields in this order:
int id number
char name of maximum length of 20 chars +1 for the null-terminator
double pay rate
short int number of dependents
enum PayType which can be Salaried or Hourly

In the main() function, create an array of these employee records and allow for a
maximum of 500 employees. The main() function should first call the LoadMaster() function to
load all of the employee master records into the array.

The LoadMaster() function is passed the EMPLOYEE array and the array bounds. It
inputs the employee master file from a file called Cs13-1-master.txt. The function returns the
number of employee records input. In the input file, the names are padded to twenty characters
with blanks as needed. The payType input field contains an H for Hourly workers and S for
Salaried workers.

Acme employees work on various job sites around the city. The second input file, Cs13-
1-jobsites.txt, contains the job site number and its corresponding name sorted into site number
order. Create another structure called JOBSITE that contains these members.
int job site number
char sitename string of a maximum 20 chars plus 1 for the null-terminator

With the array loaded, the main() function should next define an array of 500 JOBSITE
structures and an integer count of the number in the array. The main() function then calls the
function LoadSites() to get that array filled. The LoadSites() function is passed the array of job
sites and the maximum number. It returns the number of job sites actually input from the file.
Structures 619

With these initial setup actions done, main() is now ready to process the transactions file.
The third file, Cs13-1-transactions.txt, contains the pay period information. Each line contains
the following fields
int employee id number
int job site number
double hours worked
The transactions file is sorted into employee number order.

Your task in the main() function is to open this file and write a loop that inputs each
employee and calculates their pay. The report is written to output file pay.txt. The report consists
of one line per valid employee showing the id number, name, job site name, gross pay and net
pay. Hence, two table look-ups must be done: one to obtain the employee’s name and the other to
get the job site name.

Create a pair of functions to do the table look ups: MatchEmployee() and MatchSite().
Each should return the index (integer) of the corresponding EMPLOYEE or JOBSITE structure
for this employee’s id or site number. Both functions must return a –1 for no matching record
(invalid id or site number). If either the employee’s id or the site number are invalid, then the
main() function should print an error message to the error log file called errors.txt. It should
print either
Error bad Id: then the employee’s transaction record information
or
Error bad site Id: then the employee’s transaction record information

An employee’s pay is more complex. For Salaried workers, gross pay is the amount
contained in the employee record’s pay rate. For Hourly workers, the pay rate should be
multiplied by the hours worked. However, they also get time and a half for all hours worked
more than 40.

The net pay for any employee is gross pay minus tax. Here are the taxation rules:
if the number of dependents is 1, the tax rate is 15%
else the tax is the larger of these two amounts
gross pay x 2.5%
gross pay x 15% x (1 – number dependents/(number dependents – 6))

Finally, the last line of this report should list the number of employees paid and the total
gross payroll.
Structures 620

Problem Cs13-2 — Friends’ Phone Number Database

The objective of this program is to aid you in maintaining a file of friends’ phone numbers. The
phone numbers contain three separate parts. For example, we are used to seeing phone numbers
as (309) 699-9999. However, the program must store the phone number as three separate
numbers: the area code (309), the prefix (699) and the main number (9999). Make a structure
template called FRIEND that contains the following fields in this order.
Last name — up to 20 characters maximum + 1 for the null terminator
First name — up to 15 characters maximum + 1 for the null terminator
area code stored as a short
phone number prefix stored as a short
phone number main stored as a short
The main() function defines an array called friends; allow for a maximum of 200 entries.

The main() function begins by calling a LoadFriends() function that is passed the array
and the maximum limit. LoadFriends() returns the number of friends actually input. In
LoadFriends(), input all the data from the file friends.txt. Initially, the file does not exist, so no
friend records are input and the function returns 0 elements are in the array.

With the array loaded, main() presents a short menu of options, similar to the following
Enter the number of your choice
1. Add new friend
2. Update a friend
3. Save all the data
4. Make a friends listing in Last Name order
5. Make a friends listing in Area Code order
6. Make a friends listing in Prefix order
7. Quit the program
Enter your choice: _
Now get the user’s choice.

The friends array is always saved on disk sorted into alphabetical order on last name and
then first name, if there are two friends with the same last name. When the friends array is
actually in memory, the array can be sorted into another order.

When the choice is “Add new friend,” prompt the user for the needed information and
input the data. Then, place that data into the array maintaining the correct alphabetical order.

When the choice is “Update a friend,” prompt the user for the last and first names.
Perform a match on last and first names to find that friend in the data base. If the entry is found,
display the phone number three fields. Prompt and input the changes. Store the updated
information in that friend’s record.
Structures 621

When “Save” is selected, rewrite the friends.txt file from the friends array. Make sure
that the array is sorted into order by last name and then first name, if two entries have the same
last name.

For the three “Make a listing” actions, sort the array into the correct order. Sorting by area
code means sort on area code first, but, when duplicate area codes occur, also sort on last name
and then first name if the last names are the same. Sorting by prefix means to first sort on the
prefix, but, when two prefixes are the same, use the main number portion next. If two main
numbers are the same, do anything you desire. Note that the Sort on Name function must be done
before a Save, Add or an Update operation can be done. To avoid needless sorting, have the
main() program maintain a bool, isInNameOrder. It is true when the original input file is
loaded. It is changed to false when any of the numerical sorts are done. It is reset back to true
when the Sort on Name is subsequently done.

When the user chooses any of the Make a Friends Listing, prompt the user for the name
of the output report file. Then sort the array as needed before producing the report.

Problem Cs13-3 — Your Pet’s Rabies Shot Is Due

Acme Animal Control maintains a database of all registered pets in the city. Each month, they
send out a notice to all pet owners reminding them that the pet’s rabies shot is due this month,
but only of course if that pet is due for its rabies shot that month. Rabies shots are given yearly.

Define a structure template, PETREC, that contains the needed fields as follows. Be sure
to add one to each of the character strings to get the needed array bounds. Each listed character
string length represents the maximum number of characters it could hold.
Pet name — 10 characters
Owner’s name — 40 characters
Address — 30 characters
City — 20 characters
State — 2 characters
Zip — 5 characters
Year of last vaccination
Month of last vaccination

The main() function defines an array of 100 pet records and calls the LoadPets()
function. As usual, LoadPets() is passed the array of pet records and the array limit. It inputs the
pet records in the file and returns the number actually input. Use the data file Cs13-3-pets.txt.

Next, the main() function prompts the user to enter the year and the month for the
reminders. Given that reminder date, the main() function then checks each record in the array to
see if that pet is due for its yearly rabies shot. If a pet is due, main calls PrintReminder() passing
Structures 622

it a constant reference to the pet record and the output file stream on which to display the
reminder.

PrintReminder() should display the reminder notice formatted as follows. (Assume that
10/2000 was the due date that was entered.)
Acme Animal Control
Spot is due for its annual rabies shot this month (10/2000).
John Smith
123 Somerville Ave
North Plains, TX 59685
Use the pet’s name where I have used Spot. Insert the due date where I have 10/2000. Insert the
owner’s name, address, city, state and zip. Print each reminder notice on a new page. (Remember
that the escape sequence \f causes the printer to eject to a new page.)

Problem Cs13-4 — Acme Grocery Store Sales Analysis

The Acme Grocery Store has inventory records that consist of these fields.
unsigned long productId;
char description[21];
unsigned short quantityOnHand;
double cost;
Create an INV_REC structure to contain these four fields. In the main() function, create an array
of a maximum of 100 INV_REC records. Then call a function, loadArray(), to input a file with
up to 100 records in it. Use the provided file called cs13-4-master.txt.

Next, call a function, findMin(), that returns the subscript of the record that costs the
least. Then, call a function, findMax(), that returns the subscript of the record that costs the
most. Save these subscripts for later use in the final report.

Next, the main() function prompts the user to enter the mark up percentage, such as 20.0,
for 20%. For testing purposes, use 20.0 for the mark up percentage.

Now print a report similar to the one shown below. There are a title line and two column
heading lines.
Acme Grocery Store Sales Analysis: Mark up Percentage is 20%

Product Description Quantity Wholesale Retail Total

ID Cost Cost Profit

9999999 Potatoes 10 $ 6.00 $ 7.20 $ 12.00

The retail cost is the inputted cost that has been marked up by the mark up percentage. The profit
is the difference between the retail and wholesale costs multiplied by the quantity.
Structures 623

Print a final summary set of lines. The first line shows the total expected profit if
everything is sold. Follow that with two lines as follows:
The least expensive item is: nnnnnnnn $ nn.nn
The most expensive item is: mmmmmmmm $ nn.nn

Finally, again prompt the user for another mark up percentage and repeat the report using
the new percentage. Continue until the user signals EOF by a ctrl-Z code. Begin each report on a
new page.

Problem Engr13-1 — Problem Engr07-2 Revisited

(Electrical Circuit Design)
In Problem Engr07-2, an electronic circuit involving a resistor, capacitor and an inductor was
designed according to the formula

The resistance required was found using the bisection method. However, the problem
with the F(R) function was the remaining variables, L, C, t and q/q0 . Those four values were
input from the user in a function called Input(). However, they are actually used in F(R). This
was solved using four global variables.

A major use of structures is reducing the number of parameters needed to be passed or

made available to functions. Create a structure to contain those four values. Then, create a global
instance of that structure. Remove the other original four global variables. Rewrite the Input()
function to fill up the global instance. Rewrite the F(R) function to use this global instance.
Verify that all is working properly by running the same four test cases in the original problem.

Problem Engr13-2 — Problem Engr07-1 Revisited — Vibration

Studies
In Problem Engr07-1, various root solving methods were used to find the first three roots of the
spring dampening function. The variables input by the user consisted of the car’s mass and the
depth of the pothole. Those were made global variables to facilitate communications to the F(t)
function.

Remove those global variables. Create a structure whose two members are the mass of the
car and the pot hole depth. Then, create a global instance of the structure. Change the program to
Structures 624

input the user’s choices into the global instance. Modify the F(t) function to also access the two
data members from the global instance. Then test the program in the same manner as given in
Engr07-1.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 625

Appendix A: How to Use Microsoft’s Visual Studio

.NET 2005 Compiler
Microsoft’s Visual Studio .NET 2005 is designed to make state of the art Windows and .NET
applications. Microsoft provides one and only one way to emulate an older DOS C++ program, a
Win32 Console Application. Even this form of program is really a full-fledged Windows
application, but can appear as if it were an ordinary C++ DOS program. This Appendix explains
how to create and work with a DOS Console Application.

A Windows Application is a C++ program that runs in a windows with a graphical user
interface, such as dialogs, edit controls, list boxes, pictures, and so on. These types of programs
generally require the knowledge of this ebook, plus Advanced Data Structures or its equivalent,
C++ Object Oriented Programming.

A DOS Console application is a basic C++ program that can be run from a command
prompt in a DOS window. However, the Integrated Development Environment or IDE is
normally used to run our applications under development. Please note carefully that a Console
Application is not a Windows Application and does not have a graphical user interface such as
dialogs, edit controls, and list boxes. All output is simply text lines of 80 characters maximum.
Console applications are used in all of the beginning C++ programming courses.

C++ DOS Console Applications

Normally, we think of a program as having one or more source files (with the .cpp or .h file
extensions) and the compiler created executable, the .exe file. However, Visual Studio always
thinks in terms of programming “solutions.” A “solution” represents the solving of a user
programming problem or application and consists of all of the files necessary to fully implement
it.

Suppose that we wish a “solution” that displays the message “Hello World!” on the
screen so that we can learn how to use the Visual Studio package. Visual Studio is always going
to place all of the files that make up that solution into a folder of your choice on some drive of
your choice. The key point is that all solution files reside in a folder. Normally, for beginning
programmers, each program or “solution” should be stored in its own separate folder.

Where are these program folders located? If you are working at home, my suggestion is to
begin by making a top-most folder to store all of the program or “solution” folders for a course.
Use Explorer to make a folder such as this.
C:\Cs125
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 626

If you are working in the labs at school, you should place all of your program or
“solution” folders beneath the folder they have designated, such as
D:\Student
Note: do not use your local H: drive. It only holds 20M which translates to just a couple complete
programming assignments.

Another, perhaps better, school solution is to purchase a thumb USB port drive which can
store much larger volumes of data.

With your top-most folder created, you are ready to begin making the “Hello World”
program or “solution.” There are three approaches that are used: “I am beginning a brand new
program” or “I wish to continue working on an existing program” or “I am bringing my files
from home to School and I want to work on my program here at School for a while.” Let’s
examine these three situation in turn.

Making a New Programming Solution —

I Am Building a New Program
Begin by launching the VC compiler. Choose the “File|New|Project” menus shown in Figure 1.

Figure 108 - File-New Project

Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 627

The next series of steps is critical. In the New Project dialog, click on “Visual C++
Projects” in the “Project Types” list and then click “Win32 Console Project” in the
“Templates” list on the right side of the dialog as shown in Figure 2.

Figure 109 Creating a Win32 Console Application

Next, enter a program name in the “Name” edit control. In the above figure, I called it
Pgm1. Then set the “Location” of the project. Use the Browse button to select the top-most
folder which contains all of your assignments. Here it is D:\Cs125. The “Create Directory for
Solution” is not checked. Normally, the compiler will make a subfolder whose name is the same
name as what you entered in the “Name” of the project edit control just above. It is vital that you
verify where it is about to build the project and in what folder. So look at the last line, “Project
will be created at:” — verify that this is what you want. If you do not see this line, click on the
bottom left “More” button. When everything is just the way you desire, click Ok.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 628

You are then presented with the Win32 Application Wizard. We must make some
changes in the Application Settings because the wizard still assumes that you really want to make
a Windows Application with a fancy graphical user interface! So click on Application Settings.

Figure 110 The Win32 Application Wizard

Figure 111 Making an Empty Project

Here you must make two choices. You must click on the “Console Application” radio
button and check the “Empty Project” box. Failure to click these two choices results in VC
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 629

reverting back to making a Windows application which will fail utterly when you begin to
compile your program forcing you to start over at the beginning of this sequence to make a new
Win32 Console project. When you have made these two choices, click “Finish.”

At this point, Visual Studio has created a new subfolder for the project and built some
housekeeping files. The file with the extension .sln is the “solution” file and the file with the
.vcproj extension is the project. At this point there are no C++ source files making up the project.
So the next step is to make a new C++ Source file.

In the Solution Explorer window, click on the icon representing the program, here
“Pgm1" and then right click to bring up the context sensitive menu and choose Add and then
Add New Item. This is shown in Figure 5.

Figure 112 Adding a New C++ File to the

Project
Ne xt, in the Add New Item Dialog, choose
“C++ File,” enter the name of the cpp file. Here I called it Pgm1. The compiler will append the
.cpp file extension for you. Then click Open as shown in Figure 6.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 630

Figure 113 Entering the Name of the New C++ File

The compiler then creates the new cpp file and you are ready to begin typing in the C++
coding. Your screen should look similar to mine. Note that I have dragged the Solution Explorer
over to the right side and docked it there instead of on the left side. This is shown in Figure 7.

Figure 114 Pgm1.cpp Ready to Code

Please double check that the new cpp file, Pgm1.cpp is located beneath the Source Files
icon as shown below. If it is not so located, drag the icon representing the Pgm1.cpp file to the
Source Files. The Source Files represents the cpp file(s) that make up the project or solution.

Tip: Any time that you want to open an existing file for editing, you can just double click
its icon in the Solution Explorer window.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 631

Finally, you type in your program in the pgm1.cpp editor window. In Figure 8, I have
entered the “Hello World!” program coding. We are now ready to compile and execute the
program. Before we examine how to run the program, let’s look at the second method of building
a solution.

Figure 115 Pgm1 Ready to Compile and Execute

Continue to Work on an Existing Program — Starting Visual Studio

You want to continue working on a previous project or solution. But either Visual Studio is not
even running yet or you have some other program or solution open in the Studio. There are two
ways to begin, depending upon which is easiest for you to do.

1. If Visual Studio is not opened, you can launch Visual Studio from its icon or from
Start. Then, choose File|Open Solution menus. In the Open Solution dialog, navigate to the drive
and folder that contains the desired project or solution. The dialog is automatically looking for all
.sln (solution) files. When you have found the desired solution (sln) file, select it and press the
Open button. Your previous project or solution is then opened and you are ready to resume.

2. Using the Explorer, navigate to the desired drive and folder that contains the program
and solution (.sln) file you want to work on next. Double click on the .sln file. This launches
Visual Studio which then open the project or solution. If the Studio is already opened, it switches
to this solution automatically.

3. You can also use File|Open Solution to reopen an existing project.

Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 632

Bringing Files From Home to School

or
Building a New Project in Which the Cpp Files Already Exist
This method is used when you are bringing in files from home to work on a program at school,
when you want to build a new project but already have existing cpp files, or when you have to
build a new project or solution because the existing solution is not a Win32 Console Application
project.

The first step is to copy the cpp file(s) to a new folder on the hard disk. At school, this
should be beneath D:\Student. For example, using the Explorer, you have made
D:\Student\Pgm1
Then copy the existing cpp file(s) into that folder.

Bring up Visual Studio and go through the steps listed above to make a new project. In
short, these steps are File|New|Project followed by selecting Visual C++ Projects and Win32
Project. Enter the new name of the project which is usually the same name as the folder in
which the file(s) are located. Then, use the Browse button to point the location and “Project
Will Be Created At” to be the desired folder in which the files are currently located. Be very
sure that this location is where the files are.

Then click Ok and in the Application Wizard which appears next, select Application
Settings, choose Console Application and check Empty Project.

Remember, that any deviation from these settings produces a Windows graphical
application not a DOS console application. If you goof, when you try to build the program you
will get a linker error saying that it cannot find the WinMain function.

When the project or solution is built, in the Solution Explorer window, select the icon
representing the entire project, right click on it, and choose Add|Add an Existing Item. This is
shown in Figure 9.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 633

Figure 116 Adding Existing Cpp File to a

Project
In the Open dialog, navigate to the folder
what contains the file(s) if it is not already pointing there and select the file(s) to be added into
this new project. Click Ok. The newly added files should appear under the Source Files Icon in
the Solution Explorer (or under Header Files if their extension is .h). That is, at this point, the
Solution Explorer should look exactly like Figure 7 above.

If an icon representing a cpp file is not located under the Source Files icon, drag its icon
there.

When you are done working on your program at school, close Visual Studio. Use the
Explorer to copy your files back onto your disk. Then, delete your program folder from
D:\Student. Do not leave your files on the computers in the labs. If you do so, you run the risk of
having some other student copy your work which can have very bad consequences for you.

Compiling and Running Your Program

In a later section below, I will show you how you can put some short cut buttons for these actions
onto your toolbar. For now, to compile your program, choose the menus Build|Build Solution or
Build|Build Pgm1.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 634

The difference between these options is moot at this point. A solution can consist of more
than one project or program. Build Solution rebuilds all of the programs in a solution. Build
Pgm1 is specifically requesting that this particular program within the complete solution package
be built. Since we are only putting a single program into a solution, you can use either choice. (In
my Data Structures and Object Oriented Programming ebooks, I show you how to create several
projects within a solution.)

When a build is requested, the compiler translates your source statements into the
corresponding machine instructions and runs the linker program to merge in the Microsoft
provided routines for input, output, math and so on building the resulting executable file, the .exe
file.

If there are any errors, the program executable cannot be built and the errors are shown in
the Build Window located at the bottom of the screen. You should correct all errors so that it
reports no errors and no warnings. This is shown in Figure 10.

Figure 117 Successful Build

Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 635

Tip: If you have an error, double clicking on the error in the Build window automatically
repositions the source cpp file to the offending statement, when it is known or is possible to
determine.

Executing a DOS Console Program

There are several ways the program can be executed. The choice of method rather depends upon
the needs at hand. From within the Visual Studio interface, a program can be run by using the
Execute or Debug options available from the Debug menu.

When Start Without Debugging is chosen (previously in version 6.0 it was known as
Execute), a DOS console window opens and the program runs. When the program is finished, the
window remains open so that you may view the results at your leisure. When Start (previously
known as Debug) is chosen, again a DOS console window opens and the program runs.
However, the full capabilities of debugging are available. See the Debug Section below. But
when the program finishes, the DOS window promptly closes. Thus, if all you want to do is see
the program run, you use Start Without Debugging, because Start (debugging) closes the output
results window when the program finishes and you cannot review the results of the run.

Visual Studio has built the executable (.exe) file in the Debug folder beneath your project
folder. Thus, using Explorer you can navigate to that Debug folder and run the program by
double clicking the exe file. However, as soon as the program finishes, the DOS window closes,
so you cannot easily review the results.

You can also open a DOS command prompt window and use CD (change directory)
commands to make the current default folder be the Debug folder containing the .exe file. Then,
you can run the program at the command line by typing its name, such as Pgm1. This window
stays open until you explicitly close it.

Getting Source File Printouts

Usually, your instructor requires hard copy printouts of the source program. This is easily done
by having the cpp window open and choosing File|Print. Note, the source editing window needs
to be the currently active window. If it is not, then click anywhere in the source window to make
it the currently active window before choosing File|Print.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 636

Getting a Printed Copy of the Program Execution Output

Instructors usually require a copy of the execution run(s) of a program to show that the program
works as expected. This is done in one of several ways, depending upon the volume of the lines
that the program displays and upon your operating system.

Case 1: The Entire Output Fits on One Screen Without Scrolling

If the entire results are visible on the screen (or if you can scroll the DOS window and see all of
the program’s output), then you can take a screen shot. On the system menu (icon in the upper
left corner of the caption of the DOS window, choose Edit|Mark. Then, in the DOS window,
drag down and to the right to highlight (select) all of the lines to be copied. With the lines
selected and highlighted, again click on the system menu and choose Edit|Copy. Launch an
instance of Notepad and Edit|Paste into Notepad and do a File|Print.

Some DOS windows have a toolbar with Edit|Mark and Edit|Copy buttons on them.
Those can be used instead of the system menu.

Note that this method is only effective when there is a limited number of output lines that
the program produces.

Case 2: Using cout and There Are Too Many Lines To Capture With a Screen
Shot

When there are too many lines to just do a screen shot, then DOS redirection can be used. The
redirection can be done from within the Visual Studio setup or from a DOS command prompt
window. The first way is the easiest.

Right click on the icon representing the program in the Solution Explorer window and
choose Properties. When the properties window appears, select Debugging. Then in the
Command Arguments control, enter
> results.txt
If you use a file extension of .txt, then when you wish to view the output or print it, you merely
have to double click on this file in the Explorer to launch Notepad. This is shown in Figure 11.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 637

Figure 11 Setting DOS Output Redirection in the Program’s Properties Window

Now with the command argument set for output DOS redirection, you can run the
program either by Debug|Start or Debug|Start without Debugging. Of course, you will see
nothing on the screen since all screen output is being redirected to this new file. When the
program is finished, you can then open up the results.txt file and view or print it. The file is
located in the same folder as the rest of the project files.

You could also run the program from a DOS Command prompt, but it is not
recommended because the executable is located in the \Debug or \Release subfolder beneath the
project. Thus, at the DOS prompt, you must use the CD (change directory) command to set the
current default folder to the \Debug folder that contains the exe file. Then at the command
prompt you run the program this way.
D:\Student\Pgm1>Pgm1 > results.txt
Of course, this time, the results.txt file is located in the Debug folder. This is generally
considered awkward.

Case 3: Using an Output File Stream

Perhaps the best way to create an output report is to use an output file instead of using the cout
screen stream.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 638

ofstream outfile ("results.txt");

outfile << "Hello\n";
outfile.close ();
This simple sequence opens an output file called results.txt which is located in the same folder as
the project files. It then displays “Hello” to that file. Don’t forget to close the file when you are
done. Again, you can double click on results.txt in the Explorer to open up Notepad to view or
print the file.

Visual Studio Operational Tips

Tip 1: If you should accidentally close a window such as the Solution Explorer, you can
reopen any of them using the View menu items for the main windows.

Tip 2: Tools|Options allows you to customize your setup.

Tip 3: In your source window, tabs and blank spaces are not the same thing. The
observed symptoms are in the editor, all lines are nicely aligned and indented properly. However,
when you print the source, a jagged edge appears. The reason for the jagged edge is that tabs are
expanded differently when going to the screen versus going to the printer. Thus, I always set the
Options|TextEditor|Tabs as shown in Figure 12. You can set the amount of indentation you
prefer. I like to indent only one space. But some may consider this too small. However, the key
item is to not use tabs, but have the editor insert spaces or blanks instead.

Figure 119 Using Spaces Instead of Tabs and the Amount to Indent
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 639

Tip 4: Double clicking on error messages in the Build Results window positions the
editor to the offending line of code, where known. Selecting the error message and pressing F1
for help sometimes yields valuable clues to fix an error.

Tip 5: Under Options|Customize all of the possible toolbars are shown. Those with a
check mark beside them are currently visible. Thus, if you accidentally close a toolbar, you can
make it visible once again here by rechecking it.

Tip 6: Under Options|Customize you can also select what toolbar buttons are available
for your use. Click on the Commands tab. The commands are shown in groups. Simply find a
command that you want to have available on your toolbar and drag that item from this dialog to
the desired location you want that button to occupy on your toolbar and drop it there. To remove
a button from the toolbar, drag it back into this dialog window.

Debug Versus Release Builds

The compiler can create both a debug and a release version of a program. The release form is
highly optimized for speed of execution. Typically, the release exe file and its associated files are
located in the \Release subfolder below the project folder. The debug version of a program has
no speed optimizations, rather it is setup to provide maximum debugging information as the
program executes. Typically, the debug form is stored in the \Debug subfolder of the project.

When you make a new project, the debug version is the default. When you have all of the
errors fixed, you could then make a release build for distribution to production. To do so, you
need to choose Build|Configuration Manager menus. This opens the Configuration Manager
dialog. Just select Release in the top combo box labeled Active Solution Configuration and
click Close. Then, rebuild the project.

From the toolbar, there is a faster shortcut to switching between Release and Debug
builds. In the middle portion of the toolbar there is a combo box which initially says “Debug.”
Simply switch it to “Release” and rebuild the project.

Beginning with .NET 2005, by default your exe files are no longer stand alone programs
but require three Microsoft DLL files in order to execute. What this means is that if you put your
exe onto another computer, it cannot run unless you also install those three dll files. However,
there is a way around this. Bring up the Project Options and then the C++ tab and select Code
Generation. On the right side of the dialog, Runtime Library to Multithreaded /MT. Make sure it
is the Release build you are modifying. Then, rebuild the program and it can run without the
three Microsoft dll files. See Figure 13.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 640

Figure 120 Making a Program’s Exe Able to Run Without any DLL Files

A Primer on Using the Debugger

Gaining an ability to use the Debugger is vital to accurately and speedy finding of program logic
errors. This section explores some of the basic debugging methods. The debugger is launched in
several ways, depending upon the debugging circumstances you face. Let’s begin with a simple
situation.

Suppose that a program to calculate a person’s wages is producing the wrong results. One
way to quickly find the errors is to step through the program line by line, examining the key
variables’ contents as we go. This is done by using the F10 (Debug|Step Over) key. F10 executes
one line of the program and then stops allowing you to examine the contents of key variables and
such.

In Figure 14, I have pressed the F10 key a couple times. The yellow arrow in the source
window is pointing to the next instruction to be executed if I should press F10 again. At the
bottom is a tabbed window that is showing the contents of the automatic storage variables and
their data types and current contents. All show core garbage as their contents.
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 641

Figure 14 Case 1 Running the Debugger After a Couple F10 Key Presses

When the debugger is waiting on us to do something, we can float the mouse over a
variable and pause for a second or so. The debugger then pops up a small window around the
mouse displaying the current contents of the variable the over which the mouse is hovering. One
could also click on the Watch tab. In this window, click on an open line and type in the name of a
variable. The Watch window will then keep track of the current contents of that variable as the
program runs.

Next, I press F10 a couple times so that we get to the while statement and perform the
input operation. This while statement illustrates why we are using F10 and not F11 (Step Into).
F10 steps over function calls such as the cin extraction operation. That is, it executes the function
call and continues when the function returns back to our program. F11, on the other hand, steps
into the functions that we call. Here, had we pressed F11, we would suddenly find ourselves
tracing through the iostream coding! Should that happen accidentally, use Debug|Step Out to get
back out of the iostream coding. Later on, when we write our own functions, if we suspect a
function is not working properly, the when the debugger gets to that function call, press F11 to
step into that function and begin executing it line by line.

When entering test data, use easily recognizable and simple values. I entered “1234" for
the id number and “10" for the hourly rate and “5" for the hours worked. This way, we can easily
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 642

spot errors. In Figure 15, we can see that thus far the program is working correctly with the
correct values in the variables.

Figure 15 Case 1 With Correct Input Values Shown in Red

Next, I pressed F10 two more times to execute the two calculations. Figure 19 shows the
results. By comparing Figures 15 and 16, it is obvious that the pay is incorrect (it looks like 102
and not 50) and that we have failed to provide an initial value for the totalPay field and are
simply copying pay into it instead of adding pay to totalPay.

Figure 16 Case 1 With Incorrect Pay and totalPay Variables

Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 643

Next, let’s examine a more complex situation. In Case 2, two functions have been added
to handle the input operations and to perform the calculations. However, when the program is
executed, it produces only the following output.
enter id, name, rate and hr
1234 sam spade 10 5

Total Company Pay: 0

Press any key to continue

For your reference, here is the Case 2 program.

#include <iostream>
#include <iomanip>
using namespace std;

istream& InputData (istream& is, long& id, double& rate,

int& hrs, char name[]);
double CalcPay (double rate, int hrs);

const int NAMELEN = 10;

int main () {

long idNumber;
char name[NAMELEN];
double hourlyRate;
int hours;
double pay;
double totalPay = 0;

while (InputData (cin, idNumber, hourlyRate, hours, name)) {

pay = CalcPay (hourlyRate, hours);
totalPay += pay;
cout << setw (10) << idNumber << setw (10) << pay << endl
<< endl;
cout << "Enter the worker's id number, hourly rate and hours"
" worked\n";
}
cout << endl << "Total Company Pay: " << totalPay << endl;

return 0;
}

istream& InputData (istream& is, long& id, double& rate,

int& hrs, char name[]) {
cout << "enter id, name, rate and hrs\n";
is >> id;
is.get (name, sizeof (name));
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 644

is >> rate >> hrs;

return is;
}

double CalcPay (double rate, int hrs) {

double pay = rate * hrs;
return pay;
}

Since the while loop is controlling the repetitive operations, evidently for the loop to
terminate prematurely, the input stream must not be in the good state after attempting to input the
first set of data. While we could use F10 to step our way through the program and then use F11
to step Into the InputData function, it is quicker to set a “breakpoint” inside InputData on the
first line of coding that does something. A breakpoint is a mark on a line of coding where the
debugger will automatically stop before it executes that line if the debugger ever gets to that line.

Breakpoints are set and unset rather like a toggle switch. Right click on the line on which
you want the debugger to stop and choose Insert Breakpoint. A red dot appears to the left of the
line and if Go, or F5, is pressed, the debugger executes the program until it gets to the
breakpoint(s) at which point it stops as usual. To remove a breakpoint, right click on the line with
the red dot and choose Remove Breakpoint.

In this case, I set a breakpoint on the cout prompt before it inputs the first portion of the
data the user types. With the breakpoint set, simply choose Debug|Start or press F5. The
debugger executes the program until the breakpoint is reached. Figure 17 shows the debugger
after it has reached the breakpoint I have set.

Figure 17 Case 2 Initial Breakpoint Has Been Reached in InputData

Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 645

Notice that the local variables contain core garbage. This is expected since we are about
to fill these fields up with the user’s input values. Press F10 a couple times and watch what
happens to the variables. Figure 18 shows the results after the couple F10 key presses.

Figure 18 Case 2 The Results Showing an Abbreviated name

Looking over the input results in Figure 18, the id number is correct. The name was
inputted next, but it only contains a small portion of the 10-byte string. Why? We have found the
first error. In main, the variable name is an array of 10 characters. It was passed to InputData.
The name of an array is always a constant memory address of where that array is located, the
address of the first element. Thus, when we use the sizeof (name) here in the InputData
function, name is now a pointer to the character array and the sizeof (pointer) is always 4 bytes.
Here we need to use the const int value, NAMELEN to specify how many characters to input.

Ok. So I fixed this error, rebuilt the program and pressed F5 (Go Start Debugging). When
the debugger popped up at the breakpoint in InputData, I then pressed F10 a few times. Figure
19 shows that the data still is not being properly input. Notice that theoretically hours and the pay
rate should have been inputted, but that their values are still core garbage, indicating that the
stream is in the fail state once more. Why?
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compiler 646

Figure 19 Case 2 After the get Function Fix

Take a close look at the values that have been inputted as shown in the Locals window
above. The id is correct. But there is something wrong with the name field. The last letter has
been truncated and there is a blank before the first letter S. Ah. This blank is the whitespace
separating the name from the id number. And we have now found the next error.

There is another situation in which the debugger is invaluable. Suppose that you are
executing the program (Start without Debugging) and one of the famous Windows message
boxes pops up saying “This program has done an illegal thing.” When this occurs, restart the
program by using the Go (Start with Debugging). Then when the error message pops up, choose
Retry and the debugger breaks showing you the line at which the crash occurred. Very often this
is not in your program. You may be seeing some C++ coding from the library routines, such as
the iostreams, or it may appear to be microassembler machine instructions. This is where the
program is currently executing. What we need to know is how did we get to this spot. So choose
in the bottom tabbed window that usually shows the local variable contents, click on the Call
Stack tab. This window shows you the call list leading up to the current routine that is failing.
Look back down the call stack until you find one of your functions or your main function. Click
on it and you are shown the line in your program that called into the system coding that is failing.
Now you can examine what’s what at this point in your function or main procedure.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 647

Appendix B
Using Microsoft’s VC 7 (.NET) Compiler
Microsoft’s Visual Studio Version 7 (.NET) is designed to make state of the art Windows and
.NET applications. Microsoft provides one and only one way to emulate an older DOS C++
program, a Win32 Console Application. Even this form of program is really a full-fledged
Windows application, but can appear as if it were an ordinary C++ DOS program. This Appendix
explains how to create and work with a DOS Console Application.

A Windows Application is a C++ program that runs in a windows with a graphical user
interface, such as dialogs, edit controls, list boxes, pictures, and so on. These types of programs
generally require the knowledge of this ebook plus Advanced Data Structures or its equivalent,
C++ Object Oriented Programming.

A DOS Console application is a basic C++ program that can be run from a command
prompt in a DOS window. However, the Integrated Development Environment or IDE is
normally used to run our applications under development. Please note carefully that a Console
Application is not a Windows Application and does not have a graphical user interface such as
dialogs, edit controls, and list boxes. All output is simply text lines of 80 characters maximum.
Console applications are used in all of the beginning C++ programming courses.

C++ DOS Console Applications

Normally, we think of a program as having one or more source files (with the .cpp or .h file
extensions) and the compiler created executable, the .exe file. However, Visual Studio always
thinks in terms of programming “solutions.” A “solution” represents the solving of a user
programming problem or application and consists of all of the files necessary to fully implement
it.

Suppose that we wish a “solution” that displays the message “Hello World!” on the
screen so that we can learn how to use the Visual Studio package. Visual Studio is always going
to place all of the files that make up that solution into a folder of your choice on some drive of
your choice. The key point is that all solution reside in a folder. Normally, for beginning
programmers, each program or “solution” should be stored in its own separate folder.

Where are these program folders located? If you are working at home, my suggestion is to
begin by making a top-most folder to store all of the program or “solution” folders for a course.
Use Explorer to make a folder such as this.
C:\Cs125
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 648

If you are working in the labs at school, you should place all of your program or
“solution” folders beneath the folder they have designated, such as
D:\Student

With your top-most folder created, you are ready to begin making the “Hello World”
program or “solution.” There are three approaches that are used: “I am beginning a brand new
program” or “I wish to continue working on an existing program” or “I am bringing my files
from home to School and I want to work on my program here at School for a while.” Let’s
examine these three situation in turn.

Making a New Programming Solution —

I Am Building a New Program
Begin by launching the VC compiler. Choose the “File|New|Project” menus shown in Figure 1.

Figure 1 — Making a New Project

The next series of steps is critical. In the New Project dialog, click on “Visual C++
Projects” in the “Project Types” list and then click “Win32 Project” at the very bottom of the
“Templates” list on the right side of the dialog as shown in Figure 2.

Figure 2 — Making a New DOS Console Application

Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 649

Next, enter a program name in the “Name” edit control. In the above figure, I called it
Pgm1. Then set the “Location” of the project. Use the Browse button to select the top-most
folder which contains all of your assignments. Here it is C:\Cs125. The “Create Directory for
Solution” is not checked. Normally, the compiler will make a subfolder whose name is the same
name as what you entered in the “Name” of the project edit control just above. It is vital that you
verify where it is about to build the project and in what folder. So look at the last line, “Project
will be created at:” — verify that this is what you want. If you do not see this line, click on the
bottom left “More” button. When everything is just the way you desire, click Ok.

You are then presented with the Win32 Application Wizard. We must make some
changes in the Application Settings because the wizard still assumes that you really want to make
a Windows Application with a fancy graphical user interface! So click on Application Settings.

Figure 3 — The Win32 Application Wizard

Figure 4 — Setting the Empty and Console Application Properties

Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 650

Here you must make two choices. You must click on the “Console Application” radio
button and check the “Empty Project” box. Failure to click these two choices results in VC
reverting back to making a Windows application which will fail utterly when you begin to
compile your program forcing you to start over at the beginning of this sequence to make a new
Win32 Console project. When you have made these two choices, click “Finish.”

At this point, Visual Studio has created a new subfolder for the project and built some
housekeeping files. The file with the extension .sln is the “solution” file and the file with the
.vcproj extension is the project. At this point there are no C++ source files making up the project.
So the next step is to make a new C++ Source file.

In the Solution Explorer window, click on the icon representing the program, here
“Pgm1" and then right click to bring up the context sensitive menu and choose Add and then
Add New Item. This is shown in Figure 5.

Figure 5 — Adding a New C++ File to the Project

Next, in the Add New Item Dialog, choose “C++ File,” enter the name of the cpp file.
Here I called it Pgm1. The compiler will append the .cpp file extension for you. Then click Open
as shown in Figure 6.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 651

Figure 6 Entering the Name of the Solution and Folders

The compiler then creates the new cpp file and you are ready to begin typing in the C++
coding. Your screen should look similar to mine. Note that I have dragged the Solution Explorer
over to the right side and docked it there instead of on the left side. This is shown in Figure 7.

Figure 7 — The New Cpp File (in the bottom right empty window)
Ready For Work
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 652

Please double check that the new cpp file, Pgm1.cpp is located beneath the Source Files
icon as shown below. If it is not so located, drag the icon representing the Pgm1.cpp file to the
Source Files. The Source Files represents the cpp file(s) that make up the project or solution.

Tip: Any time that you want to open an existing file for editing, you can just double click
its icon in the Solution Explorer window.

Finally, you type in your program in the pgm1.cpp editor window. In Figure 8, I have
entered the “Hello World!” program coding. We are now ready to compile and execute the
program. Before we examine how to run the program, let’s look at the second method of building
a solution.

Figure 8 The Hello World C++ Coding

Continue to Work on an Existing Program — Starting Visual Studio

You want to continue working on a previous project or solution. But either Visual Studio is not
even running yet or you have some other program or solution open in the Studio. There are two
ways to begin, depending upon which is easiest for you to do.

1. If Visual Studio is not opened, you can launch Visual Studio from its icon or from
Start. Then, choose File|Open Solution menus. In the Open Solution dialog, navigate to the drive
and folder that contains the desired project or solution. The dialog is automatically looking for all
.sln (solution) files. When you have found the desired solution (sln) file, select it and press the
Open button. Your previous project or solution is then opened and you are ready to resume.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 653

2. Using the Explorer, navigate to the desired drive and folder that contains the program
and solution (.sln) file you want to work on next. Double click on the .sln file. This launches
Visual Studio which then open the project or solution. If the Studio is already opened, it switches
to this solution automatically.

3. You can also use File|Open Solution to reopen an existing project.

Bringing Files From Home to School

or
Building a New Project in Which the Cpp Files Already Exist
This method is used when you are bringing in files from home to work on a program at school,
when you want to build a new project but already have existing cpp files, or when you have to
build a new project or solution because the existing solution is not a Win32 Console Application
project.

The first step is to copy the cpp file(s) to a new folder on the hard disk. At school, this
should be beneath D:\Student. For example, using the Explorer, you have made
D:\Student\Pgm1
Then copy the existing cpp file(s) into that folder.

Bring up Visual Studio and go through the steps listed above to make a new project. In
short, these steps are File|New|Project followed by selecting Visual C++ Projects and Win32
Project. Enter the new name of the project which is usually the same name as the folder in
which the file(s) are located. Then, use the Browse button to point the location and “Project
Will Be Created At” to be the desired folder in which the files are currently located. Be very
sure that this location is where the files are.

Then click Ok and in the Application Wizard which appears next, select Application
Settings, choose Console Application and check Empty Project.

Remember, that any deviation from these settings produces a Windows graphical
application not a DOS console application. If you goof, when you try to build the program you
will get a linker error saying that it cannot find the WinMain function.

When the project or solution is built, in the Solution Explorer window, select the icon
representing the entire project, right click on it, and choose Add|Add an Existing Item. This is
shown in Figure 9.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 654

Figure 9 Adding an Existing File to a Project

In the Open dialog, navigate to the folder what contains the file(s) if it is not already
pointing there and select the file(s) to be added into this new project. Click Ok. The newly added
files should appear under the Source Files Icon in the Solution Explorer (or under Header Files if
their extension is .h). That is, at this point, the Solution Explorer should look exactly like Figure
7 above.

If an icon representing a cpp file is not located under the Source Files icon, drag its icon
there.

When you are done working on your program at school, close Visual Studio. Use the
Explorer to copy your files back onto your disk. Then, delete your program folder from
D:\Student. Do not leave your files on the computers in the labs. If you do so, you run the risk of
having some other student copy your work which can have very bad consequences for you.

Compiling and Running Your Program

In a later section below, I will show you how you can put some short cut buttons for these actions
onto your toolbar. For now, to compile your program, choose the menus Build|Build Solution or
Build|Build Pgm1.

The difference between these options is moot at this point. A solution can consist of more
than one project or program. Build Solution rebuilds all of the programs in a solution. Build
Pgm1 is specifically requesting that this particular program within the complete solution package
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 655

be built. Since we are only putting a single program into a solution, you can use either choice. (In
my Data Structures and Object Oriented Programming ebooks, I show you how to create several
projects within a solution.)

When a build is requested, the compiler translates your source statements into the
corresponding machine instructions and runs the linker program to merge in the Microsoft
provided routines for input, output, math and so on building the resulting executable file, the .exe
file.

If there are any errors, the program executable cannot be built and the errors are shown in
the Build Window located at the bottom of the screen. You should correct all errors so that it
reports no errors and no warnings. This is shown in Figure 10.

Figure 10 The Build Results — No Errors or Warnings

Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 656

Tip: If you have an error, double clicking on the error in the Build window automatically
repositions the source cpp file to the offending statement, when it is known or is possible to
determine.

Executing a DOS Console Program

There are several ways the program can be executed. The choice of method rather depends upon
the needs at hand. From within the Visual Studio interface, a program can be run by using the
Execute or Debug options available from the Debug menu.

When Start Without Debugging is chosen (previously in version 6.0 it was known as
Execute), a DOS console window opens and the program runs. When the program is finished, the
window remains open so that you may view the results at your leisure. When Start (previously
known as Debug) is chosen, again a DOS console window opens and the program runs.
However, the full capabilities of debugging are available. See the Debug Section below. But
when the program finishes, the DOS window promptly closes. Thus, if all you want to do is see
the program run, you use Start Without Debugging, because Start (debugging) closes the output
results window when the program finishes and you cannot review the results of the run.

Visual Studio has built the executable (.exe) file in the Debug folder beneath your project
folder. Thus, using Explorer you can navigate to that Debug folder and run the program by
double clicking the exe file. However, as soon as the program finishes, the DOS window closes,
so you cannot easily review the results.

You can also open a DOS command prompt window and use CD (change directory)
commands to make the current default folder be the Debug folder containing the .exe file. Then,
you can run the program at the command line by typing its name, such as Pgm1. This window
stays open until you explicitly close it.

Getting Source File Printouts

Usually, your instructor requires hard copy printouts of the source program. This is easily done
by having the cpp window open and choosing File|Print. Note, the source editing window needs
to be the currently active window. If it is not, then click anywhere in the source window to make
it the currently active window before choosing File|Print.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 657

Getting a Printed Copy of the Program Execution Output

Instructors usually require a copy of the execution run(s) of a program to show that the program
works as expected. This is done in one of several ways, depending upon the volume of the lines
that the program displays and upon your operating system.

Case 1: The Entire Output Fits on One Screen Without Scrolling

If the entire results are visible on the screen (or if you can scroll the DOS window and see all of
the program’s output), then you can take a screen shot. On the system menu (icon in the upper
left corner of the caption of the DOS window, choose Edit|Mark. Then, in the DOS window,
drag down and to the right to highlight (select) all of the lines to be copied. With the lines
selected and highlighted, again click on the system menu and choose Edit|Copy. Launch an
instance of Notepad and Edit|Paste into Notepad and do a File|Print.

Some DOS windows have a toolbar with Edit|Mark and Edit|Copy buttons on them.
Those can be used instead of the system menu.

Note that this method is only effective when there is a limited number of output lines that
the program produces.

Case 2: Using cout and There Are Too Many Lines To Capture With a Screen
Shot

When there are too many lines to just do a screen shot, then DOS redirection can be used. The
redirection can be done from within the Visual Studio setup or from a DOS command prompt
window. The first way is the easiest.

Right click on the icon representing the program in the Solution Explorer window and
choose Properties. When the properties window appears, select Debugging. Then in the
Command Arguments control, enter
> results.txt
If you use a file extension of .txt, then when you wish to view the output or print it, you merely
have to double click on this file in the Explorer to launch Notepad. This is shown in Figure 11.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 658

Figure 11 Setting DOS Output Redirection in the Program’s Properties Window

Now with the command argument set for output DOS redirection, you can run the
program either by Debug|Start or Debug|Start without Debugging. Of course, you will see
nothing on the screen since all screen output is being redirected to this new file. When the
program is finished, you can then open up the results.txt file and view or print it. The file is
located in the same folder as the rest of the project files.

You could also run the program from a DOS Command prompt, but it is not
recommended because the executable is located in the \Debug or \Release subfolder beneath the
project. Thus, at the DOS prompt, you must use the CD (change directory) command to set the
current default folder to the \Debug folder that contains the exe file. Then at the command
prompt you run the program this way.
D:\Student\Pgm1>Pgm1 > results.txt
Of course, this time, the results.txt file is located in the Debug folder. This is generally
considered awkward.

Case 3: Using an Output File Stream

Perhaps the best way to create an output report is to use an output file instead of using the cout
screen stream.
ofstream outfile ("results.txt");
outfile << "Hello\n";
outfile.close ();
This simple sequence opens an output file called results.txt which is located in the same folder as
the project files. It then displays “Hello” to that file. Don’t forget to close the file when you are
done. Again, you can double click on results.txt in the Explorer to open up Notepad to view or
print the file.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 659

Visual Studio Operational Tips

Tip 1: If you should accidentally close a window such as the Solution Explorer, you can
reopen any of them using the View menu items for the main windows.

Tip 2: Tools|Options allows you to customize your setup. Figure 12 shows the first
category, Environment. Notice that on the top right side is At Start Up. I have it set to
automatically load the project that I was last working on when I shut Visual Studio down.

There is another choice Show Start Page. On this page, you can elect to use a C++ style
setup for all of the windows. This is the choice I often use. The Start Page selections are shown
in Figure 13 below.

Figure 12 The Options — Environment General Set of Options

Tip 3: Double clicking on error messages in the Build Results window positions the
editor to the offending line of code, where known. Selecting the error message and pressing F1
for help sometimes yields valuable clues to fix an error.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 660

Figure 13 The Start Page C++ Options

Tip 4: In your source window, tabs and blank spaces are not the same thing. The
observed symptoms are in the editor, all lines are nicely aligned and indented properly. However,
when you print the source, a jagged edge appears. The reason for the jagged edge is that tabs are
expanded differently when going to the screen versus going to the printer. Thus, I always set the
Options|TextEditor|Tabs as shown in Figure 14. You can set the amount of indentation you
prefer. I like to indent only one space. But some may consider this too small. However, the key
item is to not use tabs, but have the editor insert spaces or blanks instead.

Figure 14 Setting Insert Blanks not Tabs and the Amount to Indent
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 661

Tip 5: Under Options|Customize all of the possible toolbars are shown. Those with a
check mark beside them are currently visible. Thus, if you accidentally close a toolbar, you can
make it visible once again here by rechecking it.

Tip 6: Under Options|Customize you can also select what toolbar buttons are available
for your use. Click on the Commands tab. This is shown in Figure 15 below. The commands are
shown in groups. Simply find a command that you want to have available on your toolbar and
drag that item from this dialog to the desired location you want that button to occupy on your
toolbar and drop it there. To remove a button from the toolbar, drag it back into this dialog
window.

Figure 15 Setting Toolbar Buttons By Dragging From

Here to the Toolbar

Figure 7 near the beginning of this Appendix shows some of the buttons I have chosen to
have on my toolbar for faster action. These include the Execute or Start Without Debugging and
Debug (Go or Start).
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 662

Debug Versus Release Builds

The compiler can create both a debug and a release version of a program. The release form is
highly optimized for speed of execution. Typically, the release exe file and its associated files are
located in the \Release subfolder below the project folder. The debug version of a program has
no speed optimizations, rather it is setup to provide maximum debugging information as the
program executes. Typically, the debug form is stored in the \Debug subfolder of the project.

When you make a new project, the debug version is the default. When you have all of the
errors fixed, you could then make a release build for distribution to production. To do so, you
need to choose Build|Configuration Manager menus. This opens the Configuration Manager
dialog. Just select Release in the top combo box labeled Active Solution Configuration and
click Close. Then, rebuild the project.

From the toolbar, there is a faster shortcut to switching between Release and Debug builds. In the
middle portion of the toolbar there is a combo box which initially says “Debug.” Simply switch it
to “Release” and rebuild the project.

The Configuration Manager and the toolbar combo box are shown in Figure 16.

Figure 16 The Configuration Manager and the Toolbar Combo Box Showing “Release”
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 663

A Primer on Using the Debugger

Gaining an ability to use the Debugger is vital to accurately and speedy finding of program logic
errors. This section explores some of the basic debugging methods. The debugger is launched in
several ways, depending upon the debugging circumstances you face. Let’s begin with a simple
situation.

Suppose that a program to calculate a person’s wages is producing the wrong results. One
way to quickly find the errors is to step through the program line by line, examining the key
variables’ contents as we go. This is done by using the F10 (Debug|Step Over) key. F10 executes
one line of the program and then stops allowing you to examine the contents of key variables and
such.

In Figure 17, I have pressed the F10 key a couple times. The yellow arrow in the source
window is pointing to the next instruction to be executed if I should press F10 again. At the
bottom is a tabbed window that is showing the contents of the automatic storage variables and
their data types and current contents. All show core garbage as their contents.

Figure 17 Case 1 Running the Debugger After a Couple F10 Key Presses

When the debugger is waiting on us to do something, we can float the mouse over a
variable and pause for a second or so. The debugger then pops up a small window around the
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 664

mouse displaying the current contents of the variable the over which the mouse is hovering. One
could also click on the Watch tab. In this window, click on an open line and type in the name of a
variable. The Watch window will then keep track of the current contents of that variable as the
program runs.

Next, I press F10 a couple times so that we get to the while statement and perform the
input operation. This while statement illustrates why we are using F10 and not F11 (Step Into).
F10 steps over function calls such as the cin extraction operation. That is, it executes the function
call and continues when the function returns back to our program. F11, on the other hand, steps
into the functions that we call. Here, had we pressed F11, we would suddenly find ourselves
tracing through the iostream coding! Should that happen accidentally, use Debug|Step Out to get
back out of the iostream coding. Later on, when we write our own functions, if we suspect a
function is not working properly, the when the debugger gets to that function call, press F11 to
step into that function and begin executing it line by line.

When entering test data, use easily recognizable and simple values. I entered “1234" for
the id number and “10" for the hourly rate and “5" for the hours worked. This way, we can easily
spot errors. In Figure 18, we can see that thus far the program is working correctly with the
correct values in the variables.

Figure 18 Case 1 With Correct Input Values Shown in Red

Next, I pressed F10 two more times to execute the two calculations. Figure 19 shows the
results. By comparing Figures 18 and 19, it is obvious that the pay is incorrect (it looks like 102
and not 50) and that we have failed to provide an initial value for the totalPay field and are
simply copying pay into it instead of adding pay to totalPay.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 665

Figure 19 Case 1 With Incorrect Pay and totalPay Variables

Next, let’s examine a more complex situation. In Case 2, two functions have been added
to handle the input operations and to perform the calculations. However, when the program is
executed, it produces only the following output.
enter id, name, rate and hr
1234 sam spade 10 5

Total Company Pay: 0

Press any key to continue

For your reference, here is the Case 2 program.

#include <iostream>
#include <iomanip>
using namespace std;

istream& InputData (istream& is, long& id, double& rate,

int& hrs, char name[]);
double CalcPay (double rate, int hrs);

const int NAMELEN = 10;

int main () {

long idNumber;
char name[NAMELEN];
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 666

double hourlyRate;
int hours;
double pay;
double totalPay = 0;

while (InputData (cin, idNumber, hourlyRate, hours, name)) {

pay = CalcPay (hourlyRate, hours);
totalPay += pay;
cout << setw (10) << idNumber << setw (10) << pay << endl
<< endl;
cout << "Enter the worker's id number, hourly rate and hours"
" worked\n";
}
cout << endl << "Total Company Pay: " << totalPay << endl;

return 0;
}

istream& InputData (istream& is, long& id, double& rate,

int& hrs, char name[]) {
cout << "enter id, name, rate and hrs\n";
is >> id;
is.get (name, sizeof (name));
is >> rate >> hrs;
return is;
}

double CalcPay (double rate, int hrs) {

double pay = rate * hrs;
return pay;
}

Since the while loop is controlling the repetitive operations, evidently for the loop to
terminate prematurely, the input stream must not be in the good state after attempting to input the
first set of data. While we could use F10 to step our way through the program and then use F11
to step Into the InputData function, it is quicker to set a “breakpoint” inside InputData on the
first line of coding that does something. A breakpoint is a mark on a line of coding where the
debugger will automatically stop before it executes that line if the debugger ever gets to that line.

Breakpoints are set and unset rather like a toggle switch. Right click on the line on which
you want the debugger to stop and choose Insert Breakpoint. A red dot appears to the left of the
line and if Go, or F5, is pressed, the debugger executes the program until it gets to the
breakpoint(s) at which point it stops as usual. To remove a breakpoint, right click on the line with
the red dot and choose Remove Breakpoint.
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 667

In this case, I set a breakpoint on the cout prompt before it inputs the first portion of the
data the user types. With the breakpoint set, simply choose Debug|Start or press F5. The
debugger executes the program until the breakpoint is reached. Figure 20 shows the debugger
after it has reached the breakpoint I have set.

Figure 20 Case 2 Initial Breakpoint Has Been Reached in InputData

Notice that the local variables contain core garbage. This is expected since we are about
to fill these fields up with the user’s input values. Press F10 a couple times and watch what
happens to the variables. Figure 21 shows the results after the couple F10 key presses.

Figure 21 Case 2 The Results Showing an Abbreviated name

Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 668

Looking over the input results in Figure 21, the id number is correct. The name was
inputted next, but it only contains a small portion of the 10-byte string. Why? We have found the
first error. In main, the variable name is an array of 10 characters. It was passed to InputData.
The name of an array is always a constant memory address of where that array is located, the
address of the first element. Thus, when we use the sizeof (name) here in the InputData
function, name is now a pointer to the character array and the sizeof (pointer) is always 4 bytes.
Here we need to use the const int value, NAMELEN to specify how many characters to input.

Ok. So I fixed this error, rebuilt the program and pressed F5 (Go Start Debugging). When
the debugger popped up at the breakpoint in InputData, I then pressed F10 a few times. Figure
22 shows that the data still is not being properly input. Notice that theoretically hours and the pay
rate should have been inputted, but that their values are still core garbage, indicating that the
stream is in the fail state once more. Why?

Figure 22 Case 2 After the get Function Fix

Take a close look at the values that have been inputted as shown in the Locals window
above. The id is correct. But there is something wrong with the name field. The last letter has
been truncated and there is a blank before the first letter S. Ah. This blank is the whitespace
separating the name from the id number. And we have now found the next error.

There is another situation in which the debugger is invaluable. Suppose that you are
executing the program (Start without Debugging) and one of the famous Windows message
boxes pops up saying “This program has done an illegal thing.” When this occurs, restart the
program by using the Go (Start with Debugging). Then when the error message pops up, choose
Retry and the debugger breaks showing you the line at which the crash occurred. Very often this
Appendix B: How To Use Microsoft’s .NET 2002-3 C++ 7.0 Compiler 669

is not in your program. You may be seeing some C++ coding from the library routines, such as
the iostreams, or it may appear to be microassembler machine instructions. This is where the
program is currently executing. What we need to know is how did we get to this spot. So choose
in the bottom tabbed window that usually shows the local variable contents, click on the Call
Stack tab. This window shows you the call list leading up to the current routine that is failing.
Look back down the call stack until you find one of your functions or your main function. Click
on it and you are shown the line in your program that called into the system coding that is failing.
Now you can examine what’s what at this point in your function or main procedure.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 670

Appendix C — How to Use Microsoft’s Visual C++ 6

This section is designed for all levels of C/C++ programming — beginning to advanced levels.
Review the sections that are crucial for your needs at the moment. As you discover the need for
additional information, refer back to other sections.

Preface

Microsoft dropped all support for “vanilla” DOS applications a number of years ago. However,
there is full support for Console applications which are basically 32-bit applications that run as if
they were DOS applications. Console applications run under Windows 95, 98 and NT systems.
There are many benefits to using console applications and no drawbacks, unless you need to be
able to run on older DOS 6.22 platforms.

Step 0. Get Organized

Assuming you have installed your VC compiler, the first step to get organized. The VC
environment is going to force you down the path of organization. It is more optimum to begin in
an organized manner than to later have to come back and try to put in some order. What
organization is required?

The Visual C++ platform, or Integrated Development Environment, IDE for short, is
totally centered around a project. A project may be thought of as a complete programming
assignment. The project contains the source file or files; this ranges from a single cpp file for the
beginning classes (*.cpp), to several cpp files and header files (*.h) for the Object Oriented
programs, to many cpp, header and resource files for the more complex Windows programs. The
project also includes the final executable program itself (*.exe). Along the way towards creation
of the final executable (*.exe), the compiler must convert the source statements into machine
instructions which are saved in object (*.obj) files. A portion of the VC IDE called the Linker
joins these separate object files into the final executable file, merging in all the Microsoft
supplied functions, such as those to input data and output data. And there are many more
additional files the IDE creates to support the project.

The project information is stored in two files, the desktop work space file (*.dsw) and the
desktop project file (*.dsp). Stored in the project information is the location of your compiler
files as well as the exact path to your project files. Thus, when you wish to take your program to
another computer, usually the project files themselves are NOT portable because the location of
the system libraries and the path to your project files are likely to be different on different
machines. This simply means that when you wish to take your programs to another place to work
on them some more, such as bringing them from home to school labs, you should bring your
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 671

source files, header files and resource files (if any), but you must build a new project on the other
computer. (For more information, consult the Transporting Programs to Another Computer
section below.)

Perhaps the biggest advantage to the Visual C++ development platform is program
debugging capabilities. You can dynamically debug or find the errors in your programs. That is,
you can step through your program one line of code at a time and visually examine the contents
of any variable as you go along. You can see which branches are taken. It is amazing what all is
really available to you to assist you in finding the errors in a program. (See the Debugging
section below.)

Once a program is thoroughly debugged, the professional versions of VC also support the
creation of a highly optimized version (for speed usually) for production or Release, as VC calls
it. All the debugging code is stripped out and a fast executable version is created.

The VC platform keeps these two versions of your program, the Debug and Release
versions, separate from each other. That is, the intermediate object files and the final executable
file for each version must be stored in separate subfolders since they have the same file names.
By default, usually these two subfolders are called Debug and Release. (However, sometimes, the
compiler gets confused and creates some really weird subfolder names. Later on I show you how
to remedy this should it occur.) What this means in terms of organization is you must put each
programming assignment into its own separate folder. Then when you run the compiler, it builds
the Debug and Release subfolders beneath that main folder.

Rule 1: All programming assignments should be placed in their own separate folder.

So your first action should be to make a folder using Explorer to serve as a main folder
beneath which you will place subfolders for each programming assignment. For example, if your
class was CS125, you might consider making a main folder called
C:\CS125

Let’s say that you are about to start work on a new assignment, Program 1. Then, as
outlined below in detail, you might decide to place the files and the project for Program 1 in a
subfolder such as
C:\CS125\Program1

Tip: Avoid the use of blanks or spaces in the names of your folders. Why? Because
eventually you are going to have to get to a DOS prompt to obtain the printed output from your
program to turn in. DOS commands are a bit tricky when blanks occur in path names.

Rule 2. Do not make the project folder on a floppy disk in drive A: because the
intermediate output files that the compiler creates exceed the storage capacity of a floppy.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 672

Rule 3: Normally, the name of the resultant executable program (the .exe file) has
the same name as the project.

So if you prefer to have your executable named 2h.exe for example, then the project
folder should also be called 2h and would appear as
C:\CS125\2h

At our school, the preferred location for student programs is D:\Student. Thus, you might
consider making a folder called
D:\Student\2h

Step 1: Building the Program Project

Each programming assignment should be contained in its own folder. The compiler can
automatically create these new folders for you when you create the new project.

Step 1-A. Bring up Windows and launch VC6.

Tip: Personally, on my desktop I have created two shortcuts to the compiler. All I have to
do to launch Developer Studio (VC6) is double click on the desktop icon or select the VC6 item
in the Start menu. An easy way to create this shortcut is to use the Explorer and scroll to find the
Windows or WinNT Start Menu and then the Programs subfolder and then the VC 6 program
folder. In the left window, locate the Visual C++ icon in the VC 6 program folder. Drag the
Visual C++ icon and drop it on the Start button to have a new Start menu item; alternatively right
click the VC icon and drag and drop on an empty spot on your desktop and choose create
Shortcut to have a desktop icon.

Developer Studio Terms:

The Developer Studio supports a work space as the highest level grouping; a work space
represents the entire application package of one or more related programs, such as a complete
Payroll system. The work space contains one or more projects. A project usually contains an
individual program. The Payroll work space, for example, might contain projects for entering
weekly time sheets, calculation of year-to-date totals, check writing, and so on.

Files with the dsw extension represent the Work space while those with dsp extension
represent the projects within the work space.

Your First Design Decision

Thus, your first design issue is how you wish to organize your programming assignments and
thus how you will make use of the work space/project concept. It is highly recommended that
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 673

you make one work space for each programming assignment that contains the one project for
that program. This becomes the easiest system to maintain.

If you are more adventurous, you could make one work space for all the assignments for a
given chapter; that work space would then contain a variable number of projects depending upon
the number of assignments from that chapter. Note that there is a menu option to allow you to
switch between projects within a work space — to make each one in turn the currently active
project. Project | Set Active Project allows you to switch from one to the other of a multi project
work space. When you choose File | New and project as shown below, there will be a check box
for add to the current work space or not. This approach of multiple projects per work space is not
recommended for beginners.

Step 1-B. To start a new work space and subsequent project, choose File | New. This
menu choice is used to add many “new” things to work spaces, projects and resource files for
Windows applications. For a new work space, you should see a dialog similar to this one, Figure
1.

Figure 149 Building the Project

Here I am making in folder K:\cs230\2e a new project and work space both called 2e
(2e.dsw, 2e.dsp) and the executable program when finished is going to be called 2e.exe. Select
the type of project to create as a Win32 Console Application and then enter the location
and project name.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 674

Rule 4: Always choose Win32 Console Application! This is the ONLY choice that
makes a DOS application. All other choices for the type of the project are going to make
real Windows applications.

Once you select the Console Application type, next fiddle with the location and project
name. As you type in a project name, VC6 assumes that that name is also going to be the
subfolder in which the files are to be stored and appends that name to the location path. Given
the above entries, the compiler is going to create a Console project called 2e located in a folder
called 2e located beneath folder K:\cs230. When ready, click Ok; the compiler then asks you if
this is what you want to do by showing a dialog that indicates it will build an empty project. It is
asking you “What kind of Console Application do you want to create?”. The default choice, “An
Empty Project”, is the desired selection for all courses, except Windows Programming. So click
the “Finish” button. Next a “New Project Information” dialog appears summarizing what you are
about to build. Click “Ok.” The compiler now builds the project and any needed folders. It is at
this point that the dsw and dsp files are made.

Note that anything other than building a default empty project also results in a Windows
application project being constructed. If you accidentally make any other type of project or build
something other than an empty Win32 console project, when you later try to build the program, it
will fail to build and you will have to start over with a new project.

TIP: When compiling the program if you get either “This program must be run under
Windows” message or a linker error message “cannot find procedure WinMain,” you have most
likely not selected a Win32 Console Application for the project type. All other project creation
choices result in full-fledged Windows applications being created. Make a new project and make
sure you are selecting a console app.

Now one must add in the various source cpp files into the project. At this point, you face
two possibilities: either the source files already exist or they do not yet exist.

Step 1-C. Create new source cpp files to be contained in the project.

If the cpp files do not yet exist, which is the usual case, you must either click on the File|New
button or choose File |New menu item. You should see the following dialog assuming that you
wish to add a new cpp file.

In Figure 2 below, I have selected a C++ source file to add, provided its name (2e.cpp)
and specified to which project it belongs and its hard disk folder. Click OK and the compiler
creates this new 2e.cpp file and opens an editor window in which you can begin to enter the
program coding. Type in your code and use Build or Make buttons or menu commands to
compile and link the program. (Discussed below in Step 4.)
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 675

Figure 150 Building a New cpp File

Step 1-C Alternate. Adding existing files to the project

The other possibility is that you have already have your source files created. Simply copy your
existing files into the desired folder that is to contain the project. For example, suppose you have
the 2e.cpp file stored on a floppy disk. Use the Explorer to copy that file into the K:\cs230\2e
folder that has been created by Step 1A above. You should now see files 2e.cpp, 2e.dsw and
2e.dsp in that folder.

To add this existing file to the new project, in the left side pane, click on the Files tab and
select the 2e Files icon. Right click on the 2e Files icon and choose the Add Files to Project
menu item. In the normal File Open dialog, choose 2e.cpp. Click Ok and you are ready to go.
This is shown in Figure 3 below.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 676

Figure 151 Adding Existing Files to a

Project

Step 2. Transporting Programs to and from School Computers

The next issue is how to transport programs to/from school so that you may work on them both at
school and at home. You need to copy your source files, including any header files that you have
created in the advanced classes, to a floppy disk. Use the Explorer to copy them to a floppy disk.

Do NOT copy the project files (dsp and dsw) or other intermediate files. You only need to
copy cpp and perhaps h files.

When you are finished working on a program, simply copy the revised source and header
files (if any) back onto your floppy disk to take home. At home, copy the revised files back into
your original folder and launch VC to continue.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 677

Step 3. Opening an Existing Project

How do you open an existing project? Several options are available.

Use the File menu item and look in the Recent Work spaces item and select the one you
were working on last.

Alternatively, the simplest is to have the VC6 compiler automatically reopen the last
project you were working on when you closed VC6. On the Tools | Options | Work space tab
place a check in the Reload last work space at startup to make this option the default.

Alternatively, use File | Open Work space and navigate to the project you wish to open
and activate.

Alternatively, using the Explorer to get to the folder containing the project you wish to
open, double click on the dsp file in the project’s folder. This launches VC6 and opens that
project.

Step 4. Compiling the Program

Once you have the source code entered, the next step is to compile the program and build the
debug executable version of your program. From the Build menu or from the tool bar buttons,
there are three choices: Compile, Build and Rebuild All. (Note: you may want to experiment with
Tools | Customize and Commands tab and drag the Rebuild All button onto the tool bar beside
the Build button.)

Two steps are required to make an executable program: compile the source file(s) into the
machine instructions or object (obj) files and then link (create the exe file) the object files
together along with system provided object files that contain the system functions, such as those
for I/O operations. A Compile menu choice only compiles the current cpp file; it does not
perform the link step necessary to create the executable program. Thus, one normally uses either
Build or Rebuild All menu choices which do both.

If your program consists of a single cpp file, the differences between a Build and a
Rebuild All is practically non-existent. Both compile all the source files and then link the object
files to form the exe executable. However, if one has several cpp files that make up the project or
program, then there is a significant difference between Build and Rebuild All. A Build only
recompiles those source files that have changed since the last compilation.

Suppose your project contained three cpp files. Suppose further that you have made some
changes only to one of the source files. Then a Build compiles only that one cpp file that changed
and then links the new object file together with the previous object files that have not changed to
make the resultant executable. A Rebuild All, on the other hand, always recompiles every source
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 678

file. The difference is then one of speed. This difference really does not become significant until
you write Windows programs (CS250) or have a project with a lot of cpp files in it.

Step 5. Handling Compile Errors

If there are compile errors, they appear in the Output window normally located across the bottom
of the main window.

Tip: if you double click on an error message, it opens that source file (if it is not opened
already) and positions the cursor on the offending line of code, where known. However,
sometimes the compiler does not know where the real error is located, as in the case of mis-
matched begin-end {}.

Tip: if you select an error message by single clicking on it and then press the F1 function
key, a help message window appears providing more information on the nature of the error.

Tip: if you accidentally close the Output Window or even the Project View window, they
can be reopened by using the View Menu items.

Step 6. Where Is the Executable File (*.exe) Located?

Once the Build is finished, use the Explorer to look at the main project folder. You should see a
Debug folder appear beneath that folder. In the above example, the executable is in
K:\cs230\2e\Debug and is called 2e.exe.

However, occasionally, VC creates a totally weird subfolder instead of the Debug folder.
You can either live with the strange subfolder or instruct the compiler to place the object and
executables in a Debug folder. Use Project | Settings to bring up all the project options. In the left
pane of the tabbed dialog you should see “Settings for Win32Debug” and an icon for 2e project.
In the General tab on the right side of the dialog, find the lower Output Directories edit boxes.
The Intermediate and Output files edit controls show the name of the subfolder to create and use.
If it does not say Debug in both of these edit controls, simply enter Debug for both. Click Ok and
then do a Rebuild All.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 679

Step 7. Running The Program

To make test runs and debug the program, there is a wide variety of choices and circumstances.
First, let us look at all the possibilities for running the program and then we’ll look into how to
actually use the debugger to find your runtime errors.

A. No input file — I will type in the input data when the program asks for it.

This method is the simplest and satisfactory for a first run with many programs. In other words,
when your program does an input operation, you are going to type in all the requested data.

There are two ways you can just run the program — the Execute button (or Menu: Build
— Execute) and the Go (F5) button (or Menu: Build — Start Debugging — Go F5).

What’s the difference? Execute simply runs the program in a window. It gives you the
tool bar buttons to mark and copy results to the clipboard so you can paste the test run results into
Notepad and print them. Go runs the debugger and maximizes your ability to find bugs.

B. I want to use a file of provided test data but the program is not actually using an
input file — it uses cin.

Very often, the programming assignments ask you to run your program with provided test
data. Two possibilities arise centered around whether or not your program is defining an input
file and opening it or just reading the input from cin, the keyboard. Method B expects the
program to be reading the data from cin.

Locate where you have put the test files that you need. When you specify a test data
filename, you can provide either the simple filename or a fully qualified path. The simplest
approach is to just copy the test data file(s) from where they are located into the project folder
where the cpp file is located. It’s either copy the file(s) into the project folder or use a full path.

To use a file to replace input from a keyboard cin request, DOS input file redirection
must be used. This is easily done. Choose the menu Project | Settings and the Debug tab. On the
Debug tab settings page, find the Program Arguments edit control. In this edit control enter a <
(less than sign which is the DOS redirection of input command) and then put in either the short
filename with extension or the full drive and path and filename. For example, entering
< test1.txt
is specifying that the input file is called test1.txt located in the project folder. Specifying
< A:\CS125\test1.txt
is telling VC6 the file is located on the floppy in drive A: and within the \CS125 folder.

Now whenever you press the Go button or F5 key, your program will obtain all its input
data from the above test1.txt data file. This option is saved in your project so you do not need to
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 680

reset it the next time you want to run. This is shown in Figure 4.

Figure 152 Specifying an Input File for DOS Redirection to use with cin
Tip: If your program cannot seem to find the file to open it, double check several things.
First, verify with Explorer that you have the correct drive and full path to the file. Verify you
have the filename correct. Remember to use the Explorer option View-Options and uncheck
“Hide file extensions for known file types” so that you can see the complete filename. If you
have the right drive, path and filename, then go to a DOS prompt and go to that drive and use a
CD (change directory) command to get to the folder that contains the file. Then enter a DIR
command. Carefully note the case of the letters in the dos filename. Finally, double check the file
size using the Explorer and make sure it is not 0 bytes.

Pressing the Go button or F5 (Execute button alternatively) runs the program which then
inputs the data directly from the file.

C. Using an Actual File Stream, an ifstream Instance

If you are using a real file stream instance in your program, then in your coding you can specify
only the file name or the full path. That is sufficient to find the file. Nothing more needs to be
done. Pressing the Go button or F5 (Execute button alternatively) runs the program which then
inputs the data directly from the file.

Tip: it is simpler coding if you do not have to provide the full path in your program.
However, that means you must ensure that the file has been copied into your project folder.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 681

With ALL of these approaches, if you just press the Go Debug button or F5 key, VC6
opens a DOS window, runs the program and then when the program finishes, it closes the DOS
window. It gives you no time to view the results of the program execution! Thus, we must
examine some of the basic concepts of running the debugger so that we can have the output
window visible when the program is finished as well as be able to examine the contents of
variables as the program runs.

On the other hand, if you use the Execute button, the DOS window remains visible when
the program is finished. However, no debugging is available to assist you in finding errors.

Step 8. Program Debugging and Execution

The really nice feature of VC and Win32 console applications is its total runtime debugging
support. You can simply click on the debug button Go (F5) to begin the execution of your
program. If a runtime error occurs, the debugger will pop up at the point of the error with full
debugging information available. If you place the mouse cursor on top of a variable in the code,
the debugger will immediately show you the current contents of that field! The list of features
goes on and on.

However, normally you do NOT want to just press the F5 key or use the Go button. Why?
Yes, if there is a runtime error, the debugger pops back up at the offending instruction. But if
there is no hard fail, when the program terminates, the debugger closes the DOS window and
shuts down. Your output window disappears leaving you with no idea of whether or not it ran
correctly.

For simpler programs, press the F10 (Step Over) and or F11 (Step Into) key. These
buttons execute one instruction of your program and then stop, allowing you to view the results
of that single instruction or watch the paths taken when decisions are made. The difference
between F10 and F11 is simple. The difference is Step Over versus Step Into. When the
Debugger gets to a function call either a system provided one, like cin >> quantity, or a user
written function, Step Into goes on into the first instruction of that function so that the function
itself can be debugged. Step Over simply executes the entire function and then returns control to
you. You normally want to Step Over system functions and Step Into your own functions.

This approach is fine for smaller programs. But for larger programs with many functions
it is impractical if not extremely time-consuming. There is another way, using breakpoints.

The first action of a debug run is to set a break point. A break point is a flag placed on a
line of code that tells the debugger to halt the execution of your program when and if it gets to
that line of code. Technically, the line that has the break point set on it is the next instruction that
would be executed if you continue to run the program. In other words, it has not yet been
executed.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 682

How do you set break points? Click on the line where you want the execution of your
program interrupted so you can inspect things. Press either F9 or the button Insert/Remove
Breakpoints. You should see a red dot appear to the left of that line of code. Press it again and the
break point is removed. It acts like toggle switch — on — off — on — off.

Once you have set a break point, then go ahead and run the program with the Go button
or F5 key.

On which line of code should you set a break point? Ah! That depends on the errors you
have in your program and what you want to see and so on. However, for sure, you should place a
break point on the last line of the main program, on the
return 0; // terminate and return back to DOS
Figure 5 shows the setting of just such a break point in a simple C++ program that is to print
Hello 5 times and then quit.

Figure 153 Setting a Breakpoint and the Debugger Run

What Can You Examine Using the Debugger?

First, by tracing through your program one line at a time, you can verify the flow of control is
doing what you expect. Consider the following main loop that is to input a set of data and
produce a report, one line for each set of data.
int main () {
int cost;
int qty;
int total;
while (cin >> cost >> qty);{
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 683

total = cost * qty;

cout << "total: $" << total << endl;
}
return 0;
}

When it is run, the first action is to assume it works perfectly and set a break point on the
return 0; instruction. Of course, when you run the program and enter 5 sets of data, you discover
only the last set of data has its total printed. Now it’s time for serious debugging. You can
perhaps stare at the code and find the error, or you can let the debugger find it for you in just a
few quick steps.

Having discovered that the program is not working, restart the program and use the F10
Step Over option to begin tracing through the program line by line. You should see the debugger
get to the while statement. When you press F10 to execute the while and input the first set of
data, control goes to the DOS window and you enter the first pair of values. When you press F10
expecting to now execute the total calculation, you find instead you are back at the while
statement, ready to input the second set of data. As you continue pressing F10, you should
discover that only after the last set of data is entered and you have pressed the DOS end of file
key, the Ctrl-Z key, only then does control go to the total calculation. The error is now easy to
spot. There is a misplaced ; after the while clause and before the begin block of the while body.
This makes the intended while clause body execute only when the loop ends. This is a while loop
with no body because of the errant semicolon.

Checking the Contents of Variables

You can observe the contents of variables in several ways. First, just float the mouse over a
variable for a half second and a tiny window appears beside the cursor with its contents.

In the bottom pair of windows you can also watch the contents of variables. In the left
pane, the debugger displays the contents of the variable currently involved in the current or most
recent instruction or two. In the right pane is the watch window. Here you can enter a variable’s
name and have its contents permanently displayed. Simply click on an empty line and type in the
name of the variable and press enter. Typing over an existent entry replaces that variable with the
one you type in.

Step 9. The Help System

The VC6 platform has a large amount of Help topics available. There are several ways Help can
be used. The easiest is to simply select or highlight a function name and press F1 for help on that
function. You can also select a function name and choose Help — Search. Or you can just do
Help — Search and enter any word or series of words. It is dynamic in responding to keystrokes.
That is, as you start to type in say s of sqrt, the square root function, you should see in the lower
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 684

help window a list of all topics that start with s. As you type in more characters of sqrt, say sq,
Help narrows the list of topics.

If you do Help — Contents, then the left pane’s Info tab activates and you should see a
tree view of all available reference books and materials. Simply click on the + signs as usual to
expand a topic. When you get to an article you want to read, double click on that entry and the
Info Topic window opens.

Step 10. Some VC6 Options

Take a minute to look over the options — Tools | Options. You should see a large tabbed dialog
or set of property pages as they are called in Windows.

Notice that you have control over the tab amounts — see the Tab tab. Personally, I always
use the Insert Spaces option instead of using tabs. It causes fewer surprises when printing the
source files to hand in with your programs.

Look over the options that are available in each of the tabs. You may find some that you
may prefer to use. Notice that you need to use the upper right corner scroll buttons because there
are so many tabs of options.

Specifically, I call your attention to the Format Tab. When you go to print out your
programs, you can control the font size used. I use the Format tab and select Source Windows in
the Category window. Then choose a font and a font size that makes a good looking print out for
you. I have found on my system, that the font I prefer to see in the edit windows is not the same
font I prefer when I am printing the source listings. This is where you can tweak them. Figure 6
shows the selection of a different source file window font.

Figure 154 Selecting the Source Window Font

Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 685

Step 11. Getting the hardcopy documentation for programs to hand in to your
instructor — or how to run your program at the DOS prompt

The hard copy requirements may vary from instructor to instructor. However these should cover
most requirements.

First, you need to get a print out of your source file(s). The easiest way is to simply press
the Print button in the VC platform. It prints the currently active window. So if you have several
files to print out, make one active and print it. Then, make the next active or open that file and
print it.

Tip: you can open a source file (or header file) very easily. In the left most pane in the
Files tab, select the source file you wish opened in the editor and double click it. Presto, it is
opened. Also you can do a File | Open or use a tool bar button.

Tip: if you do not like the font size of the source file print out, you can temporarily
change it. Use the Options | Format tab as shown on the previous page. Select Source Window.
Then install a larger font size or a different font that has a larger size available. Reprint the
source. If the new font is too large for screen reading, reselect the old one when the printing in
done.

Tip: if your source file is precisely 60 lines long and you hit File | Print, nothing appears
on the printer. This is a known bug in VC6. Simply make your program one line longer or one
line shorter and it will print.

Getting the program’s output is trickier. How you can do it often depends on the precise
circumstances of how you are handling the input operations and how much output the program
produces. Let’s examine all the possibilities.

First, there are two modes in which you can run a DOS prompt: full screen DOS or DOS
in a window. When you click on a DOS prompt, you are in one of these two modes. To switch
from one mode to the other, press Alt-Enter. Alt-Enter switches from one mode to the other,
toggling between a window version and a full screen version. When you run a DOS prompt in a
window, you are really in a graphics mode; this means if you hit the PrintScreen button, the
window graphics image is copied onto the Windows clipboard ready to be Edit-Pasted into
Paintbrush, for example. With DOS in a window, that image is a bitmap image and can only be
pasted into image processing programs like Paintbrush. On the other hand, when you run DOS in
full screen mode, you are in a text mode; the Print Screen operation places the text lines onto the
Clipboard and which can then be Edit-Pasted into Notepad and printed using File|Print. Also
when you use the Execute button from the IDE to run your program and if you use the tool bar
buttons of the DOS window to select and copy the output text, it is in text mode so that you can
paste them into Notepad.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 686

Secondly, there is a big difference between a screen shot done using PrintScreen and
using DOS redirection to the printer ( > PRN) or redirection to a file to be later viewed (printed)
with Notepad ( > a:\results.txt). Screen shots capture the contents of the screen. Appending >
PRN or > file to your command line causes DOS to redirect all output of your program to the
printer and not to the screen.

What’s the difference? There are some side effects of which you must be aware. If your
program is producing a report that contains 30 lines of output, a screen shot is totally
unacceptable in all programming classes because you are only going to get the last 24 or so lines,
the others scroll off the top of the screen. In this case, you must use the DOS redirection method
which sends all output to the printer or a file. However, if you are entering the data values in
response to a prompt from the program to enter some data, then the screen shot is the better
approach to use because the redirection of output does not redirect a copy of the values you type
in for the input operations.

Thus, if a screen shot is desired, Use the Execute button. Choice 1: When the “Press any
key” message appears, hit the PrintScreen button and paste into PaintBrush or Notepad and print
from there. Choice 2: You can use the program’s DOS window’s toolbar or System Menu to
“Mark” the portion to be copied and then “Copy” that selected portion to the Clipboard and Edit |
Paste into Notepad. The procedure is to press the Mark button or bring down the System Menu
by pressing the system icon in the upper left corner of the DOS window and choose Edit | Mark.
You should see a rectangular blinking cursor in the upper left corner of the DOS window. Drag
down and to the right to select the desired text to be copied. The press the Copy toolbar button or
System Menu | Edit | Copy menu to copy it to the clipboard. Open Notepad and Edit | Paste into
Notepad. Then File | Print from Notepad.

If you have more output than can be captured with a screen shot, then use DOS output
redirection. Use Project | Settings - Debug Tab. See Figure 4 which shows input redirection. In
the Program Arguments edit control, type
> results.txt
Click Ok. Now use either Execute or the Go button to run the program. Remember, all screen
output is going to the results.txt file and not to the screen. When the program is done, open
results.txt located in the project folder along with your cpp file.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 687

Running Your Program from a DOS Prompt

Alternatively, you can run the program from a DOS prompt. Launch a DOS prompt. For all test
runs, the very first action at the DOS prompt is to make the current default drive and directory
be where the executable version of your program is located. In the above program 2e example,
one would issue a change directory as follows:
C>K:
K>cd \Cs125\2e\debug
K:\Cs125\2e\debug>
You can make sure you are all set by issuing a DIR command and looking for the 2e.exe file.
Next, proceed with the most appropriate choice below.

Tip: If you have put blanks in the path names, the above cd or change directory command
fails. You need to put “” around the complete path name
K>cd "\Cs125\2e\debug"

Choice 1: You are manually typing in the data that the program is requesting. This
circumstance occurs most frequently on the very beginning programming assignments. Quite
often, the program prompts you to enter a set of values and then prints a set of results. The output
completely fits on screen, that is within the 24 lines or so available to DOS before the screen
scrolls.

A. Switch to full screen mode, if you are currently in a DOS window (Alt-Enter). You
can issue a CLS (clear screen) dos command to clear the screen. Now type in the program name
and enter the values as required by the program specifications. For example, you might have the
following
K:\Cs125\2e\Debug>2e
Enter cost: 42
Enter qty : 10
Total is $420
K:\Cs125\2e\Debug>
Now press the Print Screen button. This makes a text copy of all the lines on the screen, putting
that copy on the clipboard. Now task switch back into Windows - use the Alt-Tab key. Launch
Notepad (Start | Programs | Accessories | Notepad for example). Do Edit | Paste and you should
see the entire screen’s contents. Now you can do a File | Print.

B. Switch to DOS in a window mode. Run the program as above. Now notice the tool bar
and its buttons. The far left one is Mark. Press this button and a blinking rectangle appears in the
upper left corner. You then drag using the left mouse button down and to the right, selecting that
portion of the window you wish to copy. When you have selected all the text, press the Copy
button. This places a copy of the selected text onto the clipboard. So continue with Notepad as in
A above. Note, Under Windows NT, these Mark and Copy operations are available from a
context menu activated by right clicking the title bar and choosing Edit menu items.
Appendix C: How to Use Microsoft’s Visual C++ 6.0 Compiler 688

Choice 2. A file of data is needed to be input using DOS redirection or the program
produces more lines than can be shown in a dos window. When using this choice, it does not
matter whether you are in DOS full screen mode or using DOS in a window. To use DOS input
redirection, insert a < sign and the full path to where the data file is located. DOS input
redirection works as follows. When a program asks dos for another keystroke which would
normally be come from the keyboard, the input redirection causes the next character in the data
file to be given to the program as if the corresponding key on the keyboard had been pressed. It is
a bit of DOS internal trickery.

Assume that our program 2e actually needed a test data file that you downloaded from
school called 2e1.txt and that you had placed that file in the A:\Cs125 folder. You would then
enter
K:\Cs125\2e\Debug>2e < A:\Cs125\2e.1
I would run it this way with all the output going to the screen to make sure all is well. If you get
bad command or file not found, then DOS cannot find either 2e.exe or the file 2e1.txt. Double
check your folders using Explorer. When all looks well, then get the printed copy by appending
the > PRN (redirection to the printer option) or > results.txt.
K:\Cs125\2e\Debug>2e < A:\Cs125\2e.1 > PRN
or
K:\Cs125\2e\Debug>2e < A:\Cs125\2e.1 > results.txt
Tip: if you have to do much of this redirection stuff, copy the test data file into the debug folder.
It simplifies the command to just
K:\Cs125\2e\Debug>2e < 2e.1 > PRN

Choice 3: Your program is actually using input file operations (an instance of ifstream).
This is the easiest one to handle. To run the program and see the results on screen, just enter the
name of the program
K:\Cs125\2e\Debug>2e
When all is ready for the print out, enter the > PRN or > results.txt.
K:\Cs125\2e\Debug>2e > PRN
or
K:\Cs125\2e\Debug>2e > results.txt
Note that suggested that these output files have the .txt extension. This way, simply double
clicking on them in the Explorer will launch Notepad to view and print it.
Index 689

Index
? : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
+= operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
^Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
#define . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371, 372, 415, 511, 546, 589
#include . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
To include user header files, use double quote marks . . . . . . . . . . . . . . . . . . . . . . . . . . 596
#include <iomanip.h> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
#include <iostream.h> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
#include directive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
_MAX_PATH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511, 517
A More Compact Loop That Detects End of File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
abs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Accessing Array Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
AND logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
AND operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
AND operator has a higher precedence than does the OR operato . . . . . . . . . . . . . . . . 122
Appendix A: How to Use Microsoft’s Visual Studio .NET 2005 Compi . . . . . . . . . . . . . . . . . 631
Applications of Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
binary search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
Defining Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Initializing an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
Inputting a Known Number of Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
Inputting an Unknown Number of Elements Until EOF Is Reached . . . . . . . . . . . . . . . 419
Inputting the Number of Array Elements To Be Input . . . . . . . . . . . . . . . . . . . . . . . . . 418
Inserting Another Element into an Unsorted Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
Merging Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467, 476
Methods of Inputting Data into an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
Ordered (Sorted) Lists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
Parallel Arrays and Sequential Searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
Passing Arrays to Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
Sorting an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
Straight Selection Sort . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
subscript out of bounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
the Output Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
Using an Array for Direct Lookup Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
Using Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
Working with Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
Arrays of Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 557
ASCII . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356, 357
Assigning Larger Sized Integers to Smaller Sized Integer Variabl . . . . . . . . . . . . . . . . . . . . . . . 91
Index 690

Assigning Smaller Sized Integers to Larger Sized Integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

assignment instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Assignment Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46-48
assigns a floating point number to any of the integer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
bad () . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Basic Computer Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
binary file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
Mechanics of Binary Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
Reading a Binary File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
Writing a Binary File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
binary number system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
binary search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
bisect() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255, 310-312
Bisection Method
Generic Bisection Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
Using a Generic bisect() Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
blank lines to separate key lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
bool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239, 461
Functions that Return a bool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
bool Data Type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
convert an integer relational expression result into a bool . . . . . . . . . . . . . . . . . . . . . . . 125
false . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
true . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
break Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
Breaking a Complex Calculation Down . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Bulletproofing Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173, 174, 196
C++ startup code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
calcTax() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
Calculating N! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369, 386
Case of a Variable Name . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
cctype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
cerr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
chaining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
char . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91, 351
Character Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Character Processing Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
character variable can be initialized . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
Defining Variables to Hold a Character of Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
How Are Character Data Stored? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
insertion operator can be used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
Numbers and Letters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
Output of Character Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
Index 691

The escape sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358

The Processing of Character Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
The put() function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
Using the Extraction Operator to Input a Character . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
Using the get() Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
cin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49, 50, 151
clog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
close() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156, 174
Closing a File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
coefficient of variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
Comments in C++ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
compiler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Compound Assignment Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Conditional Expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
syntax of the conditional expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
const . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458, 594
const char* . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
const int . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275, 372, 415, 458, 507, 546
Constant Data Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Constants and Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
double . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
floating point constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
int . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
long . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
suffixes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
unsigned int . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
unsigned long . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
continue Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
Controlling Leading 0's on Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
converging series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
CopyPartialString() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
cos() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Counters and Totals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
cout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28, 54
Creating Output Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
CS02-1 Ticket Prices for a Concert . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
ctype.h . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
current position within the stream pointer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Cycle of Data Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11, 157, 217
data conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48, 598
All types of char and short (signed and unsigned) are automa . . . . . . . . . . . . . . . . . . . . 89
assigning any floating point type to any of the integer . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Assigning Smaller Sized Integers to Larger Sized Integers . . . . . . . . . . . . . . . . . . . . . . . 90
Index 692

Calculations Involving Multiple Floating Point Data Types . . . . . . . . . . . . . . . . . . . . . . 93

converts the final value of an expression to the data type of th . . . . . . . . . . . . . . . . . . . . 91
converts the one of the lesser rank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Data Type Ranking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Mixed Mode Math . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Principles of Data Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
when assigning any float data type to any of the integer data ty . . . . . . . . . . . . . . . . . . . 95
data segment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
Data Type Ranking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
daysInMonth() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
debugging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
dec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
dec Manipulator Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
Decisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
decrement operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
define multiple variables of the same data type with a single st . . . . . . . . . . . . . . . . . . . . . . . . . 44
Definition of Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42, 44-46
Designing Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10, 13, 15, 40
diverging series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Do Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Do Case Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
Do Case syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
Do Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Do Until Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
syntax of the Do Until . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
What is the difference between a Do While and a Do Until . . . . . . . . . . . . . . . . . . . . . 178
Do While . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
dot (.) operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
double . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42, 61, 88, 93
Efficient Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
eject to a new page . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
End of File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
endl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Engr02a - Pressure Drop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
entering some real or floating point constant numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
enum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372-374, 376-378, 381, 386
Anonymous enums . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377, 404
Enumerated Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
enum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372-374
eof () . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Escape Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
Index 693

exp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
exp() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
extern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
extraction operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49, 154
fabs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
file can have no data in it . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
How are the input files created . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
ifstream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Input Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
findAverage() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
findRanges() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
float . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88, 93
Floating Point Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
double . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
float . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
long double . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
for Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179, 182, 183, 192
for syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
fstream.h . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
calctax() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229, 230
Code the Function’s Body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Define the Function Header . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Define the Function’s Prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
function coding must be outside of any other block . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
higher() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Invoke or Call the Function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Writing Your Own Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
function body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
function header . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
function prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
functional decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
functions
functions are often written to unclutter the main() . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Functions that Return No Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
How a Function Returns a Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
More on Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
pass parameters by address . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
variable is passed by value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Writing Your Own Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Gauss elimination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568
Gauss Jordan method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 570
Index 694

get() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272, 273, 354, 508, 510, 512

getline() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508, 511, 512
global optimize for speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
global variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
Global Variables in Other Cpp Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
Use of Global Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
Global/External Storage Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
lifetime of a global/external variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
good () . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
good() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Grand Totals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
headings() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244, 281, 282, 293
hexadecimal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243, 353
higher() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226, 233, 237
hms_to_totsec() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
How Integer Data Is Stored in Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
How much do you indent? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
How Parameters Are Passed to Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
How to Pass iostreams to Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
I/O Stream States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
IDE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
if (!infile) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153, 156
if (!x) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
if (x = y) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
if (x) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125, 126
If-then-else . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113, 118
AND or OR relational operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
a single statement can be a null statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Components of an If-then-else . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Compound Test Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Decisions can be nested . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
efficiency concern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
else . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
else-clause is optional . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
If-then-else Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Nested Decisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
nothing to do when the statement is false . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
only a single statement to do . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
provide the missing else-clause . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
The Test Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
ifstream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151-153, 508, 517
Increment and Decrement Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
increment operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98, 181
Index 695

postfix inc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
prefix inc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Initializing Variables and the Assignment Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Input of Data Values into Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
inputData() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
Inputting Integers that have Leading Zeros . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
inquiry program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
Insertion of Floating Point Numbers into an Output Stream . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
insertion operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28, 51, 52
Additional Insertion Operator Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
int . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Integer Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84, 85
Which Type of Data Do I Use in My Program? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
integer division . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Integer Versus Floating Point (Real) Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Integration Using Simpson’s Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
iomanip.h . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30, 54
ios::adjustfield . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
ios::binary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
ios::fixed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54, 55
ios::floatfield . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
ios::left . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
ios::right . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
ios::scientific . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
ios::showpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
iostream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
iostream.h . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30, 49
istream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Testing for Goodness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Iterative Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Keyboard Data Entry Sentinel Controlled Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
Labeling Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
left justification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
lineCount . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281, 300
Linker program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
list . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
literal constants. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
loadArray() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422-424, 601, 617
LoadBudget() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 550
local heap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
logic error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Logical Not Operator - ! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Index 696

long double . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
loop control variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158, 183
Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Loops That Are to Be Executed a Known Number of Times . . . . . . . . . . . . . . . . . . . . . . . . . . 158
Loops to Input All Data in a File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
main() function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25, 59
MatchId() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
Math Library Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60, 62
Math Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Matrix Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
Addition and Subtraction operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
column matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
diagonal matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
Gauss-Jordan Method of Solving Simultaneous Linear Equations . . . . . . . . . . . . . . . . 570
identity matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
Mathematical Theorems of Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 567
Matrix multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
Multiplication of a Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
row matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
square matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
The Gauss Method for Solving a System of Linear Equations . . . . . . . . . . . . . . . . . . . 568
Two matrices are said to be equal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
Maximum and Minimum Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Rule: when finding the maximum or minimum values . . . . . . . . . . . . . . . . . . . . . . . . . 170
mean value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
meaningful names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Mechanical Mouse Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Mechanical Robot Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
member variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
Menus as Sentinel Controlled Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
mixed mode math . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94, 95
Multi Dimensioned Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
All elements of every row are entered, but not all rows are inpu . . . . . . . . . . . . . . . . . 550
column offset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
Defining Multi Dimensional Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
Initialization of Multi Dimensional Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
Input - All elements of every row and all rows are entered . . . . . . . . . . . . . . . . . . . . . . 550
Loading a Multi Dimensional Array from an Input File . . . . . . . . . . . . . . . . . . . . . . . . 549
Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
Passing Multi Dimensioned Arrays to Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
Physical Memory Layout Versus Logical Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
row offset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
Index 697

Working with Multi Dimensioned Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551

Multiple Assignments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Multiple Level Control Break Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293, 299
control fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
processing steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
name of a function is the memory address . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
namespace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
namespace std . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
NameToCommaForm() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
NameToParts() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
Nesting of Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184, 185
new line code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28, 29
Newton’s Method of Root Solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
null terminated strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
Numerical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249, 251
desired degree of accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
estimate of the error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
limited range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
rapidity of convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
Numerical Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Integration Using Simpson’s Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
Simpson’s Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
trapezoid method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Numerical Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
object file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
octal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
ofstream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
open() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151, 517
OR || operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
OR logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
|| . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
ostream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28, 53, 293
output file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
How do you write data to the file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
output manipulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Output of a Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51, 56
output stream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
overflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Integer Variable Overflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87, 88
pageNum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
parallel arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
PI
Most Accurate Value of PI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Index 698

Plot() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442, 443

Plotting Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
pow() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Precedence of Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128, 129
Precedence or Priority of Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Use () to override the precedence of operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Primed Input Loops that Detect End of File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
primed loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163, 164
PrintReport() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601, 617
printResults() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
pseudocode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
put() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356, 444
quadratic equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
read() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598, 599
reference variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Reference Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Need for Reference Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
reference variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
To make a reference variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Registers and the Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Regula Falsi Root Solving Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
remainder operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Right alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
Root Solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Bisection Revisited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
Newton’s method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
Regula Falsi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
secant method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Summary of Root Solving Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
the Bisection Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251, 254
Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-27, 30, 31, 42, 43, 46, 47, 50
All automatic and parameter storage class variables are stored o . . . . . . . . . . . . . . . . . 236
Data Conversion Rule 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Data Conversion Rule 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Data Conversion Rule 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Data Conversion Rule 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
entire If-then-else must be within the loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
For local type variables, the scope is from the point of their d . . . . . . . . . . . . . . . . . . . 232
For parameter type variables, the scope is from the point of the . . . . . . . . . . . . . . . . . . 232
If a valid character to be input could contain white space . . . . . . . . . . . . . . . . . . . . . . . 355
inner loop must be entirely contained within the outer loop . . . . . . . . . . . . . . . . . . . . . 184
pass iostreams and file streams by reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Index 699

Rule. All while loops can be rewritten as a more compact for . . . . . . . . . . . . . . . . . . . 181
Rule: when finding the maximum or minimum values, initialize . . . . . . . . . . . . . . . . . 170
Rules of Use for Structure Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
The function coding must be outside of any other block of coding . . . . . . . . . . . . . . . . 223
The rules for counters and totals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
scope of a variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
variables to become hidden within blocks of coding . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
Secant Method of Root Solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Sentinel Controlled Input Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
sentinel value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
setf() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54, 161, 513
setfill() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
setprecision() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54, 55
setprecision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
setup floating point format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160, 175
setw() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53, 355
setw() Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
short . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87, 88
Simp() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
sin() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
sizeof() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510, 517, 552
SortArray() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559, 601
SortArrays() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
source program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
spreadsheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
sqrt() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236, 288
standard deviation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
static . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Static Storage Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
global/external storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
lifetime of a static variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
main storage classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
scope of a static variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
static . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Statistical Computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
stdlib.h . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
storage classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
automatic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
strcat() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517, 519
strchr() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
strcmp() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515, 518, 519
Index 700

strcpy() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516, 520, 560

stricmp() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515, 518, 519, 559
String Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
string.h . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
array of character strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 557
Character String Manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
Defining Character Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
extraction operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
get() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508
getline() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508
How Could String Functions Be Implemented . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
Input - All strings Are Delimited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
Input - All Strings Have the Same Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
Input - Is the Last Field on a Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
Inputting Character Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
Left alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
null terminator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
null-terminator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
Outputting Character Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
Passing a String to a Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514
string functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
trailing blanks can be removed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
Working with Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514
strlen() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517, 518
strlwr() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
strrev() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
strstr() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
structure variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
Structured programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23, 24
Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
arrays of structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 590
Binary Files and Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
can also contain instances of other structures and arrays of oth . . . . . . . . . . . . . . . . . . 591
can contain arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589
Creating Instances of a Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 590
Defining Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
How are Structure Members Accessed? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
How are Structures Initialized? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
pass the structure variable by reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594
struct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
structure members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589
structure tag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
Index 701

structure template . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588

structure variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 590
structure variable can be passed to a function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
Structure variables can be used for . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
Where are the structure templates located . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589
strupr() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
subscript . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
subscript out of bounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
sumcol() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 552
Summary Reports Based upon Control Break Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Summation of a Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Summation of Infinite Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
sumrow() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366-369, 386
syntax error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
table lookup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
tan() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Test conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115, 127, 157
basic format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Data Type and Value of Relational Expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
result of any test condition or relational expression . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
six comparison operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
testing floating point numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Testing for Bad Data Entry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Testing of Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
To develop a total . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
Top-down design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216-218, 245, 254, 301
totsec_to_hms() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
toupper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Trap() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
Trapezoid Method of Numerical Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Typecast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91, 92
typedef . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
types of variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
global . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
local . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Types, Scope and Storage Classes of Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
unsigned char . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
User Written Header Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
Using a Generic processFile() Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
Using Menus to Control Program Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
Index 702

variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22, 40, 46

variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
Vector Coordinate Conversions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
What is a Computer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-9
What is a program? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Where are Global and Static Variables Actually Stored . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
Where are the variable definitions placed within a program? . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Where Should Error Messages Be Displayed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Which Type of Data Do I Use in My Program? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Which Type of Data Do You Use for Which Variable? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
while . . . . . . 157-163, 165, 167, 173, 178, 179, 181, 182, 184, 196, 292, 419, 459, 510, 595, 599
while (cin >> quantity >> cost) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
while (infile) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163-165
White space, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
write() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
ws is another manipulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419