3 C++ at a Glance...................----------...............................................................................................

73
3.1................................................Intro
duction................................................................ w.................................................................................. 73
3.2.......................................................................................................... Data Encapsulation and Abstraction
73
3.3..........................................................................................................Inheritance—Derived Classes
78
3.4..........................................................................................................Polymorphism-Operator Overloadin
81
3.5..........................................................................................................Friend Functions
84
3.6..........................................................................................................Polymorphism—Virtual Functions
87
3.7..........................................................................................................Generic Classes—Class Templates
90
3.8..........................................................................................................Exception Handling
92
3.9..........................................................................................................Streams Computation
95
Review Questions............................................................................................................................. 99

4 Data Types, Operators and Expressions ....................................

100
4J____Introduction .............................................................................................
1(10
4? Character Set............................................................................ ____...............................................
100
4.3 Tokens, Identifiers, and Keywords................................................................................................. 101
4 4 Variahles......................................................................................................................................... 102
4.5 Data Types and Sizes.................................................................................................................... 103
4.6 Variahle Definition......................................................................................................................104
4.7 Variable Initialization...................................................................................................................104
4.8 Characters and Character Strings.................................................................................................. 106
4.9 Operators and Expressions............................................................................................................ 107
sdv Contents

4.10.......................................................................................................Qualifiers
109
4.11.......................................................................................................Arithmetic Operators
112
4.12.......................................................................................................Relational Operators
115
4.13.......................................................................................................Logical Operators
118
4.14.......................................................................................................Pit-wise Operators
120
4.15.......................................................................................................Compound Assignment Operators
124
4.16.......................................................................................................Increment and Decrement Operators
125
4.17.......................................................................................................Conditional Operator (Ternary Oper
126
4.18.......................................................................................................Special Operators
128
4.19.......................................................................................................typedef Statement
128
4.20.......................................................................................................Promotion and Type Conversion
129
4.21.......................................................................................................Constants
131
4.22.......................................................................................................Declaring Symbolic Constants-Liter
134
4.23.......................................................................................................Enumerated Data Types
136
4 24 Macro Functions .......................................................................................................................... 138
4.25 Operator Precedence and Associativity......................................................................................... 140
Review Questions......................................................................................................................... 142

5J___Introduction.............
.....................................................................................
............143
5.2 Statements and Block...................................................................................................................... 143
13 if Statement...................................................................................................................................... 144
S4 if-p.lse Statement............................................................................................................................ 146
S S N«ted if-else Statements...............................................................................................................148
return temp;
)
// it is just a friend function, it is not a member of
counter classes ostream & operator « ( ostream &
Out, counter & counter )
{
/
/

d
i
s
p
l
a
y

a
l

Copyrighted material
sdv Contents

i
n
t
e
r
n
a
l

d
a
t
a

o
f

c
o
u
n
t
e
r

c
l
a
s
s

c
o
u
t

c
o
u
n
t
e
r
.
v
a
l
u
e
;
//
retu
rn

Copyrighted material
sdv Contents

out
put
stre
am
Out
for
cas
cad
ing
pur
pos
e
retu
rn
Out
;
)
v
o
i
d

m
a
i
n
O

{
counter counterl;. // calls no
argument constructor
counter counter2 = 1 ; // calls 1 argument
constructor cout << "counterl initially = " «
counterl « endl; cout « "counter2 initially = " «
counter2 « endl; // increment counter ♦♦counterl;
counter2++;
cout « "counterl on increment = " « counterl «
endl; cout << "counter2 on increment = " «
counter2 « endl; // decrement counter --counter1;
counter2--;
cout « "counterl on decrement = " « counterl «
endl;
cout « "counter2 on decrement = " « counter2 «
endl;
counter counter3; // calls no
argument constructor
counter3 = counterl + counter2; // calls
operator*(counter) cout « "counter3 =
"counterl+counter2 = " << counter3;
)

Run
c
o
u
n

Copyrighted material
sdv Contents

t
e
r
l

i
n
i
t
i
a
l
l
y

c
o
u
n
t
e
r
2

i
n
i
t
i
a
l
l
y

c
o
u
n
t
e
r
l

o
n

i
n
c

Copyrighted material
sdv Contents

r
e
m
e
n
t

c
o
u
n
t
e
r
2

o
n

i
n
c
r
e
m
e
n
t

c
o
u
n
t
e
r
l

o
n

d
e
c
r
e
m
e
n

Copyrighted material
sdv Contents

c
o
u
n
t
e
r
2

o
n

d
e
c
r
e
m
e
n
t

c
o
u
n
t
e
r
3

c
o
u
n
t
e
r
l
+
c
o
u
n
t

Copyrighted material
sdv Contents

e
r
2

The contents of the object counterl can be

displayed by using the statement
c
o
u
t

c
o
u
n
t
e
r
l
;

i
n
s
t
e
a
d

o
f

u
s
i
n
g

t
h
e

Copyrighted material
sdv Contents

s
t
a
t
e
m
e
n
t
cout << counter.GetCounter(); This is the
same as the use of the stream operator to display the
contents of variables of standard data type. The
operator member function
ostream & operator « ( ostream & Out,
counter & counter ) ;
defined in the counter class displays the contents of
the objects of the counter class (see Figure 3.4). The
stream classes, istream and ostream are declared in the
iostream.h header file.

Copyrighted material
Chapter 3: C++at a Glance 91

c
l
a
s
s

m
y
c
l
a
s
s

{
int arr[5] ;

};
The template declaration of the vector class is
illustrated in the program vector. cpp. It has a data
member which is a pointer to an array of generic type
T. The type T can be changed to int, float, etc.,
depending on the type of object to be created.

/ / vector.cpp: parameterized vector class

Copyrighted material
sdv Contents

#include <iostream.h>
template <class T>
class vector
{
T * v; // changes to int *v,
float *v, etc.
int size; // size of vector v
public:
v
e
c
t
o
r
(

i
n
t

v
e
c
t
o
r
_
s
i
z
e

{
size = vector_size;
v * new T[ vector_size ]; //e.g., v=new
int[ size ],if T is int
)
-
v
e
c
t
o
r
(
)

(
delete v;
)
T & elemi int i )
{
if( i >- size )

Copyrighted material
sdv Contents

c
o
u
t

e
n
d
l

"
E
r
r
o
r
:

O
u
t

o
f

R
a
n
g
e
*
;

r
e
t
u
r
n

v
[
i
]
;
)
void show();
);
t
e
m
p
l
a
t
e

<
c

Copyrighted material
sdv Contents

l
a
s
s

T
>

v
o
i
d

v
e
c
t
o
r
<
T
>

s
h
o
w

(
)

{
for(
i
n
t

0
;

<

s
i
z
e
;

i
+
+

Copyrighted material
sdv Contents

c
o
u
t

e
l
e
m
(

"
;
>
v
o
i
d

m
a
i
n
(
)

{
int i;
v
e
c
t
o
r

<
i
n
t
>

i
n
t
_
v
e
c
t
(

Copyrighted material
sdv Contents

)
;

v
e
c
t
o
r

<
f
l
o
a
t
>

f
l
o
a
t
_
v
e
c
t
(

)
j

f
o
r
t

0
;

<

5
;

i
+
+

)
i
n

Copyrighted material
sdv Contents

t
_
v
e
c
t
.
e
l
e
m
(

1
;

f
o
r
(

0
;

<

4
;

i
+
+

>
f
l
o
a
t
_
v
e
c
t
.
e
l

Copyrighted material
sdv Contents

e
m
(

f
l
o
a
t
(

1
.
5

)
;

c
o
u
t

"
I
n
t
e
g
e
r

V
e
c
t
o
r
:

i
n
t
_
v
e
c
t
.
s
h

Copyrighted material
sdv Contents

o
w
(
)
;

Copyrighted material
Similarly, the statement
basep = new Son{45, 20); // pointer to son
creates an object of type class Son dynamically and
assigns its address to the pointer basep. The statement
cout « basep->GetAge() « endl; //
calls Son::GetAge invokes the member function
GetAge () of the class Son (see Figure 3.5). If a call to
a non-virtual function is made in this case, it will
invoke the member function of the base class Father
instead of the derived class Son. Note that the s^me
pointer is able to invoke base or derived class's
member function depending on the class's object to
which it is bound (and this is true only with virtual
functions).
It is important to note that, virtual functions must be
accessed through the pointer to a base class. However^
they can be accessed through objects instead of
pointers, but note that the runtime polymorphism is
achieved only when a virtual function is accessed
through the pointer to a base class. Also another
important aspect is that, when a function is defined as
virtual in the base class and if the same function is
redefined in the derived class, that function is also
treated as virtual function by default. Only class
member functions can be declared as virtual
functions. Regular functions and friend functions do
not qualify as virtual functions.

3.7 Generic Classes-Class

Templates
The container class (i.e., a class that holds objects of
some other type) is of considerable importance when
implementing data structures. The limitation of such
classes to hold objects of any particular data type can
be overcome by declaring that class as a template class.
It allows several classes which differ only in the data
type of their data members to be declared with a
single declaration.
A class template arguments can be of type strings,
function names, and constant expressions, in addition
to template type arguments. Consider the following
class template to illustrate, how the compiler handles
creation of objects using the class templates:
template <class T, int size>
class myclass
{
T arr[size];■

) ;
When objects of template class are created using the
statement such as,
m
y

Copyrighted material
sdv Contents

c
l
a
s
s

<
f
l
o
a
t
,
1
0
>

n
e
w
l
;

t
h
e

c
o
m
p
i
l
e
r

c
r
e
a
t
e
s

t
h
e

f
o
l
l
o
w
i
n
g

Copyrighted material
sdv Contents

c
l
a
s
s
:
class myclass
{
float arr[10J;

);
Again if a statement such as,
myclass <int, 5> new2; is encountered for
creating the object new2, the compiler creates the
following class:
Run1
E
n
t
e
r

N
u
m
b
e
r

1
:

1
0

E
n
t
e
r

N
u
m
b
e
r

2
:

2
t
r
y
i
n
g

d
i
v
i
s

Copyrighted material
sdv Contents

i
o
n

o
p
e
r
a
t
i
o
n
.
.
.
s
u
c
c
e
e
d
e
d

n
u
m
l
/
n
u
m
2

5
Run2
E
n
t
e
r

N
u
m
b
e
r

1
:

1
0

E
n
t
e
r

N
u

Copyrighted material
sdv Contents

m
b
e
r

2
:

0
t
r
y
i
n
g

d
i
v
i
s
i
o
n

o
p
e
r
a
t
i
o
n
.
.
.
f
a
i
l
e
d

E
x
c
e
p
t
i
o
n
:

D
i
v
i
d
e
-
B
y
-
Z
e

Copyrighted material
sdv Contents

r
o
In
m
a
i
n

(
)
,

t
h
e

t
r
y
-
b
l
o
c
k

t
r
y

{
result = numl.div( num2 ) ;
)
invokes the member function div {) to perform the
division operation using the function defined in the
number class. (See Figure 3.6.)

Instance of the class number

Figure 3.6: Exception handling In number class

Copyrighted material
sdv Contents

If any attempt is made to divide by zero, the

following statement in div () member function
if( numZ.num == 0 ) // check for zero
division if yes throw DIVIDEO;
// raise exception
detects the same and raises the exception by passing a
nameless object of the DIVIDE class. The following
block of code in nam () immediately after the try-
block,
Qualifiers arc also applicable 10 the char data type as
follows:
s
i
g
n
e
d

c
h
a
r

u
n
s
i
g
n
e
d

c
h
a
r
Size qualifiers (short and long) cannot be applied to
the char and float data types and sign qualifiers
(signed and unsigned) cannot be applied to float,
double, and long double.

4.6 Variable Definition

A variable must be defined before using it in a
program. It reserves memory required for data storage
and associates it with a symbolic name. The syntax
for defining variables is shown in Figure 4.3. The
variable name can be any valid C++ identifier except
the reserved words. The data type can be any
primitive or user-defined data type such as int. float,
double, and so on.

D
a
i
a
l
y
p
e
:
c
h

Copyrighted material
sdv Contents

a
r
.
i
n
t
,
f
l
o
a
t
.
e
t
c
.
Variable name

Data Type VarNamel,, .... VarNameN;

Figure 4.3: Syntax of

variable definition

The following arc some of the valid variable

definition statements:
int ••; // a is an
integer variable
int b, c, d; // b, c, and d
are valid integer variables
float total, length; // total and length are
valid real variables double suml,
product; // suml and product are valid
double variables
Variables can be defined at the point of their usage as
follows:
int c;
c
o
u
t

"
H
e
l
l
o

W
o
r
l
d
"
;

i
n
t

Copyrighted material
sdv Contents

1
0
;
for(
i
n
t

0
;

<

1
0
;

i
+
+

c
o
u
t

<
<

i
;
Note that the variables d and i are defined at the
point of their usage.

4.7 Variable Initialization

In C++, a variable can be assigned with a value
during its definition, or during the execution of a
program. The assignment operator (=) is used in both
the cases. A variable can be initialized during its
definition using any one of the following syntax:1
data-type variable-name = constant-value;
data-type variable-name( constant-value); This
syntax is most commonly used since it avoids chances
of using uninitialized variables leading to runtime
errors. The following statements initialize variables
during their definition:
int a = 20; // or
int a{20);
float c = 1920.9, d = 4.5; //or float c
(1920.9) , d(4.5) ;
double g = 123455.56;
/ / temper.cpp: conversion of centigrade to
fahrenheit and vice-versa #include <iostream.h>
void main ( )
(

Copyrighted material
sdv Contents

float c , f ;
cout < < "Enter temperature in Celsius: " ;
cin >> c ;
f = 1.8 * c + 32;
cout <<
"Equivale
nt
fahrenheit
=■«£«
endl; cout
<< "Enter
temperatu
re in
fahrenheit
: "; cin
» f;
c = (f - 32) / 1.8;
cout < < "Equivalent Celsius = " « c ;
}

Run
E
n
t
e
r

t
e
m
p
e
r
a
t
u
r
e

i
n

C
e
l
s
i
u
s
:

E
q
u
i
v
a
l
e
n
t

f
a
h
r

Copyrighted material
sdv Contents

e
n
h
e
i
t

4
1
Enter temperature in fahrenheit: ,
Equivalent Celsius = 4.444445

4.10 Qualifiers
Qualifiers modify the behavior of the variable type, to
which they are applied. Qualifiers can be classified
into two types:
« Size qualifiers
«

S
i
g
n

q
u
a
l
i
f
i
e
r
s

C
o
n
s
i
d
e
r

t
h
e

v
a
r
i
a
b

Copyrighted material
sdv Contents

l
e

d
e
f
i
n
i
t
i
o
n

s
t
a
t
e
m
e
n
t
:

i
n
t

i
;
It specifies that i is an integer, which takes both
positive and negative values. That is, i is a signed
integer by default. The above definition could also be
written as:
signed int i ;
Prefixing of the qualifier s igned explicitly, is
unnecessary, since int data type definitions arc s
igned by default. If the variable i is used to hold only
positive values (for example, if it is used to hold the
number of students in a class), it can be defined as
follows: unsigned int i ;
Here, the qualifier unsigned is applied to the data type
int. This qualifier modifies the behavior of the integer
so that a variable of this type always contains a
positive value.

int with Size Qualifiers

Size qualifiers alter the size of the basic type. There
are two size qualifiers that can be applied to integers
(i.e., to the basic type int): short and long. In any
ANSI C++ compiler, the sizes of short int, int and
long int have the following specification:
♦ The size of a short i n t is 16 bits.

Copyrighted material
sdv Contents

♦ The size of an int must be greater than or equal

to that of a short int.
♦ The size of a long int must be greater than or
equal to that of an int.
♦ The size of a long i n t is 32 bits.

Copyrighted material
110 Mastering C++

In most compilers available on DOS, the size of a short int and an int are the same (16 bits). A long int occupies
32 bits. But in 32-bit compilers such as GNU C/C++, an int and a long int are of the same size (32 bits), while a
short int is 16 bits. On the other hand, almost all UNIX compilers have the size of int as 16 bits, int and long int
being 32 bits.

sizeof operator
The operator sizeof returns the number of bytes required to represent a data type or variable. It has the following forms:
sizeofi data-type )
sizeof{ variable )
The data type can be standard or user defined data type. The following statements illustrate the usage of sizeof operator:
int i, j ; float c;
sizeof ( int ) returns 2 or 4 depending on compiler implementation
sizeof ( float ) returns 4
sizeof ( long ) returns 4
s i zeo f ( i ) returns 2 or 4 depending on compiler implementation
sizeof { c ) returns 4
The program size.cpp determines the size of an integer and its variants. It uses the function sizeof {), which gives the
size of any data type in bytes.

/ / size.cpp: size qualifiers and sizeof operator #include

<iostream.h> void mainO {
cout « "sizeof ( char ) = " « sizeof(char) « endl;
cout << "sizeof ( short ) ■ * « sizeof (short) « endl;
cout « "sizeof ( short int ) = ■ « sizeof (short int) « endl;
cout << "sizeof( int ) = ■ « sizeof(int) « endl;
cout << "sizeof ( long )' = "<< sizeof (long) << endl;
cout « "sizeof( long int ) = ■ « sizeof(long int) « endl;
cout << "sizeof( float ) = " << sizeof(float) « endl;
cout « "sizeof( double ) = " << sizeof (double) « endl;
cout « "sizeof! long double ) = " << sizeof(long double) << endl;
}

Run1
sizeof( char ) = 1 sizeof( short } = 2
sizeof( short int ) = 2 sizeof( int ) = 2
sizeof( long ) = 4 sizeof( long int ) = 4 ?
izeof( float ) = 4 sizeof( double ) = 8
sizeof( long double ) = 10

Copyrighted material
 Chapter 4: Data Types, Operators and Expressions 111

Run2
sizeoft char ) = i sizeof( shore ) = 2
sizeofi shore int ) = 2 sizeof{ int ) = 4
sizeof( long ) = 4 sizeof( long int ) = 4
sizeof( float ) = 4 sizeof( double ) = 8
sizeoft long double ) = 10

Note: The output displayed in Runl is generated by executing the program on the DOS system compiled with
Borland C++ compiler. Run2 fs generated by executing the program on the UNIX system.
The name of the variable type is passed to the sizeof operator in parentheses. The number of bytes occupied
by a variable of that type is given by the sizeof operator. The sizeof operator is very useful in developing
portable programs. It is a bad practice to assume the size of a particular type, since its size can vary from
compiler to compiler.
The sizeof operator also takes variable names and returns the size of the variable given, in bytes. For
example, the statements
int i ;
cout << "The size of i is " << sizeof (i); will output
The size of i is 2
The result of sizeof (int) or sizeof {i) will be the same, since i is an int. However, it is a better practice to pass a
variable name to the sizeof operator; if the data-type of i has to be changed later, then rest of the program need
not be modified.

Size qualifiers, as applied to double

The type qualifier long can also be applied on the data type double. A variable of type long double has more
precision when compared to a variable of type double. C++ provides three data types for real numbers: float,
double and long double. In any C++ compiler, the precision of a double is greater than or equal to that of a
float, and the precision of a-long double is greater than or equal to that of a double, i.e., precision-wise,
precision of long double > = precision of double >= precision of float
To find out the actual range that a compiler offers, a program(mer) can refer to the constants defined in the
header file float.h.

Sign Qualifiers
The keywords signed and unsigned are the two sign qualifiers which inform the compiler whether a variable
can hold both negative and positive numbers, or only positive numbers. These qualifiers can be applied to the
data types int and char only. For example, an unsigned integer can be declared as
unsigned int i;
As mentioned earlier, signed int is the same as int (i.e., int is signed by default). The char data type can be

T- True, F- False

Table 4.8: Truth table for logical operator

treated as either signed char, or unsigned char, and the exact representation is compiler dependent.
Logical OR: Consider the following example involving the | | operator, a < m || a < n
The expression is true if one of them is true, or if both of them are true i.e., if the value of a is less than that of
m, or if it is less than n. Needless to say, it evaluates to true when a is less than m and n.
Logical NOT: The ! (NOT) operator takes a single expression and evaluates to true if the expression is false, and
evaluates to false if the expression is true. In other words, it just reverses the value of the expression. For
example, consider:
! ( x >= y ) it has the same meaning as
x<y
The ! operator can be conveniently used to replace a statement such as
if{ a == 0 ) by the statement
if< !a )
The expression ! a evaluates to true if the variable a holds zero, false otherwise.

Copyrighted material
 Chapter 4: Data Types, Operators and Expressions 111

The unary negation operator (!) has a higher precedence amongst these, followed by the logical AND (&&)
operator and then the logical OR (| |) operator, and are evaluated from left to right.
The logical operator is used to connect various conditions to determine whether a given year is a leap year
or not. A year is a leap year if it is divisible by 4 but not by 100, or that is divisible by 400. The program leap.
epp illustrates the use of the modulus operator.
/ / leap.cpp: detects whether year is leap or not #include <iostream.h>
void main() I
int year;
cout << "Enter any year: "; cin » year;
if( (year % 4 == 0 && year % 100 != 0 ) || (year % 400 == 0) )
cout << year << " is a leap year"; else
cout « year « " is not a leap year";
}

Run1
Enter any year: 1996
1996 is a leap year
Run2
Enter any year: 1997
1997 is not a leap year
In main (), the statement
if( (year % 4 == 0 fcfc year % 100 != 0 ) || (year % 400 == 0) ) can be replaced by
if ( (I (year % 4) && year % 100 ! = 0 ) ] | .' (year % 400) )

4.14 Bit-wise Operators

The support of bit-wise manipulation on integer operands is useful in various applications. Table 4.9 shows the
bit-wise operators supported by C++. To illustrate these operators with examples, assume that a, b and c are
defined as integer variables as follows:
int a = 13, b = 7, c; Consider the variables a, b, and c as 16-bit integers, and the value stored in a and b
have the following representation in the binary form:
The binary representation of a is 0000 0000 0000 1101
The binary representation of b is 0000 0000 0000 0111 (The spaces appearing after every 4 bits are
only for clarity. Actually, the integers are merely 16 continuous bits.)

Operator Meaning
& Bitwise AND
1 Bitwise OR
Bitwise EX-OR

- Bitwise complement
« Shift left
>> Shift right
Table 4.9: Bit-wise operators (i) Logical

Bit-wise Operators
Logical bit-wise operators perform logical operations such as AND, OR, EX-OR, NOT between corre sponding
bits of operands (if binary) and negation of bits (if unary).
Unary Operator: One's Complement Operator (~)
The complement operator causes the bits of its operand to be inverted, i.e., 1 becomes 0 and 0 becomes I. For
instance, the largest possible number, which can be stored in an unsigned integer can be found as follows.
When one's complement operator is applied on this word holding zero, all the bits will be inverted to ones and
a new value becomes the largest possible number. The program large. cpp illustrates this conversion process.
/ ■ large.cpp: detects largest possible unsigned integer •include
<iostream.h> int mainf) {
unsigned u = 0;
cout « "Value before conversion: ■ << u << endl; u - ~u;
cout << "Value after conversion : " << u << endl; return 0;
)

Run1
Value before conversion: 0 Value after
conversion : 65535
Run2
Value before conversion: 0
Value after conversion : 4294967295

Copyrighted material
 Chapter 4: Data Types, Operators and Expressions 111

Runl is executed on the MS-DOS based machine using a 16-bit compiler.

Run2 is executed on the UNIX based machine using a 32-bit compiler.

Binary Logical Bit-wise Operators

There are three binary logical bit-wise operators: & (and), I (or) and A (exclusive or). The operations are carried
out independently on each pair of the corresponding bits in the operands, i.e. the bit 1 of operandi is logically
operated with the bit 1 of operand 2. The operations using these operators are discussed in the following
sections.
Bitwise AND: The statement
c = a & b;
makes use of the bitwise AND operator. After this statement is executed, each bit in c will be 1 only if the
corresponding bits in both a and b are 1. For example, the rightmost bit of both integers is 1, and hence the
rightmost bit in c is 1. The next bit is 0 in a and 1 in b. Hence the second bit (from the right) in c is 0. Applying
the same reasoning for all the bits in each one of the integers, the value of c after the above statement is
executed will be 0000 0000 0000 0101 which, in decimal is 5 and is illustrated below:
Bitwise AN D operator: a & b

a 0000 0000 0000 1101

b 0000 0000 0000 0111
a & b 0000 0000 0000 0101

Bitwise OR: The statement

c = a | b;
makes use of the bitwise OR operator. After this statement is executed, a bit in c will be 1 whenever at least
one of the corresponding bits in either a or b is 1. In the example given below, the value of c will oe 0000 0000

Bitwise OR operator: a I b
a 0000 0000 0000 1101

0000 1111 i.e., decimal 15 and is illustrated below;

Bitwise XOR: The statement
c = a * b;
makes use of the bitwise XOR operator. After this statement is executed, a bit in c will be 1 whenever the
corresponding bits in a and bdiffer. So in the example given below, the value of c will be 0000 0000 0000
1010 which, in decimal is 10 and is illustrated below:

Bitwise EX-OR operator: a - b

a 0000 0000 0000 1101
0000 o o o o 0000 0111
0000 0000 0000 1010
(ii) Shift Operators
There are two shift operators in C++: left shift («) and right shift (»). These are binary operators and have the
following syntax:
operand « count for left shift
operand » count for right shift
The first operand, is the value which is to be shifted. The second is the number of bits by which it is shifted.
The left shift operator moves the count number of bits to the left, whereas the right shift operator moves the
count number of bits to the right. The leftmost or the rightmost bits are shifted out and are lost.
Left Shift Operator
Consider the statement
c = a « 3;
The value in the integer a is shifted left by three bit positions. The result is assigned to the integer c. Since the
value of a is 0000 0000 0000 1101 the value of c after the execution of the above statement is 0000 0000 0110
1000 (104 in decimal), and is illustrated below:

Left-Shift «
drop off «—| 0000 0000 0000 1101 \*— insert 0's after Left-bit shift by 3 places I.e., a « 3
I 0000 0000 0110 1000 I

The three leftmost bits drop off due to the left shift (i.e., they are not present in the result). Three zeros are
inserted in the right. The effect of shifting a variable to the left by one bit position is equivalent to multiplying
the value by 2. If the initial value of a is 13, shifting left by 3 bit positions yields 13*8=104.

Copyrighted material
 Chapter 4: Data Types, Operators and Expressions 111

While multiplying a number with a power of 2, considerable savings in execution time can be achieved by
using the left bit-shift operator instead of the multiplication operator, since a shift is faster than multiplication.
Right Shift Operator
Consider the statement c = a >> 2,-The value of a is shifted right by 2 positions. Since the value of a is 0000
0000 0000 1101 the value of c after the execution of the above statement is 0000 0000 0000 0011 (3 in
decimal) and is illustrated below:
Right-Shift »
insert 0's 0000 0000 0000 11Q3~]—»drop off
after right shift by 2 places i.e., a » 2 |
0000 0000 0000 0011 I
The 2 rightmost bits drop off (are not present in the result), and zeros are inserted in the left. The effect of
shifting a variable to the right by one bit position is equivalent to dividing the value by 2 (i.e., divide by 2 and
truncate the result). As the initial value of a is 13, shifting it right by 2 bit positions yields the value 3 (the
result of dividing 13 by 4 and truncating the result). Note that if the negative number is shifted right, then 1 is
inserted at the left for every bit shifted to the right.
The program extract.cpp illustrates the binary operators. It reads an integer and prints the value of a
specified bit in the integer. The position of bits are numbered starting with 0 from right to left. For example, to
find the value of the second bit of an integer i, it is necessary to shift i to the right by two bits, and take the least
significant digit.
// extract.cpp: Fishing the nth bit ♦include
<iostream.h> void main ( )
{
/ / a is the input integer and n, the bit position to extract
int a, n, bit;
cout « "Enter an integer:
cin >> a;
cout « "Enter bit position to extract: ";
cin >> n;
bit = ta » n) & 1; / / bit is the value of the bit extracted (0 or 1) cout « "The bit is " << bit;
)

Run
Enter an integer: 10.
Enter bit position to extract: Z
The bit is 1
In main (), the statement
bit = (a » n) & 1;
first shifts a to the right by n bits and then masks (clears) all the bits of a except the least significant bit
(rightmost bit), retaining the bit which is required. Parentheses are not required in the above statement, since
the operator » has more precedence than & (i.e., in an expression, if both » and & are present, the » operator is
executed first). Since this fact is not obvious, it is always better to use parentheses in such situations, for the
sake of readability.

Copyrighted material
 Chapter 4: Data Types, Operators and Expressions 12"

The program max. cpp reads two integers and displays the value of the larger of the two numbers computed using
the ternary operator. If they are equal, then naturally, either of them can be printed.

// max.cpp: finding the maximum using the conditional operator ♦include <iostream.h> void main () {
int a, b, larger;
cout « "Enter two integers: ";
cin » a >> b;
larger = a > b ? a : b;
cout « "The larger of the two is " « larger;
}

Run
Enter two integers: XQ. 20. The larger of
the two is 20

In main (), the statement

larger = a > b ? a : b; has three components. The conditional expression (a > b ? a : b) returns a if a > b, otherwise it
returns b. It can be equivalently coded using the if statement as follows: if( a > b )
larger = a; else
larger = b;
The expressions in the ternary operator can be any valid variable, constant, or an expression. The program
oddeven. cpp checks whether the number is odd or even using the ternary operator.

/ / Oddeven.Cpp: checks whether the number is odd or even ♦include <iostream.h> void main () {
int num; char
*str;
cout « "Enter the number: "; cin » num;
cout « "The number * << num « 1 is
cout « ((num % 2) ? "Odd" : " E v e n " ) ;
cout << endl « "Enter the number: "; cin >> num;
cout << "The number " « num « " is "; (num % 2) ? cout « "Odd"
: cout « "Even";
)

Run
Enter the number: iO. The
number 10 is Even Enter the
number: 25. The number 25 is
Odd

Copyrighted material
 Chapter 4: Data Types, Operators and Expressions 133

The character 1 or L can be used to specify long double values. For example, 0 . 257L is a long double
constant.

(iii) Character Constants

A character constant is enclosed in single quotes. Examples: Valid character constants: 'a' '5' '\n\ Invalid
character constants: 'ab' '54'. Note that multiple characters can also exist within single quotes. The compiler will
not report any error. The value of the constant however, depends upon the compiler used. This notation of
having multiple characters in single quotes is practically never used.
Inside the single quotes, a backslash character starts an escape sequence. \xhh specifies a charac ter constant
in hexadecimal, where h is any hexadecimal digit. The hexadecimal digits hh gives the ASCII value of the
character. For example, the character constant \x07 represents the BELL character. The complete list of escape
sequences is shown in Table 4.12.

Operator Meaning
\a Beep
\b Backspace
\f Formfeed
\n Newline
\t Horizontal tab
\\ Backslash
\' Single quote
\■ Double quote
\v Vertical tab
\? Question mark
\0 Null
\0ooo Code specified in octal
\xhh Code specified in hexadecimal

Table 4.12: Escape sequences

The escape sequence \0XXX specifies a character constant in octal, where 0 denotes any octal digit. As
before, xxx is the ASCII value of the number specified in octal. For example, the ASCII code of K is 75. The
character constant \0113 specifies this character in octal. The program beep, epp generates the beep sound using
the escape sequence. The escape sequence \x07 (can be \7) in the cout can be replaced by \a.
/ / beep.Cpp: generating beep sound lincludt
<iostream.h> void main() (
cout « '\x07'; // computer generates sound
)

Copyrighted material
 Contents xix

15.5............................................................................. Array of Pointers to Base

Class Objects.................................................................................................... -.....>...............5E1
15.6...............................................................................................Pure
Virtual Functions....................................................................................................................... «4
15.7...............................................................................................Abstract
Classes...................................................................................................................................... 583
15.8...............................................................................................Virtual
Destructors................................................................................................................................588
15.9...............................................................................................How is
Dynamic Binding Achieved ?.................................................................................................... 591
15.10.............................................................................................Rules for
Virtual Functions....................................................................................................................... 593
Review Questions............................................................................................................ 594

16 Generic Programming with Templates................................................................................596

16.1...............................................................................................Introduct
ion 596
16.2...............................................................................................Function
Templates.................................................................................................................................. 596
16.3...............................................................................................Overload
ed Function Templates.............................................................................................................. 605
16.4...............................................................................Nesting of Function
Calls ■............................................................................................................... 606
16.5...................................................Multiple Arguments Function Template •■ •
607
16.6...............................................................................................User
Defined Template Arguments................................................................................................... 609
16.7...............................................................................................Class
Templates.................................................................................................................................. 610
16.8...............................................................................................Inheritan
ce of Class Template.................................................................................................................. 616
16.9...............................................................................................Class
Template Containership............................................................................................................. 620
16.10.............................................................................................Class
Template with Overloaded Operators........................................................................................ 626
Review Questions............................................................................................................ 627

17 Streams Computation with Console---------------------------------------..............••.................629

17.1 What are Stream?............................................................................................................629
Ml Predefined Console Streams.........................---------------------.................................631
17.3...............................................................................................Hierarch
y of Console Stream Classes...................................................................................................... 634
17.4...............................................................................................Unforma
tted I/O Operations.................................................................................................................... 635
17.5...............................................................................................Formatte
d Console I/O Operations.......................................................................................................... 640
17.6........................................... Manipulators.....................................................
648
!7.7 Custom/User-Defined Manipulators................................................................................. 655
17.8 Stream Operator with User-defined Classes..................................................................... 660
Review Questions............................................................................................................ 662

18 Streams Computation with Files-----------------------------............................-.....-............»• 664

18.1...............................................................................................Introduct
ion
18.2...............................................................................................Hierarch
y of File Stream Classes.............................................................................................................665
18.3...............................................................................................Opening
and Closing of Files................................................................................................................... 667
18.4...............................................................................................Testing
for Errors................................................................................................................................... 672
18.3 File Modes.....................................................................----.............................-..................677
18.6............................................................................................... File
Pointers and their Manipulations................................................................................................ 678
18.7............................................................................................... Sequenti
al Access to a File...................................................................................................................... 682
18.8............................................................................................... ASCII
and Binary Files......................................................................................................................... 683
18.9............................................................................................... Saving
and Retrieving of Objects........................................................................................................... 685
18.10............................................................................................. File
Input/Output with fstream Class................................................................................................. 689
18.11............................................................................................. Random
Access to a File.......................................................................................................................... 691

Copyrighted material
18.12............................................................................................. In-
Memory Buffers and Data Formatting....................................................................................... 695
18.13............................................................................................. Error
Handling During File Manipulations.......................................................................................... 696
call during execution, since the function body is substituted at the point of macro call during compila -
tion. Thereby the runtime overhead for function linking or context-switch time is reduced. The
directive #def ine, indicates the start of a macro function as shown in Figure 4.10. The macro function
specification spans for a maximum of one line only. However, macro function body can spread to
multiple lines if each new line is followed by A' character.
■---► Start of pre-processor directive indicator
■---► pre-compiler (pre-processor) directive
I- - -► name of the macro function

r
I—► macro parameters
C++ statements performing
Macro Function Operation

# define MacroFuncName (Parameters) Single line function body

Figure 4.10: Syntax for declaring macro function

Examples:
#define inc( a ) a+1
♦define add{ a, b )
(a+b)
The program maxmacro.cpp illustrates the use of macro function in the computation of the
maximum of two numbers.
/ / maxmacro.cpp: maximum of two numbers using macros
♦include <iostream.h>
♦define max( a, b ) ( a > b ? . a : b )
void main()
{
cout << "max( 2, 3 ) = " << max( 2, 3 ) << endl;
cout « "max{ 10.2, 4.5 ) = ■ « max{ 10.2, 4.5 ) « endl;
int i = 5, j = 10;
cout << "i = ' « i « endl;
cout << "j = " « j « endl;
cout « "On execution of k ■ max( ++i, ++j );...' « endl; int k = max{ + + i, ++j ); cout << "i = ■ « i
<< endl; cout << "j = " << j << endl; cout << "k = " « k « endl;
)

Run
max( 2, 3 ) = 3
max( 10.2, 4.5 ) = 10.2
i=5
j - 10
On execution of k = max( ++i, ++j ); . . . i = 6 j
= 12 k = 12
In main (), the expressions
max( 2, 3 ) max( 10.2,
4.5 )
invoke the macro function max (). Unlike normal functions, macro functions can take parameters of
any data type. If arguments are in the form of expressions, they are not evaluated at the point of call,
but at the time of their usage. Thus, the statement
int k = max( ++i, ++j ); is processed
by the preprocessor as follows:
int k = { ++i > ++j ? ++i : ++j );
It can be observed that, the variable with greater value, will be incremented twice. The macro function
body can spread across multiple lines as follows:
#define print( n ) \
for{ int i = 0; i < n; i++ ) \ cout « i;

4.25 Operator Precedence and Associativity

Every operator in C++ has a precedence associated with it. Precedence rules help in removing the
ambiguity about the order of performing operations while evaluating an expression. The associativity
of the operators is also important. Associativity specifies the direction in which the expression is
evaluated, while using a particular operator. The precedence and associativity of alt the operators
including those introduced by C++ are shown in Table 4.13. It is important to note the order of prece -
dence and evaluation (associativity) of operators. Consider the following two statements:
int a = 10, b = 15, c = 3, d;
d = a + b * c;
In the second statement, first b is multiplied by c, and then the result is added to a and the sum is
assigned to d. Multiplication is done before addition, since it has higher precedence than the addition

Copyrighted material
operator. In order to override the precedence, braces can be used. For example, the statement d = la +
b) * c;
would add a to b first, multiply the result by c and assign the product to d. Associativity of an operator
can be from left-to-right or right-to-left. For example, in the expression d = a - b - c;
the leftmost minus is evaluated first and then the second minus is evaluated, causing c to be subtracted
from the result. Thus, in case where several operators of the same type appear in an expression without
braces, the operators are normally evaluated starting from the leftmost operator, proceeding rightward,
hence the minus operator associates from left to right. On the other hand, the assignment operator
associates from right to left. For example, in the statement d = a = c;
the second (right-most) assignment operator is evaluated first. The variable c is assigned to a and then
this value is assigned to d.
Like most programming languages, C++ does not specify the order in which the operands of an
operator are evaluated. Such operators are &&, | |, ?:, and ','.) For example, in the statement such as x =
g() + h i ) ;
tj () may be evaluated before h () or vice versa; thus if g () or h () alters a variable (global) on which
the other depends, then the resultant value of x is dependent on the order of evaluation. Similarly, the
order in which function arguments are evaluated is not specified, so the statement

Copyrighted material
Operator precedence and associativity
Copyrighted material

Copyrighted material
 142 Mastering C++

add( ++n, pow( 2 , n ) ); can produce different results with different compilers. However,
most C+ + compilers evaluate function arguments from right to left. There are cases such as in function
calls, nested assignments, increment and decrement operators cause side effects—some variable is
changed as a by-product of the evaluation of the expression. The (C language and hence, C++)
standard intentionally leaves such matters unspecified. When side effects (variable modification)
take place within an expression, it is left to the discretion of the compiler, since the best order
depends strongly on the machine architecture.
The moral is that developing a code that depends on the order of evaluation, is not a good
programming practice in any language. Hence, it is necessary to know what to avoid, but if it is not
known how they are treated, the programmer should not be tempted to take advantage of a particular
implementation.

Review Questions
4.1 What are variables ? List C++ rules for variable naming.
4.2 Why output and Output are considered as different identifiers ?
4.1 What arc keywords ? List keywords specific to C++. Can these keywords be used as
variables ?
4.2 What is a data type ? What are the different data types supported by C++ ?
4.6 Wl.at is new about C++ in terms of the variable definition ?
4.7 What is the difference between a character and a character string representation ?
4.8 What is an-expression ? Is this different from a statement ? Give reasons.
4.9 List categories of operators supported by C++.
4.10 What are qualifiers ? Illustrate them with examples.
4.11 Develop an interactive program to compute simple and compound interest.
4.12 List evaluation steps for the expression (a+(b*c))*c+d/2.
4.13 Write an interactive program to find the largest of two numbers.
4.14 How C++ represents true and false values ? Are the expressions !a and a==0 have the
same meaning ? Give reasons.
4.15 Write a program to determine the type of compiler (whether 16 or 32-bit) used to compile
it.
4.16 Write a program to multiply and divide a given number by 2 without using * and / operators.
■1.17 Illustrate how compound assignment operators allow to write compact expressions ?.
. 18 What is the effect of the following expressions if i=l and j =4 ?
a) i + + b) j = c) j = ++j; d) i+++j e) i = i++++*j;
4.19 Write an interactive program to find elder among you and me using the ternary operator.
4.20 What is the outcome of the statement: a= (a=10 , a+ + , a—) ; if a holds the value 5
initially.
4.21 What is type conversion ? What are the differences between silent and explicit type
conversion? Write type conversion steps required for evaluating the statement:
z=i+b+ j-k/4; (where i and j are ints, b is float, and k is double, and z is long type). 4.2-
What are escape sequences ? Write a program to output messages in double quotes.
4.23 What ar- macros ? Write a program to find the minimum of two numbers using macros.
What is the output of the statement: a = min( ++a, ++b); (if a = 2 and b = 4).
4.24 What is operator precedence ? Arrange the following operators in the order of their
precedence:
*. +. (). !. ++, --. |. | |. &. /,and &&
4.25 What is the significance of the associativity of operators ? What is the order of evaluation of
the
operator ? : in the statement
a = i > j ? i : j;

Copyrighted material
 Chapters: Control Flow 42

Runl
Enter your age: 15
you are a teen-aged person, good!
Run2
Enter your aae:-20
In main (), the statement
if( age > 12 && age < 20 ) first evaluates the test expression and executes the if-part only when it is
true. In Runl, the input data entered is 15 which lies between 13 and 19 and hence, the statement
cout << "you are a teen-aged person, good!'; gets executed. Whereas, in Run2. the input data is 20
which does not lie within this range and hence, the control proceeds to the next statement.

Compound Statement with if

In the if construct, the if-part can be represented by a compound statement as follows: if{test- expression) i
statement 1;
statement 2;
}
In this case, when test-expression is true, the statements enclosed within the curly braces, representing a
compound statement, are executed. The program age2 . cpp illustrates the use of the compound-if statement.

/ / age2.Cpp: use of if statement and data validation ♦include

<iostream.h> void main () {
int age;
cout « "Enter your age: "; cin » age;
if ( age < 0 )
{
cout « "I am sorry!" « endl;
cout « "age can never be negative";
return; // terminate program
)
if( age > 12 && age < 20 )
cout « "you are a teen-aged person, good!";
)
Run
Enter your age: -10 I am
sorry!
age can never be negative
In main (), the statement
if( age < 0 )
validates the input data and accordingly takes action. It terminates the program after issuing the warning
message, when the input data is negative.
The program large. cpp illustrates the use of multiple decision statements to compute the maximum of three
numbers.

// large.cpp: find the largest of three numbers •include

<iostream.h> void main () {
float a, b, c, big;
cout « 'Enter three floating-point numbers: ■; cin » a » b » c;
// computing the largest of three numbers big = a; if<b > big)
big = b; if{c >
big)
big = c;
cout << "Largest of the three numbers = ■ « big;
)

Run
Enter three floating-point numbers: 10 .2 15 . 6 12_Ji Largest of the three
numbers =15.6

5.4 if -e lse Statement

The if-else statement will execute a single statement or a group of statements, when the test expression is true. It
does nothing when the test expression fails. C++ provides the i f - el se construct to perform some action even
when the test expression fails. The control flow intheif-else statement is shown in Figure 5.2.
entry------1
I .............................../.'.. Vfalse
if ( .'expression *.) v
{ -\
..............
>.......................A* true i
i

Copyrighted material
 Chapters: Control Flow 43

else » /

>

statement; * = Figure 5.2: Control flow In

If-else condition
entry

▼ .....................✓ . , r- . ■ •......."..............
for (^initialization .''condition';;-; updacion ) V

true i false

}
statement; «-

Figure 5.3: Control flow In for loop

The for statement is a compact way to express a loop. All the four pans of a loop are in close proximity with
the for statement. The initialization part is executed only once. Next the test condition is evaluated. If the test
evaluates to false, then the next statement after the for loop is executed. If the test expression evaluates to true,
then after executing the body of the loop, the update part is executed. The test is evaluated again and the whole
process is repeated as long as the test expression evaluates to true as illustrated in the program countl. cpp.
/ / countl.cpp: display numbers 1. .N using for loop •include
<iostream.h> void main() {
int n;
cout « "How many integers to be displayed: '; cin » n;
fort ir?t i = 0 ; i < n; i++ ) cout « i << endl;

Bun
How many integers to be displayed: £
0
1
2
3
4
In main (), the statement
fort int i = 0; i < n; i++ ) cout « i « endl;
has four components. The first three components enclosed in round braces and separated by semicolons are the
following:
int i = 0 i < n i++
The first component int i = 0 is called the initialization expression and is executed only once prior to the statements
within the for loop. The second component i < n is called the test expression and

Copyrighted material
 Chapters: Control Flow 44

is evaluated every time before execution of the loop body. If this expression is true, the statement in the loop gets
executed. In case it is false, the loop terminates and the control of execution is transferred to the statement following the
for loop. The third component is called update expression, and is executed after every execution of the statement in the
loop. The fourth component is the loop-body. The program sumsql. cpp, finds the sum and the sum of squares of the first
15 positive even integers.
// sumsql .Cpp: sum of first 15 even numbers and their squares' sum #include<iostream.h> void main {) {
int i;
int sum = 0, sum_of_squares = 0; for( i =2; i <=
30; i += 2 ) {
sum += i;
sum_of_squares += i*i;
}
cout « "Sum of first 15 positive even numbers = " << sum << endl; cout « "Sum of their squares = " «
sum_of_squares;
)

Run
Sum of first 15 positive even numbers = 240 Sum of their
squares = 4960

In main (), the statement

fort i = 2; i <= 30; i 2 ) increments the loop variable i by 2
using the update expression
i += 2
The body of the loop consists of multiple statements, forming a compound statement. The for loop counts from 2 to 30
in steps of two. It is just as easy for a loop to count down, from 30 to 2 , as illustrated in the program sumsq2 . cpp.
/ / sumsq2.cpp: sum of first 15 even numbers and their squares' sum #include<iostrearn.h> void main() (
int i;
int sum = 0, sum_of_squares = 0; fort i = 30; i >=
2; i -= 2 ) {
sum += i;
sum_of_squares += i*i;
)
cout << "Sum of first 15 positive even numbers = ■ << sum « endl; cout « "Sum of their squares = " «
sum_of_squares;
)

Run
Sum of first 15 positive even numbers = 240 £;um of their squaj
es = 4960

' Copyrighted material

45 Mastering C++

Run
Enter the number of lines: 5, X
2 3 23 4 5 4 34
5 6 7 6 5 4
5 6 7 8 9 8 7 6

5.7 while loop

The whi le loop is used when the number of iterations to be performed are not known .in
advance. The control flow in the while loop is shown in Figure 5.5. The statements in the loop are
executed if the test condition is true and the execution continues as long as it remains true. The
program count2 . cpp illustrates the use of the while loop to perform the same function as the for
loop.

entry
------1 false
▼ .........- < - " ~ \
-while ('.expression:) \
/{ .................................i' <
i ' true ,'

)
statement Figure 5.5: Control flow In while loop
/ / COUnt2.Cpp: display numbers 1. .N using while loop
linclude <iostream.h> void main { ) {
int n;
cout « "How many integers to be displayed:
cin » nj int i = 0;
while{ i < n ) <
cout « i « endl; i++;
)
)

Run
How many integers to be displayed: 5_
0
1
2
3
4
The while loop is often used when the number of times the loop has to be executed is unknown in
advance. It is illustrated in the program averagel. cpp.
// averagel.cpp: find the average of the marks •
include <iostream.h> void main ( ) {
int i , sum = 0, count ■ 0, marks;
cout « "Enter the marks, -1 at the e n d . . . \ n " ;
cin » marks;
while( marks ! = -1 )
{
sum + = marks;
count++;
cin > > marks;
}
float average = sum / count;-
cout « "The average is ■ « average;
)

Run
Enter the marks, -1 at the end...
afl
2
1

£
2

1
A

z
l

Copyrighted material
 Chapters: Control Flow 46

The average i s 77
The first cin statement, just before the while loop, reads the marks scored in the first subject and
stores in the variable marks, so that the statement inside the loop can have some valid data to operate.
The cin statement inside the loop reads the marks scored in the other subjects one by one. When -1 is
entered, the condition
marks ! = -1
evaluates to false in the while loop. So, the while loop terminates and the program execution proceeds
with the statement immediately after the while loop, which in the above program is
average = sum / count; Consider the case when the user inputs -1 as the first marks. The
condition in the while statement evaluates to false, and the statements inside the loop are not executed
at all. In this case, the value of count continues to be zero, so, while computing the average it leads to
division by zero causing a run-time error. This can be prevented by using the if statement as follows;
if( count != 0 )
average = sum / count;
The above statement can also be written as
iff count )
average = sum / count;
Any expression whose value is nonzero is treated as true. The program binary. cpp illustrates such
situations. It uses the while construct to convert a binary number to its decimal equivalent. The shift-
left operator«is used for shifting bits stored in a variable in this program.
do {
cout < < i « endl; i +
+;
} while( i < n ) ;
)

Run
How many integers to be displayed: 5.
0
1
2
3
4

To realize the usefulness of the do. .while construct, consider the following problem: The user
has to be prompted to press m or f. In reality, the user can press any key other than m or f. In
such a case, the message has to be shown again, and the users should be allowed to re-enter one
of the two options. An ideal construct to handle such a situation is the do. .while loop as
illustrated in the program dowhi le. cpp.
/ / dowhlle.cpp: do. .while loop for asking data until i t i s valid ♦include <iostream.h> void main
()
{
char inchar;
do
{
cout « "Enter your sex ( m / f ) : ■;
cin » inchar; ) while{ inchar ! = ' m ' && inchar ! =
' f ' ) ; i f ( inchar == 'm' )
cout « 'So you are male, good!"; else
cout « "So you are female, g o o d ! ' ;
J
Run
Enter your sex ( m / f ) : 3
Enter your sex ( m / f ) :
Enter your sex ( m / f ) : m So
you are male, good!
In main ( ) , the do. . while loop keeps prompting for the user input until the characterm for
male or f for female is entered. Such validation of data is very important while handling sensitive
and critical data.
The solution to certain problems inherently requires data validation only after some
operation is performed as illustrated in the program pal. cpp. It checks if the user entered
number is a palindrome using the do-while construct.
The program averagel - cpp discussed earlier has the following code: cin >> marks; while{ marks !
= -1 )
<
sum + = marks;
count**;
47 Mastering C++

cin >> marks;

)
It computes the sum of marks entered by the user and maintains their count. This segment of code can
be replaced by the following piece of code using the break statement: while( 1 )
(
cin >> marks;
iff marks == -1
)
break;
sum *=
marks;
count++;
)
Note that it avoids the use of two cin statements. Whenever -1 is input, the condition marks«-l
evaluates to true, and the break statement is executed, which leads to the termination of loop. Control
passes to the statement following the while construct. Observe that the condition in the while loop has
been specified as 1 (one) which is nonzero and hence is always true. The condition specifies an infinite
loop, but the break prevents such a situation. The above segment of code can also be replaced by the
following for loop segment:
for( ;; )
(
cin » marks; if(
marks == -1 )
break;
sum +=
marks;
count++;
)
Note that, when test-expression is not mentioned in the for loop, it is implicitly treated as true causing
an infinite loop condition. However, it does not lead to an infinite loop as the break statement takes
over the responsibility of loop termination. In general, the break statement causes control to pass to the
statement following the innermost enclosing for, while, do-while, or switch statement. The same action
can also be achieved by using do. .while loop as follows: do {
cin >> marks;
if( marks ■■ -1 )
break;
sum
marks; count+
+; ) whilst
1);
The program average2 .cpp illustrates the use of break in loop statements. It performs the same
operation as that of the program averagel .cpp.

Copyrighted material
 Chapter 5: Control Flow 48

/ / sumpos.cpp: sum of positive numbers •include

<iostream.h> void main {) {
int num, total = 0;
do
(
cout << "Enter a number {0 to quit): cin » num; if(
num == 0 )

cout « "end of data entry." << endl;

break; // terminates loop
)
if( num < 0 ) {
cout « "skipping this number." « endl;
continue; // skips next statements and transfers to start of loop
J
total += num; }
while(1);
cout « "Total of all +ve numbers is " « total;
J

Run
Enter a number (0 to quit) :
Enter a number (0 to quit) : 2£
Enter a number (0 to quit): ^£
skipping this number.
Enter a number (0 to quit) : XSL
Enter a number (0 to quit) : SI
end of data entry.
Total of all +ve numbers is 40

In do. . while loop of the above program, on encountering break, control is transferred outside the loop. On
encountering continue, control is transferred to the while condition which is always true (nonzero). Figure 5.10 shows
action differences between break and continue statements in loops. The break and continue statements must be
judiciously used and their indiscriminate use can hamper the clarity of the logic.

for i: { for (
i f ( expression ) i f ( expression } break;%
continue;-------------- \

}
) statement;*"
statement;

Figure 5.10: Control flow for continue and break

Copyrighted material
 2 Mastering C++

cost of software development and maintenance and declining hardware cost over several years is
depicted in Figure 1.1. Software maintenance is the process of modifying or extending the capabilities
of the existing software It requires mastery over the understanding and modifying the existing
software, and finally revalidating the modified software.
100

1955 1970 1985 2000

Time (in years) .........«
Figure 1.1: System development cost
The cost effectiveness of hardware has been growing by about three orders of magnitude every
decade and simultaneously the market for computers is also expanding. This multiplies the number of
applications of computers and in turn places greater demands on software. While demand for software
has been growing rapidly to keep pace with the growth of hardware, the actual software development
has been progressing slowly. Unfortunately, even with all the innovations in the area of languages,
programming environments, software engineering concepts, etc., there has been no significant im-
provement in the productivity of software development, leading to software crises. The term "software
crises" refers to the overrun of the cost of software development in terms of both budget and time-
target.
The software crisis, right from the beginning, is providing an impetus for the development of
software engineering principles, tools, and better programming paradigms to build more reliable and
reusable systems. The state-of-the-art solution to overcome software crisis is the Object-Oriented
Paradigm.

1.2 OOPs! a New Paradigm

Object-Oriented Programming is a new way of solving problems with computers; instead of trying to
mould the problem into something familiar to the computer, the computer is adapted to the problem.
Object-Oriented Programming is designed around the data being operated upon as opposed to the
operations themselves. Instead of making certain types of data fit to specific and rigid computer opera -
tions, these operations are designed to fit to the data. This is as it should be. because the sole purpose of
a computer program is to manipulate data.
OOP languages provide the programmer the ability to create class hierarchies, instantiate co-opera -
tive objects collectively working on a problem to produce the solution and send messages between
objects to process themselves. The power of object-oriented languages is that the programmer can
create modular, reusable code and as a result, formulate a program by composition and modification of
Any loop (for, while, or do. .while) statement can be replaced by an if statement coupled with a
goto statement. But this, of course makes the program unreadable. On the other hand, there are
situations wherein goto statement can make the flow of control more obvious. For example, the follow -
ing segment determines whether two arrays x and y have an element in common or not. The element x
has n elements and y has m elements.
for( i = 0; i < n; i + + ) for(
j = 0; j < m; j++ ) if< x[ i I «
y[ j J ) goto found; // Element
not found

found:
// Element found

Except in cases such as the one cited above, the use of the goto statement must be avoided.
It is possible to use goto statement to jump from outside a loop to inside the loop body, but it is
logically incorrect. Hence, goto jumps shown in Figure 5.11 would cause problems and therefore must
be avoided.

Copyrighted material
 2 Mastering C++

Figure 5.11: Invalid goto's

5.13 Wild Statements

It is very difficult to detect semantic errors in a program when semicolons are used improperly with
loops. One such case is illustrated in the program age5. cpp
// age5.cpp: if statement with wrong usage of syntax •include
<iostream.h> void main () {
int age;
cout « "Enter your age: "; cin
>> age;
if( age > 12 && age < 20 );
cout << "you are a teen-aged person, good!";
)

Run1
Enter your age: 14
you are a teen-aged person, good!
Run2
Enter your age: 50
you are a teen-aged person, good!
In main (), the statement
if( age > 12 && age < 20 ); effectively does nothing; observe the semicolon after the
condition statement. The program displays the same message for any type of input data. Whether the
input age lies in range of teenage or not, it produces the message
you are a teen-aged person, good! See Run2 output which
shows even 50 year aged person as teen-aged!

Equality Test
The program agecmp. cpp is written for comparing ages of two persons. It prints the illogical message
except for some typical value.

/ / agecmp.cpp: age comparison

•include <iostream.h> void
main{) {
int myage = 25, yourage;
cout << "Hi! my age is ■ « myage << endl;
cout << "What is your age ? ";
cin >> yourage;
if( myage = yourage )
cout « "We are born on the same day. Are we twins!";
>

Run1
Hi! my age is 25 What
is your age ? 21
We are born on the same day. Are we twins!
Run2
Hi! my age is 25 What
is your age ? XSL
We are born on the same day. Are we twins!
Run3
Hi! my age is 21
What is your age ?

The statement in main ()

Copyrighted material
 2 Mastering C++

i f( myage = yourage ) has the expression myage = yourage. It assigns the contents of the
variable yourage, to myage. It is evaluated to true, for all nonzero values of yourage and hence, the
program prints the same message except for zero input value. The programmer must be careful while
writing the statement, which checks for the equality of data.

Copyrighted material
 Chapter 6: Arrays and Strings 169

cout « "Enter person 1 age: "; cin »

agel; sum += agel;
cout << "Enter person 2 age: "; cin »
age2; sum += age2;
cout « "Enter person 3 age: cin »
age3; sum += age3;
cout « "Enter person 4 age: "; cin »
age4; sum += age4;
cout « "Enter person 5 age: "; cin »
age5; sum ♦= ageS;
cout « "Average age = " « sum/5;
)

Run
Enter person 1 age: 21
Enter person 2 age: 40. Enter
person 3 age: 2£ Enter person 4
age: 27 Enter person 5 age: 25
Average age =29
The above program uses distinct statements to read and add the age of each person. The resulting value of
summation is stored in the variable sum Finally, the average age is computed by dividing the sum by 5. A program
written in this style is very clumsy, and difficult to enhance. If there are a large number of individuals, the number of
statements increase proportionately. A more elegant approach is to use an array type variable to store the age of
persons, and process them using loops as illustrated in the program age2 . cpp.

/ / age2.cpp: arrays to handle data which are of the same type ♦include <iostream.h> void main () {
int age[5]; // array definition
float sum = 0;
fort int i = 0; i < 5; i++ )
{
cout << "Enter person " « i+1 « " age: "; cin » age(i);
// reading array elements
)
fort i = 0; i < 5; i++ )
sum += age[il; // array manipulation cout « "Average age =
" « sum/5;

Run
Enter person 1 age: 23

Copyrighte
d material
6.3 Array Illustrations
The program elder. cpp finds the age of the eldest and youngest person in a family. It reads the ages of all the members
of a family stores them in an array and then scans the array to find out the required information.

/ / elder.cpp: finding youngest and eldest person age •include

<iostream.h> void main (} {
int i, tif
float age[25], younger, elder;
cout « "how many persons are there in list <max-25> ? ";
cin » n;
for{ i = 0; i < n; i++ ) {
 Chapter 6: Arrays and Strings 169

cout « "Enter person" « i+1 « " age: "; cin >> age[i];
)
// finding youngest and eldest person age begins here
younger = age[0];
elder = age[0];
for (i = I; i < n; i++)
(
if( age[i] < younger )
younger = age[i]; else
if{ age[i] > elder ) elder =
age[i];
>
// finding younger and elder person ends here
cout « "Age of eldest person is " « elder « endl;
cout << 'Age of youngest person is ■ << younger;
)

Run
How many persons are there-in list <max-25> ? 7
Enter personl age: 25.
Enter person2 age: 4
Enter person3 age: £5_
Enter person4 age: lfi.
Enter person5 age: i5_
Enter person6 age: 22.
Enter person7 age: 22.
Age of eldest person is 45
Age of youngest person is 4

Bubble Sort
A classical bubble sort is the first standard sorting algorithm most programmers learn to code. It has gained popularity
because it is intuitive, easy to write and debug, and consumes little memory. In each

Copyrighte
d material
case 9: case 10:
gap = 11;
break;
)
flag = 1;
int top = size - gap;
for( i = 0; i < top; i++)
{
j = i+gap;
if( age[ij > age[j] ) {
flag = 0; // still not sorted and requires next iteration // exchange contents of age[i] and
agelj] temp = age[iJ; agelij = age[jJ; age[j] = temp;
>
}
} while( Iflag || gap > 1 ); // sorting ends here cout
<< 'Sorted list..." « endl; for( i = 0; i < n; i ++ ) cout «
age[i] « " ■;
}

Run
How many elements to sort <max-25> ? 2
Enter age[ 0 ]: 2
 Chapter 6: Arrays and Strings 169

Enter age[ 1 ] : 5
Enter age[ 2 ] : 2
Enter age[ 3 ] : 4
Enter age[ 4 ] : £
Enter age[ 5 ] : 1
Enter age[ 6 ] : £
Sorted list...
1234569
Although the algorithm for comb son and shell sort appear to be very similar (both use a gap and a shrink factor),
they do in fact perform differently. The shell sort does a complete sort (until there are no more swaps to be made) for
each gap size, comb sort makes only a single pass for each gap size-it can be thought of as a more optimistic version of
the shell sort. There are other differences that result from this optimism: The ideal shrink factor for shell sort is 1.7.
compared with 1.3 of comb sort. The complexity obtained by plotting sorting time against the list of size n, for shell
sort, appears as a step function of (n*log 2n*log2n), whereas for comb sort it approximates to a flatter curve of (n *
logjn).

6.4 Multi-dimensional Arrays

Most of the scientific data can be easily modeled using multi-dimensional arrays. Such representations allow
manipulation of data easily and even allow the programmer to write simple and efficient programs. Matrix is a two
dimensional array and two subscripts are required to access each element.
Run
Please note the following message: C to C++ C++
to Java Java to ...
Note that the characters \ and n used in the string definition char msg[] = "C
to C++\nC++ to JavaXnJava to are treated as a new line character.

6.6 Strings Manipulations

C++ has several built-in functions such as strlen (), strcat (), strlwr (), etc., for string manipulation. To use these
functions, the header file string.h must be included in the program using the statement
#include <string.h>
String Length
The string function strlen {) returns the length of a given string. A string constant or an array of characters can be
passed as an argument. The length of the string excludes the end-of-string character (NULL). The strlen.cpp illustrates
the use of strlen{) and user defined function to find the length of the string.
/ / strlen.cpp: computing length of string
♦include <iostream.h>
♦include <string.h>
void main()
{
char si[251 ;
cout « "Enter your name: "; cin >> si;
cout « "strlen( si ): " << strlen(sl) « endl;
)

Run
Enter your name: Smrithi strlen( si ) :
7

String Copy
The string function strcpy() copies the contents of one string to another. It takes two arguments, the first argument is the
destination string array and the second argument is the source string array. The source string is copied into the
destination string. The program strcpy.cpp illustrates the use of strcpy () to copy a string.

/ / strcpy.cpp: copying string *>iiiclude

<iostrearn.h> fci:iclude <string.h> vcid main() {
char sl[25], s2 [25J;
 Chapter 6: Arrays and Strings 185

cout << 'Enter a string: "; cin >>

si; strcpyf s2, si ) ;
cout <<. "strcpyt s2, si ); " << s2;
}

Run
Enter a string: Garbage. strcpy( s2,
si ): Garbage

String Concatenation
The siring function strcat () concatenates two strings resulting in a single string. It takes two arguments, which
are the destination and source strings. The destination and source strings are concatenated and the resultant string
is stored in the destination (first) string. The program strcat. cpp illustrates the use of strcat {) to concatenate two
strings.
// Strcat.cpp: string concatenation •include
<iostream.h> •include <string.h> void main()

char si[40), s2 [25] ;

cout << "Enter string si: ";
cin >> si;
cout « "Enter string s2: "; cin > > s 2;
strcat( si, s2 ) ;
cout << "strcatl si, s2 ): " << si;
)

Run
Enter string si: £ Enter string
s2: ++ strcat( si, s2 ) : C+ +

String Comparison
The string function strcmp () compares two strings, character by character. It accepts two strings as parameters
and returns an integer, whose value is
♦<0 if the first string is less than the second
♦ == 0 if both are identical
♦>0 if the first string is greater than the second
Whenever two corresponding characters in the string differ, the string which has the character with the
higher ASCII value is greater. For example, consider the strings hello and Hello! !. The first character itself
differs. The ASCII code forh is 104. while the ASCII code forH is 72. Since the ASCII code of h is greater, the
string hello is greater than the string Hello!. Once a differing character is found, there is no need to compare
remaining characters in the string. The program strcmp.cpp illustrates the use of strcmp () to compare two
strings.

Copyrighted material
 186 Mastering C++

/ / strcmp.cpp: string concatenation

•include <iostream.h>
•include <string.h>
void main( )
{
char si[25], s2[25];
cout << "Enter string si: " ;
cin >> si;
cout << "Enter string s2: " ; cin » s2;
int status = strcmp! si, s2 ) ; cout << "strcmpf
si, s2 > : " ; if{ status = = 0 )
cout << si << " is equal to " « s2; else
i f ( status > 0 )
cout << si << " is greater than " << s2;
else
cout << si << " is less than " << s2;
}

Run
Enter string si: Computer Enter string s2:
Computing
strcmp( si, s2 ) : Computer is less than Computing

String to Upper/Lower Case

The functions strlwr ( ) and strupr ( ) convert a string to lower-case and upper-case respectively and return the address of the
converted string. The program uprlwr .cpp illustrates the conversion of string to lower and upper cases.

/ / uprlwr.cpp: converting string to upper or lower case

•include <iostream.h>
•include <string.h>
void main( )
(
char sl[25], temp[25]; cout << "Enter a
string: ■; cin >> si; strcpyi temp, si ) ,-
cout << "strupr( temp ) : " << struprt temp ) « endl; cout << "strlwr{ temp ) : "
<< strlwr( temp ) << endl;
)

Run
Enter a string: Smrithi
strupr( temp ): SMRITHI
strlwr( temp ) : smrithi

Copyrighted
material
6.7 Arrays of Strings
An array of strings is a two dimensional array of characters and is defined as follows:
char array-name\row_size][column_size}; For instance, the statement
char person[10][15] ; defines an array of string which can store names of 10 persons and each name cannot exceed 14
characters; 1 character is used to represent the end of a string. The name of the first person is accessed by the expression person[0],
and the second person by person[l], and so on. The individual characters of a string can also be accessed. For instance, the first
 186 Mastering C++

character of the first person is accessed by the expression person[0] [0] and the fifth character in the 3* person's name is accessed
by person [21 [4 ]. The program names . cpp illustrates the manipulation of an array of strings.
// names.cpp: array of strings storing names of the persons
♦include <iostream.h>
♦include <string.h>
const int LEN = 15;
void main ()
{
int i, n;
char person[10][LENJ;
cout << "How many persons ? ";
cin >> n;
fort i = 0; i < n; i + + )
{
cout << "Enter person" « i + 1 << " name: "; cin » person[i];
}
cout<< "-------- cout<< ----------------------------------------------\n-
cout « " P * Length In lower case In UPPER case\n"
Person Name ..............-----------------------------------\n"
for ! i - i < n; i++ J {
cout.width( 2 ) ;
cout « i + 1;
cout.width( LEN ) ;
cout << person[i] << " *;
cout .width ( 2 ) ,-
cout << strlen{ person[il ) « " cout.width{ LEN > ; cout <<
strlwr(person[i]); cout.width< LEN ) ;
cout << strupr(person[i]) << endl;
}
cout« "-----------------------------------------------------------------------------------------------------\n";
 .90
Mastering C++

a::.:
a[1001 = 200; cout «
a[101J ;
= 5; cout
« aC-1 J ;
Does the compiler reports an error when illegal accesses are made to an array ?
6.4 What are multi-dimensional arrays ? Explain their syntax and mechanism for accessing their ele-
ments.
6.5 Write an interactive program for calculating grades of N students from 3 tests and present the
result in the following format:

S I . NO. SCORES AVERAGE GRADE

XX XX XX XX XX X

6.6 Write a program for computing the norm of the matrix.

6.7 Can arrays be initialized at the point of their definition ? If yes, explain its syntax with suitable
examples ?
6.8 Write a program to find the symmetry of the matrix.
6.9 What are strings ? Are they standard or derived daia type ? Write an interactive program to check
whether a given string is palindrome or not. What happens if the end-of-string character is
missing ?
6.10 Write a program to sort integer numbers using shell sort and compare its time complexity with
that of the comb sort.
6.11 Write a program for computing mean(m), variance, and standard deviation(s) of a set of
numbers using the following formulae:

mean - m * -i- 2jxt

variance = ~ J) (xi -

m) s = V variance
6.12 Write a program to find the transpose of a matrix. (The transpose can be obtained by inter-
changing the elements of rows and columns).
6.13 Write a program to find the saddle points in a matrix. It is computed as follows: Find out the
smallest element in a row. The saddle point exists in a row if an element is the largest element in
that corresponding column. For instance, consider the following matrix:
7 5 6 10 2 3 1 3 3 The saddle point results are as listed below: In row 1,
saddle point exists at column 2. In row 2, saddle point does not exist. In row 3, saddle point does not
exist. 6 14 Write an interactive program to multiply two matrices and print the result in a matrix form.

i
I
Copyrighted material
6 Mastering C++

Monolithic Programming
The programs written in these languages exhibit relatively flat physical structure as shown in
Figure 1.3. They consist of only global data and sequential code. Program flow control is
achieved through the use of jumps and the program code is duplicated each time it is to be used,
since there is no support of the subroutine concept and hence, it is suitable for developing small
and simple applications. Practically, there is no support for data abstraction and it is difficult to
maintain or enhance the program code.
Examples: Assembly language and BASIC

2 ..........
goto 10&

goto 55

goto 3

goto T5
goto 6
100 ..............

Figure 1.3: Monolithic programming

Procedural Programming
Programs were considered as important intermediate points between the problem and the
computer in the mid-1960s. Initially, software abstraction achieved through procedural
abstraction grew directly out of this pragmatic view of software. Subprograms were originally
seen as labor-saving devices but very quickly appreciated as a way to abstract program functions
as shown in Figure 1.4.
The following are the important features of procedural programming:
♦ Programs are organized in the form of subroutines and all data items are global
♦ Program controls are through jumps (gotos) and calls to subroutines
♦ Subroutines are abstracted to avoid repetitions
♦ Suitable for medium sized software applications
♦ Difficult to maintain and enhance the program code
Examples: FORTRAN and COBOL

Global Data

Subprograms

Figure 1.4: Procedural programming

■ ■■■ swapl.cpp: swap integer
values by value • include
<iostream.h> void swap( int x, int
y)
(
int t; / / temporary used in swapping
cout«"Value of x and y in swap before exchange: •<< x «■ "<< y << endl; t = x;
x

y
;

Copyrighted material
 6 Mastering C++

t
;
cout<<"Value of x and y in swap after exchange: "<< x <<" " << y << endl;
)
v
o
i
'
d

m
a
i
n

{
)

{
int a, b;
cout << "Enter two integers
<a, b>: " ; cin » a >> b; swap £
a, b ) ;
cout « "Value of a and b on swapt a, b I in mainO : ■ << a << " " « b;
)

Run
Enter two integers <a, b>: 10 20 Value of x
and y in swap before exchange: 10 20 Value
of x and y in swap after exchange: 20 10
Value of a and b on swapt a, b ) in mainO : 10
20

In main
(),
the
state
ment
swa
p( x,
y)
invokes the function swap ( ) and assigns the contents of the actual parameters a and b to the
formal parameters x and y respectively. In the swap ( ) function, the input parameters arc
exchanged, however it is not reflected in the caller; actual parameters a and b do not get modified
(see Figure 7.8).

Copyrighted material
6 Mastering C++

Co
pyrig
hted
mate
rial
Pass by Address
C++ provides another means of passing values to a function known as pass-by-address. Instead of
passing the value, the address of the variable is passed. In the function, the address of the
argument is copied into a memory location instead of the value. Thfe de-referencing operator is
used to access the variable in the called function.

// swap2.cpp: swap integer values by pointers

♦include <iostream.h>
void swap( int * x, int * y )
{
int t; // temporary used in swapping t =
*x; *x = *y; *y = t;
}
void
main!)
{
int a, b;
cout « "Enter two integers <a, b>: ■; cin
>> a >> b; swap( &a, tb );
cout << "Value of a and b on swapt a, b ): " « a « " " « b;
)

Run
Enter two integers <a, b>: 10 20
Value of a and b on swap ( a, b ) : 20 10
In main (), the
statement
swapt &x,
&y )
invokes the function swap and assigns the address of the actual parameters a and b to the formal
parameters x and y respectively.

void main (void) void swap(int *x, int *y)

{

int , b; int t;
a

Copyrighted material
6 Mastering C++

t=*x;
swap(&a, &b); 20,

/ >
)

/?
1552
<----10------
s ■-1552-- X
y
/1554 1 ----20------ r-1554-

Figure 7.g: Parameter passing by address

In swap (), the statement
t = *x;
assigns the contents of the memory location pointed to by the pointer (address) stored in the
variable x (It is effectively accessing the contents of the actual variable a in the caller. Similarly,
the parameter y holds the address of the parameter b. Any modification to the memory contents
using these addresses will be reflected in the caller; the actual parameters a and b get modified
(see Figure 7.9).

Pass by Reference
Passing parameters by reference has the functionality of pass-by-pointer and the syntax of call-
by-value. Any modifications made through the formal pointer parameter is also reflected in the
actual parameter. Therefore, the function body and the call to it is identical to that of call-by-
value, but has the effect of call-by-pointer.
To pass an argument by reference, the function call is similar to that of call by value. In the
function declarator, those parameters, which are to be received by reference must be preceded by
the & operator. The reference type formal parameters are accessed in the same way as normal
value parameters. However, any modification to them will also be reflected in the actual
parameters. The program swap3. cpp illustrates the mechanism of passing parameters by
reference.
// swap3.cpp: swap integer values by reference
•include <iostreara.h>
void swap{ int & x, int & y }
{
int t; // temporary used in swapping t = x;
x = y; y = t;
)
void
main() (
int a, b;
cout << "Enter two integers <a, b>: "; cin » a
» b; swap( a, b );
cout « "Value of a and b on swapt a, b ) : ■ << a « " " « b;
)

Run
Enter two integers <a, b>: 1Q_
Value of a and b on swapt a, b ) : 20 10
In main (), the statement
swap t a,
b ) ; is translated into
swap ( & a, & b ) ; internally during
compilation. The function declarator
void swapt int & a, int & b ) indicates that the formal parameters are of reference type
and hence, they must be bound to the memory

Copyrighted material
6 Mastering C++

location of the actual parameter. Thus, any access made to the reference formal parameters in the
swap () function refers to the actual parameters. The following statements in the body of the swap
() function:
t
=
x;
x
=
y;
y
=
t;
(as treated by the compiler) have the following interpretation
internally: t = *x; // store the value pointed by x into t
*x = *y; // store the value pointed by y into location pointed by x *y = t ; // store the
value hold by ' t ' into location pointed by y
This is because, the formal parameters are of reference type and therefore the compiler treats
them similar to pointers but does not allow the modification of the pointer value (cannot be made
to point to some other variable). Changes made to the formal parameters x and y reflect on the
actual parameters a and b (see Figure 7.10).

void main (void) { void swap{int &x, int &y) {

int a, b; int t; t=x;
y;
--y=t; )
swap(a, b)

Figure 7.10: Parameter passing by reference

Copyrighted material
6 Mastering C++

--1552--

±334 — •

The following points can be noted about reference parameters:

♦ A reference can never be null, it must always refer to a legitimate object (variable).
♦ Once established, a reference can never be changed so as to make it point to a different
object.
♦ A reference does not require any explicit mechanism to dereference the memory
address and access the actual data value.
Note
Procedures can be implemented using functions. A function with no return value can be treated
similar to a procedure of Pascal. The main difference between using functions and procedures in
C++ (or C) is that function can be placed on the right side of the '=* (assignment) and on either
side of = (equal) operator. Procedures (function with no return values) cannot be used with, these
operators. The return "alue from the function can be directly passed to cout for display, whereas
procedures cannot be used tn the cout Statement.
The value of a and b on execution of max(a, b ) = 4 2 5 ; . . . a = 1 b = 425
Run2
Enter two integers < a , b>: 2 1
The value of a and b on execution of max{a, b ) = 4 2 5 ; . . . a = 425 b = 1
In main ( ) , the
statement max{ a ,
b ) = 425;
invokes the function max. It returns the reference to the variable holding the maximum
value and assigns the value 425 to it (see Run2). Since the return type of the m a x ( ) is int &,
it implies that the call to m a x ( ) can appear on the left-hand side of an assignment
statement. Therefore, the above statement is valid and assigns 4 2 5 to a if it is larger,
otherwise, it is assigned to b.

7.8 Default Arguments

Normally, a function call should specify all the arguments used in the function definition. In a C+
+ function call, when one or more arguments are omitted, the function may be defined to take
default values for the omitted arguments by providing the default values in the function prototype.
Parameters without default arguments are placed first, and those with default values are
placed later (because of the C++ convention of storing the arguments on the stack from right to
left). Hence the feature of default arguments allows the same function to be called with fewer
arguments than defined in the function prototype.
To establish a default value, the function prototype or the function definition (when functions
are defined before being called) must be used. The compiler checks the function
prototype/declarator with the arguments in the function call to provide default values (if
available) to those arguments, which are omitted. The arguments specified in the function call
explicitly .always override the default values specified in the function prototype/declarator. In a
function call, all the trailing missing arguments are replaced by default arguments as shown in
Figure 7.11.

void PrintLine( c h a r = ' - ' , i n t = 7 0 ) ;

void m a i n t ) {
PrintLine( ) ; ................
PrintLine( ' ! ' ) ; ...........
PrintLine{ ■ * ' , 4 0 J ;■
PrintLine! * R ' , • 5 5 ) ;.
)

Copyrighted material
6 Mastering C++

s
void main{ ) \
( \
> PrintLine( 70) ;
•> PrintLine(1! ' , 1 0 ) ; A'
> PrintLine{1 * ' , 40) ;
> PrintLine(' R ' , ) 5 5 ) ;

Figure 7.11: Preprocessor handling missing arguments at function call using

default arguments
When a function is called by omitting some arguments, they are supplied by the compiler
implicitly. The code of the program by no means becomes shorter or more efficient, but it
provides high flexibility
1. Therefore, the old code referring to this function need not be modified and new statements can
be added without affecting the functionality. The program defarg2.cpp extends the capability of
defargl.cpp program.
/ / defarg2.cpp: extending the functionality without modifying old calls •include <iostream.h>
void PrintLinet char = int = 70, int = 1 );
void main()
{
PrintLinet); // uses both default arguments
PrintLinet ' ' ■ ' ) ; I I assumes 2nd argument as default
PrintLinet '*', 40 ); // ignores default arguments PrintLinet 'R', 55 ); // ignores default
arguments // new code, Note: old code listed above is unaffected PrintLinet , 25, 2 ) ;
)
void PrintLinet char ch, int RepeatCount, int nLines ) {
int i, j;
fort j = 0 ; j < nLines; j++ )
{
cout « endl;
fort i = 0; i < RepeatCount; i++ )
cout « ch;
)
)

Run

[ i [ ■ j j ! ! J ■ J ! ! ! ! ! ■ ! ! ! ! I I I I J I I I I t 11 11111111 11JI I 1 I ! I!111111H1 I I 11 1

I I 111
a***************************************

RRRRRRRRRRRRRRRRRRRRRRRRRRRRBRRRRRRRRRRRRRRRRRRRRRRRRRR
&&&&&&&&&&&&&&&&&&&&&&&&&
&&&&&&&&&&&&&&&&&&&&&&&&&

The following statements in the above two programs

PrintLinet); // uses both default arguments
PrintLinet '!' ); // assumes 2nd argument as default
PrintLinet 40 ); // ignores default arguments
PrintLinet 'R', 55 ); // ignores default arguments
are the same. Though the functionality of PrintLine () is enhanced in the program def arg2 . cpp,
the old code referring to it remains unaffected in terms of its functionality; the compiler supplies
the last argument as 1, and thereby the new function does the same operation as that of the old
one. Thus, the C++ feature of default arguments can be potentially utilized in extending a
function without modifying the old code.
A default argument can appear either in the function prototype or definition. Once it is
defined, it cannot be redefined. It is advisable to define default arguments in the function
prototype so that it is known to the compiler at the time of compilation. Variable names may be
omitted while assigning default \ alues in the prototype.
7.9 Inline Functions
Function calls involve branching to a specified address, and returning to the instruction following
the function call. That is. when the program executes a function call instruction, the CPU stores

Copyrighted material
6 Mastering C++

the memory address of the instruction following the function call, copies the arguments of the
function call onto the stack, and finally transfers control to the specified function. The CPU then
executes the function code, stores the function return value in a predefined memory
location/register, and returns control to the calling function. This constitutes an overhead in the
execution lime of the program. This overhead is relatively large if the time required to execute a
function is less than the context switch time.
C++ provides an alternative to normal function calls in the form of inline functions. Inline
functions arc those whose function body is inserted in place of the function call statement during
the compilation process. With the inline code, the program will not incur any context switching
overhead. The concept of inline functions is similar to macro functions of C. Hence, inline
functions enjoy both the flexibility and power offered by normal functions and macro functions
respectively.
An inline function definition is similar to an ordinary function except that the keyword
inline precedes the function definition. The syntax for defining an inline function is shown in
Figure 7.12..

inline int
sqr( int num)
I
return num*num;
I

void main 0 ------- void main()

K
p-ep-ccesso' ^>
b-sqr(n); - - - > a *5*5; b=n-n;
) J

Figure 7.12: Inline function and its expansion

The significant feature of inline functions is: there is no explicit function call and body is
substituted at the point of inline function call, thereby, the run-time overhead for function linkage
mechanism is reduced. The program square, cpp uses inline function to compute the square of a
number.

/ / square.cpp: square of a number using inline function

((include <iostream.h> inline int sqr( int num ) {
return num*num;
)
vo
id
ma
in
{)
{
float n;
cout « 'Enter a number:
"; cin >> n;
cout << "Its Square = ■ « sqr( n ) « endl;

Copyrighted material
 Chapter 1: Object-Oriented Paradigm

Structured Programming
Structured programming has evolved as a mechanism to address the growing issues of
programming-in-the-large. Larger programming projects consist of large development teams,
developing different parts of the same project independently. The usage of separately compiled
modules (algorithmic decomposition) was the answer for managing large development teams (see
Figure 1.5). Programs consist of multiple modules and in turn, each module has a set of functions

Subprograms

Module 1 of related types. Module 3

Module 2

Figure 1.5: Structured programming

The following are the important features of structured programming:

♦ Emphasis on algorithm rather than data
♦ Programs are divided into individual procedures that perform discrete tasks
♦ Procedures are independent of each other as far as possible
♦ Procedures have their own local data and processing logic
♦ Parameter passing facility between the procedures for information communication
♦ Controlled scope of data
♦ Introduction of the concepts of user defined data types
♦ Support for modular programming
♦ Projects can be broken up into modules and programmed independently
♦ Scope of data items is further controlled across modules
♦ A rich set of control structures are available to further abstract the procedures
♦ Co-ordination among multiple programmers is required for handling the changes made to
mutually shared data items
♦ Maintenance of a large software system is tedious and costly
Examples: Pascal and C

Object Oriented Programming

The easy way to master the management of complexity in the development of a software system
is through the use of data abstraction. Procedure abstraction is suitable for the description of
abstract operations, but it is not suitable for the description of abstract objects. This is a serious
drawback in many applications since, the complexity of the data objects to be manipulated
contribute substantially to the overall complexity of the problem.

Copyrighted material
 Chapter 7: Modular Programming with Functions 68

float c. d;
cout << "Enter two floats <c, d>: ■; cin » c » d;
swapt c, d ) ; / / compiler calls swapt float &a, float &b ) ; cout << "On swapping <c, d>: " << c << " ■
<< d;

Run
Enter two Characters <chl, ch2>: R K
On swapping <chl, ch2>: K R
Enter two integers <a, b>: 5 10
On swapping <a, b>: 10 5
Enter two floats <c, d>: 20.5 99.5
On swapping <c, d>: 99.5 20.5
In the above program, three functions named swap ( ) are defined, which only differ in their argument data
types: char, int, or float. In maint), when the statement swap( chl, ch2 ) ;
is encountered, the compiler invokes the swap ( ) function which takes character type arguments. This decision
is based on the data type of the arguments, (see Figure 7.13).

void swaptfloat &x, float £y) ; ..

s
void swaptint &x, int &y) ; .
1
void swap (char fix, char iy); \

void main() ' * '

i y * /
char chl, ch2; „" * /
int a, b;
float x, y; „ - "* „-"
swaptchl, ch2) ; - " "* ^+
swap (a, b);--------------" ' '
swap(x, y) ;------------------
>

Figure 7.13: Function overloading

It is interesting to note the way in which the C++ compiler implements function overloading, Although the
functions share the same name in the source text (as in the example above, swap), the compiler (and hence the
linker) uses quite different names. The conversion of a name in the source file to an internally used name is
called name mangling. It can be performed as follows: the C++ compiler might convert the name void swaptint
&,int &) to the internal name say Vshowl, while an analogous function with a char* argument might be called
VswapCP. The actual names, which arc internally used, depend on the compiler and are transparent to the
programmer.

Another typical example program of function overloading is illustrated in show. cpp.

' / show.cpp: display different types of information with same function •include <iostream.h> void show( int
val ) (
cout « "Integer: ■ « val « endl;
>
void show( double val ) {
cout << "Double: " << val << endl;
>
void showt char *val ) (
cout « "String: " << val << endl;
)
int main ( )
{
showt 420 ) ; / / calls showt int val ) ;
show( 3.1415 ) ; / / calls showt double val ) ;
showt "Hello World\n!" ) ; / / calls showt char *val I : return( 0 ) ;
)

Bun
Integer: 420 Double: 3.1415
String: Hello World
i

The following remarks can be made on function overloading:

* The use of more than one function with the same name, but having different actions should be avoided. In the
above example, the functions show ( ) are somewhat related (they print information on the screen). However,
it is also possible to define two functions, say lookup (); one of which would find a name in a list, while the
other would determine the video mode. In this case, the two functions have nothing in common except their

Copyrighted material
 name. It would, therefore, be more practical to use names such as f indname ( ) and getvidmode ( ) , which
suggest the action they perform.
* C++ docs not permit overloading of functions differing only in their return value. The reason is that it is
always the programmer's choice to inspect or ignore the return value of a function. For instance, the fragment
printf("Hello World!\n"); holds no information concerning the return value of the function printf ( ) .
(The return value in this case is an integer, which states the number of printed characters. This return value is
practically never inspected). Two functions print f ( ) , which would only differ in their return types and hence
they are not distinguished by the compiler.
* Function overloading can lead to surprises. For instance, imagine the usage of statements
show( 0 J ;
show{ NULL ) ;
where there are multiple overloaded functions as in the program show.cpp. The zero could be interpreted here
as a NULL pointer to a char, i.e., a t char*) 0, or as an integer with the value zero. C++ will invoke the function
expecting an integer argument, which might not be what the user expects.
fort int i = 0 ; ( i < size - 1 > USt swapped; i++ )
r
swapped = false;
f or{ int j = 0 ; j < ( s i z e - 1 ) - i; j++ ) i f f a [ j ] >
a[ j * 1 ] ) (
swapped = true;
swap t a[ j ] , a [ j + 1 ] ) ;
)
)
)
void main( void )
(
int a [25] ; int i,
size;
cout « "Program to sort elements..." « endl;
cout « "Enter the size of the integer vector <max-25>:
cin » size;
cout << "Enter the elements of the integer vector..." << endl; fort i = 0; i < size; i++ )
cin » a[i] ; BubbleSort ( a, size ) ; cout «
"Sorted Vector:" « endl; for( i = 0; i < size; 1+
+)
cout « a[ij « "

Hun
Program to sort elements...
Enter the size of the integer vector <max-25>: 5 Enter the elements
of the integer vector. . .

£
I
1
2
Sorted Vector: 2 3
6 8 9
In main ( ) , the statement
BubbleSort ( a, size ) ; invokes the sorting function by passing the address of the array variable a and
the value of the variable size to it. Hence, any modification made to the elements of the airay a will be reflected
in the caller.

7.13 C++Stack
The medium for communication between a caller and the callee is the stack, which is used to store function
parameters, return address, local variables, etc. When the function is invoked, the information such as return
address and parameters, are pushed onto the stack by the function linkage mechanism;

Copyrighted material
 70 Mastering C++

block.cpp: illustration of the variables scope in blocks 'include <iostream.h> int main( void )

int i = 144;
cout << "i = " << i;

/" nested block*/ int k =

12; cin » k; i = i % k ;
)
i f ( i == 0 )
cout << * i is a divisor of " « k; / / Error: k undefined in mainO return 0;
>

Reference lo variable k in the main block results in a compile-time error: Undefined symbol k in the
function m a x n O i the variable k is declared inside the nested block within mainO. The memory space for
the variable k is allocated when the execution of the block starts, and released when execution reaches the
end of the nested block. When a variable is accessed, the compiler first checks for its existence in the
current block, and then moves outwards if it does not exist in the current block; this process continues
until the global definition. The function can access the identifiers in the parameter list, the local definitions
and the global definitions (if any).

7.15 Storage Classes

The period of time during which memory is associated with a variable is called the extent of the variable.
It is characterized by storage classes. The storage class of a variable indicates the allocation of storage
space to the variable by the compiler. Storage classes define the extent of a variable. C++ supports the
following four types of storage classes:
♦ auto
♦ register
♦ extern
♦ static
The syntax for defining variables with explicit storage class is shown in Figure 7.16. The storage
classes except extern are used for defining variables; extern is used for declaration of variables. The scope
and extent of auto and register storage class is the same. The scope of static
variables is limited to its block (maximum to a file), but its extent is throughout the execution time of the
program (docs not matter whether it is local or global type).

auto, register, s t a t i c , or extern

StorageClass DataType Variablel,

Figure 7.16: Storage classes and variable declaration

Declaration Versus Definition

A declaration informs the compiler about the existence of the data or a function some where in the
program. A definition allocates the storage location. In C++, apiece of data or function can be declared in
several different places, but there must only be one definition. Otherwise, the linker will complain
(generates multiple definition error) while uniting all the object modules, if it encounters more than one
definition for the same function or piece of data. Almost all C and C++ programs require declarations.
Therefore, it is essential for the programmer to understand the correct way to write a declaration. As far as
data is concerned, except extern storage class, all others define data i.e., they not only direct the compiler,
but also allocate resource for a variable.

Auto Variables
By default, all the variables are defined as auto variables. They arc created when the function/block is
entered and destroyed when the function/block is terminated. The memory space for local auto variables is
allocated on the stack. The global auto variables are visible to all the modules of a program, and hence,
they cannot be defined many times unlike the declarations.

Copyrighted material
 Chapter 7: Modular Programming with Functions 71

Register Variables
The allocation of CPU (processor) registers to variables, speeds up the execution of a program; memory is
not referred when such variables are accessed. The number of variables, which can be declared as register
are limited (typically two or three), within any function or as global variables (else they are treated as auto
variables). A program that uses register variables executes faster when compared to a similar program
without register variables. It is possible to find out the allocation of register variables only by executing
and comparing the timing performance of the program (perceptible in large programs). It is the
responsibility of the compiler to allot register variables. In case the compiler is unable to do so. these
variables are treated as auto variables. It is advisable to define frequently used variables, such as loop
indices, as register variables. It is illustrated in the program regvar. cpp.

/ / regvar.Cpp: use of register variable as loop index

•include <iostream.h>
•include <string.h>
void main ( )
{
char name[30];
register int i ; / / register variable
cout « "Enter a string: " ; cin >>
name;
cout << "The reverse of the string is: " ; f o r t i =
strlen( name ) - l ; i >= 0; i-- ) cout << name[i] ;
)

Runl
Enter a string: mahatma
The reverse of the string is: amtaham
Hun2
Enter a string: malavalam
va_start(ap, msg);
while ((arg = va_arg(ap,int)) != 0) { total += arg;
}
cout << msg « total; va_end(ap) ;
}

int main(void) {
sum("The total of 1+2+3+4 is *. 1 , 2 , 3 , 4 , 0 ) ; return 0;
>

Run
The total of 1+2+3+4 is 10
In main ( ) , the statement
sum("The total of 1+2+3+4 is •, 1 , 2 , 3 , 4 , 0 ) ; invokes the variable argument function. The
function sum ( ) is designed such that when a zero valued argument is encountered, it is understood that
no more arguments exists for further processing. Hence, the last argument 0 (zero) in this case, is the end-
of-argument indicator. The programmer has full freedom for selecting suitable end-of-argument indicator.

7.17 Recursive Functions

Many of the scientific operations are expressed using recurrence relations. C++ allows the programmers to
express such a relation using functions. A function that contains a function call to itself, or a function call
to a second function which eventually calls the first function is known as a recursive function. The
recursive definition for computing the factorial of a number can be expressed as follows:

1 if n = 0 fact(n) =
n * f a c t ( n-1), otherwise
Recursion, as the name suggests, revolves around a function recalling itself. Recursive functions are
those, in which there is atleast one function call to itself (there can be more than one call to itself as in the
tower of hanoi algorithm). The recursive approach of problem solving substitutes the given problem with
another problem of the same form in such a way that the new problem is simpler than the original.
Two important conditions which must be satisfied by any recursive function are:
1. Each time a function calls itself it must be nearer, in some sense, to a solution.
2. There must be a decision criterion for stopping the process or computation.
Recursive functions involve the overhead of saving the return address, formal parameters, local
variables upon entry, and restore these parameters and variables on completion.
Factorial of a Number

Copyrighted material
 72 Mastering C++

The program r f a c t . cpp computes the factorial of a number. It has a recursive function f a c t ( ) which
implements the above stated definition of recursion.
The data items enclosed between flower brackets in the above structure declaration are called structure
elements or structure members. Student is the name of the structure and is called structure tag. Note that,
some members of Student structure are integer type and some are character array type. The description of
various components of the structure Student is shown in Figure 8.2.

structure name

struct Student {
int roll_no; char name(251; char branch[15]; int marks;

members of the structure

Figure 8.2: Declaration of structure Student

The individual members of a structure can be variables of built-in data types, pointers, arrays, or even
other structures. All member names within a particular structure must be different. However, member
names may be the same as those of variables declared outside the structure. The individual members
cannot be initialized inside the structure declaration. For example, the following declaration is invalid:
struct Student
{
int roll_no = 0 ; / / Error: initialization not allowed here char name[2 5 ] ; char
branch[151; int marks;
);

8.3 Structure Definition

The declaration of a structure will not serve any purpose without its definition. It only acts as a blueprint
for the creation of variables of type struct (structure). The structure definition creates structure variables
and allocates storage space for them. Structure variables can be created at the point of structure declaration
itself, or by using the structure tag explicitly as and when required. The most commonly used syntax for
structure definition is shown in Figure 8.3.

keyword struct is optional

[struct] StructureName varl, var2, structure variables

...;

Figure 8.3: Syntax of structure definition

The use of the keyword struct in the structure definition statement is optional. The following
siatements create variables of the structure student declared earlier:
struct Student
si; or
Student si ;

Figure 8.4 shows the storage of the members of the structure student.

-A 2 bytes
-A 25 byles
15 bytes
struct Student ( 2 bytes
int roll_no; char
name[ 2 5 ] ; char
branch[15]; int marks; > -

Student s i ;

Figure 8.4: Storage organisation when structure variable is defined

Copyrighted material
 Chapter 7: Modular Programming with Functions 73

The structure variables can be created during the declaration of a structure as follows:
struct Student
{
int roll_no; char
name[25]; char
branch[15] ;
int.marks; } s i ;
In the above declaration, Student is the structure tag, while si is a variable of type Student. If variables of
this structure type are not defined later in the program, then the tag name Student can be omitted as shown
below:
struct
{
int roll_no; char
name[ 2 5 ] ; char
branch[15J; int
marks; ) Si;
It is not a good practice, to have both declaration and definition in the same statement.
Multiple variables of a structure can be created using a single statement as follpws:
struct Student s i , s 3 , s4; or
Student si, s 3 , s 4 ;
All these instances are allocated separate memory locations and hence, each one of them are indepen dent
variables of the same structure type as shown in Figure 8.5.

struct Student {
int roll_no; char
name[ 2 5 1 ; char
branch[15); int
marks;
J ;

Student s l , s 2 , s 3 ;

si s2 s3

Figure 8.5: Variables of type Student

8.4 Accessing Structure Members

C++ provides the period or dot (.) operator to access the members of a structure independently. The dot
operator connects a structure variable and its member. The syntax for accessing members of a structure
variable is shown in Figure 8.6.
structure variable
structvar .membernajne structure member
Figure 8.6: Accessing a structure member using dot operator

Here, structvar is a structure variable and membername is one of its members. Thus, the dot operator
must have a structure variable on its left and a legal member name on its right. Consider the following
statement:
Student s i ;
Each member of the structure variable si can be accessed using the dot operator as follows:
sl.roll_no will access si's roll_no
si.name will access si's name
s i . branch will access si's branch
s i . marks will access si's marks

Copyrighted material
 Chapters.* Structures and Unions 241

The following are valid operations on the structure

variable si:
si.roll_no = 5;
cin >> sl.roll_no;
strcpy{ s i . name, "Mangala" ) ;
cout « sl.name;
strcpyt s i . branch, "Computer" ) ;
Accessing members of a structure using structure tag is not allowed. Hence, a statement such as
Student.roll_no = 5 ; / / Error: Student is not a structure variable is invalid; structure name Student is a
data type like int, and not a variable. Just as int = 10 is invalid, Student.roll_no = 5 is invalid.
The program student 1 .cpp illustrates the various concepts discussed in the earlier sections such as
structure declaration, definition, and accessing members of a structure.

/ / studentl.Cpp: processing o f student data using structures

•include <iostream.h>
/ / structure declaration
struct Student
{
int roll_no; char
name[ 2 5 ] ; char
branch[ 1 5 ] ; int
marks;
) ;
void m a i n ( )
{
Student s i ; / / structure definition
cout « "Enter data for s t u d e n t . . . " « endl;
cout « "Roll Number ? ' ;
cin » sl.roll_no; / / accessing structure member
cout « "Name ? " ; c i n > >
sl.name; cout « "Branch ?
" ; c i n » s i . branch;
cout « "Total Marks <max-J25> ? ■ ;
cin » s i . marks; >'
cout « "Student Report" « endl;
cout « "---------------------------" « endl;
/ / process student data
cout « "Roll Number: " « s l . r o l l _ n o « endl;
cout << "Name: ■ << sl.name < < endl;
cout « "Branch: ' « s i . branch « endl;
cout « "Percentage: " « s i .marks* (100.0 / 3 2 5 ) « endl;
}

Run
Enter data for s t u d e n t . . .
Roll Number ? 5 Name ?
Mflflqftla

Copyrighted
material
 242 Mastering C++

Branch ? Computer
Total Marks <max-325> ? 290
Student Report

Roll Number: 5 Name:

Mangala Branch: Computer
Percentage: 89.230769

Precedence of the DOT operator

The dot operator is a member of the highest precedence group, and its associativity is from left to right.
Hence, the expression such as ++scvar.membern is equivalent to ++(stvar.membern), imply ing that the
unary operator will act only on a particular member of the structure and not the entire structure.

8.5 Structure Initialization

Similar to the standard data types, structure variables can be initialized at the point of their definition.
Consider the following structure declaration:
struct Student
{
int- roll_no; char
name[ 2 5 ] ; char
branch[151; int marks;
>;

A variable of the structure Student can be initialized during its definition as follows:
Student s i ={ 5 , "Mangala", "Computer", 2 9 0 } ; The initial values for the components of
the structure are placed in curly braces and separated by commas. The members of the variable si, roll_no,
name, branch, and marks are initialized to 5, "Mangala", "Computer", and 290 respectively (see Figure
8.7).
The program days. cpp illustrates the initialization of the members of a structure at the point of a
structure variable definition.
/ / days.cpp: structure members initialization a t the point of definition
•include <iostream.h>
/ / structure declaration
struct date
{
int day; int
month; int year;
);
void main( ) {
date dl = { 14, 4 , 1971 } ; date d2 =
{ 3 , 7 , 1996 } ;
cout << "Birth date: ■;
cout « dl.day «■-"« dl.month «■-•« dl.year; cout « endl «
"Today date: ";
cout « d2.day «"-"« d2.month «■-"« d2.year;
)
Run
Birth date: 14-4-1971
Today date: 3-7-1996

Copyrighted material
 242 Mastering C++

Student si = {5, "Mangala", "Computer", 290};

Figure 8.7: Structure members' Initialization during definition 8.6

Nesting of Structures
A member of a structure may itself be a structure. Such nesting enables building of very powerful data
structures. The Student structure can be enhanced to accomodate the date of birth of a student. The new
member birthday is a structure of type date by itself as shown below: struct date
(
int day;
int month; int
year;
instruct Student
{
int roll_no; char
name[251; struct date
birthday; char
branch[15]; int marks;
>;

The structure to be embedded must be declared before its use. Another way of declaring a nested
structure is to embed member structure declaration within the declaration of a new structure as
follows:
struct Student {
int roll_no; char
name[25]; struct date
{
int day;
int month;
int year; } birthday;
char branch[15]; int marks;
);

The embedded structure date is declared within the enclosing structure declaration. A variable
of type Student can be defined as follows:
Student si;
The year in which the student si was born can be accessed as follows:
si-birthday.year The following are the some of the valid operations on the variable si:
si.roll_no - 5; cin » si. roll_.no; si.
birthday, day = 2; si.birthday.month =
2; si.birthday.year = 1972;

name

Figure 8.8: Accessing members of nested structures

Copyrighted material
 Chapter 8: Structures and Unions 245

The dot operator accessing a member of the nested structure birthday using the statement
si.birthday.year = 1972; is shown in Figure 8.8. The program student2 . cpp illustrates the declaration,
definition, and processing of nested structure members.
A statement such as
si.date.day = 2 ; / / error
is invalid, because a member of the nested structure must be accessed using its variable name.

/ / Student2.cpp: processing of student data using structures #include <iostream.h> //' structure declaration struct
date
i
int day; int
month; int
year;
Instruct Student
{
int roll.no; char
name[2 5 J ;
struct date birthday; // structure within a structure
char branch[15]; int
marks;
);
void main () {
Student si; / / structure definition
cout << "Enter data for student..." << endl;
cout « "Roll Number ? ";
cin » sl.roll_no; / / accessing structure member
cout « "Name ? "; cin »
si.name;
cout « "Enter date of birth <day month year>: ";
cin >> si.birthday.day » si-birthday.month » si.birthday.year;
cout « "Branch ? ■;
cin >> si.branch;
cout « "Total Marks <max-325> ? "; cin »
si.marks;
cout « "Student Report" « endl;
cout << "-----------------------------" « endl;
/ / process student data
cout « "Roll Number: " « sl.roll_no « endl; cout << "Name: " «
si.name « endl; cout « "Birth day: •;
1
cout<<sl .birthday. day «"-■<< si .birthday .month<<" - "«sl .birthday .year ; cout << endl « "Branch: «
si.branch « endl; cout « "Percentage: " << si .marks*(100.0/325) « endl;
cin >> s[ij.name; cout
<< "Branch ? "; cin »
s[i].branch;
cout << "Total Marks <max-325> ? "; cin
» s [ i ] .marks;
>
cout « "Students Report" « endl;
cout « "---------------------------------" « endl;
// process student data
for( i = 0; i < n; i++ ) (
cout << "Roll Number: " << s[i].roll_.no « endl;
cout << "Name: " « s[i].name << endl;
cout << "Branch: " « s[i].branch « endl;
cout « "Percentage: ■ « s[ij.marks*(100.0/325) « endl;
)
}

Run
How many students to be processed <max-10>: _,
Enter data for student 1... Roll Number ? 5 Name ?
Manga la
Branch ? Computer
Total Marks <max-325> ? 290
Enter data for student 2 . . .
Roll Number ? %
Name ? Shivakumar
Branch ? Electronics

Copyrighted material
 Chapter 8: Structures and Unions 245

Total Marks <max-325> ? ___!

Students Report

Roll Number: 5 Name:

Mangala Branch:
Computer Percentage:
89.230769 Roll Number:
9 Name: Shivakumar
Branch: Electronics
Percentage: 76.923077

Initialization of Array of Structures

An array of structures can be initialized in the same way as a single structure and hence, the
discussion regarding the initialization of a single structure is still relevant. This is illustrated
by the following
example:
Student s[5] - {
2, "Tejaswi", "CS", 200,
3, "Laxmi H", "IT", 215,
5, "Bhavani", "Electronics", 250, 7 ,
"Anil", "Civil- 215,
i

i
Run
Students Report

Roll Number: 2 Name: Tejaswi

Branch: CS Percentage: 87.6923 Roll
Number: 3 Name: Laxmi Branch: IT
Percentage: 66.1538 Roll Number: 5
Name: Bhavani Branch: Electronics
Percentage: 76.9231 Roll Number: 7
Name: Anil Branch: Civil
Percentage: 66.1538 Roll Number: 9
Name: Savithri Branch: Electrical
Percentage: 89.2308
Operations Involving the Assignment Operator
The individual structure member can be used in an assignment statement just like any other ordinary variable. It is
illustrated in the following statements:
s f l j -marks = 290; / / marks set to 290
s[l].marks += 5; / / marks is incremented by 5
Notice that only the individual structure members are accessed, and not the entire structure. If the structure member is
itself a structure, then the embedded structure's member is accessed as follows: s[1].birthday.day
It accesses the member day of the structure variable birthday embedded in the 2 nd element of the array of structure
variable s. The assignment operator can also be used to copy variables of the same structure. For instance, the statement.
si = s2;
copies contents of s2 to si, which are variables of the student structure. It is performed by copying each member
transparently. Array of structure elements can also be copied as follows:
s[2] = s[l] ;
s[i] = s[j);
If a structure has members of type pointers, then only the address stored in that pointer member is copied and hence, such
members still point to that pointed to by the source variable. In such a situation, make sure that memory is allocated, and
explicitly copy the elements pointed to by the pointers. If this is not done, it might result in a dangling reference. It
happens when the destination variable releases memory and the source variable continue to exist. (Dangling reference: it
refers to a situation when a pointer to the memory item continues to exist, but memory allocated to that item is released.
Garbage memory'- it indicates that the memory item continues to exist but the pointer to it is lost; it happens when memory
is not released explicitly.) void main () {
union emp e; //or emp e; strcpy{e.name,
"Rajkumar"); show( e ) ; e.idno = 10; showt e ) ;
e.salary = 9000; show{ e ) ;
)

Run
Employee Details ...
The name is Rajkumar
The idno is 24914
The salary is 2.83348e*26
Employee Details . . .
The name is

Copyrighted material
 Chapter 8: Structures and Unions 245

The idno is 10
The salary is 2.82889e+26
Employee Details . . .
The name is
The idno is -24576
The salary is 9000
The status of the variable e after execuion of each one of the following: 1.s trcpy{e.name, ■Raj kumar");
2. e.idno = 10; and
3. e. salary = 9000;
is shown in Figure 8.14a, 8.14b, 8.14c respectively. Note that, access of non active members will lead to meaningless
values.

(a) strcpy(e.name, "Rajkumar");

union emp e;(b) e.idno=10; (c) e.salary=9000;

Figure 8.14: Union variable initialization
9-4 Pointer Variables
Pointers are also variables and hence, they must be defined in a program like any other variable. Rules for variable names
and declaring pointers are the same as for any other data type. This naturally gives rise to questions about the data type of
a pointer, size of memory allocated to a pointer and the format for defining different types of pointers.

Pointer Definition
When a pointer variable is defined, the C++ compiler needs to know the type of variable the pointer points to. The syntax
of pointer variable definition is shown in Figure 9.2.

standard or user defined data type: asterisk followed by a pointer variable name
char, short, int, float, etc.

M DataType * PtrVar,

Figure 9.2: Syntax of pointer definition

DataType could be a primitive data type or user defined structure (such as structures and classes). The PtrVar could
be any valid C++ variable name. The character star (*) following the DataType informs the compiler that the variable
PtrVar is a pointer variable. The pointer so created can hold the address of any variable of the specified type. Some
typical pointer definitions are:
int *int_ptr; // int_ptr is a pointer to an integer char *ch_ptr; // ch_ptr is a pointer to a character
Date *d_ptr; // d_ptr is a pointer to user defined data type The pointer variable must be bound to a memory
location. It can be achieved cither by assigning the address of a variable, or by assigning the address of the memory
allocated dynamically. The address of a variable can be assigned to a pointer variable as follows:
int_ptr = &marks; where the variable marks is
of type integer.
Pointer to characters (a string) can be defined as follows: char *msg;
It can be initialized at the point of definition as follows:
char 'msg = " abed . . xyz" ;
Or, it can also be initialized during execution as follows: msg =
"abed..xyz";

Dereferencing of Pointers
Dereferencing is the process of accessing and manipulating data stored in the memory location pointed to by a pointer.
The operator * (asterisk) is used to dereference pointers in addition to creating them. A oointer variable is dereferenced
when the unary operator * (in this case, it is called as the indirection opi rator) is prefixed to the pointer variable or pointer

Copyrighted material
 Chapter 8: Structures and Unions 245

expression. Any operation that is performed on th; dereferenced pointer directly affects the value of the variable it points
to. The syntax for dereferencing pointers is shown in Figure 9.3.
means of pointers. The power of pointers becomes evident in situations, where indirect access is the
only way to access variables in memory. Figure 9.4 gives a pictorial representation of accessing a
variable using a pointer.

Pointers and Parameter Passing

Pointers provide a two way communication between the service requester and service provider. It is
achieved by passing the address of the actual parameters instead of their contents. Any modification
done to formal variables in the function will be automatically reflected in the actual parameters when
they are passed by address. A program to swap two numbers is listed in swap. cpp.
// swap.cpp: swap
2 numbers using pointers
#include <iostream.h>
void swap(float *, float *);
void main ()
{
float a, b;
cout « "Enter real number <a>: cin » a;
cout « "Enter real number <b>: "; cin » b;
// Pass address of the variables whose values are to be swapped swap(&a, &b) ; // figure 9.6a
cout << "After swapping............\n';
cout « "a contains " << a << endl; cout « "b
contains " « b;
)
void swapt float *pa, float *pb ) // function to swap two numbers
<
float temp;
temp = *pa; // figure 9.6b *pa = *pb; //
figure 9.6c *pb = temp; // figure
9.6d
)

Run
Enter real number <a>: 10.5 Enter real
number <b>: 20.9
After swapping .........
a contains 20.9 b
contains 10.5

In main ( ) , the statement

swap(&a, &b) ;
assigns addresses of the actual parameters to the formal parameters, which are of type pointers. How-
ever, they are manipulated differently (see Figure 9.5). Inmain ( ) , the parameters are accessed directly
with their names whereas in swap { ) , they are accessed using the indirection operator.

Copyrighted material
 Chapter 1: Object-Oriented Paradigm 13

Object Name Object Name

Attribute 1
Attribute 2

Attribute N

Attribute 1 Attribute 2

Attribute N

Operation 1 Operation 2

Operation N

Copyrighted material
Ope ation 1

Operation 2

Ope ration
N

(a) (b) (c|

Copyrighted material
 AccountNumber AccountType
Name Balance

Deposit() Withdraw*) Enquire()

AccountNumber AccountType
Name Balance

Deposit()
Withdraw*) Enquire*)

(a) (b) (c)

Figure 1.11: Different styles of representing the account object 1.7 Classes
The objects with the same data structure (attributes) and behavior (operations) are grouped into a class.
All those objects possessing similar properties are grouped into the same unit. The concept of class-ing
the real world objects is demonstrated in Figure 1.12. It consists oflhcPerson class. Vehicle class, and
Polygon class. In the case of Person class, all objects have similar attributes like Name. Age. Sex -md
similar operations like Speak, Listen, Walk. So boy and girl objects are grouped into the Person class.
Similarly, other related objects such as triangle, hexagon, and so on. are grouped into the Polygon class.
Consider the following statements:
float 'sum; char
*name; A statement
such as
sum++; or ++sum;
advances the pointer variable sum to point to the next element. If the pointer variable sum holds the
address 1000, on execution of the above statement, the variable sum will hold the address (1000+4) =
1004 since the size of float is 4 bytes. Similarly, when a statement such as name++; or ++name;
is executed, and if the pointer variable name points to address 2000 earlier, then it will hold the address
(2000*1), since the size of char is one byte. This concept applies to all"arithmetic operations performed
on pointer variables.
When a pointer variable is incremented, its value actually gets incremented by the size of the type to
which it points. For example, let pi be a pointer to an integer defined with the statement
int* pi;
Also, let pi point to the memory location 1020. i.e., the number 1020 is stored in the pointerpi. Now, a
statement which increments pi, such as
pi++;
will add two topi, making it 1022 (assuming that the size of an integer is 2 bytes). This makes pi point
to the next integer. Similarly, the statement
pi--;
will decrement the value of pi by 2. The pointer arithmetic on different types is shown in Table 9.2.

Table 9.2: Pointer arithmetic

ecomes significant for accessing and processing array elements efficiently (a more detailed account of array processing with pointers is taken up later i
int *
data; int
size;
set_new_handler{ 0 );

Copyrighted material
 cout << "How many bytes to allocate: "; cin »
size;
iff (data = new int[ size ]) )
cout « "Memory allocation success, address = " « data; else
{
cout << "Could not allocate. Bye ...'; exit(1);
;.
delete data;
J
Run1
How many bytes to allocate: 100
Memory allocation success, address = 0xl6be
Hun2
How many bytes to allocate: 30000
Could not allocate. Bye . ..

Note: A request for allocation of 0 bytes returns a non-null pointer. Repeated requests for zero-size
allocations return distinct, non-null pointers.

9.8 Pointers to Pointers

C++ allows programmers to define a pointer to pointers, which offers flexibility in handling arrays,
passing pointer variables to functions, etc. The syntax for defining a pointer to pointer is:
DataType **PtrToPtr; which uses two * symbols (placed one beside the other). It implies
that PtrToPtr is a pointer-to-a-pointcr addressing a data object of type DataType. This feature is often
used for representing two dimensional arrays. The program pt r2pt r. cpp illustrates the format for
defining and using a pointer to another pointer.
// ptr2ptr.cpp: definition and use of pointers to pointers •include
<iostream.h> void main(void) {
int *iptr; // iptr as a pointer to an integer, figure 9.8a
int "*ptriptr; // Defines pointer to int pointer, figure 9.8b
int data; // Some integer location, figure 9.8c
iptr = &data; // iptr now points to data, figure 9.8d
ptriptr = iiptr; // ptriptr points to iptr, figure 9.8e
"iptr = 100; // Same as data = 100, figure 9.8f
cout << "The variable 'data' contains " « data << endl;
•"ptriptr = 200; // Same as data = 200, figure 9.8g
cout << "The variable 'data' contains " << data << endl;
data = 300; // figure 9.8h
Passing Address of a Pointer
When a pointer variable is defined, a memory location for the pointer is allocated, but it will not be
initialised. Before using the pointer variable, it should be initialized. If the pointer variable has to be
initialized in a function other than where it is defined, then the pointer's address has to be passed to the
function. The contents of the pointer to a pointer variable can be used to access or modify the pointer
type formal variable. The program big. cpp illustrates passing the address of a pointer, so that the
pointer can be made to point to a desired variable (in this program, it is the biggest of two integers).
big.Cpp: program Co find Che biggesc number using pointers "include <iosCream.h>
void rindBigfint "pa. int "pb, int "pbig)

■ ■ compare the contents of *pa and *pb and assign their address to pbig ift *pa >
'pb )
•pbi
g = pa;
else
"pb i g = pb;
}
void main()

int a. b. 'big;
cout << "Enter two integers: ";
cin >> a >> b;
FindBig( &a, &b, tbig );
cout << "The value as obtained from the pointer: " << *big;
)

Run
Enter two integers: 10. £0.
The value as obtained from the pointer: 20 In
main (). the statement
FindBigt &a, &b, fcbig ); passes a, b, and big variables by address. It assigns the address of the
variable a or b to the pointer variable big. In FindBig (). the statement •pbig = pa;
effectively stores the address of the variable a in the pointer variable big, which is defined in the main()
function.

Copyrighted material
9.9 Array of Pointers
An array of pointers is similar to an array of any predefined data type. As a pointer variable always
contains an address, an array of pointers is a collection of addresses. These can be addresses of ordinary
isolated variables or addresses of array elements. The elements of an array of pointers are stored in the
memory just like the elements of any other kind of array. All rules that apply to other arrays also apply
to the array of pointers.
The syntax for defining an array of pointers is the same as array definition, except that the array name is
preceded by the star symbol during definition as follows: DataType *ArrayName[ARRSIZE\;

Copyrighted material
 Salient Features:
□ Clearty explains the language constructs using syntax,
lllustratlons,code segments^nd simple examples.

□ Important concepts such as classes and objects, object

Initialization and cleanup, dynamic objects, polymorphism
-operator overloading and virtual fun net ions, Inheritance etc are
covered exhaustively with well - designed programming
examples.

□ Details topics like templates and exception handling.

□ A chapter exclusively devoted to object-oriented analysis,
design, and development Explains the OOP concepts and their
Implementation In relevant chapters too.

Related Books of Interest from Tata McGraw-Hill

□ Venugopal: Programming with PASCAL and C
□ Venugopal: Programming with FORTRAN
□ Venugopal Be Prasad: Programming with C (THB Outline
Series)
□ venugopal & Maya: Programming with Pascal (TMH Outline
Series)

□ Balagurusamy: Object - Oriented Programming with C++

□ Ravtchandran: Programming with C++
The McGraw-Hill Compmm

I
JH Tata McGraw-Hill
Publishing Company Limited
7 West Patel Nagar, New Delhi 110008
 Chapter 9: Pointers and Runtime Binding 287

cout « "Unsorted list: "; fori

i = 0 ; i < n ; i + + )
cout << endl << person t i l ;
SortByPtrExchangeI person, n ) ; cout <<
endl « "Sorted list: "; fori i = 0; i < n; i +
+)
cout << endl « person[ i ] ; / /
release memory allocated fort i = 0; i <
n; i++ )
delete person[i];
)

Run
Enter Name: TeiflSWi Enter
another (y/n) ? y_ Enter Name:
Prasad Enter another (y/n) ? y.
Enter Name: PrakflSh Enter
another (y/n) ? y_ Enter Name:
Sudeep Enter another (y/n) ? y
Enter Name: Anand Enter
another ( y / n ) ? n Unsorted
list: Tejaswi Prasad Prakash
Sudeep Anand
Sorted list:
Anand
Prakash
Prasad
Sudeep
Tejaswi
In main ( ) , the statement
person[n] = new char(40); allocates 40 bytes of memory to the (n+ l)lh element and stores its
memory address in the array of pointers to strings indexed by n. The statement
SortByPtrExchange( person, n ) ; invokes the sorting function by passing the array of pointers
and data count as actual parameters. Note that, array is passed to a function just by mentioning its name.
This is equivalent to passing an entire array; the address of the first clement of an array can be used to
access any element in the array by using offset values. The data sorted by SortByPtrExchange ( ) do not
change their physical location (see Figure 9.9). The effect of sorting is seen when strings are accessed
using pointers in a sequence.

Copyrighted material
 288 Mastering C++

char "person[100]

1000 1002 1004 1540 >1500 ♦ 1540 >1580 Prasad

Anand
1006 1008 1560 person(0 person[1
+-1620 -►1660 Prakash
1500 1620 Sudeep person[2 person[3
Teiaswi
1660 person[4

unsorted pointers

char -person[100]

lejasw person[0] person(1] person[2] person[3] person[4]

Prasad
Prakash sorted pointers
Sudeep
Anand

Figure 9.9: Sorting using pointers

Precedence of * and [ ] Operators

In C++, the notations *p [ 3 1 and (*p) [ 3 ] arc different since * operator has a lower precedence than [ 1
operator. The following examples illustrate the difference between these two notations:
1. int *data[10] ;
It defines an array of 10 pointers. The increment operation such as data++; or
+*data; is invalid; the array variable data is a constant pointer.
2. int <*data> [10] ;
It defines a pointer to an array of 10 elements. The increment operation such as data++; or ++data;
is invalid; the variable data will point beyond 10 integers, i.e., 10 *sizeof (int) will be added to the variable
data. The program show, cpp illustrates the use of defining a pointer to a matrix having arbitrary number of
rows and fixed number of columns.
show.cpp: matrix of unknown number of rows and known number of columns ♦include <iostream.h> void
show{ int a[J [31 , int m ) {
int (*c)[3]; // pointer to an array of 3 elements c = a;
for( int i = 0; i < m; i++ j {

Copyrighted material
290 Mastering C++

sizeof (int) = 2 bytes sizeof (int*) = 2 bytes,

near pointer = 4 bytes, far pointer

Figure 9.10: Model of dynamic matrix

/ / matrix.cpp: matrix
manipulation and dynamically resource al
♦include <iostream.h>
♦include <process.h>
int **MatAlloc( int row, int col )
{
int **p;
p = new int *( row ];
fort int i = 0; i < row; i++ )
p[i] = new int[ col J; return p;
}
void MatRelease( int **p, int row ) {
for ( int i.= 0; i < row; i++ )
delete p(i); delete p;
}
void MatRead(int **a, int row, int col ) {
int i, j ;
for( i = 0; i < row; i++ ) fort j = 0: j < col;
j++ )
(
cout « "Matrix[" « i << ■,■ « j << "] = ? ";
cin >> a[i][j];
}
)
// multiplication of matrices, c3.mul(cl, c2) : c3 = cl*c2
void MatMul ( int **a, int m, int n, int **b, int p, int q, int **c )
I
i
n
t

i
,

j
,

;
290 Mastering C++

i
f
<

!
=

(
cout << "Error: Invalid matrix order for multiplication"; exitt 1 );
}
fort i =
0; i
< m;
i++ J
fort j
= 0;
j<
q; j+
+){
c[i] [j] = 0;
fort k = 0; k < n; k++ )
c[ijljj += a[ij(k] * blfclEj];
}
}
void MatShowt int **a,
int row, int col ) (
int i, j;
fort
i=
0; i
<
row;
i++ )
{
cout
<<
endl;
fort j
= 0; j
<
col;
j++ )
cout
« a[i]
[j] «
"
)
)
v
o
i
d

m
a
i
n
290 Mastering C++

(
)

{
i
n
t

*
*
a
,

*
*
b
,

»
*
c
;

i
n
t

m
,

n
,

p
,

q
;
cout << "Enter Matrix A
details. . . " << endl; cout «
"How many rows ? ";
cin » m;
cout « "How
many
columns ? ";
cin » n;
a
»
M
a
t
A
l
l
o
c
t
m
,
n

)
;
M
a
t
R
e
290 Mastering C++

a
d
t
a
,
m
,
n

)
;
cout << "Enter Matrix B
details.. . " « endl; cout <<
"How many rows ? "; cin » p;
cout «
"How
many
columns
? cin » q;
b = MatAlloct
p, q ) ;
MatReadt b,
p, q ) ; c =
MatAlloct m,
q );
MatMul( a,
m, n, b, p, q, c
);
void my_strcat( char *s2, char *sl )
{
// move end of
string while*
"s2 != '\0' )
s2++; //
append si to s2
whilet *sl !=
"\0* )
*s2 + + = *sl++; *s2 =
'\0'; // copy end of string
)
int my_strcmp( char *sl, char *s2 )
(
// compare as long as they are equal
whilet *sl == *s2 && (*sl != NULL || *s2 != NULL) )
(
S
l
*
*
:

S
2

+
;
}
return *sl - 's2;
>

Run
Enter stringl:
Object Enter
string2: Oriented
Length of stringl: 6
Strings' on concatenation: ObjectOriented
String comparison using ...
Library function: -16
User's function: -16

9.13 Environment Specific Issues

290 Mastering C++

Pointer variables, like other variables are also allocated memory whenever they are defined. The
size of the memory allocated (in bytes) to a pointer variable depends on whether the pointer just
holds the offset part of the address, or both the segment and offset values. The memory model in
wnich the program is compiled also influences the size of the pointer variables used in that
program. C++ compilers (such as Borland or Microsoft C++) running under DOS environment
support six different memory models, each of which determines the amount of memory allocated
to the program's data and code (see Table 9.3).
Normally, all pointers defined in a program in the small model contain only the offset part of
the address. Such pointers are known as near pointers, for which two bytes of memory are
allocated. The use of near pointers limits the programmer to access only those memory locations,
which lie within a single segment only. (The maximum size of a segment is 64 KB). This
limitation can be overcome by the use of pointers, which arc capable of holding both the segment
as well as the offset part of an address. Such pointers arc called far pointers, for which four bytes
of memory is allocated. It is possible to access any memory location, using far pointers. The far
pointers can be defined (even in a small memory model) by using the keyword far as follows:
int far "ifarptr; // defines a far pointer to int
This declaration is similar to the structure declaration in C. It enables the creation of the class
variables called objects. For example, the following statements,
account savings_account;
account current_account;
account FD_account; create instances of the class account. They define savings_account,
current_account. and FD_account as the objects of the class account. From this, it can be inferred
that, the account class groups objects such as saving account, current account, etc. Thus, objects
having the same structural and behavioral properties are grouped together to form a class.
Each class describes a possibly infinite set of individual objects; each object is said to be an
instance of its class and each instance of the class has its own value for each attribute but shares
the attribute name and operations with other instances of the class. The following points on
classes can be noted:
♦ A class is a template that unites data and operations.
♦ A class is an abstraction of the real world entities with similar properties.
♦ A class identifies a set of similar objects.
♦ Ideally, the class is an implementation of abstract data type.

1.8 Multiple Views of the Same Object

A commonly accepted notion about objects is illustrated through the definition of a tree. In this
classical model, a tree is defined as a class, in terms of internal state information and methods
that can be applied. The designer of such an object-oriented tree, ideally works with the intrinsic
properties and behavior of the tree. In the real world, properties of a tree like its height, cell-
count, density, leaf-mass, etc., are intrinsic properties. Intrinsic behavior includes like growth,
photosynthesis, etc., that affect the intrinsic properties. This idea of a classical model is
inadequate to deal with the construction of large and growing suites of applications that
manipulate the objects. Every observer (for instance, a tax-assessor, a woods man and a bird) of
the tree, with different backgrounds, has his own view on the ideal model of a tree as shown in
Figure 1.13.
A tax-assessor has his own view of the features and behavior associated with a tree. The
characteristics include its contribution to the assessed value of the property on which it grows.
The behavior includes the methods, by which this contribution is derived. These methods vary
from tree-type to tree-type. In fact, such methods may form a pan of a tax assessor's view of all
objects, tree and non-tree alike. These characteristics and behaviors are extrinsic to the tree. They
form the part of a tax-assessor's subjective view of the object-oriented iree.
Figure 1.13 reminds that the tax-assessor is merely one of a suite (type) of applications, each
having its own subjective view, its own extrinsic state and behavior for the tree. The views of a
woodsman and a bird on the same object.arc also different compared to the tax-assessor's view.
A woodsman's view of the tree, is in terms of sales price and time required to cut the tree with
capital profit as a method. A bird's view of the same tree is different and its view characteristics
include FoodValue and ComputeFlight {). Thus, a woodsman views the tree in terms of the
amount of time required to cut the tree and the price it would fetch. The bird views it in terms of
the food value and the amour.t of energy required to carry the food from the tree to its nest.

Struct
first
LIST '
data i

I-------1
: "next Figure 9.14: Linked list with self-referential structures
l
290 Mastering C++

The program l i s t , cpp illustrates the manipulation of a linked list. It supports

create, delete, and display operations on the linked list. The structure LIST is a
self referential structure, since it has a pointer to the next node as one of the data
items

/ / Jist.cpp: self referential structure-linked l i s t

tinclude <iostream.h>
♦include <new.h>
♦include <process.h>
♦define SUCC( node ) node->next
struct LIST
{
int data; / / node data
LIST *next; / / pointer to next node
};
/ / creates node using data and returns pointer to first node of the list
LIST * InsertNodet int data, LIST * first )
{
LIST "newnode;
newnode = new LIST; / / allocate memory for node
i f ( newnode = = NULL )
(
cout << "Error: Out-
of-memory" << endl;
exitf 1 ) ;
>
newnode->data = data; / / Initialize list data member
SUCC{ newnode ) - first; / / new node becomes f i r s t node return
newnode;
}
/ / deletes node whose data matches input data and returns updated l i s t
LIST * DeleteNode( int data, LIST * f i r s t )
(
LIST *current, *pred; / / work space for insertion
i f ( ! first )
(
cout «
"Empty
list" < <
endl;
return
first;
J
for{pred=current=first;current; pred=current,current = SUCC( current ) }
i f ( current->data == data ) {
/ / node found, release this node
i f ( current == first ) / / i f node to be deleted is first
node
first = SUCCt current ) ; / / then update list pointer
else
SUCC( pred ) = SUCC{ current ) ; / / bypass the node / / release
delete current; return allocated memory
first;
•

return( first ) ;

' Display list void

DisplayList( LIST *first )
{
LIST 'list;
fori list = first; list;, list ^ SUCC { list ) )
cout << "->" << list-
>data; cout << endl;
;
v
o
i
290 Mastering C++

m
a
i
n

(
)

{
LIST *list = NULL; / / list is
empty int choice, data;
set_new_handler t 0 ) ; / / makes new to return to NULL i f it fails cout <<
"Linked-list manipulations program...\n";
w
h
i
l
e
(
l
)

(
cout << "List operation, 1- Insert, 2-Display, 3-Delete, 4-Quit: "; cin >>
choice; switch( choice ) (
case 1:
cout << "Enter data for node to be created: * ; cin >> data;
list = InsertNode( data, list ),-
b
r
e
a
k
;

c
a
s
e

2
:
cout << "List contents: ";
DisplayList( list ) ;
b
r
e
a
k
;

c
a
s
e

3
:
cout << "Enter data for node to be delete: "; cin » data;
list = DeleteNode( data, list ) ;
break; case 4:
cout « "End of Linked List Computation ! ! . \ n " ; return; default:
290 Mastering C++

cout « "Bad Option

Selected\n"; break;
)
}
}

Another important point to be noted is that, avoid storing the address of a variable or an
object into a pointer in the inner block, and using the same in the outer block. The program wild4
. cpp illustrates the wild pointer accessing garbage location.

wild4.cpp: out of scope of a block variable access

#include <iostream.h> #include <string.h> void main{)
i
c
h
a
r

"
p
i
;

C
char name I] =
"Savithri "; pi = name;
J
// do some processing here
cout « "Name = " << pi << endl;
}
Run
Name = SavithQS!
In main f), the statement
cout « "Name = " « pi << endl; accesses the data pointed to by the pointer variable pi.
The variable pi is assigned to point to the variable name defined within an inner block. When the
execution of this block is completed, all the variables are destroyed and hence, accessing of data
stored in the variable name becomes invalid data. In some situation, the programs might execute
properly, but they may corrupt other program's data and lead to system crash.
The above discussion also holds good for pointer to objects. Like variables, whenever objects
goes out of scope, they are destroyed. Referencing such objects is like accessing invalid-data
variable and hence, such reference should be avoided.

Review Questions
9.1 What are pointers ? What are the advantages of using pointers in programming ? Explain
addressing mode required to access memory locations using pointers.
9.2 Under what situations, the use of pointers is indispensable ?
9.3 Write a program to print address of the variables defined by the following statement:
int a, b =
10; float
c = 2, d;
9.4 Explain the syntax for defining pointer variables. How different arc these from normal
variables?
9.4 What is dereferencing of pointers ? Write a program to dereference the pointer variables
in the following statements (print value pointed to by pointer variables):
int *a; double
*b; a = &i; b
= &f;
9.6 What are the differences between passing parameters by value and by pointers ? Give
examples.
Classes are the basic language construct of C++ for creating the user defined data types. They
arc syntactically an extension of structures. The difference is that, all the members of structures
are public by default whereas, the members of classes are private by default. Class follows the
principle that the information about a module should be private to the module unless it is specifically
declared public.

10.2 Class Specif ication

C++ provides support for defining classes, which is a significant feature that makes C++ an
object oriented language. In C terms, a class is the natural evolution of a structure. Classes
290 Mastering C++

contain not only data but also functions. The functions are called member functions and define
the set of operations that can be performed on the data members of a class. Thus, a class can be
described as a collection of data members along with member functions. This property of C++,
which allows association of data and functions into a single unit is called encapsulation.
Sometimes, classes may not contain any data members or member functions (and such classes
are called as empty classes). The syntax of a class specification is shown in Figure 10.2.

Na
me
of
the
user
defi
ned
clas
s
Keyword

class
ClassNa
me {
// body of a class

\/ Semicolon required here

Figure 10.2: Syntax of class specification

The class specifies the type and scope of its members. The keyword class indicates that the
name which follows (ClassName) is an abstract data type. The body of a class is enclosed within
the curly braces followed by a semicolon—the end of a class specification. The body of a class
contains declaration of variables and functions, collectively known as members. The variables
declared inside a class are known asdata members, and functions are known as member functions.
These members are usually grouped under two sections, private and public, which define the
visibility of members.
The private members are accessible only to their own class's members. On the other hand,
public members arc not only accessible to their own members, but also from outside the class.
The members in the beginning of class without any access specifier are private by default.
Hence, the first use of the keyword private in a class is optional. A class which is totally private is
hidden from the external world and will not serve any useful purpose.
The following declaration illustrates the specification of a class called s tudent having rol l_no
and name as its data members: class student {
int roll_no; // roll number
char name! 20 J; // name of a student
public:
 Chapter 10: Classes and Objects 315

void setdata( int roll_no_in, char *name_in ) {

roll_no = roll_no_in; strcpy( name,
name_in );
}
void outdata(> {
cout << "Roll No = " « roll_no « endl; cout « "Name = 1 « name « endl;
)
);
A class should be given some meaningful name, (for instance, student) reflecting the information it holds.
The class name s t uden t becomes a new data type identifier, which satisfies the properties of abstraction', it
can be used to define instances of class data type. The class student contains two data members and two
member functions. The data members are private by default while both the member functions are public as
specified. The member function setdata () can be used to assign values to the date members roll_no and name.
The member function outdata () can be used for displaying the value of data members. The data members of
the class student cannot be accessed by any other function except member functions of the student class. It is a
general practice to declare data members as private and member functions as public. Three different notations
for representation of the student class is shown in Figure 10.3.

int rollno

char name[20]

setdata()

outdata()

Figure 10.3: Different representations of the class student

(a) (b) (c)

The name of data and member functions of a class can be the same as those in other classes; the members of
different classes do not conflict with each other. Essentially, a class identifies all the data members associated
with its declaration. The following example illustrates this concept: class Person {
private:
char name[20]7
int age;
The data member name appears in the student class and in the Person class declarations, but their scope is
limited to their respective classes. However, more than one class with the same class-name in a program is an error,
whether the declarations are identical or not. A class can have multiple member functions (but not data members)
with the same name as long as they differ in terms of signature ; this feature is known as method overloading.
Like structures, the data members of the class cannot be initialized during their declaration, but they can be
initialized by its member functions as follows:
class GeoObject {

float x, y = 5; // Error: data members cannot be initialized here void SetOriginO // set point
to origin
{
x = y = 0.0;
)
};
The data members x ory of the class GeoObject cannot be initialized at the point of their declaration, but, they
can be initialized in member functions as indicated in the SetOrigin () member function.

10.3 Class Objects

A class specification only declares the structure of objects and it must be instantiated in order to make use of
the services provided by it. This process of creating objects (variables) of the class is called class instantiation. It
is the definition of an object that actually creates objects in the program by setting aside memory space for its
storage. Hence, a class is like ^blueprint of a house and it indicates how the data and functions are used when the
class is instantiated. The necessary resources are allocated only when a class is instantiated. The syntax for
defining objects of a class is shown in Figure 10.4. Note that the keyword class is optional.

Name of the user defined class

Copyrighted material
 Chapter 10: Classes and Objects 315

Name of the user-defined object

1/
class ClassName ObjectName, ..-;

An example of class instantiation for creating objects is shown below:

class student si; or
student si;
It creates the object si of the class student. More than one object can be created with a single statement as
follows:
class student si, s2, s3, s4; or
student si, s2, s3, s4; Ii creates multiple objects of
the class student.

Copyrighted material
 C
h
a
p
t
e
r
1
:
O
b
j
e
c
t-
O
ri
e
n
t
e
d
P
a
r
a
d
i
g
m
17

Data abstraction is supported by several other modern programming languages such as Smalltalk.
Ada. etc. In these languages, and in C++ as well, a programmer can define a new abstract data type by
specifying a data structure, together with the operations permissible on that data structure as shown in
Figure 1.14. The important feature o*"C++. the class declaration, allows encapsulation and creation ol
abstract data types.
The use of encapsulation in protecting the members (data and code) of a class from unauthorized
access is a good programming practice; it enforces the separation between the specification and imj.
lementation of abstract data typos, and it enables the debugging of programs easily.

1.10 Inheritance
Inheritance is the process, by which one object can acquire the properties of another. It allows the
declaration and implementation of one class to be based on an existing class. Inheritance is the most
promising concept of OOP, which helps realize the goal of constructing software systems from reusable
parts, rather than hand coding every system from scratch. Inheritance not only supports reuse across
systems, but also directly facilitates extensibility within a given system. Inheritance coupled with
polymorphism and dynamic binding, minimizes the amount of existing code to be modified while
enhancing a system.
To understand inheritance, consider the simple example shown in Figure 1.15. When the class Child,
inherits the class Parent, the class Child is referred to as derived class (sub-class), and the class Parent as a
base class (super-class). In this case, the class Child has two parts, a derived part and an incremental part.
The derived part is inherited from the class Parent. The incremental part is the new code written
specifically for the class Child. In general, a feature of Parent may be renamed, re-implementcd.
duplicated, voided (nullified), have its visibility status changed or subjected to almost any other kind of
transformation when it is mapped from Parent to Child. The inheritance relation is often called the is-a
relation. This is because when the class Child inherits the base class Parent, it acquires all the properties
of the Parent class. It can also have its own properties, in addition to those acquired from its Parent. This
is an example of single inheritance; the child class has inherited properties from only one base class.

Parent Base or Super Class

Child Derived or Sub Class

Figure 1.15: Single inheritance

The inheritance relation is often used to reflect the elements present in an application domain. For
example, consider a rectangle which is a special kind of polygon as shown in Figure 1.16. This relation-
ship is easily captured by the inheritance relation When the rectangle is inherited from the polygon, it

Copyrighted material
 In mainO, the statements
student si; // first object/variable of class student student s2; // second
object/variable of class student create two objects called si and s2 of the student class. The
statements
si.setdata( 1, "Tejaswi" ); //object si calls member function setdata
s2.setdata( 10,"Rajkumar" ); //object s2 calls member function setdata
initialize the data members of the objects si and s2. The object si's data member roll_no is assigned 1
and name is assigned Tejaswi. Similarly, the object s2's data member roll_no is assigned 10 and
name is assigned Rajkumar.

Instance of the class student

Figure 10.6: Student object and member access

The statements
si.outdata(); // object si calls member function outdata
s2.outdata(); // object s2 calls member function outdata
call their member outdata () to display the contents of data members namely, rol l_no and name of
student objects si and s2 in succession. Thus, the two objects si and s2 of the class student have different
data values as shown in Figure 10.7.

Client-Server Model
In conventional programming languages, a function is invoked on a piece of data (function-driven
communication), whereas in an OOPL (object-oriented programming language), a message is sent to an
object (message-driven communication) i.e., conventional programming is based on function abstraction
whercrs, object oriented programming is based on data abstraction.
The object accessing its class members resembles a client-server model. A client seeks a service
whereas, a server provides services requested by a client. In the above example, the class student
Similar to the real world objects. OO objects also have a life cycle. They can be created and de-
stroyed automatically whenever necessary. Communication between the objects can take place as long as
they are alive (active). Communication among the objects takes place in the same way as people pass
messages to one another. The concept of programming with message passing model is an efficient way of
modeling real-world problems on computers.
student s2;

Client

Information
Message

s2.setdata ( 10, "Rajkumar'

Object

Figure 10.8: Client-Server model for message communication

Copyrighted material
 A message for an object is interpreted as a request for execution of a procciurc. The subroutine or
function is invoked soon after receiving the message and the desired results are generated within an
object. It comprises the name of an object, the name of a function, and the information to be sent to an
object.

10.5 Defining Member Functions

The data members of a class must be declared within the body of the class, whereas the member
functions of the class can be defined in any one of the following ways:
♦ Inside the class specification
* Outside the class specification
The syntax of a member function definition changes depending on whether it is defined inside or
outside the class specification. However, irrespective of the location of their definition (inside or outside
a class), the member function must perform the same operation. Therefore, the code inside the function
body would be identical in both the cases. The compiler treats these two types of function definitions in
a different manner.
Member Functions Inside the Class Body
The syntax for specifying a member function declaration is similar to a normal function definition except
that it is enclosed within the body of a class and is shown in Figure 10.9. All the member functions
defined within the body of a class are treated as inline by default except those members having looping
statements such as for, while, etc.. and it also depends on compilers.
class ClassName
{
private: int
age;
int SetAge<int agein) ----------------- Member function
{
age = agein; // body of the function
)

public:
in
t
b;
void myfunco--------------- Member function
(

// body of a function
)
};

Figure 10.9: Member function defined within a class

The program datel. cpp demonstrating the definition of member functions with the class specification
of the date class. It has private data members day, month, year and inline member functions, set () which
initializes data members and show {), which displays the value stored in the data members.
// datel.cpp: date class with member functions defined inside a class ♦include <iostream.h> class date {
private:
int day;
int month;
int
year;
public:
void set( int Dayln, int Monthln, int Yearln ) {
day = Dayln;
month = Monthln;
year = Yearln;
)
void
show() {
cout << day << "-" << month << "-" << year << endl;
}
J i inform the compiler to treat the member functions set and show as inline functions. The method
of invoking inline member functions is the same as those of the normal functions. In main (J, the
statements
d l . s e t t 2 6 , 3 , 1958 )
; d2.show( ) ;
will be replaced by the function itself since the function is an inline function. Note that, the inline
qualifier is tagged to the inline member function at the point of its definition.
The feature of inline member functions is useful only when they are short. Declaring a function
having many statements as inline is not advisable, since it will make the object code of a program
very large. However, some C++ compilers judge (determine) whether a given function can be
appropriately sized to inline expanded. If the function is too large to be expanded, it will not be
treated as inline. In this case, declaring a function inline will not guarantee that the compiler will
consider it as an inline function.

Copyrighted material
When to Use inline Functions
The following are simple thumb rules in deciding as to when inline functions should be used:

♦ In general, inline functions should not be used.

♦ Defining inline functions can be considered once a fully developed and tested program runs too
slowly and shows bottlenecks in certain functions. A profiler (which runs the program and deter-
mines where most of the execution time is spent) can be used in deciding such an optimization.
♦ Inline functions can be used when member functions consist of one very simple statement such
as the return statement in date : :getday { ) . which can be implemented as follows:
inline int date::getday{ ) / / definition (
return day;
)
♦ It is only useful to implement an inline function if the time spent during a function call is more
compared to the function body execution time. An example, where an inline function has no
effect at all is the following:
inline void date::show{) / / definition {
cout « day « "-" « month < < ■-■ « year < < endl;
)
The above function, which is presumed to be a member of the class date for the sake of argument,
contains only one statement: but takes relatively a long time to execute. In general, functions
which perform input and output operation spend a considerable amount of time. The effect of
conversion of the function show ( ) to inline would lead to reduction in execution time.
Inline functions have one disadvantage: the actual code is inserted by the compiler and
therefore it should be known at compile-time. Hence, an inline function cannot be located in a run-
time library. Practically, an inl ine function is placed near the declaration of a class, usually in the
same header file. It results in a header file having the declaration of a class with its
implementation visible to the user.
Number:
24 3 Cost:
2354.75

In the above program, the data fields ModelNum. PartNum, and cose of the class part cannot be
accessed by direct references using pi.ModelNum, pi.PartNum, and pi.cost respectively. When a class is
used, its declaration must be available. Thus, a user of the class is presented with a description of the
class. The internal details of the class, which are not essential to the user arc not presented to him. This is
the Concept of information hiding or data encapsulation. As far as the user is concerned, the knowledge of
accessible data and member functions of a class is enough. These interfaces, usually called the user
interface methods, specify their abstracted functionality. Thus, to the user, a class is like a black box with
a characterized behavior.
The purpose of data encapsulation is to prevent accidental modification of information of a class. It
is achieved by imposing a set of rules—the manner in which a class is to be manipulated and the data
and functions of the class can be accessed. The following are the three kinds of users of a class:
♦ A class member, which can access all the data members and functions of its class.
♦ Generic users, which define the instance of a class.
♦ Derived classes, which can access members based on privileges.
Each user has different access privileges to the object. A class differentiates between access privi -
leges by partitioning its contents and associating each one of them with any one of the following
keywords:
♦ private

♦ public
♦ protected
These keywords arc called access-control specifiers. All the members that follow a keyword (upto another
keyword) belong to that type. If no keyword is specified, then the members are assumed to have private
privilege. The following specification of a class illustrates these concepts:
class
PiggyBank {
int Money; void Display{) Private by default Private by
default

}
private: int AccNumber;
Private by declaration
public:
int code; Public Public
void SetData(int a, int b) {

}
protected:
int PolicyCode; void GetPolicyCode{) { Protected Protected

Copyrighted material
 )
In (he above declaration, the members Money, AccNumber, and Display {) will be
of type private; the members code and SetData () will be of typepublic; and the
members PolicyCode and GetDataO will be of type protected.
Data hiding is mainly designed to protect well-intentioned programmers from honest mistakes. It
protects access to the data according to the design decision made while designing a class. Programmers
who really want to figure out a way to access highly protected data such as private, will find it hard to do
so even by accident. There are mechanisms to access even private data using friends, pointer to members, etc.
from outside the class.

Private Members
The private members of a class have strict access control. Only the member functions of the same class
can access these members. The private members of a class are inaccessible outside the class, thus,
providing a mechanism for preventing accidental modifications of the data members. It is illustrated in
Figure 10.12. Strictly speaking, information hiding is implemented only partially, the private members
can still be accessed. Access control in C++ has the objective of reducing the likelihood of bugs and
enhancing consistency. Since the basic intention of declaring a class is to use it in a program, the class
should have atleast one member that is not private.

class Person
{ ^^J* Note: colon here
private : ^ access specifier
// private members

int age; — —• private data

int getage () ; - private function

);

Person pi;
a=pl. age; X cannot access private data
pi. getage {); X cannot access private function
Figure 10.12: Private members accessibility

rhe following example illustrates the situation when all the members of a class are declared as private:
class Inaccessible <
int x;
void Display()
(
cout « "\nData
} }
;
void main ()
i
// Creating an object. //
Inaccessible objl; objl.x = 5;
Error: Invalid access. //
objl.Display{J ;
Error: Invalid access.
)

Copyrighted material
Table 10.1: Visibility of class members
class C;
Member (unctions of class C public:
can access both private and ✓
public members Object of class C can access
only public members of C

C objc;

Figure 10.15: Class member accessibility

The following declaration of a class illustrates the visibility limit of the various class members:
class MyClass

private: int
a; void
fl()

//can refer to data members a , b, c and functions f l , f 2 , and f j

protected
: int
b;
voi
d
f2

//can refer to data member a , b, c and functions f l , f 2 , and f 3

public:
int c;
void
£3 0 {
//can refer to data member a, b, c and functions f l , f2, and f ?
)

int * v ; / / pointer o
to a veotor i n t s z ; i
/ / s i z e o f a vector d
public:
s
void VectorSizet int s i z e ) / / allocate memory
h
dynamically { o
sz = size; w
v = new int [ size ] ; / / dynamically allocate vector _
} s
v u
o m
i (
d )
;
r
e
a
d
(
)
;

Copyrighted material
 void release( ) / / release memory allocated { vl.release(
delete v ; ); / / free
) vector
}' resources
void vector: : read ( ) J

f o r t int i = 0 ; i Run
< sz; i++ ) { How many
cout « 'Enter v e c t o r ! ■ « 1 « ■ ] ? " ; cin > > v ( i ] ; elements are
J there i n
) v e c t o r : 5_
void Enter vector[ 0
J?1
vector:
Enter vector[ 1
:show_s
)?2
um() (
Enter vector[ 2
int sum = 0 ;
]?1
f o r t int i = 0 ; i < s z ; i + + )
Enter
sum + =
vector! 3 ] ?
v[i] ; cout «
"Vector Sum = " 4
« sum; Enter
vector! 4 ]
? 5_
v
Vector Sum ~
o 15
i
d

m
a
i
n
(
)

{
v
e
c
t
o
r

v
l
;

i
n
t

c
o
u
n
t
;
cout < < "How many elements are there i n vector: " ; c i n > >
count;
vl.VectorSize{ count ) ; / / set vector s i z e
vl.read( ) ;
vl. show_sum { ) ;

Copyrighted material
 Copyrighted
material
In main (), the statement
vector vl ;
creates an object vl of the class vector and the statement
vl.VectorSize( count ); // set vector size allocates the required amount (specified by the
parameter count) of memory, dynamically for vector elements storage. The last statement
vl.release();
releases the memory allocated to the pointer data member v of the vector class. The operation of
dynamic allocation of memory to data members can be at best realized by defining constructor and
destructor functions. (More details can be found in the chapter Object Initialization and Cleanup).

10.12 Passing Objects as Arguments

It is possible to have functions which accept objects of a class as arguments, just as there are functions
which accept other variables as arguments. Like any other data type, an object can be passed as an
argument to a function by the following ways:
♦ pass-by-value, a copy of the entire object is passed to the function
♦ pass-by-reference, only the address of the object is passed implicitly to the function
♦ pass-by-pointer, the address of the object is passed explicitly to the function
In the case of pass-by-value, a copy of the object is passed to the function and any modifications made to
the object inside the function is not reflected in the object used to call the function. Whereas, inpass-by-
reference or pointer, an address of the object is passed to the function and any changes made to the object
inside the function is reflected in the actual object. The parameter passing by reference or pointer is more
efficient since, only the address of the object is passed and not a copy of the entire object.

Passing Objects by Value

The program distance.cpp illustrates the use of objects as function arguments in pass-by-value
mechanism. It performs the addition of distance in feet and inches format.
/ / dlstance.cpp: distance manipulation in feet and inches #include
<iostream.h> class distance {
private:
float feet;
float inches;
public:
void init( float ft, float in ) {
feet = ft;
inches = in;
)

Copyrighted material
 void read()
{
cout « "Enter feet: "; cin >> feet; cout « "Enter
inches: "; cin » inches;
)
Account number is: 20
Balance is: 7 5 0 . 5
How much money is to be transferred from acc3 to accl: 200
Updated Information about accounts...
Account number i s : 1
Balance is: 300
Account number i s : 10
Balance i s : 0
Account number i s : 20
Balance is: 550.5
In main ( ) . the statement
acc3-MoneyTransfer( accl, trans_money ) ; transfers the object accl by reference to the
member function MoneyTransf er ( ) . It is to be noted that when the MoneyTransf er ( ) is
invoked with accl as the object parameter, the data members of acc3 are accessed without the use
of the class member access operator, while the data members of accl are accessed by using their
names in association with the name of the object to which they belong. An object can also be
passed to a non-member function of the class and that can have access to the public members only
through the objects passed as arguments to it.

Passing Objects by Pointer

The members of objects passed by pointer are accessed by using the - > operator, and they have
similar effect as those passed by value. The above program requires the following changes if
parameters are to be passed by pointer:
1. The prototype of the member function MoneyTransf er { ) has to be changed to:
void MoneyTransfer{ AccClass * acc, float amount ) ;
2. The definition of the member function MoneyTransfer ( ) has to be changed to:
void AccClass: : MoneyTransfer( AccClass & acc, float amount ) {
balance = balance - amount; / / deduct money from source acc->balance = acc-
>balance + amount; / / add money to destination
}
3. The statement invoking the member function MoneyTransfer ( ) has to be changed to:
acc3-MoneyTransfer( fcaccl, trans_money ) ;

10.13 Returning Objects from Functions

Similar to sending objects as parameters to functions, it is also possible to return objects from functions.
The syntax used is similar to that of returning variables from functions. The return type of the function is
declared as the return object type. It is illustrated in the program complex. cpp.
// complex.cpp: Addition of Complex Numbers, class complex as data type ♦include <iostream.h>
•include <math.h>
class complex

private:
float real; float // real part of complex number
imag; // imaginary part of complex number

Copyrighted material
342 Mastering C++

10.14 Friend Functions and Friend Classes

The concept of encapsulation and data hiding dictate that non-member functions should not be al lowed to
access an object's private and protected members. The policy is, if you are not a member you cannot get it.
Sometimes this feature leads to considerable inconvenience in programming. Imagine that the user wants a
function to operate on objects of two different classes. At such times, it is required to allow functions
outside a class to access and manipulate the private members of the class. In C++, this is achieved by using
the concept of friends.
One of the convenient and a controversial feature of C++ is allowing non-member functions to access
even the private members of a class using friend functions or friend classes. It permits a function or all the
functions of another class to access a different class's private members. The accessibility of class members
in various forms is shown in Figure 10.17.

class X classY

private:

data or function fyl()

protected:

data or function

friend class Y. friend

functionl(); friend Z:fzl{);

class Z

fzl()
functionl()

friend of X
fz2()

Figure 10.17: Class members accessibility in various forms

The function declaration must be prefixed by the keyword friend whereas the
function definition must not. The function could be defined anywhere in the
program similar to any normal C++ function. The functions that are declared
with the keyword friend are called friend functions. A function can be a friend to
multiple classes. A friend function possesses the following special
characteristics:
/ ' friend function of class one and two int
add_both( one a, two b )
{
return a.datal + b.data2; // a.datal and b.data2 are private
}
void main ( )
(
one
a;
two
b;
a.setdata( 5 ) ; b.
setdata( 10 > ;
cout << "Sum of one and two: " << add_both( a, b ) ;
}

Copyrighted material
342 Mastering C++

Run
Sum of one and two: 15
The above program, contains two classes named one and two. To allow the normal function add_both()
to have an access to private data members of objects of these classes, it must be declared as a friend
function. It has been declared with the friend keyword in both the classes as:
friend int add_both( one a, two b ) ; This declaration can be placed either in the private or the
public section of the class.
An object of each class has been passed as an argument to the function add_both{). Being a friend
function, it can access the private members of both classes through these arguments.
Observe the following declaration at the beginning of the program
class two; / / advance declaration like function prototype It is necessary, since a class cannot be
referred until it has been declared before the class one. It informs the compiler that the class two's
specification will appear later.
Though friend functions add flexibility to the language and make programming convenient in certain
situations, they are controversial; it goes against the philosophy that only member functions can access a
class's private data. Friend functions should be used sparingly. If a program uses many friend functions, it
can easily be concluded that there is a basic flaw in the design of a program and it would he better to
redesign such programs. However, friend functions are very useful in certain situations. One such example
is when a friend is used to increase the versatility of overloaded operators, which will be discussed in the
chapter Operator Overloading.
Friend functions are useful in the following situations:
♦ Function operating on objects of two different classes. This is the ideal situation where the friend function
can be used to bridge two classes.
♦ Friend functions can be used to increase the versatility of overloaded operators.
♦ Sometimes, a friend allows a more obvious syntax for calling a function, rather than what a member
function can do.

Friend Classes
Friend functions permit an exception to the rules of data encapsulation. The friend keyword allows a
function, or all the functions of another class to manipulate the private members of the original class. The
syntax of dcclaring/rii/u/ class is shown in Figure 10.19.

Copyrighted material
346 Mastering C++

class boy {
private: - ' private specifier
int income1;
int income2;
public: - public specifier
int gettotalO
{
return incomel * income2;
}
friend class girl; //class girl can access private members
};
class girl {
/ / all the members of class girl can access attributes of boy
public: private data of
int girlfunctboy bl) ^—class boy
result = bl.incomel+bl.income2; return
result;
* private data of
™ld show{) /classboy
boy bl; *
cout « "Incomel: " << bl.incomel; / / private data of boy
>

Figure 10.19: girl class is a friend of class boy

All the member functions of one class can be friend functions of another class. The program
friend2 .cpp demonstrates the method of bridging classes using friend class.
/ / friend2.cpp: classgirl is declared as a friend of class boy #include <iostream.h>
/ / forward declaration of class girl; is optional
class boy
t
private: / / private members
int incomel;
int income2;
public:
void setdata ( int inl, int in2 ) {
incomel = inl;
income2 = in2;
)
friend class girl; / / class girl can access private data

11
Object Initialization and Cleanup

11.1 Class Revisited

A class encapsulates both data and functions manipulating them into a single unit. It can be further used as an abstract
data type for defining a class instance called object. As with standard data types, there must exist a provision to initialize
objects of a class during their definition itself. Consider an example of the class bag having two data members: contents
to hold fruits and ItemCount to hold the number of items currently stored in the bag. It has interface functions such as
SetEmptyO. put ( ) , and show ( ) whose usage is illustrated in the program bag. cpp.

/ / bag.Cpp: Bag into which fruits can be placed •include

<iostream.h>
const int MAX_ITEMS = 25; / / Maximum number of items that a bag can hold
class Bag
{
private:
int contentsfMAX_ITEMS]; / / bag memory area
int ItemCount; / / Number of items present in a bag
public:
/ / sets ItemCount to empty void
SetEmptyO (

Copyrighted material
346 Mastering C++

ItemCount = 0; //When you purchase a bag, it will be empty

}
void put( int item ) / / puts item into bag
{
contents( ItemCcunt++ I = item; // counter update
)
void show{);
);
/ / display contents of a bag
void Bag: :show()
(
fori int i = 0; i < ItemCount; i++ )
cout « contents[i] « ' "; cout << endl;
)
void main ()
{
int item; Bag
bag;
bag. SetEmpty ( ) ; / / set bag to empty
public:
/ / sets ItemCount to empty
Bag () // constructor
(
ItemCount = 0 ; / / When you purchase a bag, i t w i l l be empty
)
void p u t t int item ) / / puts item into bag
t
contents! ItemCount++ ] = item; / / item into bag, counter update
}
void showt) ;
>;
/ / display contents o f the bag
void B a g : : s h o w ( )
{
f o r i int i = 0 ; i < ItemCount; i + + )
cout « contents[ i ) << " " ;
cout « endl;
)
void
main( )
(
int
item;
Bag
bag;
while
t 1 ) {
cout < < "Enter Item Number to be put into the bag <0-no item>: ■ ; cin » item;
i f t item == 0 ) / / items ends, break
break;

bag. p u t ( item ) ;
cout « "Items in Bag: " ;
bag.s h o w t ) ;
}
)

Run
Enter Item Number to be put into the bag <0-no item>: .1 Items in
Bag: 1
Enter Item Number to.be put into the bag <0-no item>: 2 Items in
Bag: 1 3
Enter Item Number to be put into the bag <0-no item>: 2. Items in
Bag: 1 3 2
Enter Item Number to be put into the bag <0-no item>: 4 Items in
Bag: 1 3 2 4
Enter Item Number to be put into the bag <0-no item>: 0.

Copyrighted material
346 Mastering C++

In m a i n ( ) , the class instantiation

statement Bag bag;
creates the object bag and initializes ItemCount to zero by invoking the no-argument constructor
Bag: : Bag( )
automatically. In the earlier program bag. cpp, these actions are performed by the following statements
Bag bag;
bag. SetEmpty ( ) ; // set bag to empty First, the object bag is created and then, SetEmptyO is explicitly
invoked to initialize the data member ItemCount to zero.
When an object is a local non-static variable in a function, the constructor Bag ( ) is called when the function is
invoked. When an object is a global or a static variable, the constructor Bag ( ) is invokec before the execution of main
{) as illustrated in the program testl. cpp.

/ / testl.cpp: a class Test with a constructor function •include <iostream.h> class

Test {
public: // 'public* function:
TestO; / / the constructor
J;
Test: :TestO // here is the definition
{
cout « "constructor of class Test called" « endl;
)
/ / and here is the test program:
Test G; // global object
void func()
(
Test L; // local object in function funcO
cout « "here's function funcO" « endl;
)
void main ( ) {
Test X; / / local object in function mainO
cout « "mainO function" << endl; func { ) ;
)

Run
constructor of class Test called (globalobject G)
constructor of class Test called (objectX in main())
mainO function
constructor of class Test called (object L in func())
here's function funcO
The output produced by the program is as desired (see Run - the text in parentheses indicates comment). The
program shows how a class Test is defined, which consists of only one function: the constructor. The constructor
performs only one action; a message is printed. The program contains three objects of the class Test: first a global
object, second a local object in main ( ) , and thirtf arother local object in func ( ) .
Chapter 1: Object-Oriented Paradigm 23

gradually to programming in C+ +, in the first step just feeding their existing C code through the C* + translator and checking if
some small modifications would be necessary". However, some consider this as a disadvantage. They claim that an abrupt
change of paradigm is necessary to make programmers think in an object-oriented fashion.
C++ is evolved from a dialect of C known as C with Classes as a language for writing effective event-driven
simulations. Several key ideas were borrowed from the Simula67 and ALGOL68 programming languages. The heritage
of C++ is shown in Figure 1.20. Earlier version of the language, collectively known as "C with Classes" has been in use
since 1980. It lacked operator overloading, references, virtual functions, and all these are overcome in C++. The name
C++ (pronounced as C plus plus) was coined by Rick Mascitti in the summer of 1983. The name signifies the
evolutionary nature of the changes from C. "++" is the C increment operator. The slightly short name C+ is a syntax
error; it has also been used as the name of an unrelated language. Connoisseurs of C semantics find C++ inferior to ++C
The language is not called D, because it is an extension of C, and does not attempt to remedy the problems by removing
features.

Simula 67 Algol 68

* N
C with AT&T C++
* Classes"

AT&T
C

Figure 1.20: Heritage of C++

Copyrighted material
346 Mastering C++

The C++ language corrects most of the deficiencies of C by offering improved compile-time type checking and
support for modular and object-oriented programming. Some of the most prominent features of C++ are classes, operator
and function overloading, free store management, constant types, references, inline functions, inheritance, virtual functions,
streams for console and file manipulation, templates, and exception handling.
In C++, both attributes (data) and methods (functions) are members of a class. The members must be declared either
as private or public. Public members can be accessed by any function; private members can only be accessed by methods
of the same class. C++ has a special constructor function to initialize new instances and a destructor function to perform
necessary cleanup when an object is to be destroyed. C++ provides three kinds of memory allocation for objects: static
(preallocated by the compiler in fixed global memory); automatic (allocated on the stack) anddynamic (allocated from a
heap). Static storage is obtained by defining a variable outside any function using the static keyword . Local variables
within functions normally use automatic storage. Dynamic storage is allocated from a heap on an explicit request from
the programmer and it must be explicitly released since, standard implementations of C++ do not have a garbage
collector.
The superclass of a class is specified as a part of class declarations. A superclass is known as base crass and a
subclass is known as derived class. Attributes once declared in the superclass, which are inherited by its subclasses, need
not be repealed. They can be accessed from any subclass unless they assigns 1.5 toreal_in and 2.0 to imag_in. If the
actual parameter is explicitly specified, it overrides the default value. As stated earlier, the missing arguments must be
the trailing ones. The invocation of a constructor with default arguments while creating objects of the class complex is
shown in Figure 11.6.
class complex {

public: - ■ ►complex () ;
complex(float real_in,float imag_in=6.0);

complex cKl.5,2.0);------------------------....................
complex c2(2.2);-------------------------------------------
complex c3;

Figure 11.6: Default arguments to constructor

Suppose the specification of the constructor complex (float, float) is changed to, complex! float real_in = 0.0, float
imag_in = 0 . 0 ) in the above program, it causes ambiguity while using a statement such as, complex cl;
The confusion is whether to call the no-argument constructor,
complex::complex( ) or the two
argument default constructor
complex::complex( float = 0.0, float = 0.0) Hence, such a specification should be avoided. If no constructors
are defined, the compiler tries to generate a default constructor. This default constructor simply allocates storage to
build an object of its class. A constructor that has all default arguments is similar to a default (no-argument) constructor,
because it can be called without any explicit arguments. This may also lead to errors as shown in the following program
segment:
class X {
int value; public:
X()
{
value=0;
)
X(int i=0) (
value=i;
)
);
-vector{) / / destructor
{
delete v; / / release vector memory
)
void readO ; void
show_sum<);
};
void vector: : read ( )
{
for( int i = 0; i < sz; i++ ) (
cout « "Enter vector! ■ « i « ■ ] ? ■ ; cin >> v[i J ;
}
)
void vector::show_sum()
{
int sum = 0;
for( int i = 0; i < sz; i++ )
sum += v ( i J ; cout << "Vector Sum = " «
sum;
>
void main ( ) {
int count;

Copyrighted material
346 Mastering C++

cout « "How many elements are in the vector: " ; cin » count;
// create an object of vector class and compute sum of vector elements
vector vl{ count ) ;
vl.readt) ;
vl.show_sum{);
}

Run
How many elements are in the vector: 5_
Enter vector! 0 ] ? 1
Enter vector! 1 1 ? 2.
Enter vector! 2 ) ? 2
Enter vector! 3 ] ? 4
Enter vector! 4 ] ? 5_
Vector Sum = 15

In main ( ) , the statement,

vector vl { count ) ;
creates the object vl of the class vector dynamically of the size specified by the variable count (it is also read at runtime).
The function read ( ) accepts elements of the vector from the console and show_sum { ) computes the sum of all the
vector elements and prints the same on the console.
The following points can be emphasized on dynamic initialization of objects.
♦ A constructor of the class makes sure that the data members arc initially 0-pointers (NULL).
void vector::show{) {
for( int i = 0; i < size; i++ ) cout << elem( i ) << ",
■;
)
void main ( ) I
int i ;
vector vl( 5 ) , v2( 5 ) ;
for( i = 0 ; i < 5 ; i + + )
v2.elem( i ) = i + 1;
vl ■ v2; // copy constructor is not invoked
vector v3 = v2; / / copy constructor is invoked, vector v3(v2) cout << "\nvector vl: "; vl.show{);
cout << "\nvector v2: "; v2.show{);
cout « "\nvector v2: "; v3. showf.) ;
)

Run
Copy constructor invoked vector vl: 1, 2,
3, 4 , 5, vector v2: 1, 2, 3, 4 , 5, vector
v2: 1, 2 , 3, 4 , 5,

A copy constructor copies the data members from one object to another. The function also prints the message (copy
constructor invoked) to assist the user in keeping track of its execution. The copy constructor takes only one argument,
an object of the type vector, passed by reference. The prototype is
vectorf vector &v2 ) ; It is essential to use a reference in the argument of a copy constructor. It should not be
passed as a value; if an argument is passed by value, its copy constructor would call itself to copy the actual parameter
to the formal parameter. This process would go on-and-on until the system runs out of memory. Hence, in a copy
constructor, the argument must always be passed by reference, preventing creation of copies. A copy constmctoi also
gets invoked when the arguments are passed by value to functions, and when values arc returned from functions. If an
object is passed by value, the argument on which the function operates is created using a copy constructor. If an object
is passed by address, or reference, the copy constructor would not be invoked, since, in such a case, copies of the objects
need not be created. When an object is returned from a function, the copy constructor is invoked to create a copy of the
value returned by the function.

11.12 Constructors for Two-dimensional Arrays

A class can have multidimensional arrays as data members. Their size can be either statically defined or dynamically
varied during runtime. A matrix class can be designed to contain data members for storing matrix elements, which are
created dynamically. The program matrix. cpp illustrates the method of
return 1;
J
void matrix: : read
() {
int i , j ;
f o r t i = 0 ; i < MaxRow; i++ )
f o r t j = 0 ; j < MaxCol; j++ ) (
cout « "Matrix[1 « i « « j « ■ ] = ? ■ ; cin » p [ i ] [ j J ;
)
)

Copyrighted material
346 Mastering C++

void matrix: : show

() {
int i , j ;
f o r t i = 0 ; i < MaxRow; i + + )
{
cout « endl;
for( j = 0 ; j < MaxCol; j++ )
cout « p [ i ] [ j ] « " ';
}
}
void
maint)
{
int m, n, p, q;
cout « "Enter Matrix A d e t a i l s . . . " « endl; cout < <
"How many rows ? ■; cin >> m;
cout « "How many columns ? " ;
cin » n;
matrix a t m, n ) ;
a.read( ) ;
cout « "Enter Matrix B d e t a i l s . . . * « endl; cout «
"How many rows ? ' ; cin » p;
cout << "How many columns ? " ;
cin >> q;
matrix b( p , q ) ;
b.readt) ;
cout « "Matrix A is
a. showt) ;
cout « endl « "Matrix B i s . . . ' ;
b.show( ) ;
matrix c( m( n ) ;
c.addt a , b ) ;
cout « endl « ■ C = A + B. . . " ;
c . showt) ;
matrix d( m, n ) ;
d.subt a , b ) ;
cout « endl « "D = A - B . . . " ;
public: date() i
. . . / / initializing members of date class
)
read( ) ;
......... / / other member functions of date class
} birthday; / / instance of the nested date class
Student{)
{
. . . . / / initialize members of Student class
}
-Student( ) (

}
read( ) (
//read members of Student class's object including 'birthday' cin >> roll_no;

birthday.read( ) ; / / accessing member of a nested class date

)
. . . . / / other member functions of Student class
J;
The embedded class date is declared within the enclosing class declaration. An object of type Student can be defined as
follows: Student si;
The year in which the student si was born can be accessed as follows: si.birthday.year
A statement such as,
si.date.day = 2 ; / / error
is invalid, because members of the nested class must be accessed using its object name.
The feature of nesting of classes is useful while implementing powerful data structures such as linked lists and trees.
For instance, the stack data structure can be implemented having a node data member which is an instance of another
class (node class).

Review Questions

Copyrighted material
346 Mastering C++

11.1 What are constructors and destructors ? Explain how they differ from normal functions.
11.2 W^hat are the differences between default and parameterized constructors ?
11.3 What are copy constructors and explain their need ?
11.4 What is the order of construction and destruction of objects ?
11.5 What are read-only objects ? What is the role of constructor in creating such objects ?
11.6 State which of the following statements are TRUE or FALSE. Give reasons.
Figure 12.2). The program ptrobjl.cpp illustrates the definition of pointers to objects and their usage in accessing
members of a class.

class XYZ {
private:
int a ;

int b; public:
int c , d ;
int fund ( ) ;
};
XYZ objl; XYZ *ptr;

ptr = tobjl; ptr->c ^

ptr->funcl();

Figure 12.2: Pointer accessing class members

/ / ptrobjl.cpp: pointer to object, pointing to statically created objects #include <iostream.h> class
someclass {
public:
int datal; char
data2; someclass()

cout « "Constructor someclass() is invoked\n"; datal = 1, data2 = *A';

-someclass()

cout « "Destructor -someclass() is invoked\n";

void showt)

cout << "datal = " « datal;

cout « * data2 = " << data2 « endl;

);
void main (void) {
someclass *ptr; // define a pointer to object of class someclass
someclass objectl; // object of type someclass created statically ptr = fcobjectl;
cout << "Accessing object through objectl.show()...' << endl;

The program ptrob j 2 . cpp illustrates the binding of dynamic objects' address to a pointer vari-
able. The pointer defined is initialized with the address returned by the new operator, which actually
creates the object.

/ / ptrobj2.cpp: pointer to object, pointing to dynamically created objects # include <iostream.h> class
someclass
(
public:
int datal; char
data2;
someclass( )
(
cout « "Constructor someclass{ ) is invokedNn'; datal = 1, data2 = * A ' ;
)
-someclass{ ) {
cout « "Destructor -someclass( ) i s invokedXn";
>

void show() {
cout « "datal = " « datal;
cout « " data2 = " « data2 « endl;

Copyrighted material
346 Mastering C++

)
};
void main (void) {
someclass *ptr; / / define a pointer to object of class someclass
cout « "Creating dynamic o b j e c t . . . " « endl;
ptr = new someclass; / / object created dynamically
cout « "Accessing dynamic object through p t r - > s h o w ( ) . . . " « endl;
ptr->show();
cout < < "Destroying dynamic o b j e c t . . . " « endl;
delete ptr; / / object destroyed dynamically
)

Run
Creating dynamic o b j e c t . . .
Constructor someclass ( ) is invoked
Accessing dynamic object through p t r - > s h o w ( ) . . .
datal = 1 data2 = A
Destroying dynamic object...
Destructor -someclass( ) is invoked
In main { ) , the statement
ptr = new someclass; / / object created dynamically creates the nameless object of the class
someclass dynamically and assigns its address to the object pointer ptr. It executes the constructor of
the class someclass automatically during the creation of dynamic objects. The default argument
constructor initializes the data members datal and data2.
/ / refobj.cpp: reference co dynamic objects
•include <iostream.h>
♦include <string.h>
class student
{
private:
int roll_no; / / roll number
char name[ 20 ] ; / / name of a student
public:
/ / initializing data members void setdata( int roll_no_in,
char *name_in ) (
roll_no = roll_no_in; strcpyf
name, name_in ) ;
)
/ / display data members on the console screen void
outdata f) (
cout « "Roll No = ■ « roll_no « endl; cout « "Name
= " « name « endl;
)
);
void main ( )
{
student &sl = "(new student); / / reference to a dynamic object s i . setdata( 1 , "Savithri" ) ;
si.outdata{ ) ;
student &s2 = *{new student); / / reference to a dynamic object s2.setdata( 2 , "Bhavani" ) ;
s 2 . outdata( ) ; student s 3 ;
s 3 . setdata( 3 , "Vani" ) ;
student &s4 = s 3 ; / / reference to static object s 3 .outdata ( ) ;
s4.outdata( ) ;
}

Bun
Roll No = 1
Nam = Savithri
e
Roll No = 2
Nam = Bhavani
e
Roll No = 3
Nam = Vani
e
Roll No = 3
Nam = Vani
e
In main ( ) , the
statement

Copyrighted material
346 Mastering C++

student &sl = "(new student); creates a dynamic object of the class student and binds it to the reference
variable si. The expression * (new student) creates a dynamic object. The memory allocated to such objects cannot l>c
12.4 Array of Objects
C++ allows the user to create an array of any data type including user-defined data types. Thus, an array of variables of
a class data type can also be defined, and such variables are called an array of objects. An array of objects is often used
to handle a group of objects, which reside contiguously in the memory. Consider the following class specification:
class student
(
private:
int roll_no; / / roll number
char namef 20 ] ; / / name of a student
public:
void setdatat int roll_no_in, char *name_in ) ; void outdata( ) ;
)i
The identifier student is a user-defined data type and can be used to create objects that relate to students of different
courses. The following definition creates an array of objects of the student class:
student science[10J; / / array of science course students student medical[ 5 ] ; / / array of medical course students
student engg[25]; / / array c f engineering course students The array science contains ten objects, namely
science[ 0 ] , . . . science[ 9 ] of type student class, the medical array contains 5 objects and the engg array contains 25
objects.

An array of objects is stored in the memory in the same way as a multidimensional array created at compile time.
The representation of an array of engg objects is shown in Figure 12.5. Note that, only the memory space for data
members of the objects is created; member functions are stored separately and shared by all the objects of student class.

roll_no name roll_no name

enggI U ]

engg[1]

roll_no name

Figure 12.5: Storage for data items in an array of objects

An array of objects behaves similar to any other data-type array. The individual element of an array of
objects is referenced by using its index, and member of an object is engg[24] accessed using the dot operator
For instance, the statement
engg[i).setdata( 10, "Rajkumar" ) ; sets the data members of the ith element of the array engg. Similarly, the
statement
engg[i].outdata( ) ;
will display the data of the ith element of the array engg[i]. The program student 1. cpp illustrates the use of the array of
objects.

/ / Student!.cpp: array of student data type

#include <iostream.h>
#include <string.h>
class student
{
private: / / roll number
int roll_no; char name[ 2 0 1 ; / / name of a student
public:
/ / initializing data members
void setdata( int roll_no_in, char *name_in }
{
roll_no = roll_no_in; strcpy( name,
name_in ) ;
)
/ / display data members on the console screen void outdata( ) {
cout « "Roll No = " « roll_no << endl; cout « "Name = " « name
« endl;
)
);
void main( )
{
int i , roll_no, count;
char response, name[ 2 0 ] ;
student s [ 1 0 J ; / / array of 10 objects
count = 0 ;
f o r ( i = 0 ; i "< 10; i++ ) {

Copyrighted material
346 Mastering C++

cout << "Initialize student object { y / n ) : ■ ; cin > >

response;
i f ( response = = ' y ' | | response == ' Y ' )
{
cout « "Enter Roll no. of student: • ; cin » roll_.no;
cout « "Enter Name of student: " ; cin » name;
s [ i ) .setdata( roll__no, name ) ;
count++;
}
class student
(
private:
int roll_no; / / roll number
char name[ 20 ] ; public: / / name of a student
/ / initializing data members
void setdata( int roll__no_in, char *name_in ) {
roll__no = roll__no_in;
strcpyt name, nameLin
);
)
/ / display data members on the console screen void
outdata( ) {
cout « " R o l l No = " « roll_no « endl; cout «
"Name = " « name << endl;
)
);
void main{ )
{
int i , roll_no, count;
char response,
name[20J;
student * s [ 1 0 ) ; / / array o f pointers to objects count = 0 ;
f o r ( i = 0; i < 10; i + + ) {
cout « "Create student object ( y / n ) : " ; c i n »
response;
i f ( response == ' y ' | | response == ' Y ' ) {
cout « "Enter Roll no. of student: " ; c i n »
roll_no;
cout « "Enter Name of student: "; cin »
name;
s [ i ] = new student; / / dynamically creating objects s [ i l ->setdata( roll_no, name ) ;
count+ + ;
)
else
break;
>
cout << "Student d e t a i l s . . . " « endl; for ( i
= 0 ; i < count; i-*+ ) s [ i ] - > o u t d a t a ( ) ;
f o r t i = 0 ; i < count; i++ ) / / release memory allocated to a l l objects delete s [ i ] ;
)

Km
Create student object ( y / n ) : y

Copyrighted material
 Chapter 1: Object-Oriented Paradigm 27

mer must switch from designing programs based on actions to designing programs around data types and
their interactions.
The designer of C++. Bjarne Stroustrup. recommends that the shift from C to C++ should be a gradual
one. with programmers learning the improvements a small step at a time. With C++, quite often, people, as a
first exercise, write a string class and as a second exercise, try to implement a graphics system. That is very
challenging and might be good for a professional programmer, but it's not the best way of teaching an
average student programmer. What we need is an environment that has a very good string class that you can
take apart and look at one which has a very nice graphics system, so that you never care about MS-windows
or X-windows again, unless you absolutely want to. So, the two components needed to start OO
programming are an environment and a library supporting resuability.

1.17 Software Reuse

Programmers have to write code from scratch, when a new software is being developed, using tradi tional
languages, because there is hardly any reuse of the existing components. Software systems have become so
complex that even coding is considered as a liability today. Reusing existing software components is treated as a
key element in improving software development productivity. It facilitates the use of existing well tested and
proven software code as a base module and then develop on it, instead of developing from scratch. The
simplest approach in this direction involves the development and use of libraries of software components.
Once a class has been developed, implemented, and tested, it can be distributed to other programmers for
use in their programs (called reusability). It is similar to the way library functions are used in different
programs. However, in OOP, the concept of inheritance provides an important extension to the idea of
reusability. A programmer can use an existing class without modifying it and add new additional features and
capabilities to build a new class. A newly created derived class has all the inherited features of the old one
with additional features of its own. The ease with which the existing software can be reused is a major
benefit of OOP.
Reuse is becoming one of the key areas in dealing with the cost and quality of the software systems. The
b? s for reuse is the reliability of the components intended for reuse and gains achieved through its
application. The components developed for reuse must have a quality stamp, for example, concerned with
reliability and performance. Object-Oriented techniques make it possible to develop components in general,
and to develop reusable components in particular.
One of the important problems of the software component reuse consists of their localization and retrieval
from a large collection. In fact, reuse implies the following three actions: (i) Retrieve needed component, (ii)
Understand them, and (iii) Use them.
A method to reduce the effort of reusable components' search, comprehension, and adaptation consists of
developing a reuse, strategy which defines a component classification, a component structure, and scarch-
and-use mechanism. The OO concepts such as classes and inheritance provide a better mechanism for
grouping related entities and simplifying the identification of reusable components.
Reuse through Inheritance and its Quantification
Inheritance is considered as an excellent way to organize abstraction, and as a tool to support reuse. The use
of inheritance does have some trade-offs (costs) - inheritance increases the complexity of the ay. tern and the
coupling between classes. Booch recommends that inheritance hierarchies be built as balanced lattices and
that the maximum number of levels and their width be limited to 7 ± 2 classes.
Member functions can also be accessed using" the operator ->* as follows.
(pobj.*init_ptr)( 5 ) ;
int k = (pobj->*init_ptr) ( 5 ) ;
The general format can be deduced to the following
(pointer-to-object->*pointer-to-member-function)(arguments); In such calls, the
parentheses must be used explicitly, since the precedence of ( ) is higher than the
dereferencing - > * operator. The program ptrmemb.cpp illustrates the concept of a pointer
to class members.
/ / ptrmemb.cpp: pointer to class
members #include <iostream.h> class X (
private:
int y; / / through pointer it cannot be accessed public: / / all public members can
be accessed through pointers int a; int b;
int i n i t ( int
z){

Copyrighted material
 Chapter 1: Object-Oriented Paradigm 27

a
=
z;
re
tu
rn
z;
)
);
void main ( )
{
X Obj;
int X : : * i p ; / / pointer to data member
ip = & X : : a ; / / address of data member a i s assigned to pointer / / access through
object obj.*ip = 10;
cout « "a in obj, after obj.*ip = 10 is " « obj.*ip « endl; X *pobj; / / pointer to object of
the class X pobj = &obj;
/ / access through object
pointer pobj->*ip = 10;
cout « "a in obj, after pobj->*ip = 10 is ■ « pobj->*ip « endl; int ( X : : * p t r _ i n i t )
( i n t ) ; / / pointer to member function ptr_init = & X : : i n i t ; / / pointer to member
function i n i t O / / access through object <obj.*ptr_init)( 5 ) ;
cout « "a in obj, after (obj. *ptr_init) ( 5 ) = * « obj.a « endl; / / access through object
pointer (pobj->*ptr_init) ( 5 ) ;
cout « "a in o b j , after (pobj->*ptr_init) ( 5 ) = " « obj.a < < endl;
)

Ad
a in o b j , a f t e r o b j . * i p = 10 is 10
a in ob j, a f t e r pobj->*.ip = 10 is 10 a in o b j ,
a f t e r (obj.*ptr_init)( 5 ) = 5 a in obj, a f t e r
(pobj->*ptr_init)(5) = 5

Access Through Friend Functions

The friend functions can access private data members of a class although it is not in the scope of the class.
Similarly, members of any access privilege can be accessed using pointers to members. Both the
dereferencing operators .* and - > * can be used to access class members. The program friend, cpp illustrates
the concept of accessing class members through pointers from friend functions.

/ / frlend.cpp: friend functions and pointer to members ♦include

<iostream.h> class X
(
private:
int a ;
in
t b;
public:
X{)
{
a = b = 0;
)
void SetMemberst int a l , int bl )
(
a

a
l
;

Copyrighted material
 Chapter 1: Object-Oriented Paradigm 27

b
l
;
)
friend int sum( X x ) ;

int sum( X objx )

{
int X : : * p a = & X : ; a ; / / pointer to member a
int X : : * p b = & X : : b ; / / pointer to member b
X "pobjx = &objx; / / pointer to object of the class X
int result;
/ / the member a is accessed through objects / J and the member b is accessed through object pointer
result = objx.*pa + pobjx->*pb; / / sum a and b; return result;
)
void
main{
) {
X objx;
void ( X : : * p f u n c ) U n t ,
i n t ) ; pfunc = &X::SetMembers;
(objx.*pfunc)( 5 , 6 ) ; / / equivalent to objx.SetMembers(5, 6 )
cout < < "Sum = " « sum( objx ) « endl;
X "pobjx; / / pointer to object o f the class X
pobjx = &objx;
(pobjx->"pfunc){7, 8 ) ; / / equivalent to pobjx->SetMembers(5, 6 )
cout « "Sum = " << sum( objx ) < < endl;
)

Run
Sum =
11 Sum
= 15

12.7 Function set_new_handler ( )

Tne C++ run-time system makes sure that when memory allocation fails, an error function is activated. By
default, this function returns the value 0 to the caller of new, so that the pointer which is assigned by new is
set to zero. The error function can be redefined, but it must comply with a few prerequisites, which are.
unfortunately, compiler-dependent.
The function set_new_handler t) . sets the function to be called when a request for memory allocation
through the operator new{} function cannot be satisfied. lis prototype is
void ( * set_new_handler (void ( * my_handler) ( ) ) ) ( ) ; If new ( ) cannot allocate the
requested memory, it invokes the handler set by set_new_handler ( ) . The user defined function,
my_handler() should specify the actions to be taken when n e w ( ) cannot satisfy a request for memory
allocation.
If my_handler { ) returns, new { ) will again attempt to satisfy the request. Ideally, my_handler would
release the memory and return, new ( ) would then be able to satisfy the request and the program would
continue. However, if my_handler() cannot provide memory for n e w { ) , my__handler must terminate the
program. Otherwise, an infinite loop will be created.
The default handler is reset by set_new_handler (0) . Preferably, it is advisable to overload the new (J to
take appropriate actions as per the application requirement.
The function set_new_handler returns the old handler, if it has been defined. By default, no handler is
installed. The user-defined argument function, my_handler, should not return a value.
The program memhnd. cpp demonstrates the implementation of user-defined function (in Borland C++)
to handle memory resource shortage error.

/ / memhnd.cpp:user-defined handler to handle out-of-memory issue

♦include <iostream.h>
♦include <new.h>

Copyrighted material
 Chapter 1: Object-Oriented Paradigm 27

♦ include <process.h>
void out_of_memory ( )
{
cout « "Memory exhausted, cannot allocate"; e x i t t 1 ) ;
/ / terminate the program
)
void main () (
int *ip;
long total_allocated = OL; / / install error
function set_new_handler( out_of_memory
);
void display! l i s t
"first ) {
l i s t "traverse;
cout << "List traversal yields: " ;
/ / scan f o r a l l the elements
f o r t traverse = f i r s t ; traverse; traverse = traverse->next )
cout < < traverse->data < <
" , " ; cout << endl;
)
void main
(void) {
int choice, d a t a ;
l i s t " f i r s t = NULL; / / i n i t i a l l y points to NULL l i s t "node; / / pointer to
new node to be created while( 1 ) {
cout « "Linked L i s t . . . " « endl;
cout < < " 1 . Insert" << endl;
cout < < " 2 . Display" << endl;
cout << " 3 . Quit" « endl; cout
<< "Enter Choice: " ; c i n >>
.choice-switch (choice) {
case 1:
cout « "Enter
Data: " ; cin »
data;
node = new
l i s t ( data ) ; i f f
f i r s t = = NULL )
first =
node; else
first->insert( node ) ;
b
re
ak
;
ca
se
2:
display! f i r s t ) ;
break; / / Display
l i s t , case 3:
ex
it( 1 )
;
default
:
cout < < "Bad option selected" « endl;
continue;
)
)

Copyrighted material
 Chapter 1: Object-Oriented Paradigm 27

Run
Linke
d
List.
. .
1.Inse
rt
2.Disp
lay
3 . Qu
it
Enter Choice: 1
Enter Data: 2
Linked L i s t . . .
1.Insert 2.Display
3.Quit
Enter Choice: 2
L i s t traversal yields: 2 ,
Linked L i s t . . .
1.Insert
2.Display
3-Quit
Enter Choice: I
Enter Data: 2
Linked L i s t . . .
1.Insert 2.Display
3.Quit
Enter Choice: X
Enter Data: 4
Linked L i s t . . .
1.Insert 2 . Display
3.Quit
Enter Choice: 2
List traversal yields: 2 , 3, 4 , Linked
List. . . 1.Insert 2.Display 3.Quit
Enter Choice: 1

The use of a self-referential class is inevitable in the above program, since each node in the slack has a
pointer to another node of its own type, which is its predecessor (in the case of the stack).
Several problems whose solutions are based on the use of data structures like trees, graphs and lists make
extensive use of self-referential class.

Doubly Linked List

Using this pointer when referring to a member of its own class is often unnecessary, as illustrated earlier; the
major use of the this pointer is for writing member functions that manipulate pointers dircctiy. The doubly
linked list has two pointer nodes: one pointing to the next node i n the list and another pointing to the
previous node in the list. The pictorial representation of a doubly linked list is shown in Figure 12.11.
The program dl 1. cpp makes use of the data structure, doubly linked list, illustrating the typical us? of
the thi s pointer at relevant points. The t h i s pointer is particularly used as a pointer to the first node while
traversing through the entire list.

Copyrighted material
 Chapter 1: Object-Oriented Paradigm 27

class doubly
linked list first

Figure 12.11: Doubly linked list representation

/ / dll.cpp: doubly linked list £include
<iostream.h> ♦include <process.h>
class d l l / / doubly linked l i s t class
(
private:
int data; / / data of a node d l l *prev; / / pointer to previous node dll "next; //
pointer to next node public: d l l ( ) {
data = 0 ;
prev = next = NULL;
)
dll( i n t data_in )
{
data = data_in; prev = next
= NULL;
}
-aim {
cout << * - > ■ << data;
>
int get( ) { return data; )
void i n s e r t f d l l "node); / / Inserts new node at l i s t
friend void displaytdll * ) ; / / Display list
void FreeAllNodes( ) ;
);
/ / Inserts node. I f l i s t empty the f i r s t node is created else the / / new node is inserted
immediately after the first node, void-d l l : : insert( d l l "node ) C
dll "last;
/ / f i n d out last node, this points to f i r s t node f o r { last = this; last->next; last = last-
>next ) ; / / insert new node at the end of list

Copyrighted material
 Chapt er 12: Dynam i c Obj ect s 429

node->prev = last: node-

>next = last->next; last-
>next = node;

void dll::FreeAllNodes<)
i
cout << 'Freeing the node with data: ";
// this points to first node, use it to release all the nodes for< dll 'first
= this; first; first = first->next ) delete first;
)
// "Displays the doubly linked list in both forward and reverse order making >'i

use of the series of next and prev pointers, void displayl dll *first )
{
dll 'traverse = first;
if( traverse == NULL)
(
cout « "Nothing to display !" << endl; // along the list, return;
)
els
e {
cout << "Processing with forward -> pointer:
"; // scan for all the elements in forward
direction for( ; traverse->next; traverse =
traverse->next )
cout << ■->■ « traverse-
>data; // display last element
cout << "->" << traverse->data << endl; cout <<
"Processing with backward <- pointer: "; // scan
for all the elements in reverse direction for( ;
traverse->prev; traverse = traverse->prev )
cout << "->" << traverse->data; //
display first element cout << "->" <<
traverse->data << endl;
)
)
dll * InsertNodet dll 'first, int
data ) (
dll 'node;
node = new dll{ data );
if( first == NULL )
first = node;
else
first->insert( node );
return first;
}
void main( void >
(
int choice, data;

Copyrighted material
dll "first = NULL; //' initially points to NULL cout « "Double Linked List
Manipulation..." << endl; while( 1 ) (
cout << "Enter Choice ([1] Insert, (2] Display, [3J Quit): cin >> choice;
switch (choice) {
 Chapt er 12: Dynam i c Obj ect s 429

case 1:
cout « "Enter Data: "; cin >>
data;
first = InsertNode( first, data );
break; case 2:
display( first ) ;
break; // Display list, case 3:
first->FreeAllNodes(); // release all nodes exit( 1 ); default:
cout « "Bad option selected" « endl; continue;
)
)
)

Run
Double Linked List Manipulation...
Enter Choice ([1] Insert,' [2] Display, [3] Quit) : I
Enter Data: 2
Enter Choice ((1) Insert, [2J Display, [3] Quit): 2
Processing with forward -> pointer: ->3 Processing with
backward <- pointer: ->3 Enter Choice ([1] Insert, [2]
Display, [3] Quit):! Enter Data: 2
Enter Choice ([1] Insert, [2) Display, [3J Quit): 2.
Processing with forward -> pointer: ->3->7 Processing with
backward <- pointer: *>7->3 Enter Choice ([1] Insert,
[2) Display, [3] Quit): 1 Enter Data: £
Enter Choice ([1] Insert, [2] Display, [31 Quit): 2
Processing with forward -> pointer: ->3->7->5 Processing
with backward <- pointer: ->5->7->3 Enter Choice ([1]
Insert, [2) Display, [3J Quit) : 0. Bad option
selected
Enter Choice ([1] Insert, [21 Display, [3) Quit): 2
Freeing the node with data: ->3->7->5
Besides handling dynamic data structures, the this pointer finds extensive
application in the following contexts:
♦ Member functions returning pointers to their respective objects.
13
Operator Overloading

13.1 Introduction
The operators such as +, +=. >, ». etc., are designed to operate only on standard data types in structured
programming languages such as C. The + operator can be used to perform the addition operation on integer,
floating-point, or mixed data types as indicated in the expression {a+b). In this expression, the data type of the
operands a and b on which the + operator is operating, is not mentioned explicitly. In such cases, the compiler
implicitly selects suitable addition operation (integer, floating-point, double, etc.,) depending on the data type
of operands without any assistance from the programmer. Consider the following statements:
inc a, b, c; float
x, y, 2 ;
c = a + b; // 1: integer addition and assignment
z = x + y; 1/ 2 : floating-point addition and assignment
x = a + b; // 3: integer addition and floating point assignment
The operators = and + behave quite differently in the above statements: the first statement does integer
addition and assigns the result to c, the second performs floating-point addition and assigns the result to 2, and
the last performs integer addition and assigns the result to the floating-point variable x. It indicates that, the +
operator is overloaded implicitly to operate on operands of any standard data type supported by the language.
Unlike C, in C++, such operators can also be overloaded explicitly to operate on operands of user-defined data
types. For instance, the statement
c3 = AddComplext cl, c2 );
performs the addition of operands cl and c2 belonging to the user defined data type and assigns the result to
c3 (which is also operand of the user defined data type). In C++, by overloading the + operator, the above
statement can be changed to an easily readable form:
c3 = cl + c2;
It tries to make the user-defined data types behave in a manner similar (and have the same look and feet) to the
built-in data types, thereby allowing the user to redefine the language itself. Operator overloading, thus allows
to provide additional meaning to operators such as +,*, >=,+=,etc, when they are applied to user defined data
types. It allows the user to program (develop solution to) the problems as perceived in the real world.
The operator overloading feature of C++ is one of the methods of realizing polymorphism. The word
polymorphism is derived from the Greek words poly and morphism (polymorphism = poly + morphism). Here,poly
refers to many or multiple anamorphism refers to actions, i.e., performing many actions with a single operator.
As staled earlier, the + operator performs integer addition if the operands are of integer type and floating point
addition if the operands are of real type.
The concept of operator overloading can also be applied to data conversion. C++ offers automatic
conversion of primitive data types. For example, in the statement x=a+b, the compiler implicitly con verts the
integer result to floating-point representation and then assigns to the float variable x. But the conversion of
user defined data types requires some effort on the part of the programmer. Thus, operator overloading
concepts are applied to the following two principle areas:
•Extending capability of operators to operate on user defined data.
♦ Data conversion.
Operator overloading extends the semantics of an operator without changing its syntax. The grammatical
rules defined by C++ that govern its use such as the number of operands, precedence, and associativity of the
operator remain the same for overloaded operators. Therefore, it should be remembered that the overloaded
operator should not change its original meaning. However, semantics (meaning) can be changed, but it is
advisable to retain the predefined logical meaning.

13.2 Over loadable Operators

C++ provides a wide variety of operators to perform operations on various operands. The operators are
classified into unary and binary operators based on the number of arguments on which they operate. C++

Copyrighted material
allows almost all operators to be overloaded in which case atleast one operand must be an instance of a class
(object). It allows overloading of the operators listed in Table 1 3.1.
The precedence relation of overloadable operators and their expression syntax remains the same even after
overloading. Even if there is a provision to change the operator precedence or the expression syntax, it does
not offer any advantage. For instance, it is improper to define a unary division (/) or a binary complement (-),
since the change of precedence or syntax leads to ambiguity. For example, defining an operator * * to
represent exponentiation as in the case of Fortran language, appears to be obvious, however, interpretation of
the expression a* *b, leads to confusion; whether to interpret it as a*(*b) or (a) * Mb), because, C++already
interprets itasa*(*b).

Table 13.1: C++ overloadable operators

The variables idxl and idx2 are the objects of the class index. The index value is advanced by using
statements such as ++idxl; idx2++; instead of explicitly invoking the member function Nextlndex {) as in the
earlier program. The operator is applied to objects of the Index class. Yet. operator function ++. takes no
arguments. It increments the data member value of the Index class's objects. Figure 13.1 shows the index class
representation and invocation of its member functions when they are accessed implicitly (constructor
function) or explicitly (other members).

instances of the class index

Figure 13.1: Index class and ++ operator overloading

13.4 operator Keyword
The keyword operator facilitates overloading of the C++ operators. The general format of operator
overloading is shown in Figure 13.2. The keyword operator indicates that the operator symbol following it, is
the C++ operator to be overloaded to operate on members of its class. The operator overloaded in a class is
known as overloaded operator function.
Function return type: primitive, void, or user defined Keyword

Copyrighted material
 Operator to be overloaded (see Table 13.1)

r
Arguments to Operator Function

ReturnType operator OperatorSymbol ([argl,[arg2 t

/ / body of Operator function

Figure 13.2: Syntax of operator overloading

public:
Index( ) / / N o argument constructor
{ value = 0 ; )
Index( int val ) / / Constructor with one argument
{
value = val;
)
int GetlndexO // Index Access
{
return value;
}
Index operator ++() // Returns nameless object of class Index {
value = value + 1;
return Index( value ) ; // calls one-argument constructor
)
> ;

void
main
() {
Index idxl, idx2; / / idxl and idx2 are the objects of Index
cout « "\nlndexl = ■ << idxl.Getlndex( ) ; cout
« "\nlndex2 = 1 « idx2 .Getlndex( ) ,-
idxl = idx2++; / / return object idx2++ is assigned to object idxl
idx2++; / / return object idx2++ is unused
cout « ■Nnlndexl = " « idxl.Getlndex( ) ; cout
<< "\nlndex2 = " « idx2 .GetlndexO ;
)

Run
Ind
exl
= 0
Ind
ex2
= 0
Ind
exl
= 1
Ind
ex2
=2

In the program index3 . cpp, the statements used to return an object are the following: Index temp;
value = value + 1; temp.value = value; return temp; In this program, the statements,
value = value + 1;
return Index( value ) ;
perform the same operation as achieved b; ihe above four statements. It creates a nameless object by
passing an initialization value. To perform this operation, the following parameterized constructor is
added as the constructor member function to the Index class:
Index( int val )

Copyrighted material
 (
val"~ - val;
cout << "(" « real;
cout « ", " « imag « ■)■;
}

// adds default and c2 complex objects

complex complex::AddComplex( complex c2 )

c m
o p
m ;
p temp.real = real + c2.real; temp.imag = imag +
l c2.imag;
e
return( temp );
x
// object temp of complex class
// add real parts
t
// add imaginary parts
e
// return complex object
)
void
main
() (
complex cl, c2, c3; // cl, c2, c3 are object of complex class cout « "Enter Complex
Number cl .." « endl; cl.getdata();
cout « "Enter Complex Number c2 . . " « endl;
c2.getdata();
c3 = cl. AddComplex ( c2 ); // add cl and c2 and assign the result to c3
c3.outdata ( "c3 = cl .AddComplex ( c2 ): ");
)

Run
Enter Complex Number cl . .
Real Part ? 2^3.
Imag Part ? 2 . 0
Enter Complex Number c2 . .
Real Part ? XJ2.
Imag Part ? 1.5
c3 = cl.AddComplex( c2 ): (5.5, 3.5)

In main(). the statement

c3 = cl.AddComplex{ c2 ); invokes the member function AddComplex (} of the cl

object's class and adds c2 to it and then the returned result object is assigned to c3. By
overloading the + operator, this clumsy and dense-looking statement can be represented in the
simplified standard (usual) form as follows:
c3 = cl + c2;
The program complex2 . cpp illustrates the overloading of the binary operator + in order to
perform addition of complex numbers.

/ / complex2.cpp: Complex Numbers operations with operator overloading finclude <iostream.h>

class complex (
private:
float real; float // real part of complex number
imag; // imaginary part of complex number

complex operator*{complex c2)

F
t
complex temp;

Copyrighted material
 temp.x=

temp.y=
temp

5.5
3.5

Copyrighted material
 X + c2.x
y + c2.y
return(temp);

c3
2.5 2.0 3.0 1.5
C2;
5.5 3.5 cl

Figure 13.6: Operator overloading in class complex

// COmplex3 .Cpp: Manipulation of Complex Numbers

#include <iostream.h> class complex {
private:
float real;
float
imag;
public:
complex()
{
real = imag = 0;
)
void getdataO // read complex number
{
cout « "Real Part ? "; cin »
real; cout « "Imag Part ? ";
cin » imag;
}
void outdata( char *msg ) // display complex number
{
cout « endl « msg; cout « " (" «
real; cout « *, ■ « imag « *)";
}
complex operator + ( complex c2 ) ;
Index( int val ) ( // Constructor with one argument
value = val;
}
int GetlndexO // Index Access
(
return value;
)
boolean operators ( Index idx ) //compare indexes {
return( value < idx.value ? TRUE : FALSE );
}
);
void
main
() (
Index idxl
= 5 ; Index
idx2 = 10;
cout « "\nlndexl = " « idxl .Getlndex () ;
cout « "\nlndex2 = " « idx2.Getlndex{);
if ( idxl < idx2 )
cout « '\nlndexl is less than Index2"; else
cout « "\nlndexl is not less than Index2";
}

Copyrighted material
Bun
Indexl
= 5
Index2
= 10
Indexl is less than Index2

The concept of overloading the comparison operator < in the above program is similar to
overloading arithmetic operators. The operator function < () returns TRUE or FALSE depending
on the magnitudes of the Index operands.

Strings Comparison
The relational operators such as <, >, ==, etc., can be overloaded to operate on strings.
These operators return TRUE or FALSE depending on the contents of the string arguments. The
program strcmp. cpp illustrates the overloading of relational operators in a string-class*.

/ / strcmp.cpp: Comparison of strings

♦include <iostream.h> ♦include
<string.h>
const int BUFF_SIZE = 50; // length of string
enum boolean { FALSE, TRUE );
class string // user defined string class
{
private:
Char
str[BUFF_SIZE];
public:
instance, a statement such as

cl = cl + c2;
can be replaced by
cl += c2;

The program complex4. cpp illustrates the overloading of arithmetic assignment operators to
manipulate complex numbers.

// complex4.cpp: Overloading of +=, - = , * = , /= operators for complex class #include

<iostream.h> class complex (
private:
float real;
float
imag; public:
complex() // constructorl
{
real = imag = 0;
)
void getdataO // read complex number
{
cout « "Real Part ? "; cin »
real; cout << "Imag Part
? "; cin » imag;
)
void outdata( char *msg ) // display complex.number
(
cout << endl << msg; cout « "("
« real; cout « " « imag
« ")";
)
void operator += ( complex c2 )
void operator -= { complex c2 )

Copyrighted material
 void operator *= { complex c2 )
void operator /= { complex c2 )
);
// addition of complex numbers, cl += c2 instead of cl = cl + c2; void complex;:operator +=
( complex c2 ) {
real = real + c2.real;
imag = imag + c2.imag;
}
// subtraction of complex numbers, cl -= c2, i.e., cl = cl - c2; void complex::operator -= <
complex c2 ) {
real = real - c2.real;
imag = imag - c2.imag;

Copyrighted material
460 Mastering C++

// Multiplication of complex numbers, cl *= c2, instead of cl =

cl*c void complex: :operator *= ( complex c2 )
{
complex old = *this; // *this is an object of type complex
real = old.real * c2.real - old.imag * c2.imag; imag =
old.real * c2.imag + old.imag * c2.real;
)
// Division of complex numbers, cl /= c2, i.e., cl = cl / c2
void complex::operator /= ( complex c2 J
{
complex old = *this;
float qt;
qt = c2.real*c2.real+c2.imag*c2.imag;
real = (old.real * c2.real + old.imag * c2.imag)/qt;
imag = (old.imag * c2.real - old.real * c2.imag) /qt;
)
void main()
(
complex cl, c2, c3;
// read complex numbers cl and c2
cout << "Enter Complex Number cl . ." << endl;
cl.getdata();
cout « "Enter Complex Number c2 .." « endl; c2.getdata();
cout << "Entered Complex Numbers are..."; cl.outdata( "cl = '
J ; c2.outdata( "c2 = " ) ;
cout « endl « "Computational results are..."; // "c3 = cl +
c2 c3 ,= cl; c3 += c2; c3.outdata("let
// c3 = cl -
c2 c3 = cl; c3
c3 = cl, c3 *= c2: ") ;
- = c2; c3-
outdata("let C3 = Cl, c3 -= c2: ");
// c3 = cl *
c2 c3 = cl; c3
*= c2;
c3.outdata("le c3 = cl, c3 *= c2: -);
t
// c3 = cl /
c2 c3 = cl; c3
/= c2;
c3.outdata("le c3 = cl, c3 /= c2: ");
t

Copyrighted material
 Chapter 13: Operator Overloading

Run
Enter Complex Number cl . .
Real Part ? 2^1
Imag Part ? 2.0
Enter Complex Number c2 . .
Real Part ? 3.0
Imag Part ? 1. 5
Entered Complex Numbers are... cl = (2.5,
2) c2 = (3, 1.5)
Computational results are... let c3 = cl, c3 *= c2:
(5.5, 3.5) let c3 * cl, c3 -= c2: (-0.5, 0.5) let c3 = cl,
c3 *= c2: (4.5, 9.75) let c3 = cl, c3 /= c2:
(0.933333, 0.2)
Observe the difference between the operator function + () defined in the program complex3 . cpp and
operator function += () defined in the program complex4 . cpp. In the former, a new temporary object of
complex type must be created and returned by the function, so that the resultant object can be assigned to a
third complex object, as in the statement
c3 = cl + c2;
In the latter, the function operator += () is a member function of the object (destination object's class), which
receives the result of computation. Hence, the function operator +=() has no return value; it returns void
type. Normally, the result of the assignment operation is not required. In a statement, such as,
c3 ♦= c2;
the operator alone is used without bothering about the return value.
The use of the arithmetic assignment operator in a complicated statement such as, c3 = cl += c2;
requires a return value. Such requirements can be satisfied by having the function operator += (), which
terminates with the statement such as
return! *this );*orreturn complex( real, imag ) ,-
In the first case, the current object is returned and in the latter case, a nameless object is created with
initialization and is returned as illustrated in the program complexS. cpp.

// complex5.cpp: Overloading of += operator for complex expressions •include <iostream.h> class complex

private:
float real;
float imag;
public:
complex( ) { / / no argument constructor
real = i imag 0.0;
void getdata() // read complex number
■"

cout << "Real Pare ?

"; cin >> real; cout
<<-"Imag Part ? "; cin
>> imag;
)
complex operator + ( complex c2 ) ; // complex addition
void outdata( char "msg ) // display complex number
{
endl << msg;
cout << coat <<
"(" << real; cout « ' .. "
<< imag << " ) " :

complex operator += ( complex c2 );

};
// addition of complex numbers, cl += c2 instead of cl = cl +
c2; // return complex object *this or build temporary object
and return complex complex::operator += ( complex c2 ) {
real = real +
c2.real; imag =
imag + c2.imag;
return( *this ); // *this is current object
>
void main ()
(
complex cl, c2, c3;
 Mastering C++

cout << "Enter Complex Number cl .." << endl; cl.getdata();

cout « "Enter Complex Number c2 ..' « endl; c2.getdata();
// Performs 1. cl += c2 and 2. c3 = cl
c3 = cl += c2; // cl += c2 is evaluated first, and assigned
to c3 cout « "\nOn execution of c3 = cl += c2 ..";
cl.outdata('Complex cl: "); c2.outdata("Complex c2: ") ;
c3.outdata("Complex c3: ");
}

Run
Enter Complex Number cl . .
Real Part ?
Imag Part ? 2.0
Enter Complex Number c2 . .
Real Part ?
Imag Part ? 1.5
On execution of c3 = cl c2
Complex ci: (5.5, 3.5)
Complex c2: (3, 1.5)
Complex c3: (5.5. 3.5)
tor function is called recursively leading to stack overflow. Hence, prefixing of the scope resolution operator
to the new operator forces to use the standard new operator supported by the language instead of the one
defined in the program. The class vector has a data item of type dynamic array, defined by int *array.
Another statement in the above function
my_vector->array = new intFARRAY_SIZE); // calls : :new creates an array and dynamically
allocates memory to it.
Similar to the overloaded new operator function, the overloaded delete operator function handles the
process of releasing memory that has been allocated during the dynamic object creation by the new operator;
it also releases the memory allocated to the internal data-item array through the function call
delete my_vector; It invokes the
overloaded operator function
void operator delete( void* vec ) to release the entire memory resource allocated to the my_vector
object and its data members.

13.14 Data Conversion

Representing the same data in multiple forms is a common practice in scientific computations. It involves the
conversion of data from one form to another, for instance, conversion from radian to degree, polar to
rectangular, and vice versa. Implicit invocation of the conversion procedure in C++ is achieved by
overloading the assignment operator, =. The assignment operator assigns the contents of a variable, the result
of an expression, or a constant, to another variable. For example,
varl = var2; // varl and var2 are defined as integer variables
assigns the value of var2 to varl which are of the same data-type. User defined objects of the same class can
also be assigned to one another. In a statement such as
c3 = cl + c2; // cl, c2, and c3 are objects of complex class
the result of addition, which is of type complex is assigned to another object c3 of complex class. The
assignment of one variable/object to another variable/object, which are of the same data-type is achieved by
copying the contents of all member data-items from source object to the destination object. Such operations
do not require any conversion procedure for the data-type conversion. In the above expression, the result of
(cl+c2) is of the same data-type as that of the destination object c3. Hence, the compiler does not require any
special instruction from the user to perform the assignment of objects.
Thus, assignment of data items are handled by the compiler with no effort on the part of the user, whether
they are basic or user defined provided both source and destination data items are of the same data-type. In
case the data items are of different types, data conversion interface function must be explicitly specified by
the user. These include conversions between basic and user-defined types or between the user-defined data
items of different types.

13.15 Conversion between Basic Data Types

Consider the statement
weight = age; // weight is of float type and age is of integer type when weight is of type float and
age is of type integer. Here, the compiler calls a special routine to cor v-crt the value of age, which is
represented in an integer format, to a floating-point format, so thai
object to basic data types and returns a basic data-item. The syntax of such a conversion function is
shown in Figure 13.8.

Keyword operator Primitive tfata type: char, int, float, etc.

Copyrighted material
 Chapter 13: Operator Overloading

operator
BasicTypeO (
/ / steps for converting
/ / Object attributes to BasicType
}

Figure 13.8: Conversion function: user-defined to basic

In the above syntax, it can be observed that the conversion operator function has no return type
specification. However, it should return BasicType value. The program meter. cpp illustrates the
conversion of the Meter class's object to float representing centimeter and vice-versa.

/ / meter.cpp: Conversion from Meter to Centimeter and vice-versa #include <iostream.h>

/ / Meter class for MKS measurement system
class Meter
(
private:
float length; / / length in meter
public:
Meter( ) // constructorO, no arguments
(
length = 0 . 0 ;
)
// Conversion from Basic data-item to user-defined type / / InitLength is in centimeter unit
Meter( float InitLength ) / / constructorl, one argument
(
length = InitLength / 100.0; / / centimeter to meter
}
/ / Conversion from user-defined type to Basic data-item // i . e . , from meter to centimeter
operator float( ) (
float LengthCms;
LengthCms = length * 100.0; / / meter to centimeter return( LengthCms ) ;
>
void
GetLengthO (
cout « "\nEnter Length (in meters): " ; cin »
length;
)
void ShowLengthO
{
Radiant float InitRad ) // constructorl, one argument { rad = InitRad; }
float GetRadianO // Access function
{
return( rad );
J
void Input() // Read radian
{
cout << "Enter Radian:
"; cin >> rad;
}
void Output() // Display of radian
{
cout « "Radian = " « GetRadianO « endl;

};
class Degree
{
private:
float degree; // Degree
public:
Degree() // constructorO, no arguments
{
degree = 0.0;
)
// degree = radian: Conversion routine at the destination Degree( Radian
rad ) // constructorl, one-argument constructor {
 Mastering C++

degree = rad.GetRadianO * 180.0 / PI;

)
float GetDegreeO //Access function
(
return{ degree );
)
// radian = degree; conversion routine at the source operator Radian() {
// convert degree to radian and create an object radian // and then return, here
radian constructor 1 is called return( Radiant degree * PI / 180.0 J );
)
void Input() // Read degree
{
cout « "Enter Degree:
*; cin » degree;
>
void OutputO // Display output
{
cout « "Degree = " << degree << endl;

>;
Chapter 2: Moving from C to C++ 33

/* hello.c: printing Hello World message */

♦include <stdio.h> void main ( ) {
p r i n t f t -Hello World- ) ;

Run:
Hello World
The standard C library function printf () sends characters to the standard output device. The Heilo World
program will also work in C++, since it supports the ANSI-C function library. However, the program could
be rewritten using C++ streams. The C++ equivalent of the Hello World program is listed in the program
hello. cpp.

/ / hello.cpp: printing Hello World message ♦include

<iostream.h> void main{) {
cout « "Hello World";
)

Run:
Hello Wor^d
The header file iostream.h supports streams programming features by including predefined stream
objects. The C++*s stream insertion operator, « sends the message "Hello World" to the predefined console
object, cout.which in turn prints on the console. The Hello World program in C++ is shown in Figure 2.1 for
the purpose of comparative analysis.

// hello.cpp: printing Hello World message '"»< comment

♦ include <iostream.h> — preprocessor directive

void main ( ) ■■- function declarator

{ function begin

cout « "Hello world"; - body of the function main

} ' function end

Figure 2.1: Hello World program in C++

The various components of the program hello, cpp.shown in Figure 2.1, are discussed in the following
section:
First Line - Comment Line
The statement which starts with symbols // (i.e., two slash characters one after another without a spa«) is
treated as comment. Hence the compiler ignores the complete line starting from the // character pa ;r.
/ / complex8.cpp: Addition of Complex Numbers with friend feature •include <iostream.h> class
complex C

Copyrighted material
 Chapter 13: Operator Overloading

private:
float real;
f
loat
imag;
publi
c:
complex()
()
complex{ int
realpart ) (
real = realpart;
)
void
readd
ataO
(
cout « "Real Part ?
cin » real; cout «
"Imag Part ? "; cin
>> imag;
)
void outdata { char *msg ) // display complex number
{
cout « endl « msg; cout
« ■(■ « real; cout « ",
" « imag « ")";
)
friend complex operator + ( complex cl, complex c2 );
);
// note that friend keyword and scope resolution operator are not used
complex operator + ( complex cl, complex c2 )
(
complex c;
c.real = cl.real +
c2.real; c. imag =
cl.imag + c2.imag;
returnf c ) ;
)
v
o
i
d

m
a
i
n

(
)

{
complex cl, c2, c3 = 3.0;
cout << 'Enter Complexl cl..:" « endl;
cl.readdata{);
cout « "Enter Complex2 c2..:" « endl;
c2.readdata(); c3 » cl + c2;
c3.outdata( "Result of c3 = cl + c2: " ); // 2.0 is
considered as real part of complex c3 = cl + 2.0; // c3
 Mastering C++

= cl + complex(2.0) c3.outdata( "Result of c3 = cl + 2.0:

" ); // 3.0 is considered as real part of complex
complex() { }
complex( float InReal )
{
real
=
InRea
l;
imag
= 0;
}
void outdata();
friend complex operator + ( complex cl, complex c2 ) {
complex c;
c.real = cl.real +
c2.real; c.imag =
cl.imag + c2.imag;
return( c ) ;
)
friend istream & operator » ( istream &In, complex &c ) ; friend
ostream & operator « { ostream &0ut, complex &c );
};
istream & operator » ( istream & In, complex &
c ) {
cout « "Real Part ? ■;
In » c.real; // cin » c.real;
cout << "Imag Part ? ■;
In » c.imag; // cin » c.imag;
return in;
}
o«cream & operator « ( ostream &Out, complex &
c ) {
Out « '(" << c.real; // or cout « "Real = ■ « c.real;
Out « ", ' « c.imag « ")'; // cout in place of
Out return Out;
>

void
main
() {
complex cl, c2, c3 = 3;
cout « "Enter Complexl cl..:" «
endl; cin » cl;
cout « "Enter Complex2 c2..:" «
endl; cin » c2; c3 - cl + c2;
cout « -Result of c3 = cl + c2:
"; cout « c3;
// 2.0 is considered as real part of
complex c3 = cl + 2.0; // c3 = cl +
complex(2.0)
cout«endl«"Result of c3 ■ cl + 2.0: "; //c3scl + complex(2 .0) ; cout « c3;
// 3.0 is considered as real part of
complex c3 = 3.0 + c2;
cout<< endl«'Result of c3 ■ 3.0 + c2 : "; //c3=complex {3 .0) * c2 ; cout «
c3;
constructor also gets invoked when arguments are passed by value to functions and when values are returned
from functions. When an object is passed by value, the argument on which the function operates is created
using a copy constructor. If an object is passed by its address or reference, the copy constructor of course
would not be invoked, and the copies of the objects are not created. When an object is returned from a
function, the copy constructor is invoked to create a copy of the value returned by the function.

13.21 Tracing Memory Leaks

Memory fragmentation can affect program performance, but memory leaks frequently cause programs to
crash. A memory leak occurs when the user program fails to free an allocated memory block. The new
operator can be overloaded to write signature bytes for the blocks it allocates. The meaning of memory leak is
that dynamic memory being allocated (newod) without being releasing (deleted). The executable size quickly

Copyrighted material
 Chapter 13: Operator Overloading

outgrows the size of memory in the machine, requiring an undesirable amount of s vapping activity. The first
step in attacking this problem is to find where memory is being requested, used, and not returned.

Approach
In C++, it is easy to overload the built-in new and delete operators with user-supplied versions and thereby
determine when the memory is requested and to which memory location it is bounded. The program
mleak.cpp overloads new and delete operators and records the memory location to which the request is
bound, in the disk file space. raw. It also records all those bindings that are released using explicit memory
free request command.
/ / mleak.cpp: Memory leak tracing • include
<iostream.h> •include <stdio.h> ♦include
<process.h> •include <alloc.h> ♦include
<string.h>
//global information
static space_debug = 1; // space_debug switch, ON
FILE * fp_space = NULL; // file pointer to the debug info
void * operator new( size_t size )
{
void *ptr;
if( space_debug ) (
if( fp_space == NULL ) // first time call to new or delete (
// open leak debug info file which is unopened if( (fp_space = fopen( "space.raw -, "w" )) ==
NULL ) {
cout « "Error opening space.raw in write mode"; exit( 1 ) ;
)
)
The space_debug variable allows the programmer to decide whether to trace a particular portion
of code or not. When tracing is desired it must be set to a nonzero (debug ON) value. When the
following statements:
vector = (int *) new int [ 10 ]; buffer =
(char *) new char( 6 3;
are invoked in the program, the overloaded new operator allocates the requested amount of memory and
returns a pointer to the memory location to which it is bound. In addition, it records this memory address to
which it is bound, in the disk file space. raw. Similarly, the overloaded delete operator releases the memory
pointed to by the input pointer and also records the memory address in the disk file.-In the above Run, the
information recorded in space. raw file is the following:
newt 36 ) -> bd2 newt
516 ) -> bfa newt 36 )
-> e02 newt 516 ) ->
e2a newt 36 ) -> 1032
newt 516 ) -> 105a
newt 10 ) -> 1262
newt 6 ) -> 127a free
<- 1262 free <- bfa
free <- bd2 free <- e2a
free <- e02 free <-
105a .'free <- 1032
The first six requests are made by the program execution start-up routine. They can be discarded in the
memory leak tracing analysis. The seventh and eighth requests are made in the program explicitly. Similarly,
the last six memory free requests made by the system, can be discarded during analysis. These requests vary
from system to system. The first request to free memory is made by the statement
delete vector; // vector is deallocated The pointer
returned for the requests
vector = (int *) new int[ 10 ];
buffer = (char *) new chart 6 1; are the
following
newt 10 ) -> 1262
newt 6 ) -> 127a
By tracing the above allocation address information in the free list, it can be detected that new (6) pointer
address is not released, leading to memory leak. In the program it can be observed that, the memory
allocated for the variable vector is released explicitly whereas, the memory allocated for the variable buffer
is not released. It can also be noticed from the trace of memory debug information.

13.22 Niceties of Operator Overloading and Conversions

 Mastering C++

Operator overloading and data conversion features of C++ provide an opportunity to the user to redefine the
C++ language. Polymorphism feature of C++ is a bonus for the user to customize C++ to their taste. Of
course, it can be misused, since C++ does not restrict the user from misusing (exploiting)
Enter Number 2 : 10.
sum = numl + num2
Run2
Enter Number 1: £ Enter Number 2: 10. sum = numl + num2 ■

-5

In main (), the statement sum = numl + num2 ; // addition of number

is supposed to perform addition of two numbers numl and num2, but instead it performs subtraction. The
statement in the body of the overloaded operator function number operator* (. .)
sum.num = num - num2.num; // instead of addition, subtraction is done performs subtraction
instead of addition. Such neglected use of operator overloading is not taken care by the C++ compiler, but it
is the responsibility of the programmer.
As operator overloading is only a notational convenience, the language should try to prevent its misuse
(but C++ does not prevent). It is indeed said that the meaning of operators applied to standard data types cannot
be redefined. The intent is to make C+ + extensible, but not mutable. Hence, operators cannot be overloaded for
enumerations, although it would be sometimes desirable and fully sensible.

Guidelines
It is essential to follow syntax and semantic rules of the language while extending the power of C++ using
operator overloading. In fact, operator overloading feature opens up a vast vistas of opportunities for creative
programmers (for instance, new and delete can be overloaded to detect memory leaks as illustrated earlier).
The following are some guidelines that needs to be kept in mind whiie overloading any operators to support
user defined data types:
1. Retain Meaning
Overloaded operators must perform operations similar to those defined for primitive/basic data types. The
operator + can be overloaded to perform subtraction; operator * can be overloaded to perform division
operation. However, such definitions should be avoided to retain the intuitive meaning of the operators. For
example, the overloaded operator + () function operating on user-defined data-items must retain a meaning
similar to addition The operator + could perform the union operation on set data type, concatenation on
string data type, etc.

2. Retain Syntax
The syntactic characteristics and operator hierarchy cannot be changed by overloading. Therefore,
overloaded operators must be used in the same way they are used for basic data types. For example, i f cl and
c2 arc the objects of complex class, the arithmetic assignment operator in the statement
Cl + = c2;
sets cl to the sum of cl and c2. The overloaded version of any operator should do something analogous to the
standard definition of the language The above statement should perform an operation similar to the statement
cl = cl * c2;

Copyrighted material
 Chapter 14: Inheritance 507

The program union. cpp demonstrates the.mechanism of extending the Bag class by using the feature of inheritance.
In this case, a new class Set is derived from the existing class Bag without any modifications. A derived class Set
inherits all the properties of the class Bag and extends itself by adding some of its own features to support set assignment
and union operation.

/ / unfon.cpp: Union of sets. Set class by inheritance of Bag class

♦include <iostream.h>
enum boolean { FALSE, TRUE );
const int MAX_ITEMS = 25; // Maximum number of items that bag can hold
class Bag
{
protected: // Note: not private
int contents [MAX_ITEMS] ; // bag memory area
int ItemCount; // number of items present in the bag
public:
BagO // no-argument constructor
(
ItemCount = 0; // When you purchase a bag, it will be empty
)
void put( int item ) // puts item into bag
{
contents! ItemCount++ 1 = item; // item into bag,counter update
}
boolean IsEmptyO // 1, if bag is empty, 0, otherwise
{
return ItemCount == 0 ? TRUE : FALSE;
)
boolean isFullO // 1, if bag is full, 0, otherwise
{
return ItemCount == MAX_ITEMS ? TRUE : FALSE;
}
boolean IsExistt int item ); void
show();
);
// returns 1, if item is in bag, 0, otherwise
boolean Bag::IsExist( int item )
{
for( int i = 0; i < ItemCount; i++ ) if( contents[i] ==
item ) return TRUE; return FALSE;
)
// display contents of a bag
void Bag: :show()
{
for{ int i = 0; i < ItemCount; i++ )
cout « contents[ij « ■ "; cout « endl;

Copyrighted material
508 Mastering C++

class Set: public Bag (

public:
void add( int element ) {
if( !IsExist( element > && !IsFull() )
put( element );
// element does not exist in set and it is not full
)
void readt);
void operator = (Set si);
friend Set operator + ( Set si. Set s2 ) ;
); .
void Set::read() (
int element;
while( TRUE ) {
cout « "Enter Set Element <0- end>: "; cin »
element; ift element == 0 )
break; add( element
);
)
)
void Set:: operator = ( Set s2 )
(
fort int i = 0 ; i < s2. ItemCount; i++ )
contents[i] = s2.contents[i]; // access Bag::contents
ItemCount = s2.ItemCount;
)
Set operator + ( Set si. Set s2 )
{
Set temp;
temp = si; // copy all elements of set si to temp
// copy those elements of set s 2 into temp, those not exist in set si
fort int i = 0 ; i < s2.ItemCount; i++ )
{
if{ isl.IsExistt s 2 .contents [i] ) ) //if element of s2 is not in si
temp.addt s2.contents[i] ); // copy the unique element
)
return( temp );
J
void maint) {
Set slf s 2 , s 3 ; // uses no-argument constructor of Bag class cout « "Enter Set 1 elements .."
<< endl; si.readt);
cout « "Enter Set 2 elements .." « endl; s 2 . readt);
S3 = si + s 2 ;
Base Class Unchanged
It may be recalled that the base class remains unchanged even if other classes have been derived from it. In main () of
the program union. cpp, objects of type Bag could be defined as,
Bag bag; // object of the base class
Behaviors of such objects remain the same irrespective of the existence of a derived class such as Set.
It should also be noted that inheritance does not work in reverse. The base class and its objects do not know about
any classes derived from it. In the example union. cpp, the objects of the base class Bag, cannot use the function,
operator* () of the derived class Set.

Accessing Base Class Member Functions

The object s3 of class Set also uses the function show() from the base class Bag. The statement
s3.show»); // uses Bag::show{) base class in the main (), refers to the function show { 1, which does not
exist in the derived class Set. It is resolved by the compiler by selecting the member function show () defined in the base
class Bag.

14.4 Forms of Inheritance

The derived class inherits some or all the features of the base class depending on the visibility mode and level of
inheritance. Level of inheritance refers to the length of its (derived class) path from the root (top base class). A base
class itself might have been derived from other classes in the hierarchy. Inheritance is classified into the following forms
based on the levels of inheritance and interrelation *mong the classes involved in the inheritance process: « Single
Inheritance
♦ Multiple Inheritance
♦ Hierarchical Inheritance

Copyrighted material
508 Mastering C++

♦ Multilevel Inheritance
♦ Hybrid Inheritance
♦ Multipath Inheritance
The differtnt forms of inheritance relationship is depicted in Figure 14.6. The pictorial representa tion of inheritance
showing the interrelationship among the classes involved is known as the inherit ance tree or class hierarchy. Base
classes are represented at higher levels (top of the hierarchy, say >"oot) and derived classes at the bottom of the
hierarchy. The arrow directed from the derived class towards the base class, indicates that the derived class accesses
features of the base class without modifying it, but not vice versa (Some use convention of representing the arrow in the
opposite direction to indicate inherited from or derived from).
Single Inheritance: Derivation of a class from only one base class is called single inheritance. The sample program,
union.cpp, discussed above falls under this category. Figure 14.6a depicts single inheritance.
Multiple Inheritance: Derivation of a class from several (two or more) base classes is called multiple inheritance. Figure
14.6b depicts multiple inheritance.
Hierarchical Inheritance: Derivation of several classes from a single base class i.e., the traits of one class may be
inherited by more than one class, is called hierarchical inheritance. Figure 14.6c depicts hierarchical inheritance.
B
O{
cout<<"No-argument constructor of the base class B executed first\n"; )
);
class D: public B / / publicly derived class
{
public: DO {
cout«"No-argument constructor of the derived class D executed next"; )
J;
void main{ )
{
D objd; / / access both base constructor
)

Run
No-argument constructor of the base class B executed first No-argument constructor of the derived class D
executed next

5. Multiple constructors in base class and a single constructor in derived class

/ / COns5 .Cpp: multiple constructors in base and single in derived classes •include <iostream.h> class B //
base class {
public:
B O { cout « "No-argument constructor of the base class B"; } B(int a) { cout <<"One-argument
constructor of the base class B"; }
};
class D: public B / / publicly derived class {
public:
D ( int a )
{ cout << " \nOne-argument constructor of the derived class D"; }
);
void main
() {
D objdt 3 ) ;
)

Bun
No-argument constructor of the base class B One-argument
constructor of the derived class D

6. Constructor in base and derived classes without default constructor

The compiler looks for the no-argument constructor by default in the base class. If there is a constructor in the base
class, the following conditions must be met: ♦ The base class must have a no-argument constructor

Copyrighted material
 522 Mastering C++

void main ( ) (
D objd; / / base constructor

Run
No-argument constructor of a base class B 2 No-argument constructor of
the base class Bl No-argument constructor of the derived class D

The siatemenl
class D: public Bl, virtual B 2 / / public Bl, private virtual B 2
specifies thai Ihe class D is derived from the base classes Bl and B 2 . The statement
DO : B 1 ( ) . B 2 ( )
in the derived class D, specifies that, the base class constructors must be called. However, the con structors are invoked in the order
B2 ( ) , Bl, and D (), instead of the order in which base classes appear in the declaration of the derived class, since, the virtual base
class constructors are invoked first followed by an orderly invocation of constructors of other classes.

11. Constructor in multilevel inheritance

/ / Cons11.cpp: constructor in base and derived classes, order of invocation •include <iostream.h> class B // base class
{
public:
BO { cout « " \nNo-argument constructor of a base class B"; )
};
class Dl: public B // derived class
{
public:
Dl O { cout « " \nNo-argument constructor of a base class Dl"; }
);
class D2: public Dl // publicly derived class
{
public: D2 0
{ cout « ■\nNo-argument constructor of a derived class D2"; )
);
void main() {
D2 objd; // base constructor

Run
No-argument No-argument No- constructor of a base class B constructor of a base class Dl constructor of a derived class D2
argument

Copyrighted material
528 Mastering C + +

void main ()
r
i
B b( 2 , 3 ) ;
b.print();
}

Run
4211
2
The compiler converts the constructor of the class B into the following form;
B( int a, int b ) t
x = <y+b);
y = a;
)
In the above converted constructor, it should be noted that the statement
x = (y+b);
refers to the data membery which is still not initialized. Hence, the program produces the wrong result.

14.9 Overloaded Member Functions

The members of a derived class can have the same name as those defined in the base class. An object of a
derived class refers to its own functions even if they are defined in both the base class and the derived class.
The program consl4 . cpp illustrates the overloaded data and member functions in the base and derived
classes.

// cons14.cpp: overloaded members in base and derived classes •include <iostream.h> class B // base
class {
protected:
int x;
int y;
public:
B() {)
void read{)
(
cout « "X in class B ? " ; cin » x;
cout « "Y in class B ? ■ ; cin » y;
);
void showO {
cout « "X in class B ■ ■ « x << endl; cout « "Y in
class B = 1 « y << endl;
}
>;

Input Streams
The inpul streams allow to perform read operation with input devices such as keyboard, disk, etc. Input from
the standard stream is performed using the cin object. C++ uses the bit-wise right-shift operator for
performing console input operation. The syntax for standard input streams operation is as follows: cin >>
variable;
The wordcin is followed by the symbol»(extraction operator) and then with the variable, into which the input
data is to be stored. The use of cin in performing an input operation is shown in Figure 2.3.

object cin
—► extraction or get operator
variable of standard or user defined data type
cin » variable;

Variable
Figure 2.3: Input with cin operator
The following are examples of stream input operations:

Copyrighted material
528 Mastering C + +

int age; cin >>

age;
float weight; cin »
weight,
double area; cin
>> area;
char code; cout
>> code;
char name [201;
cin >> name-
Input of more than one item can also be performed using the cin input stream object. Such input operations in
C++ are called cascaded input operations. For example, reading the name of a person followed by the age, can
be performed by the cin as follows: cin >> name » age;
The cin object will read all the items from left to right, Hence, in the above case, it reads the name of the
person as a string (until first blank) first, and then the age of person into the variable age. C++ does not
impose any restrictions on the number of items to be read. The complete syntax of the standard input streams
operation is as follows:
cin >> variablel >> variable2 >> . . >> variableN; The object cin, must be associated with at least
one argument. Like scanf (), constant values cannot be sent as an argument to the cin object. Following are
some valid input statements:
14.12 Multiple Inheritance
A class can be derived by inheriting the traits of two or more base classes. Multiple inheritance refers to the
derivation of a class from several (two or more) base classes. It allows the combination of the features of
several existing, tested, and well proven classes as a starting point for defining new classes. Multiple
inheritance model is shown in Figure 14.14a and its syntax is shown in Figure 14.14b.

Base class 1 Base class 1 Base class 1

Derived class

(a) Mutiple inheritance model

derived class name
is derived from
Inheritance type

r
base class I

class DerivedClass:[VisibilityMode) BaseClassl {

/ / members of derived class
/ / and can access members of base class

(b) Syntax of mutiple inheritance Figure

14.14: Mutiple inheritance

The default visibility mode isprivate. If visibility mode is specified, it must be either public or
private. In multiple inheritance also, the inheritance of base classes with visibility mode publ ic, implies
that the public members of the base class become public members of the derived class and protected
members of the base class become protected members of the derived class. Inheritance of base classes
with visibility mode private causes both the public and protected members of the base class to become
private members of the derived class. However, in both the cases private members of the base class are
not inherited and they can be accessed through member functions of the base class.
The following declaration illustrates the concept of multiple inheritance: class D: public Bl, public
B2 / / multiple inheritance {
p-ivate:

Copyrighted material
 Chapter 14: Inheritance 152

These situations create ambiguity in deciding which of the base class's function has to be referred. This
problem is resolved using the scope resolution operator as shown in Figure 14.15. The program mul_inh4 . cpp
illustrates the same.

instance of the derived class member specifier

base class in which function is defined

Function to be invoked
in base class

ObjectName . BaseClassName : : MemberName( . . )

Figure 14.15: Syntax of handling ambiguity in multiple inheritance

/ / mul.Jnh4.cpp: overloaded functions in base classes ♦include <iostream.h> class A / / base classl
t
char ch; / / private data, default public:
A t char c )
{ ch = c; }
void showt)
{
cout << ch;
)
);
class B // base class2
{
char ch; // private data, default public:
B( char b )
f ch = b; }
void showO
{
cout « ch;
)
);
class C: public A, public B
{
char ch; / / private data, default public:
C ( char cl, char c2, char c3) : A( cl ) , B( c2 ) {
Ch = c3;

} ;

Price = 10 Notice Publication Data . . .

Title = General-Meeting Price = 100
Type of Distributor = Retail

14.13 Hierarchical Inheritance

A well established method of program design is the hierarchical model, which can be modeled better using the
concepts of inheritance. Many programming problems fall into this category. Hierarchical model follows a lop
down approach by breaking up a complex class into simpler constituent classes. In other words, in the
hierarchical model, a complex class is conceptualized as being made up of simpler classes. Figure 14.17
illustrates the hierarchical classification of vehicles in a vehicle license department respectively. Hierarchical
inheritance resembles the multilevel inheritance, in which only one derived class path is taken into
consideration.
In C++, hierarchical programs can be easily converted into class hierarchies. The superclass (base class)
includes the features that are common to all the subclasses (derived classes). A subclass is created by inheriting
the properties of the base class and adding some of its own features. The subclass can serve as a superclass for
the lower level classes again and so on. The program vehicle.cpp models the class hierarchy in C++ for the
problem shown in Figure 14.17.

Vehicle

T
Light Motor Heavy Motor

Copyrighted material
 Chapter 14: Inheritance 153

I I
Gear Motor Non Gear Motor Passenger Goods

Figure 14.17: Classification of vehicles

/ / vehlcle.cpp: Vehicle Database Hierarchical Model ♦include

<iostream.h>
const MAX__LEN= 25; / / length of string class
Vehicle
{
protected:
char name[ MAX_LEN] ; / / name of the vehicle
int WheelsCount; / / number of wheels to vehicle
public:
void GetDataO

cout << "Name of the Vehicle ?

class Goods: public

HeavyMotor (
public:
void
GetData()
{
HeavyMotor::GetData( ) ;
)
void DisplayData( )
(
HeavyMotor::DisplayData( ) ;
)
);
void
main
() {
GearMotor vehil;
Passenger vehi2;
/ / read vehicle
data
cout « "Enter Data for Gear Motor Vehicle . . . " « endl; vehil.GetData( ) ;
cout « "Enter Data for Passenger Motor Vehicle . . . " « endl; vehi2.GetData{ ) ; ■'/
display vehicle data
cout « ."Data of Gear Motor Vehicle . . . " « endl;
vehil.DisplayData( ) ;
cout « "Data of Passenger Motor Vehicle . . . " << endl; vehi2.DisplayData( ) ;
)
Run
Enter Data for Gear Motor Vehicle . . .
Name of the Vehicle ? Maruti-Car
Wheels ? A Speed Limit ? 4 No. of
Gears ? 5
Enter Data for Passenger Motor Vehicle . . .
Name of the Vehicle ? KSRTC-BUS Wheels ?
£
Load Carrying Capacity ? £0.
Permit Type ? National
Maximum Seats ? AS.
Maximum Standing ?
Data of Gear Motor Vehicle . . .
Name of the Vehicle : Maruti-Car
Wheels : 4
Speed LimiC : 4
Gears: 5
Data of Passenger Motor Vehicle . . .
Name of the Vehicle : KSRTC-
BUS /.'heels : 4
)

Copyrighted material
 Chapter 14: Inheritance 154

'Load Carrying Capacity : 60

class B2: public virtual A
I
/ / body of class B2
J;

class D: public Bl, public B2 ■'; body of

class D

Consider the statement

objd.funci) ;
where objd is the object of class D and invokes the func ( ) defined in the class A. If the keyword virtual does
not exist in the declaration of classes Bl and B2, a call to func {) leads to the following compilation error:
Error: Member is ambiguous: ' A : : func' and ' A : : func' C++ takes necessary care to see that only
one copy of the class is inherited, when a class is inherited as virtual irrespective of the number of paths that
exist between the virtual base class and the derived class. The keywords virtual and public or protected may be
used in any order.
Consider the processing of the result of student depicted in the Figure 14.19. In this case, the result class is
derived from the classes internalExam and ExternalExam, which are derived classes of the common class
student. The program int_ext.cpp implements the concepts of virtual classes.

student T
Internal Exam ii External Exam
i A
•

Result

Figure 14.19: Virtual base classes

/■/ int_ext.cpp: Student result based on internal and external marks

// Multipath Inheritance with virtual classes
♦include <iostream.h>
const int MAX_LEN = 25; / / maximum length of name
class student
{
protected:
int RollNo; / / student roll number in a class
char branch[20); / / branch or subject student is studying public:
void ReadStudentData( ) {
cout « 'Roll Number ? " ;

Copyrighted material
 S60 Mastering C++

char branch(20]; // branch or subject student is studying public:

void ReadDataO {
cout « "Roll Number ? "; cin >>
RollNo;
cout « "Branch Studying ? "; cin »
branch;
)
void DisplayData() {
cout « "Roll Number: " « RollNo « endl; cout <<
"Branch: ' « branch « endl;
>

class exam: public student {

protected:
int SublMarks;
int Sub2Marks;
public:
void ReadData()
{
cout « "Marks Scored in Subject 1 < Max:100> ? "; cin » SublMarks;
cout << "Marks Scored in Subject 2 < Max:100> ? ";
cin » Sub2Marks;
} *
void DisplayData() {
student::DisplayData{); // uses DisplayData() of student class cout « "Marks Scored in
Subject 1: " « SublMarks « endl; cout << "Marks Scored in Subject 2: " << Sub2Marks <<
endl; cout « "Total.Marks Scored: " « TotalMarksO « endl;
}
int TotalMarksO {
return SublMarks + Sub2Marks;
}
};
class result: public exam, public sports {
private:
int total;
public:
void ReadData()
{
ReadPersonO; / / arn*»*s nprann r-\a<i>s member
student::ReadData();

Copyrighted material
562 Mastering C + +

14.16 Object Composition-Delegation

Most of us understand concepts such as objects, interfaces, classes, and inheritance. The challenge lies in
applying them to build flexible and reusable software. The two most common techniques for reusing
functionality in object-oriented systems are class inheritance and object composition. As explained, inheritance
is a mechanism of building a new class by deriving certain properties from other classes. In inheritance, if the
class D is derived from the class B, it is said that D is a kind of B\ the class D has all the properties of B in
addition to the features of its own.
A commonly recurring situation is one where objects are used as data members in a class. The use of
objects in a class as data members is referred to as object composition. Object composition is an alternative to
class inheritance. Here, new functionality is obtained by assembling or composing objects to support more
powerful functionality. This new approach takes a view that an object can be a collection of many other
objects and the relationship is called ahas-a relationship or containership. In OOP, the has-a relationship occurs
when an object of one class is contained in another class as a data member. In other words, a class can
contain objects of other classes as its members (see Figure 14.21).

class B { class B

};

class D class B

B obj b; B objb;

};

Figure 14.21: Object composition

In the case of inheritance (kind of relationship), the constructors of base class are first invoked before the
constructor of the derived class. Whereas, in the case of has-a relationship, the constructor of the class D is
invoked first and then the object of B is created. The concept of creating the member objects first using
respective member constructors and then the other ordinary members can also be accomplished in has-a
relationship by using an initialization-list in the constructor of the nested class. Consider the following class
declarations class B (
/ / body of a class
};
class D {
B ObjectB; / / b is a object of class B
public:
D( arg-list ) : ObjectBt arg-listl );
};
The classes publication and sales have the same declaration as in inheritance relation but they are used in
a different way by the book class. Although containcrship is an alternative to inheritance and offers the
functionality of inheritance, it does not provide flexibility of ownership. Inheritance relationship is simpler to
implement and offers a clearer conceptual framework.

/ / publish2.cpp: Publication, Sales details. Objects inside a class •include <iostream.h>

class publication / / base class, appears as abstract class {
private:
char title[401; float // name of the publication work // price of a
price; publication
public:
void getdata()
{
cout « "\tEnter
Title: cin » title;
cout « "\tEnter
Price: '; cin »
price;
)

Copyrighted material
562 Mastering C + +

void
displ
ay()
(
cout << "\tTitle = " << title <<
endl; cout << "\tPrice = " «
price « endl;
)
);
class sales //
base class {
private:
float PublishSales[3]; //sales of publication for the last 3 months
public:
void
getdat
a();
void
displa
y{);

void sales::getdata()
{
int i ,-
for( i = 0; i < 3;
i++ J {
cout << "\tEnter Sales of " << i+1 « "
Month: "; cin » PublishSales[i];
)
}
void
sales::display(
) {
int i;
int TotalSales = 0;
for( i = 0; i < 3;
i++ ) (
cout<<"\tSales of " << i+1 «" Month = " « PublishSales[il « endl:
Run
Enter Principle Amount: 1Q0Q Enter
Time (in years): 2 Enter Rate of Interest:
£ Simple Interest = 100 Total Amount =
1100

2.5 Literals-Constant Qualifiers

Literals arc constants, to which symbolic names arc associated for the purpose of readability and case of
handling standard constant values. C++ provides the following three ways of defining constants:
♦ # define preprocessor directive
♦ enumerated data types
♦ const keyword
The variables in C can be created and initialized with a constant value at the point of its definition. For
instance, the statement
float PI = 3.1452;
defincs a variable named PI, and is assigned with the floating-point numeric constant value 3.1452. It is
known that the constant value does not change. In the above case, the variable PI is considered as a constant,
whose value does not change throughout the life of the program (complete execution-time). However, an
accidental change of the value of the variable PI is not restricted by C. C++ overcomes this by supporting a
new constant qualifier for defining a variable, whose value cannot be changed once it is assigned with a value
at the time of variable definition. The qualifier used in C++ to define such variables is the const qualifier. The

Copyrighted material
562 Mastering C + +

syntax of defining variables with the constant qualifier is shown in Figure 2.5. Note that if DataType is
ommitled. it is considered as int by default.

v
standard or user defined data type: char, short, int.
float, etc.
numeric constant must be of type DataType as
specified
const [DataType] VariableName = ConstantValue;

Figure 2.5: Syntax of constant variable definition

The following examples illustrate the declaration of the constant variables:

♦ const float PI = 3.1452;
♦ const int TRUE = 1;
♦ const int FALCE = 0;
♦ const char *book_name = "OOPs with C++";
The program area. cpp, illustrates the declaration and the use of constant variables.
area.cpp: area of a circle
•include <iostream.h> const float PI
= 3.1452;

Copyrighted material
15
Virtual Functions

15.1 Introduction
Polymorphism in biology means the ability of an organism to assume a variety of forms. In C++, it
indicates the foim of a member function that can be changed at runtime. Such member functions are
called virtual functions and the corresponding class is called polymorphic class. The objects of the
polymorphic class, addressed by pointers, change at runtime and respond differently for the same
message. Such a mechanism requires postponement of binding of a function call to the member
function (declared asvirtual) until runtime.

Polymorphis
m
______________

Compile Time Run-Time

Function Operator Overloading Virtual
Overloading Functions

Figure 15.1: Types of polymorphism in C++

It has been observed that, function overloading and operator overloading features of C++ have
allowed to realize polymorphism. Yet, there is another mechanism to implement polymorphism in
C++; through dynamic binding. Figure 15.1, illustrates the taxonomy of polymorphism in C++.
Function overloading is realized by invoking a suitable function whose signature matches with the
arguments specified in the function call statement. Operator overloading is realized by allowing
operators to operate on the user defined data-typtc with the same interface as that of the standard
data types. In both cases, the compiler is aware of the complete information regarding the type and
number of operands. Hence, it is possible for the compiler to select a suitable function at the
compile-time.

Copyrighted material
572 Mastering C++

Run
Father name: Eshwarappa
Father name: Eshwarappa
In main(). the statement
Father *fp; / / pointer to the Father class's objects; defines a pointer variable f p. The statement
fp ■ &fl; / / f p points to Father class object assigns the address of the object f 1 of the class Father
to fp. After this, when statement such as
fp->show<); / / display father showt) function is executed, the member function defined in the class Father
is invoked. In C++ base class pointer is fully type-compatible with its derived class pointers and hence, the
statement such as
f p = &sl; / / valid assignment is valid. It assigns the address of the object si of the class Son to fp.
After this, when statement
fp->show(); / / guess what is the output ? Father or Son! is executed, (it is interesting to note that) it still
invokes the member function show ( ) defined in the class Father!
There must be a provision to use the member function show { ) to display the state of objects of both the
Father and Son classes using the same interface. This decision cannot be taken by the compiler, since the
prototype is identical in both the cases.
In C++, a function call can be bound to the actual function either at compile time or at runtime.
Resolving a function call at compile time is known as compile-time or early or static binding whereas, resolving a
function call at runtime is known as runtime or late oxdynamic binding. Runtime polymorphism allows to
postpone the decision of selecting the suitable member functions until runtime. In C++, this is achieved by
using virtual functions.
Virtual functions allow programmers to declare functions in a base class, which can be defined in each
derived class. A pointer to an object of a base class can also point to the objects of its derived classes. In this
case, a member function to be invoked depends on the the class's object to which the pointer is pointing. When
a call to any object is made using the same interface (irrespective of object to which pointer variable is
pointing), the function relevant to that object will be selected at run time. The program parent2 . cpp illustrates
the effect of virtual functions on an overloaded function in a class hierarchy.

/ / parentl.cpp: base class pointer and virtual function #include

<iostream.h> ♦include <string.h> class Father
{
char name [201; / / father name public:
Father{ char *fname )
{
strcpyt name, fname ) ; / / fname contains Father's name

Copyrighted material
 574 Mastering C++

15.3 Pointer to Derived Class Objects

The concept of derived classes specifies hierarchical relationship between various objects and ex -
presses the commonality between them. The properties common to different classes are placed at the
top of the hierarchy, which becomes the base class, and all other classes are derived from this base
class. A derived class is often said to inherit properties of the base class, and so, this relationship is
known as the inheritance relationship.
Pointers can be used with the objects of base classes or derived classes. Pointer to objects of a base
class are type-compatible with pointers to objects of a derived class, thus allowing a single pointer
variable to be used as pointer to objects of a base class and its derived classes. For instance, in the
above declaration having the classes Parent and Child, a pointer declared as a pointer to Parent objects
can also be used as a pointer to Child objects. C++ makes polymorphism possible through a rule that
one should memorize: a base class pointer may address an object of its own class or an object of any class
derived from the base class. (See Figure 15.2.)

e base class or objects of any derived class

Object of base class

nter Object of derived class

Derived class
pointer
Object of derived class

A derived class pointer may address objects of the derived class, but not objects of the base class

Figure 15.2: A base class pointer may address a derived class object

Consider the following definitions to illustrate type compatibility of pointers.

Father *basep; / / pointer to Parent class
Father f ; / / base class object
Son s; / / derived class object
The statement
basep = &f ;
assigns the address of the object f of the class Father to the pointer variable basep. The statement basep
= &s;
assigns the address of the object s of the class Son to the pointer variable basep. Such an assignment is
perfectly valid in C++, since the pointer to an object of a base class is fully type-compatible with pointer
to objects of its derived classes.
The use of a pointer to the objects of a base class with the objects of its derived class raises a new
problem. It does not allow access even to public members of a derived class. That is, it allows access
only to those members inherited from the base class but not to the members which are defined in the
derived class. Even in case, any member of the Son class has the same name as one of the members of
Chapter 2: Moving from C to C++ 43

void
main{ )
{
float
radius;
float
area;
cout < < "Enter Radius of Circle: " ;
cin » radius;
area = PI * radius * radius;
cout << "Area of Circle = " < < area;
)

Run
Enter Radius of Circle: 2
Area of Circle = 12.5808

Copyrighted material
 574 Mastering C++

In the above program, the use of the statement such as

PI = 2 . 3 ;
to modify a constant type variable leads to the compilation error: Cannot modify a const object
Thus, the keyword const, can be used before a type to indicate that the variable declared is
constant, and may therefore not appear on the left side of the assignment (=) operator. In C++,the const
qualifier can be used to indicate the parameters that are to be treated as read-only in the function body.
Consider the C program disp. c, having the function to display any string passed to it.

/ " dlsp.c: display message in C */

#include <stdio^h>
♦include <string.h>
void display{ char *msg )
{
printf{ "%s", msg ) ;
/* modify the message * /
strcpy( msg, "Misuse* ) ;
)
void
main { )
{
char string[15];
strcpyt string, "Hello World" J ;
display! string ),-printf( "\n%s",
string ) ;
>

Run
Hello
World
Misuse
The function display ( ) , is supposed to output the input string argument passed to it onto the
console. But accidental use of a statement such as
strcpyt msg, "Misuse" ) ; in display ( ) modifies the input argument. This modification is
also reflected in the calling function; (.;<*e the second message in the output) the string argument is a
pointer type and any modification in function will also be reflected in the calling function. Such
accidental errors can be avoided by defining

Base class pointer

Ok
Ok
Base class
Base class data members

X
Not allowed!
Pointer to
derived class
A

Figure 15.4: A base class pointer can address data members inherited by a derived class

15.4 Definition of Virtual Functions

Copyrighted material
 574 Mastering C++

C++ provides a solution to invoke the exact version of the member function, which has to be decided at
runtime using virtual functions. They are the means by which functions of the base class can be
overridden by the functions of the derived class. The keyword virtual provides a mechanism for defining
the virtual functions. When declaring the base class member functions, the keyword vir-tua 1 is used
with those functions, which are to be bound dynamically. The syntax of defining a virtual function in a
class is shown in Figure 15.5.
class MyClass {
public:
keyword

virtual ReturoTypo Funct ionName ( argument s )

Figure 15.5: Syntax of virtual function

Virtual functions should be defined in the public section of a class to realize its full potential benefits.
When such a declaration is made, it allows to decide which function to be used at runtime, based on the
type of object, pointed to by the base pointer, rather than the type of the pointer. The program f
amily2 .cpp illustrates the use of base pointer to point to different objects for executing different
implementations of the virtual functions.
{
cout < < "point" < < endl;
)
);
class line: public graphics
{
public:
v
o
i
d

d
r
a
w
(
)

{
cout < < "line" « endl;
)
);
class triangle: public graphics <
public:
void drawt)
{
cout << "triangle" < < endl;
)
);
class rectangle: public graphics {
public:
v
o
i
d

d
r
a
w

Copyrighted material
 574 Mastering C++

{
)
(
cout « "rectangle" « endl;
)
);
class c i r c l e : public
graphics {
public:
v
o
i
d

d
r
a
w

f
)
(
cout « "circle" « endl;
>

);
void main ( )
{
graphics
point_obj; line
line_obj;
triangle tri_obj;
rectangle
rect_obj; circle
circle_obj;
graphics
*basep[] = {
&point_obj, &line_obj, &tri_obj,
&rect_obj, &circle_obj
);
cout « 'Following figures are drawn with basep [ i ] - > d r a w { ) . . . " « endl; f o r { int i = 0; i < 5;
i++ ) basep[i]->draw( ) ;
Chapter 15: Virtual Functions 58:-

Run
Following figures are drawn with basep[i]->draw{) . . .
point
line
triangle
rectangle
circle
In main (}, the statement
fort int i = 0; i < 5; i+
+ ) basep[i]->draw();
invokes a different draw {) version based on the object to which the current pointer basep [ i J is
pointing. (See Figure 15.7.)
graphics Parent class
draw()
draw{)
drawn code in parent

child classes

line draw C)

Copyrighted material
 574 Mastering C++

rectangle
triangle j circle
draw(}
draw{) draw()

draw() drawO drawn ^ draw{)

draw() code in
child

graphics * basep[] = &tri_obj;

{ &line_obj.
&rect_obj, &cir_obj );
for t i=0; i < NO_PICTURES; i++J
basepfij -> drawO ;

without virtual function

with virtual function
Figure 15.7: Compile time
and runtime binding of
functions

Chapter 15: Virtual Functions 595

15.9 What are virtual destructors ? How do they differ from normal destructors ? Can constructors
I>e declared as virtual constructors ? Give reasons.
15.10 Explain how dynamic binding is achieved by the C++ compilers. What is the size of the
following classes:
class X
(
int x;
public:
void read( ) ;
);
class Y
{
int a ;
public:
virtual void readO ;
);
class Z
{
int a;
public:
virtual void read{ > ;
virtual void show() ,\
};

15.11 What are the rules that need to be kept in mind in deciding virtual functions ?
15.12 Correct the errors in the following program and include missing components:
class ABC
{
int a;
publie:
);
void main{)
{
ABC.al;
al.readt);
al.showO ;
ABC a2 =
10;
a 2 . showd ;
)

Copyrighted material
 574 Mastering C++

15.13 Consider an example of book shop which sells books and video tapes. These two classes arc
inherited from the base class called media. Themedia class has command data members such
as title and publication. The book class has data members for storing number of pages
in a book and the tape class has the playing time in a tape. Each class will have member
functions such as read ( ) and show ( ) . In the base class, these members have to be defined
as virtual functions. Write a program which models the class hierarchy for book shop and
processes objects of these classes using pointers to the base class.
Chapter 2: Moving from C to C++ 45

/ / global.cpp: global variables access through scope resolution operator •include <iostream.h> int num
= 2 0 ; void mainO

int num = 10; / / local variable / / global variable num+num; //

cout << "Local = " « num; cout << "\nGlobal both local & global use
= " « ::num; cout < < "\nGlobal+Local = " <<

Bun
Local = 1 0
Global = 20
Global+Local =
30
The program loop. cpp illustrates the accessing of local and global variables within a f o r loop. It
also shows mixing of the single-line comment statement within a single executable statement.
// (OOp.cpp: local and global variables in a loop
♦ include <iostream.h> int
counter = 50; int main ( ) global variable
{ register int // local variable this refers to the
counter; local variable
for(counter
= 1;
counter < 10;
counter**}

cout < < endl «

: : counter // / / print new line followed by global variable
/ divided by local variable
counter;
)
return{ 0 ) ;
)

Run
50
25
16
12
10
8
7
6
5

2.7 Variable Definition at the Point of Use

In C, local variables can only be defined at the top of a function, or at the beginning of a nested block.
In C++, local variables can be created at any position in the code, even between statements. Further t =
x; x = y; y = t;
)
void swapt float & x, float & y ) / / b y
reference (
float t; / / temporary variable used in
swapping t = x; x = y; y = t;
)
v
o
i
d

Copyrighted material
 574 Mastering C++

m
a
i
n

{
)
{
char chl, ch2;
cout « "Enter two Characters <chl,
ch2>: " ; cin >> chl >> ch2;
swapt chl, ch2 ) ; / / compiler invokes swapt char &a, char &b ) ; cout « "On
swapping <chl, ch2>: 1 « chl « 1 " « ch2 « endl; int a, b;
cout « "Enter two integers
<a, b>: cin » a >> b;
swapt a, b ) ; / / compiler invokes swapt int &a, int &b ) ; cout « "On
swapping <a, b>: ■ « a « ■ " « b « endl; float c , d;
cout « "Enter two floats <c, d>: ■;
cin » c » d;
swapt c, d ) ; / / compiler invokes swapt float &a, float &b ) ; cout << "On
swapping <c, d>: 1 << c << ■ 1 << d;
}

Run
Enter two Characters <chl, ch2>: E K
On swapping <chl, ch2>: K R
Enter two integers <a, b>: 5 10
On swapping <a, b>: 10 5
Enter two floats <c, d>: 20.5 99.5
On swapping <c, d>: 99.5 20.5
The above program has three
swap functions void swapt
char & x, char & y ) ;
void swapt int & x, int & y
) ; void swapt float & x,
float & y ) ;
whose logic of swapping is same and differs only in terms of data-type. Such functions
can be declared as a single function template without redefining them for each and
every data type. The C++ template feature enables substitution of a single piece of code
for all these overloaded functions with a single template function as follows: template
<class T>
void swapt T & x , T fit y ) // by reference
{
T t; / / template type temporary variable used in swapping
/ / max with integer data types int
a , b, c ;
cout < < "Enter two integers < a , b > ; " ; cin
>> a > > b; c = max( a, b ) ;
cout < < "max( a , b ) : " << c << endl; / /
max w i t h string data types char
strl[20], str2[20J;
cout << "Enter two strings < s t r l , s t r 2 > : " ; cin
» strl » s t r 2 ;
cout « "max! s t r l , str2 ) : " « max{ s t r l , str2 ) ;
)

Run
Enter two characters <chl, ch2>: A. £
max( chl, ch2 ) : Z Enter two integers
< a , b > : ii £ max{ a , b ) : 6
Enter two strings < s t r l , s t r 2 > : Teiaswi Raikumar max(
s t r l , s t r 2 ) : Tejaswi
In main (J, the statement
cout « "max( s t r l , str2 ) : ■ « maxl s t r l , str2 ) ; has the expression,
max( strl, str2 )

Copyrighted material
 574 Mastering C++

The compiler selects the user-defined normal function instead of creating a new function, since the
function call is matching with the user-defined function.

Bubble Sort Function Template

Sorting is the most commonly used operation particularly in data processing applications. These appli-
cations require function to sort data elements of different data types. Such functions can be declared as
function template and can be used to sort data items of any type. The program bsort. cpp illustrates the
declaration of function template for bubble sort and its use on integer and floating point data types.

/ / bsort.cpp: template functions f o r bubble-sort

♦include <iostream.h>
enum boolean { f a l s e , true } ;
template <class T>
void swap( T & x , T & y ) / / b y reference {
T t; / / template type temporary variable used in swapping t = x,-x = y ; y = t ;
>
template< class T >
void BubbleSort( T & SortData, i n t Size ) boolean

swapped = true;
Enter the elements of the float vector...

Sorted Vector: 3 . 2
8.5 8.9
In main ( ) , when the compiler encounters the statement y/
BubbleSort ( IntNums, size ) ; it creates the bubble sort function internally for sorting
integer numbers; the parameter IntNums is of type integer. Similarly, when the compiler
encounters the statement
BubbleSort( FloatNums, size ) ; it creates bubble sort function internally for sorting floating
point numbers. The same template function can be used to sort any other data types. Note that
the compiler creates a function internally for a particular data type only once and if there are
more requests with the same data type, the compiler accesses the old internally created function.
Usage of Template Arguments
Every template-argument specified in the template-argument-list must be used as a generic data
type for the definition of the formal parameters. If any of the generic data type is not used in the
definition of formal parameters, such function templates are treated as invalid templates. The
use of partial number of generic data types in a function defining formal parameters is also
treated as an error. All the formal parameters need not be of generic type. The following sections
show some function templates which are invalid declarations:
1. No-argument template function
template < class T >
T popt void ) / / error: T is not used as an argument
{
return *--Stack;
)
2. Template-type argument unused
template < class T >
void test ( int x ) / / error: T is not used as an argument {
T temp;
/ / . . test stuff
}
3. Usage of Partial number of template arguments
template< class T, class U >
void insert ( T & x ) / / error: U is not used in the argument
-{
U lPtr;
/ / . . test stuff
)
The template argument U is not used in argument types, and hence, the compiler reports an
error.
S2.y = b;
)
void main( void ) {
A SI; B
S2;
Assign_A{ 3, 4, SI ) ;

Copyrighted material
 574 Mastering C++

Assign_B( 3, 3.1415, S2 );//Error: no match for Assign_B(int,double,B ) *

>

In main ( ) , the statement

Assign_B{ 3 , 3.1415, S2 ) ; leads to compilation errors, since the above program is neither having
the normal function nor function template matching with its parameters data types. Both the templates
expect the first two parameters to be of the same data-type and none of them matches the above call. The
solution to the problem encountered in the above program is to declare the second template function in the
above program as follows:
template < class T, class U > void Assign_B{ T a,
U b, B & S2 ) {
S2. x * a;
S 2 . y = b;
)
The declaration of the function template is the same, except that it has an extra argument in the template-
argument-list, i.e., class u. This declaration informs the compiler that the template function A s s i g n _ B ( )
with two arguments should be instantiated. The compiler calls the appropriate instantiation. Any number of
generic data types can be declared, provided all these generic data types are used in declaring formal
parameters. The function Assign_A can also be declared as follows:
template< class T , class U > void Assign_A( T a ,
U b , A & S2 > {
Sl.x = a;
S l . y = b;
)
Since the dummy arguments T and u are same in the function call Assign_A, i t would be better to define the
function template with a single dummy argument rather than two dummy arguments.
All template arguments for a function template must be of template type-arguments, otherwise, it leads to
an error. For instance, the following declaration,
template< c l a s s T , unsigned SIZE >
void BubbleSort( T & Data, unsigned SIZE )
{
//
//
)
is not allowed. However, such declarations are allowed with class templates.

16.6 User Defined Template Arguments

In addition to primitive data-types, user defined types can be passed to function templates. Its
declaration is same as the function template processing standard data types as illustrated in the
program
student.cpp.

/ / student.cpp: student record and template with user defined data types •include
<iostream.h> struct stuRec
{
cha
r

n
a
m
e
[
3
0
J
;
i
n
t
a
g
e
;
char collegeCode;
);
template
<class

Copyrighted material
 574 Mastering C++

T> void
Display(
T& t ) {
cout « t « endl;

ostreamfic operator « ( ostream & out, stuRec & s )

{
out « "Name: ■ « s.name «
endl; out « "Age : ' « s.age «
endl;
out « "College Code: " « s.collegeCode « endl;
return out;
)
void
mai
n( v
oid )
{
stuRec si;
cout << "Enter student record details..." << endl;
cout « "Name: "; cin » sl.name; cout « "Age :
"; cin >> si.age;
cout « "College Code: "; cin » si.collegeCode;
cout << "The student record:" << endl; cout
« "Name: ■; Display( sl.name J ; cout «
"Age : ' ; Display( si.age ) ; cout «
"College Code: ";
Displayt si.collegeCode ) ; // it in turn calls operator « defined above cout « "The
student record:" « endl; Displayt si ) ;
}

Run
Enter student record details...
Name: Chinamma
Age : JJL
College Code: h
The student record:
Name: Chinamma
Age : 18
public:
DbleStackO ;
void Push( const double & element ) ; double
Pop( void ) ;
unsigned int GetSize( void ) const;
);
As seen in the above three declarations, a separate stack class is required for each and every
data type. Template declaration enables substitution of code for all the three declarations of
stacks with a single template class as follows:
template < class T >
class DataStack
{
T array[25); / / declare a stack of 25 elements of data type T
unsigned int top;
public:
DataStack( ) ;
void Pusht const T & element ) ; T
Pop( void ) ;
unsigned int GetSizet void ) const;
);
The syntax of declaring class templates and defining objects using the same is shown in Figure
16.2. The definition of a class template implies defining template data and member functions.

keyword template datatypes Tl, T2(...

template <class Tl, class T2, . . . >

class ClassName
{

Copyrighted material
 574 Mastering C++

/ / data items of template type Tl, T2, . . . Tl

datal;
/ / functions of template arguments Tl, T2, void
fund (Tl a, T2 &b) ;
T func2 (T2 *x, T2 *y);

Figure 16.2: Syntax of class template declaration

The prefix template <class T> specifics that a template is being declared, and that a type-name T
will be used in the declaration. In other words, DataStack is a parameterized class with T as its
generic data type.
Any call to the template functions and classes, needs to be associated with a data type or a
class. The compiler then instantiates a copy of the template function or template class for the
data type specified. The syntax for creating objects using the class templates is shown in Figure
16.3.
A statement to create an object of type stack that can hold integers is as follows: DataStack
<int> stack_int; / / stack of integers
datatype to be
substituted for template
datatype

ClassName <char> objectl;

ClassName <int> object2;
ClassName <some_other_class> objects,-

Figure 16.3: Syntax lor class template instantiation

Similarly, objects those hold characters, floats, and doubles can be created by the following statements:
DataStack <char> stack_char; DataStack <float> / / stack of characters / / stack of
stack_float; floats
DataStack <double> stack_double; / / stack of doubles However, the usage of these stack
objects is similar to those of normal objects. For instance, to push the integer value 10 into the
stack_int object, the following statement is used:
stack_int.push( 10 ) ;

Template Arguments
A template can have character strings, function names, and constant expressions in addition to tem plate
type arguments. Consider the following class template to illustrate, how the compiler handles the
creation of objects using class templates:
template <class T, int size>
class myclass
{
T arr[size];
);
The value supplied for a non template type argument must be a constant expression; its value must be
known at compile time. When the objects of the class template are created using a statement such as
myclass <float,10> newl; the compiler
creates the following class: class
myclass {
f l o a t arr[10] ;
);
Again if a statement such as,
myclass < i n t , 5> new2; is encountered for creating the object new2, the compiler creates the
following class:
class myclass
{
int arr[ 5 ] ;
);

Member Function Templates

A member function of a template class is implicitly treated as a template function and it can have
template <class T> void vector<T>::show() (
f o r ( int i = 0; i < size; i++ ) cout «
elem( i ) << " , ";
)

Copyrighted material
 574 Mastering C++

void main( )

int i;
vector <int> int_vect( 5 ) ; vector
<float> float_vect( 4 ) ; f o r ( i = 0; i
< 5; i + + )
int_vect.elem( i ) = i + 1 ; f o r ( i
= 0; i < 4; i + + )
f loat_veot.elem( i } = floatt i + 1 . 5 ) ; cout «
"Integer Vector: "; int_vect.show( ) ;
cout << endl « "Floating Vector: ";
f loat_vect.s h o w ( ) ;
J
Run
Integer Vector: 1 , 2 , 3 , 4 , 5 , Floating
Vector: 1 . 5 , 2 . 5 , 3 . 5 , 4 . 5 ,
Note that the class template specification is very much similar to the ordinary class
specification except for the prefix,
template <class T>
and the use of T in the place of data-type. This prefix informs the compiler that the class
declaration following it is a template and uses T as a type name in the declaration. The type T
may be substituted by any data type including the user defined types. In main { ) , the statement,
vector <int> int_vect{ 5 ) ;
vector <float> float_vect( 4 ) ; creates the vector objects int_vect and f loat_vect to hold
vectors of type integer and floating point respectively. Once objects of class template are created,
the use of those objects is the same as the non-template class objects.
Class Template with Multiple Arguments
The template class is instantiated by specifying predefined data type or user defined data classes.
The data type is specified in angular braces <>. The syntax for instantiating class template is as
follows: TemplateClassName < type > instance;
TemplateClassName < typel, type2 > instance( arguments ) ; The instantiation specifies
the objects of specified data type. If a different data type is to be specified, a new declaration
statement must be used.
The declaration of template classes with multiple arguments is similar to the function
template with multiple arguments. However, the arguments need not be of template type. These
may include character strings, addresses of objects and functions with external linkage, static
class members, and constant expressions. Consider the following declaration:
The member functions defined w i t h its class body have the same syntax as members of non-template-
typc classes. However, member function defined outside the body of a class, for instance, has the
following specification:
template <ciass T>
void
Veccor<7>::read(;
{
body of the read( I
)
Note that, the member functions of a class-template arc treated as function-template type members. The
class Vector can be instantiated as follows: Vector < i n t > vl;
In this case, the i n t specified in angular bracket is first assigned to generic data lypc in the Vector
class and then the same is also passed to its base class.
A derived class of a template based base class is not necessarily template derived class. That is. the non-
tcmplate-bascd derived classes can also be created from the template-based base classes. In this case,
the undefined template argument T has to be specified during derivation, for instance, as follows: class
Vector : public sVector< i n t >
{

>;
It creates a new class called Vector from the template-based base class sVector. The i n t is passed as
template argument type to the base class.
The program union. cpp illustrates the mechanism of extending the class template Bag by using the
feature of inheritance. In this case, a new class template S e t is derived from the existing class template
Bag without any modifications. A derived class template Set inherits all the properties of the class
template Bag and extends itself by adding some more features of its own to support set assignment and
union operation.

Copyrighted material
 574 Mastering C++

/ / unlon.cpp: Union of s e t s . Set class by inheritance o f Bag class

#include < i o s t r e a m . h >
enum boolean { FALSE, TRUE ) ;
const i n t MAX_ITEMS = 2 5 ; / / Maximum number o f items that the bag can hold
template <class T>
class Bag
{
protected: / / Note: not private
T contents[MAX_1TEMS]; / / bag memory area
i n t ItemCount; / / Number o f items present in the bag
public:
BagO / / no-argument constructor
{
ItemCount = 0 ; / / When you purchase a bag, i t w i l l be empty

Copyrighted material
 Chapter 16; Generic Programming with Templates 174

cout « "Enter Set Element <0- end>: " ; cin »

element; i f f element == 0 )
break; add( element
);
)
)
template <class S>
void Set<S>: : operator = ( Set <S> s2 ) {
for{ int i = 0; i < s2.ItemCount; i++ }
contents [ i ] = s2.contents[ i ] ;
ItemCount = s2.ItemCount;
)
template <class S>
Set<S> operator + ( Set <S> s i , Set <S> s2 ) (
Set <S> temp;
temp = si; / / copy all elements o f set si to temp
/ / copy those elements o f set s2 into temp, those not exist in set s i
f o r ( int i = 0 ; i < s 2 .ItemCount; i + + )
{
i f ( ! s l . I s E x i s t { s 2 .contents [ i ] ) ) / / i f element of s 2 is not in si
temp.addt s2.contents[i ] ) ; / / copy the unique element
}
return ( temp .) ;
)
void main()
{
Set <int> s i ; Set
<int> s 2 ; Set
<int> s 3 ;
cout « "Enter Set 1 elements . . " « endl; s i . read( ) ,-
cout < < "Enter Set 2 elements . . " « endl; s 2 . r e a d ( ) ;
s3 = s i + s 2 ;
cout < < endl « "Union of s i and s2 : " ;
s3.show(); / / uses B a g : : s h o w { ) base class
)
Bun

Copyrighted material
622 Mastering C++

TreeNode <T> * Search( T data ) {

return SearchTree( root, data ) ;
}
};
/ / Insert 'data' into tree template
<class T>
TreeNode<T> " BinaryTree<T>::InsertNode{ TreeNode <T> *tree, T data ) {
/ • I s Tree NULL */
i f ( !tree )
{
tree = new TreeNode<T>( data, NULL, NULL ) ;
return( tree ) ;
}
/ * Is data less than the parent element */ i f ( data < tree-
>data )
tree->left = InsertNodef tree->left, data ) ; else
/* Is data greater than the parent element */
i f ( data > tree->data )
tree->right = InsertNode( tree->right, data ) ; / * data already exists * / return( tree ) ;
}
/ / PreOrder Traversal template
<class T>
void BinaryTree<T>::PreOrderTraverse{ TreeNode <T> *tree ) {
i f ( tree )
{
cout « tree->data « " ■ ; / / Process node
PreOrderTraverse ( tree->left ) ; PreOrderTraverse( tree-
>right ) ;
)
)
/ / I n Order Traversal template
<class T>
void BinaryTree<T>:^InOrderTraverse( TreeNode <T> *tree ) (
i f ( tree ) (
PostOrderTraverse( tree->left ) ;
cout << tree->data << " " ; / / Process node
PostOrderTraverse{ tree->right ) ;
// Post Order
Traversal template
<class T>
void BinaryTree<T>::PostOrderTraverse( TreeNode <T> 'tree )
{
if<
tree )
{
PostOrderTraverse{ tree->left );
PostOrderTraverse( tree->right );
cout « tree->data « " // Process node
)
)
// Tree Printing in Triangle
Form template <class T>
void BinaryTree<T>::PrintTreeTriangle( TreeNode <T> 'tree, int level )
{
if< tree )
(
PrintTreeTriangle{ tree->right,
level+1 ) ; cout « "\n";
fort int i = 0 ; i < level; i++ )
cout « ■ cout
<< tree->data;
PrintTreeTriangle( tree->left, level+1 ) ;
)
}

Copyrighted material
622 Mastering C++

// Tree Printing in Diagonal

Form template <class T>
void BinaryTree<T>::PrintTreeDiagonal( TreeNode <T> *tree, int level )
{
if( tree )
{
cout << "\n";
for( int i ■ 0; i < level; i++ )
cout « ■ ';
cout << tree->data;
PrintTreeDiagonal( tree->left,
level+1 ) ; PrintTreeDiagonal( tree-
>right. level+1 ) ;
)
}
// search for data item in the
tree template <class T>
TreeNode <T> • BinaryTree<T>::SearchTree( TreeNode <T> -tree, T data )
{
whilet tree )
(
/* Is data less than the parent
element •/ iff data < tree->data )
tree = tree-
>left; else

Copyrighted material
626 Mastering C++

16.10 Class Template with Overloaded Operators

The class template can also be declared for a class having operator overloaded member functions.
The syntax for declaring operator overloaded functions is the same as class template members and
overloaded functions. The class template with operator overloading will allow to truly extend the
language and at the same time retaining the readability of object manipulation code. The program
complex. cpp illustrates the overloading of the + operator in the class template complex. In this case,
the members of the complex number (real and imaginary) can be any of the standard data types.

/ / complex.cpp: template class for operator overloaded complex class

•include <iostream.h>
template <class T>
class complex
(
private:
T real; / / real part of complex number
T imag; / / imaginary part of complex number
public:
complex( ) / / n o argument constructor
(
real = imag = 0 . 0 ;
)
void getdata( ) / / read complex number (
cout << "Real Part ?
" ; cin >> real; cout << "Imag
Part ? " ; cin » imag;
)
complex operator + ( complex c2 ) ; / / complex addition
void outdata( char *msg J / / display complex number
{
cout << msg << " ( " << real;
cout << " , " << imag << " ) " << endl;
)•
);
template <class T>
complex <T> complex<T>::operator + { complex <T>c2 ) {
complex <T> temp; / / object temp of complex class
temp.real = real + c2.real; / / add real parts
temp, imag = imag + c2.imag; / / add imaginary parts
return* temp ) ; / / return complex object
}
void
main (J {
complex <int> c l , c2, c 3 ; / / integer complex objects cout << "Addition of integer
complex o b j e c t s . . . " « endl; cout << "Enter complex number c l . . " « endl; c l .getdata{ ) ;

17
Streams Computation with Console

In general, there are several kinds of streams to form physical entities such as streams of water (rivers),
streams of electrons (electricity), streams of cars (traffic), and streams of characters (message packet). The
notion of streams and streams Computation can be visualized through the illustration of a river. It may be
the Amazon river flowing into the Atlantic ocean as shown in Figure 17.1. Drops of water collectively
form a continuous stream. Streams join to form a river. Looking over the upper river area to the lower
river area, streams converge into one stream so that a tree of streams is formed, whose root stream goes
into the ocean. One drop from one branch stream may reach the ocean a slightly earlier or later than
another in a different branch stream.

Copyrighted material
626 Mastering C++

Figure 17.1: Streams of water drops flowing into ocean

17.1 What are Streams?

Every program must go through the process of input-computation-output flow so that it can take some data
as input and generate the processed data as output. It necessitates the need for a mechanism, which
supplies the input data to a program and presents the processed data i n the desired form. In the earlier
chapters, the input and output operations were performed using c i n and cout with the stream operators »
and << respectively. Streams handling I/O operations arc different from ANSI C functions. C++ supports
a wide variety of features to control the way data is read and the output is presented.
C++ uses the concept of streams and stream classes to perform VO operations with console and disk f
ties. The stream classes supporting console-based input and output operations are discussed in this chapter
and those supporting file-based input and output operations are dircufsed in the next chapter. S: earns
Computation with Files.

Copyrighted material
630 Mastering C++

C++ streams deal with a sequence of characters and hence, ocean in the above figure can be visualized
as an object or a receiver and each drop of water as a character, flowing into the object.
Streams are classified i n t o input streams and output streams. Streams resemble the producer and
consumer model. The producer produces items to be consumed by the consumer. The producers and
consumers are connected by the C++ operators >> or < < . In C++, the I/O system is designed to operate
on a wide variety of devices including console, disks, printer etc. It is designed to provide a consistent and
device independent interface. I t allows uniform use of any I/O device—be it a disk, a terminal, or a
printer as shown in Figure 17.2a. The computer resources involved in the stream computation include
display, keyboard, files, printer, etc. The stream is an object flowing from one place to another. For
instance, in nature, a stream normally refers to the flow of water from the hills to the oceans. Similarly, in
C++, a stream is used to refer to the flow of data from a particular device to the program's variables. The
device here refers to files, memory arrays, keyboard, console, and so on. In C++, these streams arc treated
as objects to support consistent access interface.

MONITOR

PRIMER

DISK

Input EXTRACTION FROM

(a) Consistent stream interface with devices
device INPUT STREAM

INPUT STREAM

INSERTION
/ OUTPUT STREAM
INTO OUTPUT
Output STREAM
device

(b) Data streams

Figure 17.2: C++ streams

Some of the above devices exhibit the characteristics of either a producer or a consumer and others
exhibit the characteristics of both the producer and consumer depending on the operations performed on
them. For instance, the keyboard exhibits the nature of only a producer; printer or monitor screen
Chapter 2: Moving from C to C++ 49

i n t &m = *p; / / m i s bound to n cout << " n = ■

<< n << " m = ■ << m << 1 *p = ■ « *p « endl;
i n t k = 5;
p = & k ; / / pointer value is changed
k = 200;
/ / i s there change in ra value ? cout « " n = " « n << '
m = " << m << " *p = " '« *p « endl;
)

Run
n = 100 m = 100 *p = 100 n = 100 m
= 100 *p = 200
In main (J. the statement
p = &k; / / pointer value changed changes the pointer value of p, but does not effect the
reference variable m and the variable n.

2.9 Strict Type Checking

C++ is a strongly-typed language and it uses very strict type checking. A prototype must be known for
each function which is called, and the call must match the prototype. The prototype provides information
of the type and number of arguments passed and it also specifies the return type (if any) of the function.
In C++, function prototyping is compulsory if the definition is not placed before the function call

Copyrighted material
whereas, in C, it is optional. The program max. cpp for computing the maximum of two numbers
illustrates the need for the function prototype.
/ / max.cpp: maximum o f two numbers #include
<iostream.h> int main ( )
{
int x, y;
cout « "Enter two integers: ' ; cin » x » y;
cout « "Maximum = " « maxt x , y ) ; / / Error max.cpp 1 1 : . . . return 0 ;
)
i n t max( int a , i n t b ) {
if{ a > b )
return a ; else
return b;
)
Compilation of the above program produces the following errors: Error max.cpp 11: Function 'max*
should have a prototype in f u n c t i o n m a i n O C++ checks all the parameters passed to a function
against its prototype declaration during compilation. It produces errors if there is a mismatch in argument
types and this can be overcome by placing the prototype of the function max ( ) before it is invoked. The
modified program of max. cpp is listed in newmax. cpp, which is compiled without any errors.
char test[40] ; cout << "Enter string:
"; cin.getlinef test, 40 ); cout «
"Output string: ";
cout << test;
)

Run
Enter string: Hello World Output
string: Hello World
In main t). the statement
cin.getline( test, 40 ); reads a string until it encounters the new line character or maximum
number of characters (40). Now, an input of "Hello World" will produce the output as desired. The
istream: :getline member function has the following versions:
istream& getline{signed char*, int len, char = '\n*);
istreamfc getlinetunsigned char*, int len, char = '\n'); They operate in the following ways:
♦ extracts character up to the delimiter
♦ stores the characters in the buffer
♦ removes the delimiter from the input stream
♦ does not place the delimiter into the buffer
♦ maximum number of characters extracted is len-1
The terminator character can be any character. The terminator character is read but not saved into a buffer;
instead, it is replaced by the null character.
The prototype of write () functions is:
ostream: : write ( char * buffer, int size ); It displays size (second parameter) number of
characters from the input buffer. The display does not stop even when the NULL character is
encountered; If the length of the buffer is less than the indicated size, it displays beyond the bounds of
buffer. Therefore, it is the responsibility of the user to make sure that the size does not exceed the length
of the string. The program stand. cpp illustrates the use of write in string processing.
/ / stand.cpp: display stand of "Object Computing with C++"; #include <iostream.h> tinclude <string.h>
void main()

char *stringl = "Object-Computing"; char *string2

■ " with C++"; int i;
int lenl = strlen( stringl ); int len2= strlen(
string2 ); for ( i = 1; i < lenl; i + + ) <
cout.writet stringl, i ); cout <<
endl;

Copyrighted material
 C
h
a
p
t
e
r

1
7
:

S
t
e
a
m
s

C
o
m
p
u
t
a
t
i
o
n

w
i
t
h

C
o
n
s
o
l
e

>
fori i = lenl; i > 0; i-- ) (
cout.writet stringl, i ) ; cout << endl;
)
/ / print both the string cout.writet
stringl, lenl ) ; cout.write( string2,
len2 ) ; cout << endl;
/ / above two write( ) can be replaced below single statement cout.writet stringl,
lenl ) . write( string2, len2 ) ; cout << endl; cout.writet stringl, 6 ) ;
)

Run
o
Ob
Obj
Obje
Objec
Object
Object-
Object-C
Object-Co
Object-Com
Object-Comp
Obj ec t-Compu
Object-Comput
Object-Computi
Object-Computin
Ob j ec t-Comput i ng
Object-Computin
Object-Computi
Object-Comput
Obj ec t-Compu
Object-Comp
Object-Com
Object-Co
Object-C
Object-
Object
Obj ec
Obje
Obj
Ob
O
Object-Computing with C++
Object-Computing with C++
Object
 Chapter 17: Steams Computation with Console 643

Run
How many students ? 2 Enter Student 1
details.. Enter Name: Teiaswi Enter Marks
Secured: 450 Enter Student 2 details.. Enter
Name: Raikumar Enter Marks Secured: 5 2 1
Enter Student 3 details.. Enter Name: Bindu
Enter Marks Secured: 429 Student Report...
R# Student Marks Percentage
1 Tejaswi 450 75
2 Rajkumax- 525 87
3 Bindu 429 71

Setting Precision
The funciion precision ( ) is a member of the ios class and is used to specify the number of digits to be
displayed after the decimal point while printing a floating-point number. By default, the precision size is
six. This function must be accessed using objects of the ios class (commonly accessed using cout object). It
has the following two forms:
int precisionO; / / returns current precision
int precision(int d ) ;
where d is the number of digits to the right of the decimal point. It sets the floating-point precision and
returns the previous setting. For example, the statements
cout.precision( 2 ) ; cout «
2 . 2 3 « endl; cout « 5 . 1 6 9 < <
endl; cout << 3 . 5 0 5 5 < <
endl; cout « 4 . 0 0 3 « endl;
will produce the following output:
2.23 5.17 3.51 (perfect f i t )
4 (rounded)
(rounded)
(no trailing zeros, truncated)
After displaying an item, the user defined precision will not revert to the default value. Different values
can be processed with different precision by having multiple precision statements. For instance,
cout.precision( 1 ) ; cout « 2 . 2 3 < < endl; cout.precision( 3 ) ;
cout « 5 . 1 6 9 1 « endl;
will produce the following output:
2.2 (truncated)
5.169 (truncated)
Consider the statements:
cout.precision( 3 ) ;
/ / salary.cpp: f i l l i n g and padding
#include <iostream.h> void mainO {
char *desig[5J = { "CEO", "Manager", "Receptionist", "Clerk", "Peon" ) ;
int salary[5J = { 10200, 5 2 0 0 , 2950, 950, 750 } ;
cout << "Salary Structure Based on Designation" « endl;
cout << "---------------------------------------------------------------" « endl;
cout-width( 15 } ;
cout « "Designation";
cout « " " ,-
cout.width( 15 ) ;
cout < < "Salary ( i n R s . ) " << endl;
cout < < "--------------------------------------------------------------" << endl;
for( int i = 0; i < 5; i + + ) {
cout.f i l l ( ' . ' ) ;
cout.width( 1 5 ) ;
cout < < d e s i g f i ] ;
cout < < '
cout.f i l l ( ' * ' ) ;
cout.width(15) ,-
cout < < salary[ i ] < < endl;
}
cout « "-----------------------------------------------------------------■ « endl;
)

Run
Salary Structure Based on Designation
Designation Salary ( i n Rs. )

Copyrighted material
 Chapter 17: Steams Computation with Console 643

CPA mnn
.................Manager .........«*5200
...Receptionist
.....................Clerk
.......................Peon

Note that such a form of output representation is extensively used by financial institutions to represent
money transactions so that no one can modify the amount (money representation) easily.

Formatting with Flags and Bit-fields

From the earlier examples, it can be noted that, when the function width () is used, results are printed in the
right-justified form (which is not a usual practice). C++ provides a mechanism to set the printing of results in
the left-justified form, scientific notation etc. The member function of the ios class, setf () (setf stands for set
flags) is used to set flags and bit-fields that control the output. It has the following two forms:
long setf(long _setbits, long _field);
long setf(long _setbits ) ; where_setbits is one of the flags defined in the class ios. It specifics the
format action required for
the output, and _f ield specifies the group to which the formatting flag belongs. Both the forms return the
previous settings. The flags, bit-fields when set with setf () and their actions is shown in Table 17.2. There are
three bit-fields and each group has format flags that are mutually exclusive. For instance,
cout.setf't ios::right, ios::adjustfield );
cout.setf( ios::oct, ios::basefield );
cout.setf( ios::scientific, ios::floatfield );
Note that the flag argument (first) should be one of the group (bit-field) of the second argument.
Flags value Bitfield Effect produced
ios::left ios::right ios::adjustfield Left-justified output Right-adjust output
Padding occurs between the sign or base
ios::internal ios::adjustfield
indicator and the number, when the
ios::adjustfield number output fails to fill the full width
of the field.
ios::dec ios::oct ios::hex ios::basefield Decimal conversion Octal conversion
ios::basefield Hexadecimal conversion
ios::basefield
ios::scientific ios::fixed ios::floatfield Use exponential floating notation Use
ios::floatfield ordinary floating notation

Table 17.2: Flags and bit fields for setf function

Consider the following statements:
cout.setf{ ios::left, ios::adjustfield ) ;
cout.fill( );
cout.precision! 2 ); cout.width{ 6 ) ; cout « 12.53;
cout.width( 6 ); cout << 20.5; cout.width( 6 ) ; cout « 2;
The output produced by the above statements is:

The statements
cout.setf{ ios: : internal, ios:-.adjustf ield ) ; cout.fill ( 1 *' );
cout.precision! 3 )■
cin » hex >> num; // Input in hexadecimal //
output i in decimal,in a field of width 6 cout << "The
input number in decimal = "; cout « setwl 6 ) « num;
}

Run
Enter any nexadecimal number:
The incut number in decimal = 171

The manipulator hex sets the conversion base for cin to 16. So cin interprets the input characters
as digits of a hexadecimal number. The manipulator setw sets the field width as 6 for cout. Thus, the
input to the above program (ab) is converted into decimal and displayed (16*a+b = 16* 10+11 =
171).

Copyrighted material
 Chapter 17: Steams Computation with Console 643

The C++ iostream package contains a rather small handful of predefined producer consumer
manipulators, the only instance of consumer being the white-space eater, for example, ws. Other pre-
defined manipulators set stream state variables which influence processing of input and output, for
example, hex. The implementation of the ios class as well as the implementation of the insertion and
extraction operators correspond to the data type of an item they process. The list of non-
parameterized manipulators and parameterized manipulator functions are shown in Table 17.4 and
17.5 respectively. Each one of these can be used with cither thc<< or the >> operator without
incurring any compilc-time errors. But some of them affect only output streams such as cout. and
some others, only input streams such as cin. Unless otherwise mentioned, these manipulators
affect both types of streams. The first six manipulators - dec,hex, oct, ws,endl, and ends are defined
in iostream.h itself and the rest are in the header file iomanip.h.

Table 17.4: C++'s predefined non-parameterized manipulators

Copyrighted material
652 Mastering C++

sotf ill (int f char): Sets the fill character to that specified in f char. The fill character is used
to fill (or pad) the empty space in the output field when the width of the output variable is less than
the width of the output field. The default fill character is the space character.
setease (int base): Sets the conversion base according to the integer base, which can assume any
one of the following four values:
0: Base 10 is used for output;
8: Use octal for input and output.
10: Use decimal for input and output.
16: Use hexadecimal for input and output.
The base to be used for input is specified as a part of the input itself - inputs beginning with 0 are
treated as octal, those beginning with Ox arc treated as hexadecimal. Otherwise, the base is
assumed as decimal.
setiosf lags (long flags): The parameter flags can be any of the flags listed in ios stream
class. More than one flag can be set with the same manipulator by ORing the flags.
The statement
cout << setw( 8 ) << 1234; prints the value 1234 right-justified in the field width of 8
characters. The output can be left justified using the statement,
cout << setw{ 8 ) << setiosflags{ ios::left ) « 1234;
The key difference between manipulators and the ios class interface functions is in their
implementation. The ios member functions are used to read the previous format-state, which can be
used to know the current state or save for future usage, whereas, the manipulators do not return the
previous format state. The program f output. cpp illustrates the use of some of the manipulators with
output streams.
/ / foutput.cpp: various formatting flags with the « operator
•include <iostream.h>
•include <iomanip.h>
void main( )
(
int x = 100;
cout << hex << x << ' ' « dec << x << endl;
float f = 122.3434;
cout « f « endl;
cout « setprecision( 3 ) ;
cout << f << endl;
cout << setw( 6 ) « setf ill { ' 0 ' ) ; cout <<
setiosflags{ios: : internal | ios: : showcase); cout « hex « x « endl;
cout « setiosf lags { ios: : scientific ) « f « endl;
)

Run
C i 100
x22.343399
122.343
0x0064
1.223e+02

Copyrighted material
 656 Mastering C++

/ / space3.cpp: custom built manipulator

tinclude <iostream.h>
// The user-defined manipulator
ostream & sp( ostreams, os )
{
os « ' ' « flush; //or cout « ' ' « flush return os;
)
void main () {
int x=l, y=2, z=3, w=4;
cout << x « sp « y << sp << z << sp << w « endl;
>

Run
1 2 3 4

In the above program, the function

ostream & sp( ostream& os ) defines a manipulator called sp that prints a single space and flushes the same
to console for immediate display without buffering.
Another interesting use of manipulators is demonstrated in the program currency. cpp. It defines manipulators for
prefixing the currency symbol to an item cost depending on the currency used by the country which has manufactured
the item.

/ / currency.cpp: custom built manipulator for currency unit representation

•include <iostream.h>
// currency in Indian rupees
ostream & rupee ( ostreamfit os )
{
os << "Rs. " << flush; return os;
)
// currency unit in US 'dollar ostream &
dollar( ostreamfc os ) {
cout « "USS " « flush; return os;
}
void main () {
char iteml(25J, item2[251;
unsigned int costl, cost2;
cout « "Item Sales in India..." « endl;
coUt « "Enter Item Name: ";
cin.getline{ iteml, 25 );
cout « "Cost of Item: ";
cin >> costl;
cout « "Item Sales in US..." << endl;

Copyrighted material
Chapter^: Moving from C to C+ 51
+

The
above
errors are
produced
due to the
following
statement
in main ()
s
w
a
p
(
&
c
,
&
d

)
;
/
/
C
o
m
p
i
l
a
t
i
o
n

E
r
r
o
r
s
Because
the
expressio
ns &c
and &d
passed to
swap ()
are not
pointers
to integer
data type.
When a
call to a
function
is made,
the C++
compiler
checks its
parameter
s against
the
parameter
types
declared
in the
function
prototype
. The
compiler

Copyrighted material
 flags errors if improper -
arguments are passed.
*
q
2.10 Parameters ;
Passing by Reference *
A function in C++ can take q
arguments passed by value, by
pointer, or by reference. The =
arguments passed by reference is t
an enhancement over C. A copy ;
of the actual parameters in the )
function call is assigned to the A call to
formal parameters in the case of t
pass-by-value, whereas the
h
address of the actual parameters
is passed in the case of pass-by- e
pointer. In the case of pass-by-
reference, an alias (reference) of f
the actual parameters is passed. u
Mechanism of parameter linkage n
is shown in Figure 2.8. c
t
Storage Location
Variable Data Type i
Name o Value
Actual —
> Type n
parameter
s
Formal —
> Type w
Value
parameter
a
(a) p
Call-
by- {
value
)

Valu
Formal
parameter

Figure 2.8: Parameter passing mechanism

Actual parameter s
(b) Call-by-pointer/reference w
a
p
Consider an example of (
swapping two numbers to
illustrate the mechanism of &
parameter passing by reference. x
The function definition with ,
pointer type parameters is listed
below: void swapt int * p, int *
q ) //by pointers ( &
i y
n
t )
t c
; o
u
t t

= «
* "
p E
; n
* t
p e
r

Copyrighted material
 ;
I
t c
e i
m n

N >
a >
m
e c
: o
s
" t
; 2
;
c -
i c
n o
. u
i t
g
n <
o <
r
e "
( I
) t
; e
m
c
i C
n o
. s
g t
e
t S
l t
i a
n t
e i
t s
t
i i
t c
e s
m .
2 .
, .
"
2
5 <
<
)
; e
n
c d
o l
u ;
t
c
< o
< u
t
"
C <
o <
s
t "
I
o t
f e
m
I
t N
e a
m m
: e
:
*

Copyrighted material
"
"
<
< <
<
i
t i
e t
m e
l m
2
<
< <
<
e
n e
d n
l d
; l
;
c
o c
u o
t u
t
«
«
"
C "
o C
s o
t s
: t
:
■
"
«
«
r
u d
p o
e l
e l
a
« r

c «
o
s c
t o
l s
t
« 2

e «
n
d e
l n
; d
l
c ;
o )
u
t Run
Item
< Sales in
< India..
.
" Enter
I Item
t Name:
e PARAM
m Superco
mputer
N Cost o f
a Item:
m 55000
e
:

Copyrighted material
Item Sales in US. . . i
Enter Item Name: CRAY p
Supercomputer _
Cost of Item: 40500 i
Item Cost S t a t i s t i c s . . .
n
Item Name: PARAM
Supercomputer t
Cost: R s . 55000
Item Name: CRAY {
Supercomputer private:
Cost: US$ 4 0 5 0 0 ost
rea
Standard Manipulators mfi
Implementation c
(*f
The previous example was )
easy, since the manipulator did ( os
not accept any parameters in trea
the output statement. The m&
function that overloads the < < ,
operator to accept int
manipulators merely needs to );
call the manipulator with the //
output stream object (cout in Poi
this case). Manipulators nte
accepting parameters initiates r to
many actions. Consider the the
manipulator declared in act
ual
iomanip. h header file, setw
ma
(int) , to illustrate the
nip
implementation of
ulat
manipulators. The declaration or
of this manipulator is: int
ostream & w; /
setw( ostreamfc, int ) ; But in /
Wi
the output statement, setw is
dth
called with only one integer to
argument: be
cout << s e t w { 6 ) << set
i ; Another function (also p
called setw) is needed that u
accepts only one argument of b
type integer. It does not know l
i
which output object needs to
c
have its field-width set. :
Assuming the output object as
cout will unduly restrict its use /
(For instance, i t would not be /
possible to use it directly with C
files). To resolve this impasse, o
the following solution is used. n
s
A class called omanip_int is
t
declared. It has two private r
members; a pointer to function u
(the actual manipulator) and c
an integer that specifies the t
width. It has a constructor that o
sets these members, and a r
friend function that overloads
the « operator and calls the o
m
actual manipulator.
a
c n
l i
a p
s _
s i
n
t
o
(
m
a o
n s

Copyrighted material
t C
r o
e s
a .
m w
& i
d
( t
h
*
(
t
f w
)
)
( ;
o
s r
t e
r t
e u
a r
m n
f
c o
, s
i ;
n )
t // This is
) called
first from
, the
output
i statement
n .
t //It
accepts
t an integer
w and
returns an
) instance
of class
(
omanip_i
nt
f
omanip_i
= nt
setw( int
t w)
f {
; return
omanip_i
w nt( setw,
w ); //
= returns
nameless
t object
w )
;
}
Now
, the
// overloading stream
state
output operator
ment
friend ostream& operator
<< (ostreamfc os, c
omanip_int o) ( out «
return o.f( os, o.w >; setw( 6 )
« i; will
manipulator. first call
) the
); second
Two more functions are now setw
required; one that actually manipulat
manipulates the stream, and or that
another that is invoked from the remember
output statement. They are s the
declared as follows: width
//Actual manipulator passed in
ostreamfr setw( ostream& os, an
int w )

Copyrighted material
instance of the class omanip_int. •
The actual function to be called i
n
is also recorded here. This
c
instance is returned. The first << l
above now has the return value u
of setw (6) - an instance of d
omanip_int on the right, and e
cout on the left. The overloaded
<
function (defined in the class
i
omanipjnt) is invoked, which in o
turn calls the actual manipulator. s
The same concept can be utilized t
while implementing the user- r
defined manipulators. e
a
m
Parameterized Custom .
Manipulators h
Most manipulators do not accept >
parameters and are simple to
use. Sometimes it is necessary to •
pass data to the manipulators, i
n
however, as with the built-in
c
manipulator setw (int). The l
program presented in pmani. u
cpp, implements a manipulator d
that accepts three arguments - e
width, precision, and fill
character. The manipulator is <
useful as a shorthand notation i
for setting the above parameters o
m
to output floating point variables
a
with different width, precision, n
and fill characters. i
p
/ .
/ h
>
p // output
m manipulat
a or taking
n argument
l s of type
. int, int,
c char
p
p c
: l
a
P s
a
s
r
a
m m
e y
t _
e
r
m
i a
z n
e i
d p
M u
a l
n a
i t
p o
u
l r
a
t {
o priv
r ate:

Copyrighted material
 int width, precision; i
c n
h t
a |
r i
o
f s
i :
l :
l r
; i
g
p h
u t
b )
l ;
i
c r
: e
//Constructor t
my_manipulator(int tw, u
int tp, char r
tf) : width(tw),precisio n
n(tp),fill(tf) { )
//Overloaded « o
operator s
friend ostream & ;
operator « ( ostreamfc }

os, my_manipulator /
object ) ; /
}; F
//Actual manipulator called u
by overloaded operator « n
friend function c
ostream & operator « t
( ostreamfc os, i
my_manipulator object ) o
{ n
os « setw( object.width )
<< c
setprecision{ object.prec a
ision ) \ « l
setfill{ object.fill ) ;
l
o
s e
d
<
< f
i
s r
e s
t t
i
o
s f
f r
l o
a m
g
s t
( h
i e
o
s
: o
: u
s t
h p
o u
w t
p
o

Copyrighted material
s i
t d
a
t m
e a
m i
e n
n t
t
)

m
{
y
float
_ fl=1
m 23.2
a 734,
n f2=2
i 3.27
p 1,
u f3=1
l 6.16
73;
a
//
t set_f
o loat
r acce
pts
s thre
e e
t para
_ mete
rs-
f
widt
l h,
o prec
a ision
t and
( fill
char
i acte
n r
cout
t
«
set_f
w loat(
, 10,
3,
i * *'
n ) «
t fl «
endl
;
p
cout
, «
set_f
c loat(
h 9 , 2,
,Al
a )«
r f2 «
endl
f ;
cout
«
} set_f
loat(
( 8 , 3,
return my_manipulator ( *#'
w, p, f ) »• / / nameless ) «
object is returned f3 «
) endl
v ;
)
o

Copyrighted material
Run l
***123.273 l

##16.167
t
In main ( ) , the
o
statement
c
t
o
h
u
e
t
n
«
o
r
s
m
e
a
t
l
_
f
f
l
u
o
n
a
c
t
t
(
i
o
1
n
0
,
a
s
3
,
,
s
et_float(
'
10, 3,
*
■•■ )
'
which
)
in turn
<
creates
<
the
nameles
f
s object
l
of the
classmy
<
_manip
<
ulator
(and
e
initializ
n
es its
d
member
l
s) and
;
returns
the
h
same.
a
Thus,
s
the
above
t
output
h
e

c
a

Copyrighted material
 statement effectively r
becomes,
cout << e
n
my_manipulator( set_float,
d
10, 3 , ' * ' ) « f l « endl;
l
The class my_manipulator :
is a friend of the overloaded
operator function and c
hence, the mutated output o
statement invokes the u
function, t
friend ostreamfc
<
operator «
<
( ostream& os,
my_manipulator e
object ) n
Necessity of Friend d
Functions l
The function overloading the ;
operators >> and « need not in the
always be declared as friend. If previous
the data members x and y were examples
public members of the class , to insert
POINT, or, if a public member a'newline
function existed in POINT . The
which output the values of x and manipulat
y, the friend function
or endl is
declarations would be
the
unnecessary inside the class.
function
that is
How do the manipulators
declared
work with the «
as,
operator? o
Cons stream
i far &
d endl
e (ostream
r far &) ;
in the
t header
h file,
iostream.
e
h. Thus,
endl, is a
u function
s that
a accepts a
g reference
e to an
ostream
o (such as
f cout) and
returns
the same
t
(a
h reference
e to an
ostream).
m Recall
a that
n invocatio
i n of a
p function
u with its
name
l
without
a any
t parenthes
o es is

Copyrighted material
considered as a pointer to a
function. Now it is simple to
understand the appearance of the
endl on the right side of the <<
operator; the operator is
overloaded to have pointers to
functions of this type (that
accept a reference to an ostream
and returns the same).

Review Questions
17.1 What are streams ?
Explain the features of C++
stream I/O with C's I/O system.
17.2 List C++ predefined
streams and explain them with
suitable example programs.
17.3 Draw console stream
class hierarchy and explain its
members.
17.4 What is the difference
between the statements ?
cin » ch;
ch = cin.get();
17.5 Write a program to
illustrate the difference between
cin and get line while reading
strings.
17.6 What is the output of the
following statements:
(a) cout « 65;
(b) cout.put( 65 ) ;
(c) cout.putt 1A' ) ;
17.7 Write a program to print
the ASCII table using streams.
17.8 Write an interactive
program to print a string
entered in a pyramid
form. For instance, the
string "object" has to be
displayed as follows:
oobob
jobjeo
bjec
o b j e
c t
17.9 Write an interactive
program to print a
rectangle with diamond
shape gap exactly at the
centre of
that rectangle. Accept
string from standard input
device and print output on
standard output
device. Here is the
sample output when the
string "ob ject-obj ect" is
entered by the user:

Copyrighted material
 Chapter 17: Steams Computation with Console 663

object-obj ect
object object
objec bject
obj ject
ob ct
o t
ob ct
obj ect
obje ject
objec bject object
object object-
object

17.10 Write an interactive program to print the salary-slip in the following format:

Centre for Development o f Advanced Computing Bangalore, India - 560 025 Salary-
Slip for the Month of XXXXXX 1996

Employee Name: xxxxxxxxxxx Grade: xx Date: dd/mm/yy Employee No.:

No. o f days present: xx xxx Basic Salary: xxxxx.xx

<------PAYMENTS- <--------------------DEDUCTIONS----------------------> <

BASIC DA HRA XXXXX.XX PFXXX.XX ----------------------RECOVERIES----------------------->
XXX.X
CCA DDA XXXXX.XX FPF
XX .XX LIC XX. X
ARREARS XXXX.XX VPFXX.XX CCUBE CONTR. SOCIETY ADV RENT RECV x.x PF LOAN
ADHOC.ALW XXX. XX BEFUND
X. XX SALARY ADV TOUR ADV XXX.X
x.xx P.XXX.XX
TAX XXX.X
TOTAL PAY X .XX CANTEEN
XXX.XX XXXXX.XX TOTAL DED xxxx.x XXXX.X TOTAL
RECV XXX.XX WELFARE
XX.X XXXX.X XXX.X
NET PAY: xxxxx.xx

(SIGNATURE)
17.11 Explain the various methods of performing formated stream I/O operations.
17.12 What are manipulators ? List the various predefined manipulators supported by C++ I/O
streams.
17.13 How are the input and output streams tied using istream. t i e ( ) member function ?
17.14 Write a program to display numbers 0 to 10 in octal, decimal, and hexadecimal systems.
17.15 What are custom manipulators ? Write a custom manipulator for inserting 8 spaces in
output.
17.16 Explain how standard manipulators are implemented.
17.17 Illustrate parameterized custom manipulators using a suitable program.
17.18 Write a program to overload stream operators for reading and displaying the object of a
class Employee. The members of this class include name, emp_no, DateOfBirth, basic,
grade, qualification, etc.

Copyrighted material
 52 Mastering C++

has effect on the values of x and y i.e. it exchanges the contents of variables x and y. The above swap (. .}
function can be redefined by using a new parameter passing scheme, call by reference, as follows:
void swapt int & x, int & y ) //by reference {
int t; t = x; X
» y;
y = t;
}
A call to the function swap {)
swap ( x, y ) ;
with integer variables x and y, has effect on the values of x and y variables. It exchanges the contents of the
variables x and y. The body and the call to the function swap appears same as that of call-by-value case, but
has an effect of call-by-pointer. Thus, call by reference combines the flexibility (ease of programming) of call
by value and the power of call by pointer.
The complete program having swap (..) function with call-by-reference mechanism for parameter passing is
listed in swap. cpp.

// swap.cpp: swap integer values by reference •include <iostream.h>

void swapt int & x, int & y ) //by reference {
int t; // temporary variable used in swapping
t = x; x = y; y
= t;
}
void maint) {
int a, b;
cout « "Enter two integers <a, b>: "; cin » a >> b; swap( a,
b);
cout « "On swapping <a, b>: " « a << " ■ « b;
>

Run
Enter two integers <a, b>: 2 3 On swapping <a,
b>: 3 2
In main (), the statement
swap ( a, b ) ; is
translated into
swap ( Sc a, St b > ; internally during compilation; the
prototype of the function
void swapt int & x, int & y ) //by reference indicates that the formal parameters are of reference type
and hence, they must be bound to the memory

Copyrighted material
 Chapter 18: Streams Computation667
with Files

ofstream: The class of scream supports output operations. It contains open () with default output mode
and inherits put (), seekp (), tellp {}, and write () functions from ostream.
fstream: The class f stream supports simultaneous input and output operations. It contains open () with
default input mode and inherits all the functions from istream and ostream classes through iostream.

18.3 Opening and Closing of Files

In order to process a file, first, it must be opened to get a handle. The file handle serves as a pointer to the
file. Typically, manipulation of a file involves the following steps:
♦ Name the file on the disk
♦ Open the file to get the file pointer
♦ Process the file (read/write)
♦ Check for errors while processing
♦ Close the file after its complete usage
The filename is a suing of characters, with which a file is logically identified by the user. It provides a
means to communicate with the user transparently. The number and type of characters used in naming a
file depends on the operating system. Normally, a file has two parts: ^primary name and an optional
extension. If the file name has an extension, it is separated by a period from the primary name. Some of the
valid file names in the MS-DOS based machines are the following:
student.cpp
data.txt
copy.exe
student.obj
student.exe
TEMP datal
tax.in
In MS-DOS systems, the maximum size of a primary name is eight characters and that of an extension
is three characters. However, in UNIX based machines, the file name can be upto 31 characters and any
number of extensions separated by dots. Some valid file names in the UNIX system include all those
valid in the MS-DOS and in addition, it includes the following:
. login (no primary name, acts as hidden file)
xyz.txt.mine text_data_file student.8sem.raj
In order to get a file pointer, first the file must be created (if it does not exist) and linked to the file
name. A file stream can be defined using stream classes, if stream, ofstream, or fstream de pending on the
purpose (read or write). In C++, a file can be opened using the following:
• The constructor function of the class.
• The member function open () of the class.
After processing an opened file, it must be closed. It can be closed either by explicitly using the slose ()
member function of the class or it is automatically closed by the destructor of the class, when the file
stream object goes out of scope (expires).

Copyrighted material
Chapter 2: Moving from C to C++ 53

location
of the
actual
parameter
. Thus,
any
access
made to
reference
formal
parameter
s in swap
() refers
to the
actual
parameter
s. The
statement
s

y
;

t
;
in the
body of
swap ()

internally
(as
treated by
the
compiler)
have the
following
meaning,
t

"
x
;

/
/

s
t
o
r
e

Copyrighted material
 he value pointed by x into
t 2
*x = *y;
pointed by y into location .
pointed by x 9
*y = t;
hold by 't' into location :
pointed by y
because, the formal parameters are S
of reference type and therefore, the y
compiler treats them similar to
n
pointers, but does not allow the
modification of the address stored t
in them. a
x
Void Argument List
A function prototype in C with an
empty argument list, such as o
extern void func () ; f
implies that the argument list of the
declared function is not prototyped;
i
the compiler will not be able to
warn against improper argument n
usage. To declare a function in C l
which has no arguments, the i
keyword void is used, as indicated:
n
extern void func (void); In
C++, the above two declarations e
are equivalent. Because C++
maintains strict type checking, an f
empty argument list is interpreted u
as the absence of any parameter.
n
c
2.11 Inline Functions t
Function execution involves the i
overhead of jumping to and from
the calling statement. Trading of o
this overhead in execution time is n
considerably large whenever a
function is small, and hence in such
E
cases, inline functions can be used.
A function in C++ can be treated as x
a macro if the keyword inline a
precedes its definition. The syntax m
of representing the inline function
is shown in Figure 2.9. p
l
Keyword,
e
function
qualifier :

inline A
ReturnTyp n
e
FunctionN
ame i
(Parameter n
s) {
// body l
of a i
main
n
function
) e
F
i f
g u
u n
r c
e t

Copyrighted material
i function
call.
o
Thereby,
n the
t runtime
overhead
o
for
fi function
n
d
s
q
u
a
r
e
o
f
a
n
u
m
b
e
r
i
s
a
s
f
o
ll
o
w
s
:
inline float square( float x )
{
x

x
;

r
e
t
u
r
n
(

)
;
>
The significant feature of inline
functions is that there is no explicit
function call; the function body is
substituted at the point of inline

Copyrighted material
 678 Mastering C++

• For f s c r e a m class, ihc mode parameter imist bo explicitly passed.

♦ More than one value may be ORed to have a combined effect. For instance, the following statement opens a
file for reading in binary mode:
istream i n _ f i l e ( " m y f i l e " , i o s : : i n | ios::binary ) ;
The program payfile.cpp generates a payroll-like output and directs the output to the file pay.txt instead of
cout. It stores floating point data in the form of ASCII characters instead of machine representation (binary
form).

payfile: payroli l i k e output example printing results to f i l e

#inciude <fstream.h>
•include < i o m a n i p . h >
void main(i {
float fl=123.45, f2=34.65, f3=56;
/ / open f i l e " p a y . t x t " in o u t p u t mode and truncate i t s contents i f exists
ofstream out_file( "pay.txt", ios::trunc );
o u t _ f i l e « s e t i o s f l a g s (ios:sshowpoint|ios: : f i x e d )
« s e t i o s f l a g s (ios: : r i g h t ) ; o u t _ f i l e
« s e t w ( 6 ) « f l << endl; o u t _ f i l e « s e t w ( 6 )
<< f 2 << endl; o u t _ f i l e « s e t w ( 6 ) « f 3 <<
endl;
)

Run
After execution of the program, the file pay. txt contains the following:
123.45
34.6
5
56.0
0
In main (). the statement
ofstream out_file ( "pay.txt", ios::trunc ); creates the file pay. txt and truncates its contents if the file
already exists. As with the console streams, manipulators can be used with any of the file stream instances.

18.6 File Pointers and their Manipulations

The knowledge of the logical location at which the current read or write operations occur is of great importance
in achieving faster access to information stored in a file. The file management system associates two pointers
with each file, called file pointers. In C++, they arc called get pointer (input pointer) and put pointer (output
pointer). These pointers facilitate the movement across the file while reading or writing. The get pointer
specifies a location from where the current reading operation is initiated. The put pointer Specifies a location
from where the current writing operation is initiated. On completion of a read or write operation, the
appropriate pointer will be advanced automatically.

Default Actions
The file pointers arc set to a suitable location initially based on the mode in which the file is opened.
Fundamentally, a file can be opened in the read mode, write mode nr append mode. The logical location of file
pointers when a file is opened is discussed below ( see Figure 18.7.):

Copyrighted material
 Chapter 18: Streams Computation with Files 579

Read-only Mode: when a file is opened in read-only mode, ihe input (gel) pointer is initialized to
point to the beginning of the file, so that the file can bo read from the start.
Write-only Mode: when a file is opened in write-only mode, the existing contents of the file arc
deleted (if a given file already exists) and the output pointer is set to point to the beginning of the
file, so that data can be written from the start.
Append Mode: when a file is opened in append mode, the existing contents of the file remain
unaffected (if a given file already exists) and the output pointer is set to point to the end of the file
so that data can be written (appended) at the end of the existing contents.

-hellc" file

H mode
Read e 1 - 0 W 0 r 1 d
input pointer
■•
-eilo" file
1---
output pointer
Write mode

hello" file
H e mode
Append 1 1 0 W 0 r 1 d

output pointer

Figure 18.7: File pointer position on opening a file

Functions for Manipulation of File Pointers

The C++ I/O system supports four functions for setting a file pointer to any desired position
inside the file or to gel the current file pointer. These allow the programmer to have control over a
position in the file where the read or write operation takes place. The functions are listed in the

Table 18.2: File pointer control functions

Table 18.2.
/ / putget.cpp: writes and reads characters from the file
•include <fstream.h> void main()
{
char c, stringl 7 5 ] ;
fstream filet "student. txt ■, ios::m j ios::out j:
cout << "Enter String: ";
cin.getlinet string, 74 ) ;
for( int i = 0; stringtij; i + + )
file.putt stringlij ); file.seekg(0) ; //
seek to the beginning cout << "Output
String: whilet file ) {
file.get( c ); // reads a character cout
<< c;
>
)

Run
Enter String: Obiect-Computina with C++
Output String: Object-Computing with C++

Note: The file student. txt contains the entered string.

Copyrighted material
 Chapter 18: Streams Computation with Files 579

The stream fstream provides the facility to open a file in both read and write modes; so that the
file can be processed randomly by positioning the file pointers.

18.8 ASCII and Binary Files

The stream operators insertion and extraction always manipulate and deal with formatted data.
Data has to be formaitedio produce logical information. This is because, most of the I/O devices
communicate to the computer system using ASCII code, but CPU processes these data using the
binary system. Hence, it is necessary to convert data while reading from the input device or
displaying data on output device. Most visible data formatting operation is alignment of display
fields. In addition to this, data formatting operation also occur transparently while transferring
data between the program and console or a file. For example, in order to display an integer value,
the << operator converts the number into a stream of ASCII characters. Similarly, the » operator
converts the input ASCII characters to binary while reading data from the input device. For
instance, when a number, say, 120 is typed in response to an input statement such as: cin » i;
The user enters data by typing on ihe keyboard. In this case, stream operator receives ASCII codes
of the numeric characters 1,2, and 0 (which are 49,50, and 48). The » operator function converts
the input ASCII data to binary and assigns to the variable i. Similarly, the « operator in a
statement such as: cout << i;
converts the content of the variable i (say 120) into three ASCII characters. 49, 50, and 48 and
then jends the same to the standard output device. The representation of an integer in the character
form and binary form is shown in Figure 18.9.
The program obj save, cpp illustrates the flexibility gained by overloading the
insertion and extraction operators while saving objects into a file or retrieving
objects from a file.

. / objsave.Cpp: saving a object to a f i l e with stream operator overloaded •include

<fstream.h>
•include <ctype.h> / / for toupper
•include <string.h> //
for strlen •define
MAXNAME 40 class
Person {
private:
char name! MAXNAME ] ;
i
n
t

a
g
e
;

p
u
b
l
i
c
:
/ / this function writes the class's data members to the fil.e void write!
ofstream &os ) {
o s . w r i t e t name,
strleni name ) ) ;
os << ends;
o s . w r i t e t (char")&age, sizeof( age ) ) ;

Copyrighted material
 Chapter 18: Streams Computation with Files 579

}
/ / this function reads the class's date member from the f i l e . / / It returns
nonzero i f no errors were encountered while reading int read( ifstream &is
){
i s . get( na
me,
MAXNAM
E, 0 ) ;
namef
is.gcountO
] = 0;
i s . ignore! 1 ) ; / / ignore the NULL terminator in the f i l e , is.readt
(char*)&age, sizeof( age ) ) ; return i s .good( ) ;
}
/ / stream operator, < < overloading
friend ostream & operator « ( ostream &os. Person &b ) ;
/ stream operator » operator overloading friend istream &operator »
{ istream &is, Person &b } ; / / output f i l e stream operator overloading
friend ofstream &operator << ( ofstream &fos. Person &b > (
b
.
w
r
i
t
e
(

f
o
s

r
e
t
u
r
n

f
o
s
;
}
/ / output f i l e stream operator overloading
friend ifstream toperator >> ( ifstream &fos, Person &b )
{
b
.
r
e
a
d

Copyrighted material
 Chapter 18: Streams Computation with Files 579

f
o
s

)
;

r
e
t
u
r
n

f
o
s
;
)
);
istream &operator > > ( istream & i s ,
Person &b ) (
cout << 'Name: " ;
i s > > ws; / /
flush input buffer
i s . get( b.name,
MAXNAME ) ;
cout « "Age : ' ;
i s » ws >> b.age;
return is;
}
ostream ^operator « ( ostream
&os. Person &b J {
o
s
«
b
.
n
a
m
e
«
e
n
d
l;
o
s
<
<
b
.a

Copyrighted material
 Chapter 18: Streams Computation with Files 579

g
e
<
<
e
n
d
l;
r
et
u
r
n
o
s;
}
v
o
i
d

m
a
i
n

(
)

{
Person p_obj;
/ / open a f i l e in binary mode and write objects to i t ofstream of ile { "person.
txt", i'os : : trunc | ios : :binary ) ; char ch; do
(
cin >> p_obj; / / read object
of ile <<. p_obj; / / write object to the outputs f i l e
cout « "Another ? ' ;
cin » ch; )
while( toupper
{ ch ) = = * Y '
);
ofile.close( ) ;
/ / Output loop, display f i l e content ifstream i f i l e ( "person.txt",
ios::binary ) ; cout « 'The objects written to the f i l e w e r e : . . " « endl; while(
1)(
i f i l e > > p_obj; / / extract person object from f i l e i f (
ifiie.failO ) / / f i l e read f a i l , end-of-file break;
cout « p_obj; / / display person object on console
)
)

As
N
a
m

Copyrighted material
 Chapter 18: Streams Computation with Files 579

e
:

T
e
j
a
s
w
i

A
g
e

5
A
n
o
t
h
e
r

N
a
m
e
:

S
a
v
i
t
h
r
i

A
g
e

2
3

A
n

Copyrighted material
 Chapter 18: Streams Computation with Files 579

o
t
h
e
r

n
The objects written to the file were: . .
Tejaswi
5
S
a
v
i
t
h
r
i

2
3
/ / fio.cpp: input and output operations on file, random access
•include <iostream.h>
•include <fstream.h>
•define READ_SIZE 6
void main()
{
char reader! READ_SI2E ♦ 1 J;
// fstream constructor, open in binary input and output mode fstream fstr( "test.del",
ios::binary|ios::in|ios::out ); // Write the numbers 0 to 9 to file for< int i = 0; i < 10; i++ )
fstr « i; // Set the write (put) pointer, fstr.seekpt i
) ; fstr << "Hello"; // Set the read (get) pointer.
fstr.seekg( 4 ); fstr.readt reader, READ_SIZE ); reader!
READ_SIZE J = 0; // end of string cout « reader «
endl;
)

Run
llo789
Note thai an instance of fstream has l wo file pointers associated with it: a get pointer used
while reading, and a put file pointer used while writing. The statement
fstr.seekpt 2 ) ; sets
the put pointer to an offset 2.
The program first writes ASCII codes of the digits 0 to 9 to the file test - del. moves the put
pointer by an offset 2 from the beginning of the file and the overwrites the numbers 3 through 6
with the string "Hello'* It then reads 6 characters from the offset 4 into the array reader. The last
line of the program will display these 6 characters, which will be llo789. After all writes arc
completed, the contents of the file test.del will be: 01Hello789
The facility for direct file processing is essential in database applications. They perform
extensive data read, write, update, and search activities. These actions require movement of the
file pointers (get or put) from one position to another. This can be easily performed by using the
seek {), read (), and write () functions.
The location at which the mlh object is stored can be computed using a relation: location = m *
sizeoff object) This specifies Ihe offset at which the object is stored in a file. It can be used to
manipulate the m,h object by using the read() or write () functions.

Copyrighted material
 Chapter 18: Streams Computation with Files 579

The program direct. cpp illustrates the mechanism of updating a file by random access. It uses
'he file person.txt to store objects and then these objects can be updated if necessary. The file
H u n t e r s get and put are positioned based on the object to be accessed.

/ / direct.cpp:
accessing objects randomly
•include <fstream.h>
•include <ctype.h> / / For toupper
•include <string.h> / / For strlen
•
d
e
f
i
n
e

M
A
X
N
A
M
E

4
0

c
l
a
s
s

P
e
r
s
o
n

f
private:
char name[ MAXNAME ] ;
i
n
t

a
g
e
;

p
u
b
l
i

Copyrighted material
 Chapter 18: Streams Computation with Files 579

c
:
/ / this function writes the class's data members to the f i l e void
write{ ofstream kos )
i
o s . w r i t e ( na
me, strlen( name )
) ; os << ends;
o s . w r i t e t (char*)&age, sizeof( age ) ) ;
}
/ / this function reads the class's date member from the f i l e .
/ / I t returns nonzero i f no errors were encountered while reading.
int readf ifstream &is )
(
is.getf
name,
MAXNAM
E, 0 ) ;
name[ is.gc
ountO ] =
0;
is.ignore{ 1 ) ; / / ignore the NULL terminator in the f i l e ,
i s . readf (char*)&age, sizeof( age ) ) ; return i s . g o o d O ;
}
/ / stream operator, « overloading
friend ostream & operator « { ostream &os, Person &b ) ; / / stream
operator >> operator overloading friend istream ^operator » ( istream
& i s , Person &b ) ; / / output f i l e stream operator overloading
J;
istream ^operator » ( istream & i s ,
Person &b ) {
cout « "Name: " ;
is >> ws; //
flush input buffer
i s . g e t t b.name,
MAXNAME ) ;
cout « "Age : " ;
is » ws » b.age;
return is;
}
ostream &operator « { ostream tos,
Person &b ) {
os << "Name:
" << b.name <<
endl; os « "Age :
" < < b.age « endl;
return os;
void showt
double val ) (
cout « "Double: " << val << endl;
)
void showt char
*val ) {
cout << "String: " << val << endl;
)

Copyrighted material
 Chapter 18: Streams Computation with Files 579

i
nt
main
() (
showt 420 ); // calls showt int val );
showt 3.1415 }; // calls showt double val )~,
showt 'Hello World\n!" ); // calls showt char *val ); return( 0 );
)

Bun
Integer:
420 Double:
3.1415 String:
Hello World

In ihe above program, three functions named show () are defined, which only differ in their
argument lists: int, double, or char*. The functions have the same name. The definition of several
functions with the same name is called function overloading.
It is interesting to note the way in which the C++ compiler implements function overloading.
Although, the functions share the same name in the source text (as in the example above, show (J).
the compiler (and hence the linker) uses different names. The conversion of a name in the source
file to an internally used name is called name tnangling. For instance, the C++ compiler might
convert the name void showt int) to the internal name Vshowl, while an analogous function with a
char* argument might be called VshowCP. The actual names which are used internally depend on
the compiler and are not relevant to the programmer, except where these names shown in the
example, a listing of the contents of a function library.
A few remarks concerning function overloading are the following:
♦ The usage of more than one function with the same name, but quite different actions should be
avoided. In the above example, the functions show () are still somewhat related (they print
information on the screen). However, it is also quite possible to define two functions, say
lookup (), one of which would find a name in a list, while the other would determine the video
mode. In this case, the two functions have nothing in common except their name. It would
therefore be more practical to use names which suggest the action; say, findnamet) and
getvidmode ().
♦ C++ does not allow overloaded functions to only differ in their return value. The reason is that
processing (testing) of a function return value is always left to the programmer. For instance, the
fragment
printf ("Hello World!\n"); holds no information concerning the return value of the function
printf () (The return value is, in :his case, an integer value that states the number of printed
characters. This return value is practically
{
/ / f i l e does not exist
}
2. Open fail: opening read-only marked file
ofstream outfile( "myfile.dat" ) ;
i f f Unfile ) / / or i f f infile.badf) )
{
/ / f i l e already exist and marked as read only
)
3. Detecting end of file
whilef !i n f i l e .eof{ ) ) / / processes until end-of-file is reached (
/ / process f i l e
)
4. Read fail
i
nfile.
readf. .
.); iff

Copyrighted material
 Chapter 18: Streams Computation with Files 579

infile
.badf
)) {
/ / f i l e cannot be processed further
)
5. Invalid filename
i
nfile.op
enf " | -
*" )
iff !
infile
)
{
/ / invalid f i l e name
}
6. Processing unopened file
i n f i l e . readf. . ) ; / / read file
i f f infile.failf) )
{
/ / f i l e is not opened
>

The program outf ile. cpp illustrates the trapping of all possible errors, which
may be encountered during file processing.

/ / OUtfile.Cpp: writes all the input into the f i l e 'sample.out'

•include <fstream.h>
•include <process.h>
•include <string.h>
void m a i n f )
{
char buff[ 8 0 ] ;
ofstream outfile; / / output f i l e
outfile.openf"sample. o u t " ) ; / / open in output mode i f f
o u t f i l e . badf) ) / / open fail
{
cout < < "Error: sample.out unable to open";
exit f 1 ) ;
c
atch(
ObjecCl
){

// take corrective action for operation_fail

>
c
atchf
Object2
){

// take corrective action for overflow

The following sequence of steps are performed when an exception is raised:

* The program searches for a matching handler.
♦ If a handler is found, the stack is unwound to that point.

Copyrighted material
 Chapter 18: Streams Computation with Files 579

• Program control is transferred to the handler.

* If no handler is found, the program will invoke the terminate () function (explained later). If no
exceptions are thrown, the program executes in the normal fashion.
The program divz ero. cpp illustrates the mechanism for detecting errors, raising exceptions,
and handling such exceptions. It has the class number to store an integer number and the member
function read () to read a number from the console and the member function div () to perform
division operations. It raises exception if an attempt is made to perform divide-by-zero operation. It
has an empty class named DIVIDE used as the throw's expression-id.

/ / dlvzero.Cpp: Divide Operation Validation, {divide-by-zero) •include <iostream.h> class

number {
private:
in
t
num;
public
:
void readf) // read number from keyboard
{
cin >> num.-
)
class DIVIDE {}; // abstract class used in exceptions
int div( number num2 )
{
iff num2.num == 0 ) // check for zero division if yes
throw DIVIDE(); // raise exception
else
return num / num2.num; // compute and return the result
>
never inspected.)- Two functions printf () which differ in their return type could therefore, not he
distinguished by the compiler.
♦ Function overloading can lead to surprises. For instance, imagine a usage of a statement such as
showt 0 ) ;
in the program show, cpp; it is difficult to predict which one of the above three show () functions
is invoked. The zero could be interpreted here as a NULL pointer to a char. i.e.. a (char* > 0, or as
an integer with the value zero. C++ will invoke the function expecting an integer argument, which
might not be what one expects.

2.13 Default Arguments

In a C++ function call, when one or more arguments are omitted, the function may be defined to
take default values for omitted arguments by providing the default values in the function
prototype. These arguments are supplied by the compiler when they are not specified by the
programmer explicitly. The program prnstr. cpp illustrates the passing of default arguments to
function.
// prnstr.cpp: default arguments and message
printing •include <iostream.h>
void showstringf char *str = "Hello World! \n"
)(
cout << str;
)
i
nt
main
() <
showstringt "Here is an explicit argument\n" );
showstring(); // in fact this says: showstring ("Hello World!\n"); return 0;
)

Bun
Here is an explicit
argument Hello World!

Copyrighted material
 Chapter 18: Streams Computation with Files 579

In main (), when the compiler encounters the statement

showstrin
g(); it is replaced
by the statement
showstring( "Hello World! \n" ); internally. When the function parameter is missing, the
compiler substitutes the default parameter in that place.
The possibility of omitting arguments in situations where default arguments are defined is
elegant; the compiler will supply the missing arguments, when they are not specified. The code of
the program by no means becomes shorter or more efficient. Functions may be defined with more
than one default argument.
Default arguments must be known to the compiler prior to the invocation of a function with
default arguments. It reduces the burden of passing arguments explicitly at the point of a function
call. The program defargl .cpp illustrates the concept of default arguments.
Run2
Enter any number:
-10 -ve number
Run3
Enter any number: 0_
Abnormal program
termination
The prototype of the function what_sign (): void
what_sign( int num ) throw0 indicates that it does not
raise any exception, but the statement throw zeroO; //
unspecified exception
raises the exception. It invokes the default exception handler which aborts the execution of the
program (see Run3) though there exists an explicit exception handler in the caller of this function.

19.7 Catch All Exceptions

C++ supports a feature to catch all the exceptions raised in the try-block. The syntax of the catch
construct to handle all the exceptions raised in the try block is shown in Figure 19.7.
three dots:
indicate catch all
exceptions

catch
( . . . )
{
// actions for handling an exception
}

Figure 19.7: Syntax of catch all construct

The three dots in the catch(...) indicates that it catches all types of exceptions raised :n its
preceding try-block. The program catal 11. cpp illustrates the mechanism of handling all the
exceptions raised by a single handler.

/ / catall1.cpp:Ali exceptions are

caught •include <iostream.h>
class excep2 {);
void
main () {
try
{
cout « "Throwing uncaught exception" « endl;
•throw excep2();
)
catch* ... ) // catch all the exceptions (
// action for exception
cout « "Caught all exceptions" « endl;
// Array creation unsuccessful, Request > ARR_SIZE

Copyrighted material
 Chapter 18: Streams Computation with Files 579

try
{
cout « "Trying to create object a2(15)...";
array a2(15); // create array, causes
exception cout « "succeeded" « endl; a2[3] = 3;
// valid access
)
catch( arra
y::SIZE ) {
// action for exception
cout « "....Size exceeds allowable Limit" « endl;
)
catch( array::RANGE ) // true if throw is executed in try scope {
// action for exception
cout « "... .Array Reference Out of Range" « endl;
)
)

Run
Maximum array size allowed = 10
Trying to create object al(5)...succeeded
Trying to refer al[5]...succeeded..al[5] * 10
Trying to refer al[15].........Array Reference Out of Range
Trying to create object a2(15).............Size exceeds allowable Limit

The one-argument constructor of the class array,

array( int SizeRequest ) // constructor
throws an exception,
throw SIZEO ;
if an attempt is made to create an array beyond the allowable range. The statement
i f f i < 0 || i > size )
throw RANGE(); // throw abstract object throws an exception if an attempt
is made to access an array element by using invalid index (lower than minimum bound
or higher than the maximum bound).

19.9 Handling Uncaught Exceptions

The uncaught exception handling mechanism relies on two library functions, terminate
0 and unexpected (), for coping with exceptions unhandled explicitly. C++ supports
the following special functions to handle uncaught exceptions in a systematic manner:
♦ terminate{)
♦ set_terminate( )
♦ unexpected{)
♦ set_unexpected()

terminate()
The function terminate ( ) is invoked when an exception is raised and the handler is
not found. The

Copyrighted material
 Chapter 19: Exception Handling 723

the terminate (), the user should define t_func function. This t_func function can be installed by set_terminate
as the termination function. The installation of t_func allows the user to implement any action that is not taken
by abort (). The syntax of the set_terminate function declared in the header file except. h is as follows:
typedef void (*terminate_function)();
terminate_function set_terminateI terminate_function t_func ) ; / / Define your termination scheme
terminate_function my_terminate ( void ) {
// Take actions before terminating // should not
throw exceptions exit(l); // must end somehow
)
/ / Register your termination function set_terminate(
my_terminate ) ;

The program myhand. cpp handles uncaught exceptions with the user specified terminate function.
/ / myhand.cpprAll exceptions are not caught, executes MyTerminate ( )
•include <iostream.h>
•include <except.h>
Class excepl ( > ;
class excep2 {};
void MyTerminate ( )
{
cout « "My Terminate is invoked"; •xit( 1 ) ;
)
void main{) {
■exterminate{ MyTerminate ) ; // sets to our own terminate function try
{
cout « "Throwing uncaught exception\n"; throw excep2{);
)
catch( excepl ) {
/ / action for exception
cout « "Caught exception, excepl\n";
}
// program abort(} here; MyTerminate( ) will be called cout « "I am not
displayed";
)
Run
Throwing uncaught exception My
Terminate is invoked

#i defargl.cpp: Default arguments to functions nclude <iostream.h>

vo id PrintLine( char = int = 70 ) ;
vo id main (>
{
PrintLinet); -PrintLinet // uses both default arguments
'!* ) ; PrintLinet '*', 4C // assumes 2nd argument as default
> PrintLinet 'R', 55 ) // ignores default arguments
// ignores default arguments
i
VO { id PrintLinet char ch, int RepeatCount ) int i;
cout << endl;
fort i = 0; i < RepeatCount; i++ ) cout << ch;

Bm
M M ; M l M M M M M M j M M M l M M M M M M M M J ! M l M M M M M MM!!!!!! I
.......................................................... . . . . . . . , . . * . . * . , . „ * * * . » .

RRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR

The feature of default arguments can be utilized to enhance the functionality of

the program, without the need for modifying the old code referencing to functions.
For instance, the function in the above program
void PrintLinet char = int = 70 );
prints a line with default character'-' in case it is not passed explicitly. This
function can be enhanced to print multiple number of lines, whose new prototype is
void PrintLinet char = int ~ 70, int = 1 );
It may be noted that in the new function, the last parameter specifies the number of
lines to be printed and by default, it is 1. Therefore, the old code referring to
this function need not be modified and new statements can be added without affecting
the functionality. The program def arg2 . cpp has extended the capability of
defargl.cpp program.

Copyrighted material
 Chapter 19: Exception Handling 723

/ * defarg2.cpp: Default arguments to functions

Extending the functionality of defargl.cpp module */ •include
<iostream.h>
void PrintLinet char = int = 70, int = 1 ) ;
void main ()
{
PrintLinet); / / uses both default arguments
PrintLinet '!'); / / assumes 2nd argument as default
PrintLinet 1 *', 40 ); // ignores default arguments
PrintLinet 'R', 55 ); / / ignores default arguments
// new code. Note: old code listed above is unaffected
PrintLinet , 25, 2 ) ;

19.10 Exceptions in Operator Overloaded Functions

The program interact. cpp illustrates the mechanism for handling exceptions in the vector class,
while creating its objects and accessing its elements either for a read or wnte operation. It
overloads the operator [ ] to simulate the array operations on the user defined data type.

/ / Interact.cpp: interactive program raises exception for improper data •include <iostream.h>
•include <process.h>
const int VEC_SIZE = 10; / / maximum vector s i z e , that can be allocated
class vector
{
private:
int *vec; / / pointer to array for vector elements
int size; / / maximum vector s i z e
public:
class SIZE { ) ; / / Size abstract c l a s s
class RANGE { } ; / / Range abstract class
vector( int
SizeRequest ) {
i f ( SizeRequest <= 0 | j SizeRequest > VEC_SIZE )
throw
SIZEO ;
size =
SizeRequest;
vec = new
int[ size ) ;
J
-vector!) / / destructor
{
delete vec;
)
/ / subscripted operator overloading int
& operator!]( int i ) ;
);
/ / subscripted operator overloading int
& vector::operator!] ( int i ) {
i f ( i < 0 | | i >= s i z e )
throw RANGE!); / / throw abstract
object return v e c f i ] ; / / valid reference
)
void
m
ai
n
(
)
{
int s i z e , data, index;
cout « "Maximum vector size allowed = " « VEC_SIZE << endl;
try
{
cout « 'What is the size o f vector you want to create: " ; cin » s i z e ;

Copyrighted material
 Chapter 19: Exception Handling 723

cout << "Trying to create object vector vl of size = " « s i z e ; vector v l ( s i z e ) ; / / create
vector cout << " . . . s u c c e e d e d " << endl;
cout « "Which vector element you want to access (index) : " ; cin » index;
cout « "What is the new value for vl[ w << index << " ] : " ; cin >> data;
cout « "Trying to modify al [ " « index « " ] . . . " ;
vl[index] = data;
cout « "succeeded" « endl;
cout << "New Value of al [ " « index « ■ ] = " « vl[index];
)
catch( vector::SIZE ) {
/ / action for exception cout << "failed" << endl;
cout « "Vector creation size exceeds allowable limit"; exit( 1 ) ;
)
catch( vector: :RANGE ) / / true if throw is executed in try scope (
/ / action for exception cout « "...failed" « endl; cout « "Vector reference out-of-range"; exit ( 1 ) ;
}
}

Bual
Maximum vector size allowed = 10
What is the size of vector you want to create: 5
Trying to create object vector vl of size - 5 . . . succeeded
Which vector element you want to access (index): 2
What is the new value for vl[ 2 ] : 2
Trying to modify al[ 2 ]...succeeded
New Value of al( 2 ] = 7

Run2
Maximum vector size- allowed = 10
What is the size of vector you want to create: £
Trying to create object vector vl of size = 5...succeeded
Which vector element you want to access (index) : 10.
what is the new value for vl( 10 J : 2
Trying to modify al[ 10 ] ...failed
Vector reference out-of-range

Run3
Maximum vector size allowed = 10
What is the size of vector you want to create: 1£ Trying to create object vector vl of size = 15 Vector creation
s i z e exceeds allowable limit

Note:
Runl: All operations are valid, no exception is generated
s_age = m;
*.

virtual int
GetAgetvoid)
{
return s_age;
)
■■

void
m
ai
n
(
)
<
*
int
father_age
; int
son_age;
Father *basep; / / pointer to father objects cout < <
"Enter Age o f Father: " ; c i n » father_age;
try
(

Copyrighted material
 Chapter 19: Exception Handling 723

basep = new Father ( father_age ) ; / / pointer to father

)
catch( WRONG_
AGE ) {
cout < < "Error: Father's Age is < 0 " ;
exit! 1 ) ;
}
cout « "Father's Age: " ;
cout << basep->GetAge( ) << endl; / / c a l l s father::GetAge delete basep; / / remove
Father class object cout « "Enter Age o f Son: " ; cin » son_age;
try
{
basep = new Son( father_age, son_age ) ; / / pointer to son
)
catch( WR0NG_A
GE J {
cout « "Error: Father age cannot be less than son a g e ! ! ! " ; e x i t ( 1 ) ;
)
cout « "Son's Age: ■ ;
cout « basep->GetAge( ) < < endl; / / c a l l s son::GetAge( ) delete basep; / / remove Son
class object
}

Run1
Enter Age o f Father:
15. Father's Age:
4 5 Enter Age of
Son: 2£ Son's Age:
20

Bun2
Enter Age of Father:
2J2 Father's Age:
2 0 matrix <T>
operator -
( matrix <T> b ) ;
matrix <T>
operator *
( matrix <T> b ) ;
void operator -
( matrix <T> b ) ;
int operator ==
( matrix <T> b ) ;
};
template <class T>
matrix<T> matrix<T>::operator + { matrix <T> b ) {
matrix <T> c ( MaxRow, MaxCol ) ; int i , j ;
i f ( MaxRow ! = b . MaxRow | | MaxCol ! = b . MaxCol )
throw MatErrorO; f o r ( i = 0 ; i < MaxRow; i +
+ )
f o r ( j = 0 ; j < MaxCol; j++ )
c . M a t P t r [ i J [ j ] = M a t P t r ( i ) [ j ] + b.MatPtr[ i ] [ j ] ; return( c ) ;
)
template <class T>
matrix <T> matrix<T>: : operator - ( matrix <T> b ) {
matrix <T> c { MaxRow, MaxCol ) ; int i , j ;
i f ( MaxRow ! = b . MaxRow | | MaxCol • = b . MaxCol )
throw MatErrorO; f o r ( i = 0 ; i < MaxRow; i+
+}
for( j = 0 ; j < MaxCol; j++ )
c . M a t P t r t i ] [ j ] = M a t P t r [ i ] [ j ) - b . M a t P t r [ i ) [ j ] ; return( c ) ;
}
template <class T>
matrix <T> matrix<T>::operator * ( matrix <T> b ) {
matrix <T> c { MaxRow, b . MaxCol ) ; int i , j , k ;
i f ( MaxCol ! = b . MaxRow )

Copyrighted material
 Chapter 19: Exception Handling 723

throw MatError( ) ; f o r ( i = 0 ; i < c . MaxRow;

i++ )
f o r t j = 0; j < c . MaxCol; j + + J {
c.MatPtrfi][j] = 0;
f o r t k = 0; k < MaxCol; k++ )
c . M a t P t r [ i J [ j ] += MatPtr[i J ( k j * b . M a t P t r [ k ] [ j ]
)
return t c ) ;
}
template <class T>
int matrix<T>: : operator == ( matrix <T> b ) {
int i, j;
i f t MaxRow ! = b . MaxRow | | MaxCol ! = b . MaxCol ) return( FALSE ) ;
How many columns ? 2
Matrix{0.0) ~ ? 1 Macrix[0,1)
= ? 2 Enter Matrix B details. . .
How many rows ? 2. How
many columns ? 1 MatrixJO,0]
= ? I Matrix[1,0] = ? Z Matrix
A is . . . 1 2
Matrix B is . . .
1
2
Error: Invalid matrix order for addition Error: Invalid matrix order for
subtraction E = A * B . . . 5
{Is matrix A. equal to matrix B) ? No
In the definition of matrix class's member functions, it can be observed that the validity of a matrix operation
is handled by exceptions. For instance, in the overloaded member function operator +. the statement
i f { MaxRow != b.MaxRow | | MaxCol != b.MaxCol ) {
cout « 'Error: Invalid matrix order for addition"; throw MatError;
>
raises an exception MatError if there is a mismatch in the row and column count of the two matrices involved in
the addition operation. Note that, function templates can also raise exceptions.

19.13 Fault Tolerant Design Techniques

Fault tolerant software design techniques can be classified into the following:
(1) N-version programming
(1) Recovery block
These schemes correspond to the hardware fault tolerance methods, static redundancy (fault masking or voting)
and dynamic redundancy respectively.
N-Version Programming
In this technique N-programmers develop N algorithms for the same problem without interacting with each other.
All these algorithms are executed simultaneously on a multiprocessor system and the majority solution is taken as
the correct answer.
Recovery Block
The recovery block structure represents the dynamic redundancy approach to software fault tolerance. It consists
of three software elements: (1) primary routine, which executes critical software functions; (2) acceptance test,
which tests the output of the primary routine after each execution; and (3) alternate routine, which performs the
same function as the primary routine (but may be less capable or slower), and is invoked by an acceptance test
after detection of a fault.
19.14 Case-Study on Software Fault Tolerance
A simple example is chosen for the study of fault tolerance programming and the same is used
for implementation in C++. C++ does not provide any explicit constructs for fault tolerance,
however, the constructs throw, try, and catch can be suitably used to simulate the action of fault
tolerance. These exception handling constructs are suitable for implementing the recovery block
technique.
Consider a procedure (P) for computing:
sum = i ' +j ' +k* ;
The body of P is the sequential composition of the operation,
(cl) i ■ i+j; (c2) i =
i+k;
Thc behavior of the above procedure P can be examined by considering various versions of the
procedure P (proc p) for different values of the variables i, j, and k.
Version 1:
proc P signal OW begin
i = i + j [ OV -> signal OW ] ;

Copyrighted material
 Chapter 19: Exception Handling 723

i = i + k [ O V - > i = i - j ; signal OW ) ;
end
The semantic definition of the assignment operator = specifies that whenever the evaluation
of the right hand side expression terminates exceptionally (overflow occurs, OV), no new value is
assigned to the left hand side variable. Then, P will terminate exceptionally by executing the
recovery block (if it exists) and signals an OW (overflow word) exception label in the final state.
Data Case 1 : i <- MaxValue, j <- MaxValue, and k < - (-MaxValue) Operation i + j +k (as per data case 1)
is valid, but i + j exceeds the representation limit leading to an exception.

Version 2:
proc P signal OW begin
i = i + k [ OV - > signal OW ] ;
i = i + j [ OV - > i = i - k ; signal OW ] ;
end
This version terminates with a valid final state for the data case 1. DataCase2i i < - MaxValue, j
(MaxValue) Operation i + j +k (as per data case 2) is valid, but (i+k)
< - (-MaxValue), and k < -
exceeds the representation limit leading to an exception.
Version 3:
proc P signal OW begin
j = j + k [ OV - > signal OW ] ;
i = i + j [ OV - > j = j - k ; signal OW ] ;
end
This version terminates with a valid final state for the data case 1 and case 2.

Copyrighted material
Chapter 2: Moving from C to 59
C++
v
o
i
d

P
r
i
n
t
L
i
n
e
t

c
h
a
r

c
h
,

i
n
t

R
e
p
e
a
t
C
o
u
n
t
,

i
n
t

n
L
i
n
e
s

{
int i,
j;
fort j
= 0;
j <
nLin
es;
j++ )
(

Copyrighted material
 cout « endl; as 1,
fort thereby
i the new
function
=
docs the
0 same
; operation
as that of
i the old
one.
< Thus,
default
R
argument
e
p s feature
e can be
a potentiall
t y utilized
C in
o extending
u the
n function
t
without
;
modifyin
i g the old
+ code.
+ Note that
all
> argument
s in a
c multiple
o
u argument
t function
need not
« have
default
c values.
h
;
) 2.14
} Keyw
Bun ord
typed
■ I ............1 1 1 1 1 1
11 i 1 1 i 111 i
ef
11111111111 1 1
1 111111111111 1 111 i 11 i i The
11 i i i i i keyword
typedef is
R
allowed
RRRRRRRRRRRRRRRRRRRRRR
RRRRRRRRRRRRRRRRRRRRRR in C++,
RRRRRRRRRR but no
&&&&&&&&&&&&&&&&&&&&&&&&& longer
&&&&&&&&&&&&&&&&&&&&&&&&& necessary
The following statements in the , when it
above two programs is used as
PrintL / / uses both a prefix
inet);
PrintL
)default
arguments / /
in enum,
struct, or
ine t assumes 2nd union
'!
PrintL fargument as
default / /
declaratio
inet ' * 4 ignores default ns. This
PrintL 0 arguments / / is
inet ) ignores default illustrated
'R ; arguments in the
5
following
5
} example:
;
arc the same. Although, the
functionality of the function
PrintLine, is enhanced indefarg2
.cpp program, the old code
referring to it remains unaffected in
terms of its functionality; the
compiler supplies the last argument

Copyrighted material
 struct somestruct a
{ t
i
.
n
t
I
a n
;

d C
o ,
u
b
l t
e h
e
d
;
char string [ 8 0 ] ; s
a
When a struct, enum, or any other m
compound type is defined, the tag e
of this type can be used as type
name (somestruct is the tag in the
v
above example). For instance, the
a
statement
r
s
i
o
a
m
b
e
l
s
e
t
r
i
u
s
c
t
d
w
e
h
f
a
i
t;
n
d
e
e
d
f
i
a
n
s
e
s
s
truct
t
somestru
h ct what;
e Thus, the
s use of
t keyword
r struct in
u the
c structure
t variable
u is default.
In C++,
r
the
e
members
v of the
a structure
r variables
i are
a accessed
b similar to
l C. The
e statement
w wha 3.1
t.d 415
h = ;

Copyrighted material
 738 Mastering C++

DataCase3: i <- (-MaxValue), j <- MaxValue, and k <- (MaxValue)

Operation i + j +k (as per data case 3) is valid, but j +k exceeds the representation limit leading to an
exception.
Recovery Block for Procedure P: i <- i + j+k: Ensure
no exception By Version - 1 Else
B y Version - 2
Else By Version - 3 Else
FAIL
Recovery Block Implementation
The recovery block technique can be implemented by nesting the exception handling constructs of C++. To
understand the concepts of fault tolerant programming, consider a computer system having a 4-bit processor,
supporting both signed and unsigned numbers. Some of its characteristics are the following.
♦ Maximum signed number can be represented is 7 (2**1 -1).
♦ Maximum unsigned number can be represented is 15 (2* - 1).
♦ Overflow will be indicated if the result exceeds the limit of representation.
The program recovery .cpp handles all the three data cases and demonstrates the characteristics desired in a
fault tolerance program.

/ / recovery.cpp: Recover Block of sum( i , j , k ) •include

<iostream.h>
const int MAX_SIG_INT = 7; / / say, maximum signed integer number const int MAX_UNSIG_INT = 1 5 ;
/ / say, maximum unsigned integer number class OVERFLOW ( } ; / / Overflow abstract class int sum( int i ,
int j , int k ) {
int result; try
{
/ / Versionl procedure result
= i+j;
i f { result > MAX_SIG_INT )
throw OVERFLOW( ) ; result =
result+k; if ( result > MAX_SIG_INT )
throw OVERFLOW{ ) ; cout « "Version-1
succeeds" « endl;
>
catch( OVERFLOW )
{
cout « "Version-1 fails" « endl; try {
/ / Version2 procedure result
= i+k;
i f ( result > MAX_SIG_INT )
throw
OVERFLOW( ) ;
result = result*j ; i f {
result >
MAX_SIG_INT )
throw OVERFLOW f) ; cout «
"Version-2 succeeds" « endl;
)
catch( O
VERFL
OW ) {
cout << "Version-2 fails" <<
endl; try (
/ / Version3
procedure
result = j+k;
if ( result > MAX_SIG_INT )
throw
OVERFLOW( ) ;
result = result+i;
i f ( result >
MAX_S1G_INT )
throw OVERFLOW( ) ; cout «
"Version-3 succeeds" « endl;
}

Copyrighted material
 738 Mastering C++

catch( O
VERFL
OW ) (
cout << "Error: Overflow. All versions failed" « endl;
)
}
)
return result;
}
v
o
i
d

m
a
i
n

(
)

(
int result;
cout << "Sum o f 7 , - 3 , 2 computation..." « endl;
result - sum( 7 , - 3 , 2 ) ; / / versionl computes
cout << "Sum = " << result << endl;
cout « "Sum o f 7 , 2 , - 3 computation..." « endl;
result = sum( 7 , 2 , - 3 > ; / / version2 computes
cout « "Sum = " « result « endl;
/ / Device data such that version-3 succeeds
cout « "Sum of 3 , 3 , 2 computation..." « endl;
result = sum{ 3 , 3 , 2 ) ; / / all version fails
cout « "Sum = " « result « endl;
}

Bun
Sum o f 1, - 3 , 2
computation. . . Version-1
succeeds Sum = 6
Sum of 7 , 2 , - 3
computation... Version-1
fails
Version-2 succeeds Sum
=6
Sum of 3 , 3 , 2 computation...
Version-1 fails Version-2 f a i l s
Error: Overflow. All versions failed Sum =
8

19.15 Memory Allocation Failure Exception

The operator new tries to create an object of the data type Type dynamically by allocating (if
possible) sizeof (Type) bytes in free store (also called the heap). It calculates the size of Type
without the need for an explicit sizeof operator. Further, the pointer returned is of the correct
type, pointer to Type, without the need for explicit casting. The storage duration of the new object
is from the point of creation until the operator delete destroys it by deallocating its memory,
or until the end of the program. If successful, new returns a pointer to the new object. By
default, an allocation failure (such as insufficient or fragmented heap memory) results in the
predefined exception xalloc being thrown. The user program should always be prepared to
catch the xalloc exception before trying to access the new object (unless user-defined new-
handler function is defined). The program newl. cpp illustrates the simple mechanism of
handling exceptions raised by the new operator.

Copyrighted material
 738 Mastering C++

/ / newl.cpp: new operator memory allocation test

#include <except.h>
•include <iostream.h>
void main(void)
{
int * data; int
size;
cout < < "How many bytes to be allocate: " ;
cin » size;
try
(
data = new i n t [ size ] ;
cout « "Memory allocation success, address = * « data;
)
catch( xalloc ) / / new fail exception
{ / / Enter this block only of xalloc is thrown.
/ / You could request other actions before terminating
cout « "Could not allocate. Bye . . . " ;
exit(l);
)
delete data;
)

Bml
How many bytes to be allocate: 100
Memory allocation success, address = OxlSbe

Copyrighted material
 Chapter 20: OO Analysis, Design and Development 755

apart at the end of the project and individuals are scattered and reassigned to other projects or they might
change organization (which is most common in software industry). Hence, peopleware approaches to
software productivity often achieve results several times greater than technical approaches.

20.5 Software Life Cycle: Fountain-Flow Model

The conventional model requires a large amount of time to be spent in formulating the problem specifi -
cations. It delays the writing of code, and programmers may be impatient. In addition, the conventional
life cycle model permits little feedback from the end user until the coding stage, which is at the end of
the life cycle. At this point, if the system does not meet the specifications, the design and code becomes
very expensive. It is often easier to change an existing system than to redevelop the specifications and
design a new system. The conventional model fails to address software reusability hence the existing
software is not usable as a starting point.
Process

Object A Object C

Input Output

Method

Object B Object D

Figure 20.6: Objects interacting with each other

roblem into a set of abstract object types or resources in the system and a set of operations that manipulate instances in the system; and a set of oper
OO methodology allows the end-users, analysts, designers, and programmers to view various
components of the system in the same way, thus simplifying the process of mapping the customer
requirement to the implementation model. The real-world entities are represented in the form of objects.
Objects play the central role in all phases of the software development process. Therefore, there is a high
degree of overlap and interaction among the phases. The use of waterfall model in the design of OO-
based system does not allow overlap and interaction among the development phases. This problem can
be circumvented by using a model that resembles a fountain. The resultant model is called/ow/i/am all the
members of the structure have public access. The private and protected members of a structure can be
accessed only within the structure. Public members of a structure arc accessible to both member
functions and its instances (structure variables). Internal functions of a structure are privileged code and
they can see all the features of a structure, but external code can see only the public features.
A definition of the structure point is given in the code fragment below. In this structure, two int data
fields and one function draw ( ) are declared, struct point
{
int x, y; / / coordinates
void draw (void); / / drawing function
);

A similar structure could be a part of the painting program used to represent a pixel in the drawing.
The following are the points to be noted about structures:
• The function draw ( ) . which occurs in the structure body is only a declaration. The actual code of the
function, or in other words, the actions to be performed by the function arc located elsewhere in the
code section of the program. Member function can also be defined within the body of a structure.
♦ The size of the structure point is just two integers. Though a function is declared in the structure, its
size remains unaffected. The compiler implements this behavior by allowing the function draw ( ) to
be known only in the context of the point structure.
The point structure could be used as follows:
point a , b; I I two points on the screen
a.x = 0; II define f i r s t dot
a . y = 10; II and draw it
a.draw ( ) ;
b = a; I I copy a to b
b . y = 20; II redefine y-coordinate
b.draw ( ) ; I I and draw i t
The function draw (). which is a part of the structure, is selected in a manner similar to the selection of
data fields; i.e., using the field selector operator ( . ) with value structures or - > with pointers to
structures.

Copyrighted material
struct date

structure data
members

) structure
member I
function

Copyrighted material
762 Mastering C++

C Class
Certificatio
n
Figure 20.10: Steps in class design and development
A design framework states that every class should be designed so as to be amenable for use as
a component by other classes. The class design principles focuses on design for reuse and it
includes the following:
I. Design of (abstract) class rather than one shot objects.
?. Design of class interfaces (accessors, methods) rather than of attributes and transitions. ?
Standardization of interfaces, leading to the specification of interoperable subclasses and the cre-
ation of applications frameworks.
Design of reliable interaction protocols, often supplementing pure event-driven models.
values for the coordinates of the center of the circle and one value for the radius. Analogous to the
point structure, a draw() could be declared in the circle structure which would draw the circle.

The program date2 .cpp is C++ equivalent of the earlier program datel .cpp. It illustrates the
concept of associating functions operating on structure members as shown in Figure 2.10. The
structure date has both the data members and functions operating on them. The user accesses the
member functions additionally, when compared to C's structure using the dot operator.

I I date2.Cpp: displaying birth date of the

authors •include <iostream.h> struct date
{ //specifies a structure
in
t
day;
int
mont
h;
int
year;
void
show
!) {
cout < < day « " - " « month « " - " « year << endl;

Copyrighted material
762 Mastering C++

)
};
v
oid
mai
n{ )
{
date dl = ( - 2 6 ,
3 , 1958 ) ; date d2 =
( 14, 4 , 1971 ) ; date
d3 = ( 1, 9. 1973
};
cout < < "Birth Date of the First Author: " ;
dl.show( ) ;
cout « "Birth Date of the Second Author: ■ ;
d2 .showO ;
cout < < "Birth Date of the Third Author: " ;
d3.show( ) ;
)

Run
Birth Date of the First Author: 26-3-1958
Birth Date of the Second Author: 14-4-1971
Birth Date of the Third Author: 1-9-1973
In m a i n ( ) , the statements
d
l
.
s
h
o
w
(
}
;

d
2
.
s
h
o
w
(
)
;

d
3
.
s
h
o
w
{
)
;
invoke the function s h o w ' ) defined in the structure date.

2.16 Type Conversion

The basic data types can be used with great flexibility in assignments and expressions, due to the
implicit type conversion facility provided, whereas with the user-defined data types, the same can
be
Appendix A: C++ Keywords and Operators 771

Description: It causes control to pass to the statement following the innermost

enclosing while, do. for, or switch statement.

Copyrighted material
762 Mastering C++

Example:
fort i =
0 ; i < n;
i++ ) (

i f ( wants_to_terminate_loop )
break; / / transfers control to the next statement outside loop
)

case: specify actions when the switch

expression matches with it Syntax: case
<constant expressions
where <constant expression> must be a unique integer constant value.
Description: The list of possible branch points within switch <statement> is
determined by the matching case statement within the switch body. Once a value is
computed for <expression>, the list of possible <constant expression> values
determined from all case statements is searched for a match. If a match is found,
execution continues after the matching case statement until abreak statement is
encountered or till the end of switch is reached
Example:
switch( figure_type ) / / figure_type - s character variable
{
case * 1' :
draw_line( x l , y l , x 2 , y2 ) ;
break; / / transfers to next statement to switch case *C :
draw_circle( x , y , r ) ;
break; / / transfers to next statement to switch

default: / / execute i f none of the cases match with switch expression

cout « "invalid figure code"; break; / / i t can be omitted
)

catch: capture
exception thrown
Syntax:
catch( <exception-
object> )
Description: An exception thrown in the program is caught by the catch statement.
It follows try statement and is responsible for taking corrective actions in response
to an exception.
Example:
class div_by_zero { ); / / empty class

int div( int

a , int b )
{
i f ( b == 0 )
throw div_by_zero( ) ; / / divide by zero error; return a/b;

Copyrighted material
 776 Mastering C++

to be its friend, should explicitly declare it as its friend. Friend function or class cannot access members of a
class to which it is a friend directly; it has to access them using class objects.
Example:
class
stars {
private:

i n t magnitude-friend galaxy; / / galaxy i s friend

class
);
class galaxy (

void
func() {
stars s i ;
s i . magnitude = 100; / / valid since galaxy is a friend of stars

}
>;

The above declaration states that, the class galaxy can access all the members of the class s t a r s but not vice-
versa.

goto: transfer control

Syntax: goto <identifier> ;

Description: It transfers control to the specified location. Control is unconditionally transferred to the location
of a local label specified by < i d e n t i f ier>.
Example:
Again:

goto A g a i n ; Note that Labels must be

followed by a statement.

if: actions when the if condition succeeds

Syntax:
i f ( <expression> ) / / i f statement
<statementl>; i f ( <expression> ) / / i f -else
statement
<statementl>;
else
<statement2>;
Description: It transfers control conditionally to a required statement based on the conditional result. If
<expression> is nonzero when evaluated, <statementl> is executed. In the second case, <statement2> is
executed if the <expression> is zero. An optional e l s e can follow an i f template: declare generic
functions or classes
Syntax: template < template-argument-list > declaration
Description: It constructs a family of related functions or classes. It can be used to declare
function templates and class templates.
Examples:
template < class T >
swapt T & a , T & b ) / / function template {
T temp; temp
= a; a = b; b =
temp;
)
template <class T> / / class template
class myclass
{
T a;

)f

Copyrighted material
 776 Mastering C++

thia: pointer to current object___________________________________________________

Syntax:
this / / address of the class in which i t i s referenced
this->member / / access a member
Description: It is a predefined pointer variable within every class and points to the current
object, this is passed as a hidden argument in all calls to non-static member functions. The
keyword this is a local variable available in the body of any nonstatic member function. The
keyword this does not need to be declared and is rarely referred to explicitly in a function
definition. However, it is used implicitly within the function for member references.
Example:
class date (
int day;
void i n i t t )
{
this->day = 1 ; / / same as day = 1
)
};
If x.func ( y ) is called: where y is a member of x, the variable this is set to &x and y is set to
this->y, which is equivalent to x.y.
throw: raise an exception __________'__________________________________________
Syntax: throw object; / / object of a class
Run
int{10.4) = 10
int<10.99) = 10
b = 420.5
a = int(b) = 420
b = float(a)+1.5 = 421.5

2.17 Function Templates

Templates provide a mechanism for creating a single function possessing the capability of several functions,
which differ only in their parameters and local variables data type. Such a function is called function template. It
permits writing one source declaration that can produce multiple functions differing only in their data types.
The general format of a template function is depicted in Figure 2.12. A function generated from a function
template is known as template function, which is created by the compiler internally and is transparent to the user.
Keyword for declaring function template
name of the template data-type

Function parameters of type

template, primitive or user-defined
template < class Tl, class T2, ReturnType FunctionName ( Arguments of
type Tl and T2, ... ) {
// local variables of type Tl, T2, or any other
// function body, operating on variables of type Tl, T2
// and other variables
)

Figure 2.12: Syntax of function template

The syntax of template function is similar to a normal function, except that it uses variables whose data
types are not known until they are invoked. Such unknown data types (generic data types) are resolved by the
compiler and are expanded to the respective data types (depending on the data type of actual parameters in a
function call statement). A call to a template function is similar to that of a normal function. It can be called
with arguments of any data-type. The complier will create functions internally without the user intervention,
depending on the data types of the input parameters. The function template for finding the maximum of two
numbers is shown below:
template <class T>
T max{ T a, T b ) (
if( a > b )
return a;
else
return b;
)

Copyrighted material
 776 Mastering C++

The program mswap. cpp illustrates the need for function templates. I t defines multiple swap functions for
swapping the values of different data types.

Copyrighted material
Appendix B: C++ Library Functions
Function Description Include File
abort( ) abnormally terminates a program atdlib.fa
aba() return* Ihe absolute value of an integer aith.h
acoaf) calculates the arc cosine matb. b
•■ctiaat) converts dale and time 10 ASCII tiBM. b
talad calculaies the arc sine saath.b
ictii a condi::'Mi and possibly aborts aaaaxt.h
atan() calculates the arc tangent atatb.b
atan2() calculates the arc tangent of y/x n int h . h
ataxitt) registers an exit function ■tdlib.b
ato«(> converts a string to a floating-point number autb.h
atoiO converts a string to an integer ateUib.h
■toll) converts a string 10 a long integer • t d l i b. b
baaarcb( ) binary search of an array • t d l i b. b
callocO allocates main memory • t d l i b. b
cailO rounds up saath.b
claar*rr() resets error indication ■tdio.b
clock( ) determines processor time t 1M . b
coi() calculates the cosine of a value rnath.b
coab() calculates the hyperbolic cosine of a value ma t h. h
otimaO converts date and lime to a string tiaa. h
•Kit( ) terminates program • t d l i b. h
CabaO returns the absolute value of a floating-point number sutb.b
f c l o a e t) closes a stream • t d i o. b
f«cf : detects end-of-file on a stream ■tdio.b
t arror( ) detects errors in a stream • t d i o. b
£ f luaht) flushes a stream ■tdio.b
fgatcf.) gets character from stream ■tdio.b
fgatpo«<) gets OK current file pointer ■tdio.b
fsataO gets a string from a stream ■tdio.b
floorO rounds down sutb.b
fmod() calculates x modulo y. the remainder of xJy sutb.b
t opanf) opens a stream • t l i o. b
fprintf< > writes formatted output to a stream ■tdio.b
fpute( ) puts a character on a stream ■tdio.b
fputaO outputs a string on a stream ■tdio.b
Ir e a d t ) reads data from a stream ■tdio.b
fraa<) free allocated Mock alloc.b
':«-;«"■ ) associates a new file with an open stream ■tdio.b
fraxpf) splits a double number into its mantissa and exponent sutb.b
facanf< ) icans and formats mpui from a stream ■tdio.b
MO repositions a file pointer on a stream ■tdio.b
faatpoal) positions the file pointer of a stream ■tdio.h
fatat() gels file statistics ■ys\atat.h
ftalK) returns the current file pointer ■tdio.b
fwritaO writes to a stream ■tdio.b
flatcO gets a character from a stream • t d i o. b
gatcbarO gets a character from stdin ■tdio.b
gataO gets a suing from stdin ■tdio.b
iaalnuat ) character classification macro ctypa.b
laalpbaO character classification macro ctypa.b
iaaaciiO character classification macro ctypa.b
LacntrK) character classification macro ctypa.h
*adigit( ) character classification macro ctypa.b
i«graph() character classification macro ctypa.b
ii l o w a r ( ) character classification macro ctypa.h
iaprint( ) character classification macro ctypa.h
iapunct( ) character classification macro ctypa.h
iaapacef.) character classification macro ctypa.b
iauppar() character classification macro ctypa.b
iexdigit( ) character classification macro ctypa.h

7XX

Copyrighted material
Appendix B: C++Library Functions 789

Function l a b a l ) ■ a l l o c ( ) a a a e h r O r a mac r a p { ) a a m c p y ( ) a a a a e t ( ) radlrO

ldMCp ( ) -ktiMO aeaaf( )
parror( ) pow() print*() pute() putchar ( ) putaf) raiaaO •atbuf( )
logo rand(j aatjapO
loglOf) reall o c ( ) algnaK)
remoY«<) aia<)
rmaiit ( ) alnhO
aprlntf( )
Brandi )
Description Include File
aacanr( ) gives long absolute value calculates x • 2"'
aath.h
■tat( ) divides two longs, returning quotient and remainder
aath.h
atreatO calculates the natural logarithm of x
atdlib.h
atrohrO calculates logia(x)
ataeh.h
• trcmpo allocates main memorv
aath.h
atrcollO searches n byies for character c
atdlib.h
atropyO compares two blocks upto 3 length of exactly n bytes aaa.h
atrcapnf) copies a block of n bytes mam.h
atrdupf) set D bytes of a block of memory to byte c
atrarror() converts time to calendar format
aaa.h
etrftiaaO prints a system error message calculates x to the power of y
tiaa.n
atrlanl) writes formatted output to stdout
atdio.h
atrnoatt) outputs a character to a stream
aath.h
■trncmp < ) outputs character on stdout
atdio.h
atracpyO outputs a stnng to stdout
atdio.h
atrrchrO sends a software signal to the executing program
atdio.h
atrapnf) random number generator
atdio.h
atratrO reallocates main memory
aignal.h
atrtod() removes a file
atdlib.h
atrtok() renames a file
atdlib.h
atrtouK ) remove a file directory
atdio.h
■tixfiml) scans and formats input from the stdin stream atdio.h
ayataaf.) assigns a buffer to a stream dlr.h
tanh<) t i m e t ) set up for non-local goto ■tdio.h
tapni»<) tolowarO specifies signal-handling actions atdio.h
touppar( ) ungate( ) calculates sine ■atjap.b
Tfprlntf( > r p r i a t f O calculates hyperbolic sine aigaal.h
Taprlntf( ) writes formatted ouput to a stnng aath.h
initializes random number generator aath.h
scans and formats input from a string atdio.h
gets information about a file atdlib.h
appends one string to another atdio.h
scans a string for the first occurrence of a given character compares one string with ayaXatat. h
another compares two strings copies one stnng into another scans a string for initial atring.h
segment copies a stnng into a newly created location returns a pointer 10 an error ■tring.h
message stnng formats time for output calculates the length of a string appends a atring.h
portion of one stnng to another compares a portion of one stnng to a portion of atriag.h
another copies a given number of bytes from one smng into another scans a string for atring.h
the last occurrence of a given character scans a string for the first segment (a subset atriag.h
of a given stnng: scam a string for the occurrence of a given ■ubsiring converts a atring.h
stnng to a double value searches one stnng for tokens atring.h tiaa.b atring.h
converts a stnng to an unsigned long in the given radix atriag.h atring.h atdio.h
transforms a portion of a stnng atring.h atring.h atriag.h
invokes the shell in order to execute a command atdlib.h atring.h atdlib.h
calculates the hyperbolic tangent atriag.h atdlib.h aath.h
gets the time of the day tisaa.h atdio.h ctype.b
creates a unique file name otypa.h atdio.h atdio.h
translates characters to lowercase atdarg.h ■tdarg.b
translates characters to uppercase
pushes a character back into the input stream
writes formatted output to a stream
writes formatted output to stdout
writes formatted output to a string

Copyrighted material
Appendix D: ASCII Character

Continued... ►
Set

243 Copyrighted material

Appendix D: ASCII Character Set 244

Continued ... |pV

Copyrighted material
245 Mastering C++

Continued... ►
Appendix E: Bibliography
j 1 j Alan Joch, Nine ways to make your code more reliable — How Software Doesn't Work ?, Byte
Magazine 49-60p. October 1995. [2| Bernd Muller, Is Object-Oriented Programming Structured
Programming ?, ACM SIGPLAN
Notices. Volume 28, No. 9, September 1993. (3J Bjarne Stroustrup, The C+ + Programming
Language (2nd Edition), Addison Wesley, 1991. [4] Bjarne Stroustrup, The Design and Evolution of C+
+, Addison Wesley, 1994. [5] Bruce Eckel, Using C++, Osborne McGraw Hill, 1989.
[6] Capper, Colgate, Hunter, and James, The impact of object-oriented technology on software quality: Three
case histories, IBM Systems Journal, Volume 33, No. 1,1994.
[7] D. Dechanpeaur et al. The Process of Object Oriented Design, Seventh Annual Conference on Object-
Oriented Programming, System, Language, and Applications (OOPSLA), 1992.
[8] Data Quest Magazine, OOP - New Software Paradigm, 1-15 April, 1995, India.
[9] E Balagurusamy, Object-Oriented Programming with C++, Tata McGraw Hill Publications, 1996.
[10] Edmund X Dejesus.S/g OOP, No Oops, Byte, August 1995.
(11] Edward Yourdon, Object-Oriented Systems Design, Prentice Hall Inc, 1994.
[12] Harald M Muller, Ten Rules for Handling Exceptions Handling Successfully, C++ Report, Jan. 1996.
[13] Henda Hodjami, A Reuse approach based on Object-Orientation, Software Engineering Notes,
Proceedings of the Symposium on Software Reusability, August 1995.

Copyrighted material
246 Mastering C++

[14] James and Josephine, Reuse Through Inheritance, Software Engineering Notes, Proceedings of the
Symposium on Software Reusability, August 1995.
[15] Keith Gorlen, C++ Under UNIX, UNIX Papers, Waite Groups.
[ 16] Margaret A. Ellis and Bjarne Stroustrup, The Annotated C+ + Reference Manual, Addison-Wesley,
Reading, MA, 1990, ISBN 0-201-51459-1. [17] Markku Sakkinen, The Darker Side of C++
Revisited, Department of Computer Science and
Information Systems, University of Jyvaskyla, Finland. [18] Nicholas Wilt, Templates in C++,
Supplement to Dr. Dobb's Journal, December 1992. [ 19] Rajkumar, Fault Tolerant Computing, A
Seminar Report, Bangalore University, 1995. [20] Randall Hyde, Object-Oriented Programming in
Assembly Language, Dr. Dobb's Journal, Mar. 1990. [21] Tim Rentsch, Object-Oriented Programming,
SIGPLAN Notices, September, 1992. [22] Robert G Fichman and Chris F Kemerer, Object-Oriented and
Conventional Analysus and Design
Methodologies, IEEE Computer, 1992. [23] Robert Lafore, Object-Oriented Programming in Turbo
C++, Waite Group, 1992. [24] Steven J, A Technique for Tracing Memory Leaks in C++, NCR
Microelectronics, Colorado. [25] SunSoft, C + + Selected Reading, Object-Oriented Programming, August
1994. [26] Turbo C++, Library Reference Manual, Borland International Inc., 1990 [27] Venugopal K R
and Vimala H S, Programming with Fortran, Tata McGraw Hill, India, 1994. [ 28 ] Venugopal K R and
Vimala H S, Programming with Pascal and C, Tata McGraw Hill, India, 1994. [29] Venugopal K R and
Rajkumar, Microprocessorx86 Programming, BPB Publications, India, 1995. [30] Venugopal K R and
Maya, Programming with Pascal, Tata McGraw Hill, India, 1997. [31 ] Venugopal K R and Sudeep,
Programming with C, Tata McGraw Hill, India, 1997.
800 Appendix F: Index

pointers, 268-310 sorting, continue statement,

bubble sort, 174-176 comb sort, 176-178 conceptual entity, 312
strings, 182-188 assignment operators, concrete class, 584
124 auto string class, 226 copy constructor, 382

B D
Bibliography, 222 binary, 120 data, 105
mode files, 683 decomposition, 313
operator overloading, 432 binding, demotion, 122
262 entering into structures, 242
early, 22 hiding, 322=332
late, 2Q.22, 83 bits. 12Q passing to functions, 198-201
bitwise operators, 12Qrl22 bound promotion, 122
checking arrays, 121 break statement, returning from functions, 201-203 members, 313-
158-160 bubble sort, 174-176 314 datatypes, 103
charaeter(char), 103
c double precision floating point,
double, 103
calling functions, 128 enumerated(enum), 136-138 floating
function calls with arguments, 126 point(float), 103 integer(int), 103
cast operator, 122 character(char) long double precision floating point, (long
variable type, 103 character set, 100 double), 103
closing files, 662 classes, 113^6Q long integer(long), 103
declaration of, 114 short integer(short), 103
instantiation, 313.316 structure, 232
private, 314.331 void, 234 destructors, -i71-373 delete
protected, 314.332 operator, 69,403 delegation, 27.562-566 Jo-while
public, 314.332 loop, 156-158 default arguments, 210-212 default
templates, 4 class design, 761-764 constructor, 365.370 driver function, 763
class template, 610 client server model, dynamic construction, 383 dynamic destruction,
112 COBRA, 22Q 383 dynamic binding, 570. 57? dynamic objects,
command line arguments, 200 comments, 400-431
33.40-41 command line compilation, 35
composition, 562-566
conditional operator(ternary operator),
126r 128
constant object, 323
constants, 131-134
constructors, 264-3RO

Copyrighted material
804 Appendix F: Index

this pointer, 4?.?.-424 tokens, UH

typecast operators, 465 typedef
statement, 59.128

u
unary operator overloading, 417
unions, 259-266
unsigned (unsigned)variable type, 103
unsigned integer variable type, 103

v
variable, 102
definition, 104 extent, 223
initialization, 104
scope, 223 virtual destructors, 589
virtual functions, 87.570-594 virtual
pointer(VPTR), 591
vinualtable(VTBL), 591 void, 234
void pointer, 226=223

w
water falling model, 749 while loop.
154-156 wild pointer. 307-310 world
wide web, 30
 Chapter 1: Object-Oriented Paradigm 17

Data abstracter is supported by several other modern programming languages such as

Smalltalk. Ada. etc. In these languages, and i n C++ as well, a programmer can define a new
abstract data type by specifying a data structure, together with the operations permissible on thai
data Structure as shown in Figure 1.14. The important feature o f C++, the class declaration,
allows encapsulation and creation ol abstract data types.
The use of encapsulation in protecting the members (data and code) of a class from
unauthorized acc:ss is a good programming practice: i t enforces the separation between the
specification and imj lementation of abstract data typos, and i t enables the debugging of
programs easily.

1.10 Inheritance
Inheritance is the process, by which one object can acquire (he properties of another. It allows
the declaration and implementation of one class to be based on an existing class. Inheritance is
the most promising concept of OOP. which helps realize the goal of constructing software
systems from reusable parts, rather than hand coding every system from scratch. Inheritance not
only supports reuse across systems, but also directly facilitates extensibility within a given
system. Inheritance coupled with polymorphism and dynamic binding, minimizes the amount of
existing code to be modified while enhancing a system.
To understand inheritance, consider the simple example shown i n Figure 1.15. When the
class Child inherits the class Parent, the class Child is referred to as derived class (sub-class), and
the class Parent as a base class (super-class). In this case, the class Child has two parts, a derived
part and an incremental part. The derived part is inherited from (he class Parent. The incremental
part is the new code written specifically for the class Child. In general, a feature of Parent may be
renamed, rc-implemcntcd, duplicated, voided (nullified), have its visibility status changed or
subjected to almost any other kind of transformation when it is mapped from Parent to Child. The
inheritance relation is often called the is-a relation. This is because when the class Child inherits
the base class Parent, it acquires all the properties of the Parent class. It can also have its own
properties, i n addition to those acquired from its Parent. This is an example of single inheritance;
the child class has inherited proper-tics from only one base class.

Figure 1.15: Single inheritance

The inheritance relation is often used to reflect the elements present in an application domain
For example, consider a rectangle which is a special kind of polygon as shown in Figure 1.16.
This relationship is easily captured by the inheritance relation When the rectangle is inherited
from the polygon, it

Copyrighted material
 70 Mastering C++

Chapter 2: Moving from C to C+ 67

+

totally
invisible
to the
user.
Similarly,
the
compiler
translates
the
following
calls
s
w
a
p
t

a
,

)
;

/
/

c
o
m
p
i
l
e
r

c
r
e
a
t
e
s

s
w
a
p
t

i
n
t

&
x
,

i
n
c

&
y

)
;

s
w
a
p
t

i
Copyrighted material
 c, d ); // deallocati
compiler creates on
swapt float &x, respectiv
float &y );
ely.
into appropriate functions (if
necessary), and calls them based
new
on their input parameter data
types. Operator
The new
operator
Template Function
offers
Overloading dynamic
A template function can be storage
overloaded in two ways - (i) by allocation
other functions of its name or (ii) similar to
by other template functions of the
the same name. Overloading standard
resolution for functions and library
template functions can be done function
in the following three steps: malloc. It

I—►* return type, pointer to data type

number of items
to be allocated;

DataType * new DataType[size in integer]; (a) Mem

♦ If an exact match for the is
function is found, call it. particular
♦ If a function can be generated ly
from a function template designed
matching exactly, then call the keeping
generated function. OOP in
♦ If a function can be found by mind and
trying ordinary overloading throws an
resolution techniques then call it; exception
♦ If no match is found, report an if
error. memory
allocation
fails. The
2.18Runtime Memory general
Management format of
the new
Whenever an array is defined, a
operator
specified amount of memory is
is shown
set aside at compile time, which
in Figure
may not be utilized fully or may
2.13.
not be sufficient. If a situation
arises in which the amount of
memory required is unknown at voi
compile time, the memory d
"ma
allocation can be performed llo
during execution. Such a c
technique of allocating memory (si
zeo
during runtime on demand is f
known as dynamic memory (Da
allocation. taT
ype
C++ provides the following two )
special operators to perform *Si
ze
memory management in
dynamically. Int
ege
* new operator for r);
dynamic memory
allocation (b) Memory
♦ delete operator for allocation in C
dynamic memory
deallocation Figu
The memory management re
functions such as malloc ( ) , 2.13:
calloc ( ) , and f r e e ( ) in C, Synt
have been improved a n d ax of
evolved in C++ as the new and mer
delete operators to accomplish nory
dynamic memory allocation and alloc

Copyrighted mat
 70 Mastering C++

ation in C and t
C++
2 t
. o
f
d r
e e
e
l t
e (
t d
o
e
u
b
b le
; *
).
d
i )
s ;
4
e .
q delete
u c
i
i
t
v
y
a
;
l
e
i
n
s
t
e
t q
o u
freet (float i
*) b ) ; v
3 a
. l
delete e
d n
; t

i t
s o
f
r
e e
q e
u
i
v
a
l
e
n

i
Copyrighted material
 t (char * } c i t y w
); o
5
. v
e
delete date_ptr;
c
i t
s o
r
s
e
q t
u i
i n
v c
l
a
u
l d
e e
n
t <
i
o
t s
o t
f r e e t (struct r
date * ) date_ptr e
); a
The program vector, m
cpp illustrates the concept .
of dynamic allocation and h
deallocation using new and >
delete operators. void
AddVec
/ tors( int
/ * a , int
* b , int
v * c , int
size )
e
(
c
t for
o (
r
. i
c n
p t
p
: i
-
a
d =
d
i 0
t ;
i
o i
n
<
o
f s
i
t z

Copyrighted mat
 70 Mastering C++

e
; >
>
i
v
+ e
+ c
t
) o
r
c [
[ i
i ]
;
]
)
void
=
ShowVe
ctort int
a 'vector,
[ int s i z e
i )
] {
for
+ {

b i
[ n
t
i
l i
;
) =
void ReadVectort int
*vector, int s i z e I 0
{ ;
for
i i

i <
n
t s
i
i z
e
= ;

0 i
; +
+
i
)
<
c
s o
i u
z t
e
; <
<
i
+ v
+ e
c
) t
o
c r
i [
n i

i
Copyrighted material
 ]
S
< i
< z
e
" ,
■
o
;
-
)
f
v
o
V
i
e
d
c
t
m
o
a
r
i
:
n
(
"
)
;
{
c
i
i
n
n
t
>
v
>
e
c
v
_
e
s
c
i
_
z
s
e
i
;
z
e
i
;
n
//
t
allo
cate
*
me
x
mor
,
y
for
*
all
y
the
,
thre
e
*
vect
z
ors
;
x=
c
new
o
int [
u
vec_
t
size
j;
«
// x
beco
"
E
n
t
e
r

Copyrighted mat
 70 Mastering C++

mes array of s i z e
vec_size
y = new int [ vec_size ] ;
/ / y becomes array of
size vec_size
z = new int [ vec_size ] ;
/ / z becomes array o f
size vec_size
cout < < "Enter
elements of vector x :
";
ReadVector( x , vec_size
);
cout < < "Enter
elements of vector y :
";
ReadVector( y , vec_size
);
AddVectors ( x , y , z ,
vec_size ) ; / / z = x+y

i
Copyrighted material
C++ at a Glance

3.1 Introduction
The C++ language evolved as a result of extensions and enhancements to C. It has efficient memory
management techniques, provisions for building new concepts, and a new style of program analysis
and design. The reason for retaining C as a subset is its popularity among programmers, and
moreover, millions of lines of code already written in C can be directly moved to C++ without
rewriting. The other advantages are: the syntax and structure of many statements of C closely
resemble the actual operation on the computer's internal registers and allow to produce fast
executable code.
The most interesting features of C++ are those which support a new style of programming
known as object-oriented programming. I t emphasizes on data decomposition rather than algorithm
decomposition. OOP is generally useful for any kind of application, but it is particularly suited for
interactive computer graphics, simulations, databases, artificial intelligence, high-performance
computing, and system programming applications. This chapter presents the first impression of C++
with its features of object-oriented programming.
C++ as an object oriented programming language supports modular programming and enables
easy maintainability. The most prominent features of C++ that provide a foundation for data
abstraction and object-oriented programming are the following:
• Data Encapsulation and Abstraction: Classes
• Inheritance: Derived Class
♦ Polymorphism: Operator Overloading
• Friend Functions
♦ Polymorphism: Virtual Functions
• Generic Classes: Class Templates
♦ Exception Handling
♦ Streams Computation

3.2 Data Encapsulation and Abstraction—Classes

Data abstraction is the ability to create user-defined data types for modeling real world objects using
built-in data types and a set of permitted operators. Encapsulation is achieved by using the class,
which combines data and functions that operate on the data. Data hiding is achieved by restricting
the members of classes as private or protected.
The object oriented programming technique involves the representation of real world problems in
terms of objects. C++ provides a new data structure called class whose instance is called object. A
class consists of procedures or methods and data variables.
Class is the basic construct for creating user-defined data types called abstract data types: in a way

Copyrighted material
Chapter 3: C++ at a 77
Glance
Class
es are
syntactica
lly, an
extension
of
structures
. The
differenc
e is that,
all the
members
of
structures
are public
by
default,
whereas
members
of classes
are
private
by
default.
Class
follows
the
principle
of all the
informatio
n about a
module
should be
private to
the
module
unless it is
specificall
y declared
public.

Membe
r
Functio
ns
The data
members
of a class
must be
declared
within
the body
of a class,
whereas
the
member
functions
of a class
can be
defined
in one of
the
following
ways:
♦

Insid
e the
class
body

Copyrighted material
 Mastering C++

♦ .Outside the class body

The syntax of a member )
;
function definition changes
depending on whether i t is v
defined inside or outside the o
class specification. However, i
irrespective of the location of d
its definition (inside or outside
the class body), the member o
function must perform the u
same operation. Therefore, the t
d
code inside the function body
a
would be identical in both the t
cases. The compiler treats a
member functions defined {
inside a class as inline )
functions, whereas those ;
defined outside a class are not );
treated as inline functions. //
The program s t d c l a s s . c p p initiali
illustrates the mechanism of zing
defining member functions data
outside the body of the class. members
void
/ / Stdclass.cpp: member student: :
functions d e f i n e d outside a setdata{ i
body of' t h e class nt
#include <iostream.h> roll_no_i
#include < s t r i n g . h > n, char
class student *name_i
(
n)
private: name( {
int 20 ] ; r
roll_n o
o; / / roll l
char number l
/ / name o f _
a student n
public: o
v
o
=
i
d
r
s o
e l
t l
d _
a n
t o
a _
t i
n
i ;
n
t s
t
r r
o c
l
p
l
y
_
n t
o
_ n
i a
n m
, e
,
c
h n
a a
r m
e
* _
n i
a n
m
e
)
_
;
i
}
n

Copyrighted material
/ N
/ o

d *
i
s ■
p
l «
a
y r
o
d l
a l
t _
a n
o
m
e «
m
b e
e n
r d
s l
;
o
n c
o
t u
h t
e
<
c <
o
n "
s N
o a
l m
e e

s =
c
r *
e
e «
n
n
v a
o m
i e
d
<
s <
t
u e
d n
e d
n l
t ;
:
: void
o main ( )
u
t stude
d nt
a si;
t //
a first
( objec
) t/vari
able
c of
o class
u stude
t nt
stude
< nt
< s2;
//
' secon
R d
o objec
l
l

Copyrighted material
 Mastering C++

t/variable of class j
student k
sl.setdatat 1, u
"Tejaswi" ) ; // m
object s i calls member a
function setdata r
s2.setdata( 10, In the
"Rajkumar' ) ; / / class
c a l l s member function student,
setdata the
cout « "Student prototype
d e t a i l s . . . " < < endl; of
s i .outdata( ) ; // member
object si calls member functions
function outdata* setdata
s 2 .outdata( ) ; // and
o b j e c t s2 c a l l s outdata
member function outdata are
1 declared
Hun within
S the body
t of the
u class and
d they are
e defined
n outside
t
the body
d of the
e class. In
t the
a declarator
i void
l student::o
s utdataO
.
student
. : :
. indicates
that the
R function
o
outdata
l
l (),
belongs
N to the
o class
student
=
and it is a
1 member
function
N of the
a class
m student.
e

= 3.3
T Inherit
e ance-
j
a Derive
s d
w
i Class
es
R
o Inheritan
l ce is a
l technique
of
N organizin
o g
informati
=
on in the
1 hierarchic
0 al form.
It is
N similar to
a a child
m inheriting
e the
features
=
such as
R beauty of
a the

Copyrighted material
mother and intelligence of the and
father. It is an important downwa
feature of object oriented
programming that allows to rd
extend and reuse existing code countin
without requiring to rewrite it g
from scratch. Inheritance capabili
involves derivation of new ty
classes from the existing ones,
thus enabling the creation of a #includ
hierarchy of classes, similar to e
the concepts of class and <iostrea
subclass in the real world. A m.h>
new class created using an
class
existing class is called the
derived class. This process is counter
called inheritance. The {
derived class inherits the prot
members - both data and ecte
functions of the base class. It d: //
can also modify or add to the
Not
members of a base class.
Inheritance allows a hierarchy e: i t
of classes to be derived. is
priv
Derived classes, inherit data
members and member ate
functions from their base in
classes, and can be enhanced CO
by adding other data members UN
and member functions. TE
Recall that the program Rl.
counterl. cpp discussed above, CP
uses the class counter as a P
general purpose counter i
variable. A counter could be
n
incremented or decremented.
The counter class can be t
extended to support
downward counting. It can be v
achieved by either modifying a
the counter class or by l
deriving a new class called u
NewCounter from the counter
class. The program counter2 .
e
cpp is an extended version of ;
the previous program and has
two classes, one, counter as a /
base class and two, /
NewCounter as a derived
class. The private members of
a base class cannot be c
inherited. o
u
C++ supports another
access specifier called n
protected. Its access privileges t
are similar to private except e
that they are accessible to its r
derived classes. Protected
access privilege is used when v
members in base class's
a
section arc to be treated as
private and they must be l
inheritable by a derived class. u
The public members of the e
base class are accessible to the pub
derived class, but the private lic:
members of the base cldss are
c
not. However, the protected
members of the base class are o
accessible to the derived class, u
but they are private to all n
other classes.
t
e
Chapter 3: C+ 7
r
+ at a Glance 9
{
)
/ / counter2.Cpp: new
counter having upward /

Copyrighted material
 Mastering C++

/ N o argument u
constructor n
{ t
value = 0 ; / / e
i n i t i a l i z e counter r
value to zero !
)
counter! int val )
i
n
with one
t
argument
{
v
value = val;
initialize a
counter value l
) )
int GetCounterO
:
{
return value; c
} o
void u p ( ) u
/ / i n c r e m e nnt
counter t
{ e
value * value r
+ 1; !
)
); v
/ / NewCounter is a
derived from the old l
c l a s s counter publically
)
class NewCounter: public
counter
{
{
public: )
N v
e o
i
w
d
C
o
d
u
o
n w
t n
e O
r
( /
) /
:
d
c e
o c
u r
n e
t m
e e
r n
( t
)
c
{ o
u
)
n
N t
e e
w r
C {
o

Copyrighted material
 value = value
- 1; / /
decrement
counter
)
>;
v
o
i
d

m
a
i
n
{
)

(
NewCounter
counter1 ;
no argument
constructor
NewCounter counter2
- 1; / / c a l l s 1
argument constructor
cout « "counterl
initially = ■ «
counterl.GetCounter(
) < < endl;
cout « "counter2
initially = ■ «
counter2.GetCounter
{ ) « endl;
/ / increment counter
counterl. u p < ) ;
counter2. u p ( ) ;
cout < < 'counterl on
increment = ■ <<
counterl .GetCounter
( ) << endl; cout «
"counter2 on
increment = " «
counter2 .GetCounter
( ) « endl; / /
decrement counter
counterl.down( ) ;

Copyrighted material
Chapter 3: C++ at a 83
Glance
coun
ter
coun
ter2
=1;
//
calls
1
argu
men
t
cons
truct
or
cout
«
"cou
nterl
initi
ally
="«
coun
terl-
Get
Cou
nter
()
<<
endl;
cout
<<
"cou
nter
2
init
iall
y ='
«
coun
ter2.
Get
Cou
nter
O <<
endl;
//
incr
eme
nt
coun
ter
-
♦cou
nterl
;
coun
ter2
++;
cout
«
"cou
nterl
on
incre
ment
="«
coun
terl .
Get
Cou
nter
{) «
endl;
cout
«
"cou
nter
2 on
incre
 ment = " « o
counter2.GetCounter' u
) << endl; n
/ / decrement counter t
--counterl; e
counter2--; r
cout < < "counterl on l
decrement = " < < ;
counterl .GetCounter
( ) « endl; cout << c
"counter2 on o
decrement = " << u
counter2.GetCounter( n
); t
e
r
Run 2
counterl initially = 0 -
counter2 initially = 1 -
counterl on increment = ;
1 it calls
counter2 on increment = the
2
counterl on decrement = overloa
0 ded
counter2 on decrement = operato
1 r
function
The word operator is a defined
keyword. It is preceded in the
by the return type void. user-
The operator to be defined
overloaded is class. It
immediately written after can be
the keyword operator, observe
followed by the void d that
function symbol as the
operator++ ( ) . This function
declarator syntax informs body of
the compiler to call this an
member function overloa
whenever the ++ ded and
operator is encountered, a non-
provided its operand is of overloa
type counter. ded
The statement in the operato
class counter r
void // function
operator incremen is same;
++() t counter the only
overloads the increment change
operator (++■) to operate is in the
function
on the user defined data
prototy
type. When the compiler
pe and
encounters statements method
such as of
++counterl; calling.
counter2 + + ; For
it calls the overloaded instance
operator function defined , the
in the user-defined class stateme
(see Figure 3.3). The state- nt in
ment in the class counter counter
void // 2 .cpp
operator--
decremen c
O
t counter o
overloads the decrement u
operator ( - - ) to operate n
on objects of the user
t
defined data type. When
the compiler encounters e
statements such as r
- 2
- .
c u

Copyrighted material
p n
( t
) e
; r
2
c +
a +
n ;

b i
e n

r t
e h
p e
l
a a
c b
e o
d v
e
b
y p
r
a o
g
m r
o a
r m
e .

r
e
a
d
a
b
l
e

e
q
u
i
v
a
l
e
n
t

s
t
a
t
e
m
e
n
t
:
c
o
u