Oracle® Database

Data Cartridge Developer's Guide

20c
F17229-01
February 2020
Oracle Database Data Cartridge Developer's Guide, 20c

F17229-01

Copyright © 1996, 2020, Oracle and/or its affiliates.

Primary Authors: Prashant Kannan, Preethy P G, Tulika Das

Contributing Authors: Eric Belden, Timothy Chorma, Dinesh Das, Janis Greenberg, Ying Hu, Susan
Kotsovolos, Geoff Lee, Roza Leyderman, Susan Mavris, Valarie Moore, Magdi Morsi, Chuck Murray, Den
Raphaely, Helen Slattery, Seema Sundara

This software and related documentation are provided under a license agreement containing restrictions on
use and disclosure and are protected by intellectual property laws. Except as expressly permitted in your
license agreement or allowed by law, you may not use, copy, reproduce, translate, broadcast, modify,
license, transmit, distribute, exhibit, perform, publish, or display any part, in any form, or by any means.
Reverse engineering, disassembly, or decompilation of this software, unless required by law for
interoperability, is prohibited.

The information contained herein is subject to change without notice and is not warranted to be error-free. If
you find any errors, please report them to us in writing.

If this is software or related documentation that is delivered to the U.S. Government or anyone licensing it on
behalf of the U.S. Government, then the following notice is applicable:

U.S. GOVERNMENT END USERS: Oracle programs (including any operating system, integrated software,
any programs embedded, installed or activated on delivered hardware, and modifications of such programs)
and Oracle computer documentation or other Oracle data delivered to or accessed by U.S. Government end
users are "commercial computer software" or “commercial computer software documentation” pursuant to the
applicable Federal Acquisition Regulation and agency-specific supplemental regulations. As such, the use,
reproduction, duplication, release, display, disclosure, modification, preparation of derivative works, and/or
adaptation of i) Oracle programs (including any operating system, integrated software, any programs
embedded, installed or activated on delivered hardware, and modifications of such programs), ii) Oracle
computer documentation and/or iii) other Oracle data, is subject to the rights and limitations specified in the
license contained in the applicable contract. The terms governing the U.S. Government’s use of Oracle cloud
services are defined by the applicable contract for such services. No other rights are granted to the U.S.
Government.

This software or hardware is developed for general use in a variety of information management applications.
It is not developed or intended for use in any inherently dangerous applications, including applications that
may create a risk of personal injury. If you use this software or hardware in dangerous applications, then you
shall be responsible to take all appropriate fail-safe, backup, redundancy, and other measures to ensure its
safe use. Oracle Corporation and its affiliates disclaim any liability for any damages caused by use of this
software or hardware in dangerous applications.

Oracle and Java are registered trademarks of Oracle and/or its affiliates. Other names may be trademarks of
their respective owners.

Intel and Intel Inside are trademarks or registered trademarks of Intel Corporation. All SPARC trademarks are
used under license and are trademarks or registered trademarks of SPARC International, Inc. AMD, Epyc,
and the AMD logo are trademarks or registered trademarks of Advanced Micro Devices. UNIX is a registered
trademark of The Open Group.

This software or hardware and documentation may provide access to or information about content, products,
and services from third parties. Oracle Corporation and its affiliates are not responsible for and expressly
disclaim all warranties of any kind with respect to third-party content, products, and services unless otherwise
set forth in an applicable agreement between you and Oracle. Oracle Corporation and its affiliates will not be
responsible for any loss, costs, or damages incurred due to your access to or use of third-party content,
products, or services, except as set forth in an applicable agreement between you and Oracle.
Contents
Preface
Audience xxiv
Documentation Accessibility xxiv
Conventions xxiv

Changes in This Release for Oracle Database Data Cartridge

Developer's Guide
Changes in Oracle Database 20c, Version 20.1 xxv

Part I Introduction

1 Introduction to Data Cartridges

1.1 Overview of Data Cartridges 1-1
1.2 Uses of Data Cartridges 1-2
1.2.1 Data Cartridge Domains 1-2
1.3 Extending the Server: Services and Interfaces 1-3
1.3.1 Extensibility Services 1-4
1.3.1.1 Extensible Type System 1-4
1.3.1.2 Extensible Server Execution Environment 1-6
1.3.1.3 Extensible Indexing 1-7
1.3.1.4 Extensible Optimizer 1-8
1.3.2 Extensibility Interfaces 1-9

2 Roadmap to Building a Data Cartridge

2.1 Data Cartridge Development Process 2-1
2.1.1 Implement the Project 2-1
2.2 Installing and Using Data Cartridges 2-3
2.3 Requirements and Guidelines for Data Cartridge Components 2-4
2.3.1 Cartridge Schemas 2-4

iii
 2.3.2 Cartridge Globals 2-4
2.3.3 Cartridge Error Message Names or Error Codes 2-4
2.3.4 Cartridge Installation Directory 2-5
2.3.5 Cartridge Files 2-5
2.3.6 Shared Library Names for External Procedures 2-5
2.4 Data Cartridge Deployment 2-5
2.4.1 Data Cartridge Naming Conventions 2-6
2.4.1.1 Need for Naming Conventions in Data Cartridges 2-6
2.4.1.2 Unique Name Format 2-6
2.4.1.3 Data Cartridge Naming Conventions 2-7
2.4.2 Cartridge Registration 2-8
2.4.3 Cartridge Directory Structure and Standards 2-8
2.4.4 Cartridge Upgrades 2-8
2.4.5 Import and Export of Cartridge Objects 2-8
2.4.6 Cartridge Versioning 2-8
2.4.7 Cartridge Internationalization 2-9
2.4.8 Cartridge Administration 2-9
2.4.9 Data Cartridge Development Approach 2-10
2.4.9.1 Creating a Data Cartridge Plan 2-10
2.4.9.2 Developing Data Cartridges 2-10

Part II Building Data Cartridges

3 Defining Object Types for Data Cartridges

3.1 Objects and Object Types for Data Cartridges 3-1
3.1.1 Defining a DataStream Data Type 3-2
3.1.2 Defining the Type Body 3-2
3.2 Assigning an Object Identifier to an Object Type in Data Cartridges 3-2
3.2.1 Specifying an ODI for an Object Type in Data Cartridges 3-3
3.2.2 Assigning and Using OIDs in Data Cartridges 3-3
3.3 Constructor Methods in Data Cartridges 3-3
3.3.1 Creating a Type 3-4
3.3.2 Instantiating a Type Object 3-4
3.4 Object Comparison in Data Cartridges 3-4
3.4.1 Implementing a Member Function 3-4
3.4.2 Implementing Functions for Types Without a Simple Id Attribute 3-5

4 Implementing Data Cartridges in PL/SQL

4.1 Methods 4-1

iv
 4.1.1 Implementing Methods 4-1
4.1.1.1 Defining an Object Type 4-1
4.1.1.2 Defining a “Greatest Common Divisor” Function 4-2
4.1.1.3 Implementing Methods for an Object Type 4-2
4.1.2 Invoking Methods 4-3
4.1.2.1 General Syntax for Invoking Methods 4-3
4.1.2.2 SQL Syntax for Invoking Methods 4-3
4.1.2.3 PL/SQL Syntax for Invoking Methods 4-3
4.1.2.4 Using the SELF Build-In Parameter 4-3
4.1.3 Referencing Attributes in a Method 4-4
4.1.3.1 Setting Variable Values 4-4
4.2 Debugging PL/SQL Code 4-4
4.2.1 Notes for C and C++ Developers of Data Cartridges 4-5
4.2.2 Common Potential Errors 4-5
4.2.2.1 Signature Mismatches 4-5
4.2.2.2 RPC Time Out 4-6
4.2.2.3 Package Corruption 4-6

5 Implementing Data Cartridges in C, C++, and Java

5.1 Shared Libraries 5-1
5.1.1 Using Shared Libraries 5-1
5.1.1.1 Creating an Alias Library 5-2
5.1.1.2 Specifying the Location of the Library 5-2
5.1.1.3 Creating an Alias Library through Directory Objects 5-2
5.2 External Procedures 5-2
5.2.1 Registering an External Procedure 5-3
5.2.1.1 Defining the Body of a Package 5-4
5.3 How PL/SQL Calls an External Procedure 5-4
5.4 Configuring Files for External Procedures 5-5
5.4.1 Updating the Listener Configuration File 5-5
5.4.2 Directing Network to Refer to External Procedures 5-6
5.4.3 Passing Parameters to an External Procedure 5-6
5.4.4 Specifying Data Types 5-6
5.4.4.1 Conversion to External Datatypes 5-7
5.4.4.2 Conversion from External Datatypes 5-7
5.4.5 Using the Parameters Clause 5-8
5.4.6 Using the WITH CONTEXT Clause 5-9
5.5 Using Callbacks 5-9
5.5.1 Restrictions on Callbacks 5-10
5.6 Common Potential Errors 5-10

v
 5.6.1 Calls to External Functions 5-10
5.6.2 RPC Time Out 5-11
5.7 Debugging External Procedures 5-11
5.7.1 Using Package DEBUG_EXTPROC 5-11
5.7.2 Debugging C Code in DLLs on Windows NT Systems 5-11
5.8 Guidelines for Using External Procedures with Data Cartridges 5-12
5.9 Java Methods 5-12

6 Working with Unstructured Data Types

6.1 Overview of Cartridges and Unstructured Data Types 6-1
6.2 Using DDL for LOBs 6-1
6.2.1 Creating a LOB Attribute of a Type 6-2
6.2.2 Creating a LOB Object Table 6-2
6.2.3 Creating LOB Columns 6-2
6.3 LOB Locators 6-2
6.3.1 Selecting a LOBs and Assigning it to a Local Variable 6-3
6.3.2 Manipulating LOBs 6-3
6.4 Emptying LOBs 6-3
6.4.1 Using EMPTY_BLOB() in SQL 6-4
6.4.2 Using EMPTY_CLOB() in PL/SQL 6-4
6.5 Using the OCI to Manipulate LOBs 6-4
6.5.1 OCI Functions for Manipulating LOBs 6-5
6.5.2 Comparing OCI and PL/SQL Interfaces 6-6
6.5.3 Selecting a Stored LOB into a Locator 6-7
6.6 Using DBMS_LOB Package to Manipulate LOBs 6-9
6.6.1 DBMS_LOB Package Routines 6-9
6.6.2 Trimming a CLOB 6-10
6.7 LOBs in External Procedures 6-10
6.7.1 Defining an External Procedure (PL/SQL) 6-10
6.8 LOBs and Triggers 6-11
6.9 Using Open/Close as Bracketing Operations for Efficient Performance 6-11
6.9.1 Errors and Restrictions Regarding Open/Close Operations 6-11
6.9.1.1 Working with Open() and Close() Code Blocks 6-12

7 Using Extensible Indexing

7.1 Overview of Extensible Indexing 7-1
7.1.1 Purpose of Indexes 7-1
7.1.2 Purpose of Extensible Indexing 7-1
7.1.3 When to Use Extensible Indexing 7-2

vi
 7.1.4 Index Structures 7-2
7.1.4.1 B-tree 7-2
7.1.4.2 Hash 7-3
7.1.4.3 k-d tree 7-4
7.1.4.4 Point Quadtree 7-4
7.2 Extensible Indexing Framework 7-5
7.3 Using the Text Indextype 7-6
7.3.1 Defining the Indextype 7-6
7.3.1.1 Non-Index-Based Functional Implementations 7-6
7.3.1.2 Index-Based Functional Implementations 7-7
7.3.2 Using the Indextype 7-8
7.3.2.1 Declaring a New Table 7-8
7.3.2.2 Building a Text Domain Index for the Table 7-8
7.3.2.3 Querying a Table Using a Contains() Operator 7-8

8 Building Domain Indexes

8.1 Overview of Indextypes and Domain Indexes 8-1
8.2 ODCIIndex Interface 8-2
8.2.1 Index Definition Methods 8-2
8.2.1.1 ODCIIndexCreate() 8-2
8.2.1.2 ODCIIndexAlter() 8-3
8.2.1.3 ODCIIndexDrop() 8-3
8.2.2 Index Maintenance Methods 8-3
8.2.2.1 ODCIIndexInsert() 8-3
8.2.2.2 ODCIIndexDelete() 8-3
8.2.2.3 ODCIIndexUpdate() 8-3
8.2.3 Index Scan Methods 8-3
8.2.3.1 ODCIIndexStart() 8-4
8.2.3.2 ODCIIndexFetch() 8-4
8.2.3.3 ODCIIndexClose() 8-5
8.2.4 Index Metadata Method 8-5
8.2.5 Transaction Semantics During Index Method Execution 8-5
8.2.6 Transaction Semantics for Index Definition Routines 8-6
8.2.7 Consistency Semantics during Index Method Execution 8-6
8.2.8 Privileges During Index Method Execution 8-6
8.3 Creating, Dropping, and Commenting Indextypes 8-7
8.3.1 Creating Indextypes 8-7
8.3.2 Dropping Indextypes 8-7
8.3.3 Commenting Indextypes 8-8
8.3.3.1 INDEXTYPE Comments 8-8

vii
8.4 Domain Indexes 8-8
8.4.1 Domain Index Operations 8-8
8.4.1.1 Creating a Domain Index 8-9
8.4.1.2 Changing a Domain Index 8-9
8.4.1.3 Renaming a Domain Index 8-9
8.4.1.4 Rebuilding a Domain Index 8-9
8.4.1.5 Truncating a Domain Index 8-10
8.4.1.6 Dropping a Domain Index 8-10
8.4.2 Domain Indexes on Index-Organized Tables 8-10
8.4.2.1 About Rowid Storage in a UROWID Column 8-10
8.4.2.2 Determining the Size of a UROWID Column 8-11
8.4.2.3 DML on Index Storage Tables 8-11
8.4.2.4 Start, Fetch, and Close Operations on Index Storage Tables 8-11
8.4.2.5 Indexes on Non-Unique Columns 8-11
8.4.3 Domain Index Metadata 8-12
8.4.4 Moving Domain Indexes Using Export/Import 8-12
8.4.5 Moving Domain Indexes Using Transportable Tablespaces 8-12
8.4.6 Domain Index Views 8-13
8.5 Object Dependencies, Drop Semantics, and Validation 8-14
8.5.1 Object Dependencies 8-14
8.5.2 Object Drop Semantics 8-14
8.5.3 Object Validation 8-15
8.6 Indextype, Domain Index, and Operator Privileges 8-15
8.7 Partitioned Domain Indexes 8-15
8.7.1 Using Local Domain Index Methods 8-16
8.7.2 About Partitioned Indexes 8-16
8.7.3 Creating a Local Domain Index 8-16
8.7.4 Dropping a Local Domain Index 8-17
8.7.5 Altering a Local Domain Index 8-17
8.7.6 Summary of Index States 8-17
8.7.7 DML Operations with Local Domain Indexes 8-18
8.7.8 Table Operations that Affect Indexes 8-18
8.7.9 ODCIIndex Interfaces for Partitioning Domain Indexes 8-19
8.7.10 Using SQL*Loader for Domain Indexes 8-19
8.8 Using System Partitioning 8-19
8.8.1 Advantages of System Partitioned Tables 8-20
8.8.2 Implementing System Partitioning 8-20
8.8.2.1 Creating a System-Partitioned Table 8-20
8.8.2.2 Inserting Data into a System-Partitioned Table 8-20
8.8.2.3 Deleting and Updating Data in a System-Partitioned Table 8-21
8.8.3 Supporting Operations with System-Partitioned Tables 8-21

viii
 8.8.4 Running Partition Maintenance Operations 8-22
8.8.5 Altering Table Exchange Partitions with Indexes 8-22
8.9 Using System-Managed Domain Indexes 8-23
8.10 Designing System-Managed Domain Indexes 8-25
8.10.1 Methods for Non-Partitioned Domain Indexes 8-26
8.10.2 Methods for Local System-Managed Domain Indexes 8-27
8.11 Creating Local Domain Indexes 8-28
8.12 Maintaining Local Domain Indexes with INSERT, DELETE, and UPDATE 8-30
8.13 Querying Local Domain Indexes 8-30
8.14 System Managed Domain Index - Supported Schemes 8-30
8.15 Restrictions of System-Managed Domain Indexing 8-31
8.16 Migrating Non-Partitioned Indexes 8-31
8.17 Migrating Local Partitioned Indexes 8-31

9 Defining Operators
9.1 User-Defined Operators 9-1
9.1.1 Operator Bindings 9-1
9.1.2 Operator Privileges 9-2
9.1.3 Creating Operators 9-2
9.1.4 Dropping Operators 9-2
9.1.5 Altering Operators 9-2
9.1.5.1 Necessary Privileges for ALTER OPERATOR 9-3
9.1.5.2 Restrictions of ALTER OPERATOR 9-3
9.1.6 Commenting Operators 9-3
9.1.7 About Invoking Operators 9-3
9.1.7.1 Creating Contains() Operator 9-4
9.1.7.2 Using Contains() Operator in a Query 9-4
9.1.7.3 Using Contains() Operator Incorrectly 9-4
9.2 Operators and Indextypes 9-4
9.2.1 Operators in the WHERE Clause 9-5
9.2.1.1 Using Operator Predicates 9-5
9.2.1.2 Resolving Query Results with the Contains() Operator 9-5
9.2.1.3 Setting Up an Index Scan 9-6
9.2.1.4 Execution Model for Index Scan Methods 9-6
9.2.1.5 Filtering Multiple Table Queries with Contains() Operator 9-7
9.2.1.6 Invoking Indextrype Routines for the Contains() Operator 9-7
9.2.2 Using Operators Outside the WHERE Clause 9-7
9.2.2.1 Creating Index-based Functional Implementations 9-7
9.2.2.2 Implementing the Contains() Operator in Index-Based Functions 9-8
9.2.2.3 Binding the Contains() Operator to the Functional Implementation 9-8

ix
 9.2.2.4 Operator Resolution 9-8
9.2.2.5 Operator Execution 9-9
9.2.3 Operators that Return Ancillary Data 9-9
9.2.3.1 Operator Bindings that Compute Ancillary Data 9-10
9.2.3.2 Operator Bindings That Model Ancillary Data 9-10
9.2.3.3 Operator Resolution 9-11
9.2.3.4 Operator Execution 9-11

10 Using Extensible Optimizer

10.1 Overview of Query Optimization 10-1
10.1.1 Statistics 10-2
10.1.1.1 User-Defined Statistics 10-2
10.1.1.2 User-Defined Statistics for Partitioned Objects 10-3
10.1.2 Selectivity 10-3
10.1.2.1 User-Defined Selectivity 10-3
10.1.3 Cost 10-4
10.1.3.1 User-Defined Cost 10-4
10.2 Defining Statistics, Selectivity, and Cost Functions 10-5
10.2.1 Defining a Statistics Type 10-6
10.2.2 User-Defined Statistics Functions 10-7
10.2.3 User-Defined Selectivity Functions 10-7
10.2.4 User-Defined Cost Functions for Functions 10-9
10.2.5 User-Defined Cost Functions for Domain Indexes 10-10
10.2.6 Generating Statistics for System-Managed Domain Indexes 10-11
10.2.6.1 Index-Partition Statistics Storage in an Index Table 10-11
10.2.6.2 Index-Partition Statistics Storage in a Separate Table 10-12
10.2.6.3 Index-Partition Statistics Storage in a Common Table 10-12
10.3 Using User-Defined Statistics, Selectivity, and Cost 10-13
10.3.1 User-Defined Statistics 10-13
10.3.1.1 Column Statistics 10-13
10.3.1.2 Implementing Domain Index Statistics 10-14
10.3.2 User-Defined Selectivity 10-15
10.3.2.1 User-Defined Operators 10-15
10.3.2.2 Standalone Functions 10-15
10.3.2.3 Package Functions 10-15
10.3.2.4 Type Methods 10-15
10.3.2.5 Default Selectivity 10-16
10.3.3 User-Defined Cost 10-16
10.3.3.1 User-Defined Operators 10-16
10.3.3.2 Standalone Functions 10-16

x
 10.3.3.3 Package Functions 10-17
10.3.3.4 Type Methods 10-17
10.3.3.5 Default Cost 10-17
10.3.4 Declaring a NULL Association for an Index or Column 10-18
10.3.5 How DDL Operations Affect Statistics 10-18
10.4 Predicate Ordering 10-19
10.5 Dependency Model 10-19
10.6 Restrictions and Suggestions 10-20
10.6.1 Distributed Execution 10-20
10.6.2 System-Managed Storage Tables and ASSOCIATE STATISTICS 10-20
10.6.3 Aggregate Object-Level Statistics 10-21
10.6.4 System-Managed Domain Indexing 10-21
10.6.5 Collecting and Deleting User-Defined Statistics for System-Managed
Indexes 10-21
10.6.5.1 Collecting statistics for a system-managed domain index 10-21
10.6.5.2 Deleting statistics for a system-managed domain index 10-22
10.6.5.3 Collecting statistics for all partitions of a local system-managed
domain index 10-22
10.6.5.4 Deleting statistics for all partitions of a local system-managed
domain index 10-22
10.6.5.5 Collecting statistics for partition P2 of a local system-managed
domain index 10-22
10.6.5.6 Deleting statistics for partition P2 of a local system-managed
domain index 10-23
10.6.5.7 Collecting statistics for all subpartitions of a composite partition
of a local system-managed domain index 10-23
10.6.5.8 Deleting statistics for all subpartitions of a composite partition of
a local system-managed domain index 10-23
10.6.5.9 Collecting statistics for a subpartition of a local system-managed
domain index 10-24
10.6.5.10 Deleting statistics for a subpartition of a local system-managed
domain index 10-24
10.6.6 Performance 10-24

11 Using Cartridge Services

11.1 Introduction to Cartridge Services 11-1
11.2 Cartridge Handle 11-2
11.2.1 Client Side Usage 11-2
11.2.2 Cartridge Side Usage 11-2
11.2.3 Making Service Calls 11-2
11.2.4 Handling Errors 11-3
11.3 Memory Services 11-3
11.4 Maintaining Context 11-4

xi
 11.4.1 Durations 11-4
11.5 Globalization Support 11-4
11.5.1 Globalization Support Language Information Retrieval 11-5
11.5.2 String Manipulation 11-5
11.6 Parameter Manager Interface 11-5
11.6.1 Input Processing and Support for Special Characters 11-5
11.6.2 Parameter Manager Behavior Flag 11-6
11.6.3 Key Registration 11-6
11.6.4 Parameter Storage and Retrieval 11-7
11.6.5 Parameter Manager Context 11-7
11.7 File I/O 11-7
11.8 String Formatting 11-7

12 Using User-Defined Aggregate Functions

12.1 Overview of User-Defined Aggregate Functions 12-1
12.1.1 Using User-Defined Aggregate Functions 12-2
12.2 Creating a User-Defined Aggregate 12-2
12.3 Using a User-Defined Aggregate 12-3
12.3.1 Using the SELECT Statement with User-Defined Aggregate Functions 12-3
12.3.2 Using the HAVING Clause with User-Defined Aggregate Functions 12-4
12.3.3 Using Query Options with User-Defined Aggregate Functions 12-4
12.4 Evaluating User-Defined Aggregates in Parallel 12-4
12.5 Handling Large Aggregation Contexts 12-5
12.5.1 External Context and Parallel Aggregation 12-6
12.5.1.1 Using External Memory to Store Aggregate Context 12-6
12.5.2 User-Defined Aggregates and Analytic Functions 12-7
12.5.2.1 Using User-Defined Aggregates and Analytic Functions 12-7
12.5.3 Reuse of Aggregation Context for Analytic Functions 12-7
12.5.4 External Context and User-Defined Analytic Functions 12-8
12.6 Using Materialized Views with User-Defined Aggregates 12-8
12.7 Creating and Using a User-Defined Aggregate Function 12-9

13 Using Pipelined and Parallel Table Functions

13.1 Overview of Table Functions 13-1
13.2 Table Function Concepts 13-2
13.2.1 Table Functions 13-2
13.2.2 Pipelined Table Functions 13-3
13.2.3 Pipelined Table Functions with REF CURSOR Arguments 13-3
13.2.4 Parallel Execution of Table Functions 13-5

xii
 13.3 Pipelined Table Functions 13-5
13.3.1 Implementation Choices for Pipelined Table Functions 13-6
13.3.2 Declaring Pipelined Table Functions 13-6
13.3.3 Implementing the Native PL/SQL Approach 13-6
13.3.4 Pipelining Between PL/SQL Table Functions 13-7
13.3.5 Combining PIPE ROW with AUTONOMOUS_TRANSACTION 13-7
13.3.6 Implementing the Interface Approach 13-8
13.3.6.1 Scan Context 13-8
13.3.6.2 Start Routine 13-9
13.3.6.3 Fetch Routine 13-9
13.3.6.4 Close Routine 13-10
13.3.6.5 Describing Returned Data Sructures; Describe Method 13-10
13.3.6.6 Preparing a Query for Execution; Prepare Method 13-12
13.3.7 Querying Table Functions 13-13
13.3.7.1 Implementing Multiple Calls to Table Functions 13-13
13.3.7.2 Using PL/SQL REF CURSOR Variables 13-14
13.3.8 Performing DML Operations Inside Table Functions 13-14
13.3.9 Performing DML Operations on Table Functions 13-14
13.3.10 Handling Exceptions in Table Functions 13-15
13.4 Parallel Table Functions 13-15
13.4.1 Inputting Data with Cursor Variables 13-15
13.4.1.1 Using Multiple REF CURSOR Input Variables 13-16
13.4.1.2 Explicitly Opening a REF CURSOR for a Query 13-16
13.4.1.3 PL/SQL REF CURSOR Arguments to Java and C/C++ Functions
13-16
13.4.2 Input Data Partitioning 13-18
13.4.3 Parallel Execution of Leaf-Level Table Functions 13-20
13.5 Input Data Streaming for Table Functions 13-20
13.5.1 Setting up the Input Stream 13-21
13.5.2 Parallel Execution: Partitioning and Clustering 13-21
13.6 Creating Domain Indexes in Parallel 13-22
13.6.1 Loading Domain Indexes 13-23
13.7 Transient and Generic Types 13-23

14 Designing Data Cartridges

14.1 Choosing the Programming Language 14-1
14.2 Invoker's Rights 14-1
14.3 Callouts and LOBs 14-1
14.4 Saving and Passing State 14-2
14.5 Designing Indexes 14-2
14.5.1 Domain Index Performance 14-2

xiii
 14.5.2 Domain Index Component Names 14-2
14.5.3 When to Use Index-Organized Tables 14-2
14.5.4 Storing Index Structures in LOBs 14-3
14.5.5 External Index Structures 14-3
14.5.6 Multi-Row Fetch 14-3
14.6 Designing Operators 14-4
14.7 Designing for the Extensible Optimizer 14-4
14.7.1 Weighing Cost and Selectivity 14-4
14.7.2 Cost for functions 14-4
14.7.2.1 Selectivity for Functions 14-4
14.7.2.2 Statistics for Tables 14-5
14.7.2.3 Statistics for Indexes 14-5
14.8 Designing for Maintenance 14-5
14.9 Enabling Cartridge Installation 14-5
14.10 Designing for Portability 14-6

Part III Scenarios and Examples

15 Power Demand Cartridge Example

15.1 Feature Requirements 15-1
15.2 Modeling the Application 15-7
15.2.1 Sample Queries 15-8
15.3 Queries and Extensible Indexing 15-10
15.3.1 Queries Not Benefiting from Extensible Indexing 15-10
15.3.2 Queries Benefiting from Extensible Indexing 15-11
15.4 Creating the Domain Index 15-12
15.4.1 Creating the Schema to Own the Index 15-12
15.4.2 Creating the Object Types 15-12
15.4.3 Defining the Object Type Methods 15-13
15.4.4 Understanding Functions and Operators 15-14
15.4.4.1 Creating Functions and Operators 15-15
15.4.5 Creating the Indextype Implementation Methods 15-17
15.4.6 Defining theType 15-18
15.4.6.1 ODCIGetInterfaces() 15-19
15.4.6.2 ODCIIndexCreate() 15-20
15.4.6.3 ODCIIndexDrop() 15-21
15.4.6.4 ODCIIndexStart(); Specific Queries 15-22
15.4.6.5 ODCIIndexStart(); Any Queries 15-24
15.4.6.6 ODCIIndexFetch() 15-26
15.4.6.7 ODCIIndexClose() 15-27

xiv
 15.4.6.8 ODCIIndexInsert() 15-27
15.4.6.9 ODCIIndexDelete() 15-28
15.4.6.10 ODCIIndexUpdate() 15-29
15.4.6.11 ODCIIndexGetMetadata() 15-30
15.4.7 Creating the Indextype 15-32
15.5 Defining Types and Methods for Extensible Optimizing 15-32
15.5.1 Creating the Statistics Table, PowerCartUserStats 15-32
15.5.2 Creating the Extensible Optimizer Methods 15-33
15.5.2.1 Creating the Type Definition 15-34
15.5.2.2 ODCIGetInterfaces() 15-35
15.5.2.3 ODCIStatsCollect() Method for PowerDemand_Typ Columns 15-35
15.5.2.4 ODCIStatsDelete() Method for PowerDemand_Typ Columns 15-37
15.5.2.5 ODCIStatsCollect() Method for power_idxtype Domain Indexes 15-38
15.5.2.6 ODCIStatsDelete() Method for power_idxtype Domain Indexes 15-39
15.5.2.7 ODCIStatsSelectivity() Method for Specific Queries 15-39
15.5.2.8 ODCIStatsIndexCost() Method for Specific Queries 15-45
15.5.2.9 ODCIStatsIndexCost() Method for Any Queries 15-46
15.5.2.10 ODCIStatsFunctionCost() Method 15-48
15.5.3 Associating the Extensible Optimizer Methods with Database Objects 15-49
15.5.4 Analyzing the Database Objects 15-49
15.6 Testing the Domain Index 15-50
15.6.1 Creating and Populating the Power Demand Table 15-50
15.6.2 Querying Without the Index 15-52
15.6.3 Creating the Index 15-53
15.6.4 Querying with the Index 15-53

16 PSBTREE: Extensible Indexing Example

16.1 About the PSBTREE Example 16-1
16.2 Design of the Indextype 16-1
16.3 Implementing Operators 16-1
16.3.1 Functional Implementations 16-2
16.3.1.1 Implementing the EQUALS Operator 16-2
16.3.1.2 Implementing the LESS THAN Operator 16-2
16.3.1.3 Implementing the GREATER THAN Operator 16-2
16.3.2 Operators 16-3
16.4 Implementing the ODCIIndex Interfaces 16-3
16.4.1 Defining an Implementation Type for PSBTREE 16-3
16.4.2 Creating the Implementation Type Body 16-4
16.4.3 Defining PL/SQL Routines in the Implementation Body 16-4
16.4.3.1 Implementing ODCIGetInterfaces() for PBSTREE in PL/SQL 16-4

xv
 16.4.3.2 Implementing ODCIIndexCreate() for PBSTREE in PL/SQL 16-4
16.4.3.3 Implementing ODCIIndexDrop() for PBSTREE in PL/SQL 16-6
16.4.3.4 Implementing ODCIIndexAlter() for PSBTREE in PL/SQL 16-6
16.4.3.5 Implementing ODCIIndexUpdPartMetadata() for PSBTREE in
PL/SQL 16-7
16.4.3.6 Implementing ODCIIndexExchangePartition() for PSBTREE in
PL/SQL 16-7
16.4.4 Registering the C Implementation of the ODCIIndexXXX() Methods 16-8
16.4.4.1 Registering the Implementation of ODCIIndexInsert() 16-9
16.4.4.2 Registering the Implementation of ODCIIndexDelete() 16-9
16.4.4.3 Registering the Implementation of ODCIIndexUpdate() 16-9
16.4.4.4 Registering the Implementation of ODCIIndexStart() 16-10
16.4.4.5 Registering the Implementation of ODCIIndexFetch() 16-11
16.4.4.6 Registering the Implementation of ODCIIndexClose() 16-11
16.4.5 Defining Additional Structures in C Implementation 16-11
16.4.6 Defining C Methods in the Implementation Body 16-12
16.4.6.1 Implementing a Common Error Processing Routine in C 16-12
16.4.6.2 Implementing ODCIIndexInsert() for PSBTREE in C 16-13
16.4.6.3 Implementing ODCIIndexDelete() for PSBTREE in C 16-15
16.4.6.4 Implementing ODCIIndexUpdate() for PSBTree in C 16-17
16.4.6.5 Implementing ODCIIndexStart() for PSBTREE in C 16-19
16.4.6.6 Implementing ODCIIndexFetch() for PSBTREE in C 16-23
16.4.6.7 Implementing ODCIIndexClose() for PSBTREE in C 16-25
16.4.7 Implementing the Indextype for PSBTREE 16-26
16.5 Using PSBTREE 16-27
16.5.1 Creating and Populating a Partitioned Table for PSBTREE 16-27
16.5.2 Creating a PSBTREE Index on a Column 16-27
16.5.3 Using PSBTREE Operators in a Query 16-27

17 Pipelined Table Functions: Interface Approach Example

17.1 Pipelined Table Functions Example: C Implementation 17-1
17.1.1 Making SQL Declarations for C Implementation 17-1
17.1.2 Implementation ODCITable Methods in C 17-2
17.2 Pipelined Table Functions Example: Java Implementation 17-9
17.2.1 Making SQL Declarations for Java Implementation 17-9
17.2.2 Implementing the ODCITable Methods in Java 17-10

Part IV Reference

xvi
18 Cartridge Services Using C, C++ and Java
18.1 OCI Access Functions for External Procedures 18-1
18.1.1 OCIExtProcAllocCallMemory 18-1
18.1.2 OCIExtProcRaiseExcp 18-1
18.1.3 OCIExtProcRaiseExcpWithMsg 18-2
18.1.4 OCIExtProcGetEnv 18-2
18.2 Installing Java Cartridge Services Files 18-3
18.3 Cartridge Services-Maintaining Context 18-3
18.3.1 ContextManager 18-3
18.3.2 CountException() 18-4
18.3.3 CountException(String) 18-4
18.3.4 InvalidKeyException() 18-4
18.3.5 InvalidKeyException(String) 18-4

19 Extensibility Constants, Types, and Mappings

19.1 System Defined Constants 19-1
19.1.1 ODCIArgDesc.ArgType System Defined Constants 19-1
19.1.2 ODCIEnv.CallProperty System Defined Constants 19-2
19.1.3 ODCIIndexAlter System Defined Constants 19-2
19.1.4 ODCIIndexInfo.Flags System Defined Constants 19-3
19.1.5 ODCIIPartInfo.PartOp System Defined Constants 19-4
19.1.6 ODCIIPredInfo.Flags System Defined Constants 19-4
19.1.7 ODCIFuncInfo.Flags System Defined Constants 19-5
19.1.8 ODCIQueryInfo.Flags System Defined Constants 19-5
19.1.9 ODCIStatsOptions.Flags System Defined Constants 19-5
19.1.10 ODCIStatsOptions.Options System Defined Constants 19-5
19.1.11 Return Status System Defined Constants 19-6
19.1.12 ScnFlg System Defined Constants 19-6
19.2 System-Defined Types 19-6
19.2.1 ODCIArgDesc 19-6
19.2.2 ODCIArgDescList 19-7
19.2.3 ODCIRidList 19-7
19.2.4 ODCIColInfo 19-7
19.2.5 ODCIColInfoList 19-8
19.2.6 ODCICost 19-8
19.2.7 ODCIEnv 19-9
19.2.8 ODCIFuncInfo 19-9
19.2.9 ODCIIndexInfo 19-10
19.2.10 ODCIIndexCtx 19-11
19.2.11 ODCIObject 19-11

xvii
 19.2.12 ODCIObjectList 19-11
19.2.13 ODCIPartInfo 19-11
19.2.14 ODCIPartInfoList 19-12
19.2.15 ODCIPredInfo 19-12
19.2.16 ODCIQueryInfo 19-13
19.2.17 ODCIStatsOptions 19-13
19.2.18 ODCITabFuncStats 19-13
19.2.19 ODCITabStats 19-13
19.2.20 ODCIBFileList 19-14
19.2.21 ODCITabFuncInfo 19-14
19.2.22 ODCIDateList 19-14
19.2.23 ODCINumberList 19-14
19.2.24 ODCIRawList 19-14
19.2.25 ODCIVarchar2List 19-15
19.2.26 ODCIFuncCallInfo 19-15
19.3 Mappings of Constants and Types 19-15
19.3.1 Mappings in PL/SQL 19-15
19.3.2 Mappings in C 19-15

20 Extensible Indexing Interface

20.1 Extensible Indexing - System-Defined Interface Routines 20-1
20.1.1 ODCIGetInterfaces() 20-2
20.1.2 ODCIIndexAlter() 20-2
20.1.3 ODCIIndexClose() 20-5
20.1.4 ODCIIndexCreate() 20-6
20.1.5 ODCIIndexDelete() 20-8
20.1.6 ODCIIndexDrop() 20-9
20.1.7 ODCIIndexExchangePartition() 20-10
20.1.8 ODCIIndexFetch() 20-11
20.1.9 ODCIIndexGetMetadata() 20-12
20.1.10 ODCIIndexInsert() 20-14
20.1.11 ODCIIndexStart() 20-16
20.1.12 ODCIIndexUpdate() 20-18
20.1.13 ODCIIndexUpdPartMetadata() 20-19
20.1.14 ODCIIndexUtilCleanup() 20-20
20.1.15 ODCIIndexUtilGetTableNames() 20-20

21 Extensible Optimizer Interface

21.1 Extensible Optimizer Interface 21-1

xviii
 21.1.1 Using Statistics Functions in an Extensible Optimizer Interface 21-1
21.1.2 EXPLAIN PLAN 21-2
21.1.3 INDEX Hint 21-3
21.1.4 ORDERED_PREDICATES Hint 21-3
21.2 User-Defined ODCIStats Functions 21-3
21.2.1 ODCIGetInterfaces() 21-4
21.2.2 ODCIStatsCollect() 21-4
21.2.3 ODCIStatsDelete() 21-7
21.2.4 ODCIStatsFunctionCost() 21-8
21.2.5 ODCIStatsExchangePartition() 21-9
21.2.6 ODCIStatsIndexCost() 21-10
21.2.7 ODCIStatsSelectivity() 21-11
21.2.8 ODCIStatsTableFunction() 21-13
21.2.9 ODCIStatsUpdPartStatistics() 21-14

22 User-Defined Aggregate Functions Interface

22.1 User-Defined Aggregate Functions 22-1
22.1.1 ODCIAggregateDelete() 22-2
22.1.2 ODCIAggregateInitialize() 22-2
22.1.3 ODCIAggregateIterate() 22-2
22.1.4 ODCIAggregateMerge() 22-3
22.1.5 ODCIAggregateTerminate() 22-3
22.1.6 ODCIAggregateWrapContext() 22-4

23 Pipelined and Parallel Table Functions

23.1 Routines for Pipelined and Parallel Table Functions in C 23-1
23.1.1 ODCITableClose() 23-1
23.1.2 ODCITableDescribe() 23-2
23.1.3 ODCITableFetch() 23-3
23.1.4 ODCITablePrepare() 23-4
23.1.5 ODCITableStart() 23-4

A User-Managed Local Domain Indexes

A.1 Comparing User-Managed and System-Managed Domain Indexes A-1
A.2 Truncating Domain Indexes A-2
A.3 Creating Indextypes A-2
A.4 Using Domain Indexes for the Indextype A-2
A.5 Partitioning Domain Indexes A-2
A.6 APIs for User-Managed Domain Indexes A-2

xix
 A.6.1 ODCIIndexTruncate() A-2
A.6.2 ODCIIndexMergePartition() A-4
A.6.3 ODCIIndexSplitPartition() A-4

Index

xx
List of Figures
1-1 Oracle Services 1-4
1-2 External Programs Executing in a Separate Address Space 1-7
2-1 Cartridge Development Process 2-3
5-1 Calling an External Procedure 5-4
7-1 B-tree Index Structure 7-3
7-2 Hash Index Structure 7-3
7-3 2-d Index Structure 7-4
7-4 Point Quadtree Index Structure 7-5
8-1 Three-Partition Table with a Local Domain Index, and Associated Structures 8-24
8-2 A Three-Partition Table after ALTER TABLE SPLIT PARTITION 8-25
10-1 Storing Index-Specific Statistics with Index Tables 10-12
10-2 Storing Index-Specific Statistics in a Separate Table 10-12
10-3 Storing Index-Partition Statistics in a Common Table 10-13
12-1 Sequence of Calls for Parallel Evaluation of User-Defined Aggregates 12-5
13-1 Typical Data Processing with Non-Parallel, Non-Pipelined Table Functions 13-2
13-2 Data Processing Using Pipelining and Parallel Execution 13-2
13-3 Flowchart of Table Function Row Source Execution 13-10
15-1 Region Served by the Power Utility 15-1
15-2 Regional Grid Cells in Numbered Sequence 15-2
15-3 Grayscale Representation of Satellite Image 15-3
15-4 Grayscale Representation of Weather Conditions at Second Recording 15-4
15-5 Grayscale Representation of Conditions as Projected 15-5
15-6 Distribution of Power Stations Across the Region 15-6
15-7 Areas Served by Three Power Stations 15-7
15-8 Application Object Model of the Power Demand Cartridge 15-8
15-9 Implementation Model of the Power Demand Cartridge 15-9

xxi
List of Tables
1-1 Data Cartridge Domains; Content and Scope 1-3
2-1 Data Cartridge Naming Conventions 2-7
5-1 Parameter Datatype Mappings 5-7
5-2 External Data Type Mappings 5-7
6-1 Summary of OCI Functions for Manipulating LOBs 6-5
6-2 OCI and PL/SQL (DBMS_LOB) Interfaces Compared 6-6
6-3 Summary of DBMS_LOB Package Routines 6-9
8-1 Views *_INDEXTYPE_COMMENTS 8-8
8-2 Views *_SECONDARY_OBJECTS 8-13
8-3 Default and Explicit Drop Options for Operators and Index Types 8-14
8-4 Summary of Index States 8-17
8-5 Summary of Table Operations 8-18
8-6 Summary of ALTER TABLE Operations with Partition Maintenance 8-18
8-7 ODCIXXX() Methods for Non-Partitioned Domain Indexes 8-26
8-8 ODCIXXX() Methods for Local System-Managed Domain Indexes 8-27
10-1 Statistics Methods and Default Statistics for Various Schema Objects 10-6
10-2 Effects of DDL on Partition and Global Statistics 10-18
10-3 Dependency Model for DDLs 10-19
11-1 Special Characters 11-6
13-1 Generic SQL Types 13-24
15-1 Sample Power Demand Readings for an Hour 15-2
15-2 Sample Power Demand Readings for an Hour 15-10
15-3 Operators and Implementing Functions 15-15
15-4 Indextype Methods 15-17
15-5 Extensible Optimizer Methods 15-33
19-1 ODCIArgDesc.ArgType Values 19-1
19-2 ODCIEnv.CallProperty Values 19-2
19-3 ODCIIndexAlter Options 19-2
19-4 ODCIIndexInfo.Flags Bits 19-3
19-5 Description of the ODCIIPartInfo.PartOp System Defined Constant 19-4
19-6 ODCIIPredInfo.Flags Bits 19-4
19-7 ODCIFuncInfo.Flags Bits 19-5
19-8 ODCIQueryInfo.Flags Bits 19-5
19-9 ODCIStatsOptions.Flags Bits 19-5
19-10 ODCIStatsOptions.Options Bits 19-5

xxii
19-11 Return Status Values 19-6
19-12 ScnFlg Values; Function with Index Context 19-6
19-13 ODCIArgDesc Function and Operator Argument Description - Attributes 19-7
19-14 ODCIColInfo Column Related Information - Attributes 19-8
19-15 ODCICost Cost Information - Attributes 19-8
19-16 ODCIEnv Environment Variable Descriptor Information - Attributes 19-9
19-17 ODCIFuncInfo Function Information - Attributes 19-9
19-18 ODCIIndexInfo Index Related Information - Attributes 19-10
19-19 ODCIIndexCtx Index Context Related Information - Attributes 19-11
19-20 ODCIObject Index Context Related Information - Attributes 19-11
19-21 ODCIPartInfo Index-Related Information - Attributes 19-11
19-22 ODCIPredInfo Operator Related Information - Attributes 19-12
19-23 ODCIQueryInfo Index Context Related Information - Attributes 19-13
19-24 ODCIStatsOptions Cost Information - Attributes 19-13
19-25 ODCITabFuncStats Parameter 19-13
19-26 ODCITabStats - Attributes 19-14
19-27 ODCITabFuncInfo Parameters 19-14
19-28 ODCIFuncCallInfo - Attributes 19-15
20-1 Summary of System-Defined Extensible Indexing Interface Routines 20-1
21-1 Summary of User-Defined ODCIStats Functions 21-3
22-1 Summary of User-Defined Aggregate Functions 22-1
23-1 Summary of Pipelined and Parallel Table Functions for C 23-1

xxiii
 Preface

Preface
The Oracle Database Data Cartridge Developer's Guide describes how to build and
use data cartridges to create custom extensions to the Oracle server's indexing and
optimizing capabilities.

Audience
Oracle Database Data Cartridge Developer's Guide is intended for developers who
want to learn how to build and use data cartridges to customize the indexing and
optimizing functionality of the Oracle server to suit specialty data such as that
associated with chemical, genomic, or multimedia applications.
To use this document, you must be familiar with using Oracle and should have a
background in an Oracle-supported programming language such as PL/SQL, C, C++,
or Java.

Documentation Accessibility
For information about Oracle's commitment to accessibility, visit the Oracle
Accessibility Program website at http://www.oracle.com/pls/topic/lookup?
ctx=acc&id=docacc.

Access to Oracle Support

Oracle customers that have purchased support have access to electronic support
through My Oracle Support. For information, visit http://www.oracle.com/pls/topic/
lookup?ctx=acc&id=info or visit http://www.oracle.com/pls/topic/lookup?ctx=acc&id=trs
if you are hearing impaired.

Conventions
The following text conventions are used in this document:

Convention Meaning
boldface Boldface type indicates graphical user interface elements associated
with an action, or terms defined in text or the glossary.
italic Italic type indicates book titles, emphasis, or placeholder variables for
which you supply particular values.
monospace Monospace type indicates commands within a paragraph, URLs, code
in examples, text that appears on the screen, or text that you enter.

xxiv
Changes in This Release for Oracle
Database Data Cartridge Developer's
Guide
This preface contains:
• Changes in Oracle Database 20c, Version 20.1

Changes in Oracle Database 20c, Version 20.1

This section lists changes in Oracle Database Data Cartridge Developer's Guide for
Oracle Database release 20c, version 20.1.
See Oracle Database New Features Guide for a complete list of new features in this
release.

New Features
Local domain indexes can now be created on automatic list partitioned tables.

Desupported Features
DBMS_LOB.LOADFROMFILE Procedure is desupported. Use DBMS_LOB.LoadClobFromFile
or DBMS_LOB.LoadBlobFromFile instead.

The following LOB buffering subsystem APIs are desupported:

• OCILobEnableBuffering()
• OCILobDisableBuffering()
• OCILobFlushBuffer()

In place of using these LOB buffering functions, use the LOB prefetch features.

xxv
Part I
Introduction
Oracle Database provides Data Cartridges to support customized application
development in the object mode.
1
Introduction to Data Cartridges
Oracle Database provides Data Cartridges to support customized application
development in the object mode.
In addition to efficient and secure management of data ordered under the relational
model, Oracle provides support for data organized under the object model. Object
types and other features such as large objects (LOBs), external procedures, extensible
indexing, and query optimization can be used to build powerful, reusable, server-
based components called data cartridges.

1.1 Overview of Data Cartridges

Data cartridges extend the capabilities of the Oracle server by taking advantage of
Oracle Extensibility Architecture framework. This framework lets you capture
business logic and processes associated with specialized or domain-specific data in
user-defined data types. Data cartridges that provide new behavior without needing
additional attributes have the option of using packages rather than user-defined types.
Either way, you determine how the server interprets, stores, retrieves, and indexes the
application data. Data cartridges package this functionality, creating software
components that plug into a server and extend its capabilities into a new domain,
making the database itself extensible.
You can customize the indexing and query optimization mechanisms of an extensible
database management system and provide specialized services or more efficient
processing for user-defined business objects and rich types. When you register your
implementations with the server through extensibility interfaces, you direct the
server to implement your customized processing instructions instead of its own default
processes.
The extensibility interfaces consist of functions that the server calls to execute the
custom indexing or optimizing behavior implemented for a data cartridge. The
interfaces are defined by Oracle; as a cartridge developer, you must implement the
functions or interfaces that have the specialized behavior you require in your
application. In general, you implement the functions as static methods of an object
type. An object type that implements the extensible indexing interface is called an
indextype; an object type that implements the extensible optimizing interface is called
a statistics type.
Data cartridges have the following key characteristics:
• Data cartridges are server-based. Their constituents reside on the server or are
accessed from the server. The server runs all data cartridge processes, or
dispatches these processes as external procedures.
• Data cartridges extend the server. They define new types and behavior, enabling
the server to perform processes that are otherwise unavailable to it, in component
form. Data cartridges can use these new types and behaviors in their applications.
• Data cartridges are integrated with the server. The Oracle Extensibility Framework
defines a set of interfaces that integrate data cartridges with the components of

1-1
 Chapter 1
Uses of Data Cartridges

the server engine, allowing for domain-specific indexing, domain-specific

optimized access to the CPU resources, and domain-specific optimization of I/O
access to cartridge data.
• Data cartridges are packaged. A data cartridge is installed as a unit. After it is
installed, the data cartridge handles all access issues for each user, including
verification of schemas and privileges.

1.2 Uses of Data Cartridges

Most industries have evolved sophisticated models to handle complex data objects
that form the essence of their business. These data objects are both the structures
that relate different units of information and the operations that are performed on them.
The simple names given to data objects often conceal considerable complexity. For
example, the banking industry has many different types of bank accounts. Each bank
account has customer demographic information, balance information, transaction
information, and rules that embody its behavior (deposit, withdrawal, interest accrual,
and so forth). When using data cartridges and their object-relational extension,
application programmers and independent software vendors can encapsulate
business logic in software components that integrate with the Oracle server and
enhance it to support data types, processes, and logic to model business objects.
While business models have developed increasingly complex data objects, information
technology has made it necessary to work with new and complex kinds of data, such
as satellite images, X-rays, animal sounds, seismic vibrations, and chemical models.
Complex and multimedia data types are now frequently stored and retrieved, queried
and analyzed.
Web-based applications routinely include many different kinds of complex data.
Including application-specific data types and the associated business logic requires a
new class of networked, content-rich, multitiered, distributed applications. Data
cartridges help you meet this need by combining scalar and unstructured data types in
domain-specific components.

1.2.1 Data Cartridge Domains

Data cartridges are typically domain-specific, characterized by content and scope of
their target domain.
In terms of content, a data cartridge can accommodate scalar, complex, and
multimedia data. Scalar data can be modeled using native SQL types such as
INTEGER, NUMBER, or CHAR. Complex data include matrices, temperature and magnetic
grids, and compound documents. Unstructured multimedia data includes such
information as video, voice, and image data.
In terms of scope, a data cartridge can have either broad horizontal (cross-industry)
coverage, or it can be specialized for a specific type of business. For example, a data
cartridge for general storage and retrieval of text-based data is cross-industry in
scope; a data cartridge for the storage and retrieval of legal documents for litigation
support is industry-specific. Table 1-1 shows a way of classifying data cartridge
domains according to their content and scope, with some examples.

1-2
 Chapter 1
Extending the Server: Services and Interfaces

Table 1-1 Data Cartridge Domains; Content and Scope

Content Cross-Industry Uses Industry-Specific Extensions

Scalar Data Statistical conversion Financial and Petroleum
Multimedia and Complex Text Image
Unstructured Data
Audio/Video Spatial Legal
Medical Broadcasting Utilities

You can also use scalar data types to construct more complex user-defined types. The
object-relational database management system provides foundational data cartridges
that package multimedia and complex data. These data cartridges can be used in
developing applications across many different industries:
• Oracle Text uses the tokenized serial byte stream database model are used to
implement display compress, reformat, and indexing behavior.
• Oracle Spatial is for use with geometric objects (points, lines, polygons); it
implements project, rotate, transform and map behavior.
Another way of viewing the relationship of cartridges to domains is to consider basic
multimedia data types as an extensible foundation that can be customized for specific
industries. For example, medical applications can customize the Oracle Text for
records and Oracle Spatial for demographic analysis.
A cartridge that provides basic services can be deployed across many industries. A
cartridge can also leverage domain expertise across an industry. These cartridges can
be further extended for more specialized vertical applications.

1.3 Extending the Server: Services and Interfaces

The Oracle server provides services for basic data storage, query processing,
optimization, and indexing. Applications use these services to access database
capabilities. However, data cartridges have specialized needs because they
incorporate domain-specific data. To accommodate these specialized applications,
these basic services have been made extensible; where standard Oracle services are
not adequate for meeting a data cartridge's requirements, you can provide additional
services that satisfy the requirements of the specific data cartridge. Every data
cartridge can provide its own implementations of these services.
For example, if you are developing a spatial data cartridge for geographic information
systems (GIS) applications, you must to implement routines that create a spatial index,
insert an entry into the index, update the index, delete from the index, and perform
other required operations. Thus, you extend the indexing service of the server.

See Also:
Power Demand Cartridge Example

1-3
 Chapter 1
Extending the Server: Services and Interfaces

1.3.1 Extensibility Services

Consider extensible services and major Oracle capabilities as they relate to data
cartridge development. Figure 1-1 shows the standard services implemented by the
Oracle server.

Figure 1-1 Oracle Services

Data
Data
Cartridge
Cartridge

Oracle Universal
Oracle8 Universal
Data
DataServer
Server Extensibility
ExtensibilityInterfaces
Interfaces

Type Server Query Data

System Execution Processing ... Indexing

Database and Extensibility Services

1.3.1.1 Extensible Type System

The Oracle universal data server provides both native and extensible type system
services. Historically, most applications have focused on accessing and modifying
corporate data that is stored in tables composed of native SQL data types, such as
INTEGER, NUMBER, DATE, and CHAR. Oracle adds support for new types, including:

• User-defined object types

• Collections, such as VARRAY (varying length array) and nested tables
• Relationships (REFs)
• Large object types (LOBs), such as binary large objects (BLOBs), character large
objects (CLOBs), and external binary files (BFILEs)

1.3.1.1.1 User-Defined Types

A user-defined type extents the modeling capabilities of the native data types and
from them both because it is defined by a user, and because it specifies both the
underlying persistent data (attributes) and the related behaviors (methods).
With user-defined types, you can make better models of complex entities in the real
world by binding data attributes to semantic behaviors. A user-defined type can have
one or more attributes, each with a name and a type. The type of an attribute can be a
native SQL type, a LOB, a collection, another object type, or a REF type.

A method is a procedure or a function that is part of a user-defined type. Methods can

access and manipulate attributes of their type while running within the execution
environment of the Oracle server, or when they are dispatched outside the server as
part of the extensible server execution environment.

1-4
 Chapter 1
Extending the Server: Services and Interfaces

See Also:

• Defining Object Types for type definition syntax.

• Oracle Database Object-Relational Developer's Guide for more
information on user-defined types.

1.3.1.1.2 Collection Types

Collections are SQL data types that contain multiple elements. Elements, or values,
of a collection are all from the same type hierarchy. In Oracle, collections of complex
types can be VARRAYs or nested tables.

A VARRAY type contains a variable number of ordered elements and can be used for a
column of a table or an attribute of an object type. The element type of a VARRAY can
be either a native data type, such as NUMBER, or a user-defined type.

To provide the semantics of an unordered collection, you could create a nested table
using Oracle SQL As with a VARRAY, a nested table can define a column of a table or
an attribute of a user-defined type.

1.3.1.1.3 Reference Types

If you create an object table in Oracle, you can obtain a reference, REF, that behaves
like a database pointer to an associated row object. References are important for
navigating among object instances. Because REFs rely on the underlying object
identity, you can only use a REF with an object stored as a row in an object table, or
with objects composed from an object view.

See Also:

• Oracle Database SQL Language Reference for details of the REF

operator.
• Oracle Database Object-Relational Developer's Guide for more
information about objects.

1.3.1.1.4 Large Objects

Large object types, or LOBs, handle the storage demands of images, video clips,
documents, and other forms of unstructured data. LOBs storage optimizes space
requirements and efficient access.
LOBs are composed of locators and the related binary or character data. The locators
are stored inline with other table columns. Internal LOBs (BLOBs, CLOBs, and NCLOBs)
can store data in a separate database storage area. External LOBs (BFILEs) store the
data outside the database tablespaces, in operating system files. A table can contain
multiple LOB columns, in contrast to the limit of a single LONG RAW column for each table.
Each LOB column can be stored in a separate tablespace, and even on different
secondary storage devices.

1-5
 Chapter 1
Extending the Server: Services and Interfaces

You can create, modify, and delete tables and object types that contain LOBs using the
Oracle SQL data definition language (DDL) extensions. Using the Oracle SQL data
manipulation language (DML) statements, you can insert and delete complete LOBs.
There is also an extensive set of statements for piece-wise reading, writing, and
manipulating of LOBs within Java, PL/SQL, and the Oracle Call Interface.

For internal LOB types, both the locators and related data participate fully in the
transactional model of the Oracle server. The data for BFILEs does not participate in
transactions; however, BFILE locators are fully supported by Oracle server
transactions.
Unlike scalar quantities, a LOB value cannot be indexed by built-in indexing schemes.
However, you can use the various LOB APIs to build modules, including methods of
user-defined types, to access and manipulate LOB content. You can define the
semantics of data residing in LOBs and manipulate this data using the extensible
indexing framework.

See Also:

• Working with Unstructured Data Types for information on how to use

LOBs to store and manipulate binary and character data that represents
your domain.
• Oracle Database SecureFiles and Large Objects Developer's Guide for
detailed discussions of large objects.

1.3.1.2 Extensible Server Execution Environment

The Oracle type system decouples the implementation of a member method for a
user-defined type from the specification of that method. Oracle data cartridge
components can be implemented using a large number of popular programming
languages, such as PL/SQL, C, C++, or Java, extending the database server run-time
environment by user-defined methods, functions, and procedures.
Java offers data cartridge developers a powerful implementation choice for data
cartridge behavior. PL/SQL is a powerful procedural language that supports all the
object extensions for SQL. With PL/SQL, program logic can execute on the server and
perform traditional procedural language operations such as loops, if-then-else clauses,
and array access.
While PL/SQL and Java are powerful, certain computation-intensive operations such
as a Fast Fourier Transform or an image format conversion are handled more
efficiently by C programs. You can call C language programs from the server, running
them in a separate address space, thus insulating the server and protecting the
database from corruption by external procedure failures.
With certain reasonable restrictions, external procedures can callback the Oracle
Server using OCI. Callbacks are particularly useful for processing LOBs. External
procedure can use callbacks to perform piece-wise reads or writes of LOBs stored in
the database, or to manipulate domain indexes stored as index-organized tables in the
database.

1-6
 Chapter 1
Extending the Server: Services and Interfaces

Figure 1-2 External Programs Executing in a Separate Address Space

Inter-Language
Method Service

Oracle PL/SQL SQL

Server

JAVA C Listener

extproc

Oracle
Database /sh_libs/extlib.so

Oracle Address Space External Address Space

1.3.1.3 Extensible Indexing

Basic database management systems support a few types of access methods, such
as B+ trees and hash indexes, on a limited set of data types, such as numbers and
strings. For simple data types like integers and small strings, all aspects of indexing
can easily be handled by the database system. As data becomes more complex with
addition of text, spatial, image, video, and audio information, it requires complex data
types and specialized indexing techniques.
Complex data types have application-specific formats, indexing requirements, and
selection predicates. For example, there are many different means of document
encoding (ODA, XML, plain text) and information retrieval techniques (keyword, full-
text boolean, similarity, and probabilistic). Similarly, R-trees are an efficient method of
indexing spatial data. To enable you to define the index types necessary for your
business requirements, Oracle provides an extensible indexing framework.
Such user-defined indexes are called domain indexes because they index data in an
application-specific domain. The cartridge is responsible for defining the index
structure, maintaining the index content during load and update operations, and
searching the index during query processing. The physical index can be stored either
in the Oracle database as tables, or externally as a file.
A domain index is a schema object. It is created, managed, and accessed by routines
implemented as methods of a user-defined type called an indextype. The routines
that an indextype must implement, and the operations the routines must perform, are
described in Building Domain Indexes. Implementation of the routines is specific to an
application, and must therefore be completed by the cartridge developer.
With extensible indexing, the application must have the following processes:
• Define the structure of the domain index.
• Store the index data, either inside or outside the Oracle database.
• Manage, retrieve, and use the index data to evaluate user queries.

1-7
 Chapter 1
Extending the Server: Services and Interfaces

When the database system handles the physical storage of domain indexes, data
cartridges must have the following processes:
• Define the format and content of an index. Cartridges define an index structure
that can accommodate a complex data object.
• Build, delete, and update a domain index. Cartridges build and maintain the index
structures. Because indexes are modeled as collections of tuples, they directly
support in-place updates.
• Access and interpret the content of an index. Cartridges become an integral
component of query processing by handling content-related clauses for database
queries.
Typical relational and object-relational database management systems do not support
extensible indexing. Consequently, many applications maintain file-based indexes for
complex data in relational database tables. A considerable amount of code and effort
is required to complete the following tasks:
• Maintain consistency between external indexes and the related relational data.
• Support compound queries involving tabular values and external indexes.
• Manage the system, performing backup, recovery, storage allocation, and so on,
with multiple forms of persistent storage, such as files and databases.
By supporting extensible indexes, the Oracle server significantly reduces the level of
effort needed to develop solutions involving high-performance access to complex data
types.

1.3.1.4 Extensible Optimizer

The extensible optimizer lets user-defined functions and indexes collect statistical
information, such as selectivity and cost functions, and generates an execution plan
for a SQL statement. This information is used by the optimizer in choosing a query
plan, thus extending the optimizer to use the user-supplied information. The rule-
based optimizer remains unchanged.
An execution plan generated by the optimizer includes an access method for each
table in the FROM clause, and an ordering, called the join order, of the tables in the
FROM clause. System-defined access methods include indexes, hash clusters, and
table scans. For each table in the join order, the optimizer chooses a plan by
generating a set of join orders or permutations, computing the cost of each, and
selecting the one with the lowest cost. The cost of the join order is the sum of the
access method and join method costs.
The cost model is a group of algorithms used for calculating the cost of a given
operation. It can include varying levels of detail about the physical environment in
which the query runs. The current cost model includes the number of disk accesses
and estimates of network costs, with minor adjustments.
The optimizer also uses statistics about the objects referenced in the query to
calculate cost and selectivity, or the fraction of rows in a table that a query selects
(between 0 and 100, a percentage). The DBMS_STATS package contains methods for
generating these statistics.
Extensibility allows users to define new operators, index types, and domain indexes,
and enables the control of the three main components used by the optimizer to select
an execution plan: statistics, selectivity, and cost.

1-8
 Chapter 1
Extending the Server: Services and Interfaces

See Also:
Oracle Database PL/SQL Packages and Types Reference for information
about DBMS_STATS.

1.3.2 Extensibility Interfaces

There are three classes of extensibility interfaces: DBMS interfaces, cartridge basic
service interfaces, and data cartridge interfaces.

DBMS Interfaces
The DBMS interfaces offer the simplest kind of extensibility services. They can be
used through extensions to SQL or to the Oracle Call Interface (OCI). For example,
the extensible type manager uses the CREATE TYPE syntax in SQL. Similarly, extensible
indexing uses DDL and DML support for specifying and manipulating indexes.

Cartridge Basic Service Interfaces

Cartridge basic interfaces provide generic services like memory management, context
management, internationalization, and cartridge-specific management. They
implement behavior for new data types for the server's execution environment, and
provide routines that help developers implement portable and robust server-side
methods.

Data Cartridge Interfaces

When processing user-defined indextypes, Oracle calls data cartridge functions to
perform index search or fetch operations. For user-defined query optimization, the
query optimizer calls functions implemented by the data cartridge to compute the cost
of user-defined operators or functions.

1-9
2
Roadmap to Building a Data Cartridge
Consider the recommended process for developing data cartridges, including
relationships and dependencies among the steps of the process.

2.1 Data Cartridge Development Process

To understand the Data Cartridge development process, consider the project as a
whole.

Understanding the Purpose

The first step in developing a data cartridge is to establish the domain-specific value
you intend to provide by clearly defining the new capabilities of the cartridge. Specify
the objects the cartridge exposes to users.

Understand the Users

If the intended users of the cartridge are software developers, the extensibility of the
cartridge is of crucial importance. If they are end-users, the cartridge must be highly
attuned to its intended domain. The design of the cartridge should reflect a business
model that has a clear understanding of all users. Regardless of the size of the
cartridge, the development team must have a thorough understand the object-
relational database management system and apply it to the problems of the cartridge's
domain.

Plan the Project

Use a well-defined software development process, clearly identify expectations and
deliverables, and set reasonable milestones for Data Cartridge development.
Scheduling appropriate time for the project and having a realistic picture of available
resources skills makes the project more likely to succeed.

Implement the Project

A detailed outline of this phase of Data Cartridge development is described in
Implement the Project

Test and Installation

The final steps are to test the application and create the necessary installation scripts.

2.1.1 Implement the Project

• When choosing and designing objects, ensure that their names and semantics are
familiar and clearly understood by developers and end-users.
• When defining a collection of objects, consider the interface between the SQL side
of object methods and the programming language used in your application
development. Keep this interface as simple as possible by limiting the number of

2-1
 Chapter 2
Data Cartridge Development Process

methods that call library routines, avoiding numerous calls into low-level library
entry points, and writing large blocks of code that worked with pre-fetched data.
• After the interface is defined, proceed along parallel paths, as illustrated in
Figure 2-1. You can proceed on the paths in any order that suits the available
resources.
The left-most of these parallel paths packages existing 3GL code that performs
relevant operations in a run-time library such as a DLL, possibly with new entry
points on top of old code. The library routines are called by the SQL component of
the object's method code. Where possible, this code should be tested in a
standalone fashion using a 3GL test program.
The middle path defines and writes the object's type specification and the PL/SQL
components of the object's method code. Some methods can be written entirely in
PL/SQL, while others call into the external library. If your application requires an
external library, provide the library definition and the detailed bindings to library
entry routines.
The direction you take at the choice point depends on the complexity of the access
methods you must deploy to manipulate your data. If the query methods you need
are relatively simple, you can build regular indexes. If your data is complex, you
must define complex index types to make use of Oracle's extensible indexing
technology. If your project uses multi-domain queries, you should make use of
Oracle's extensible optimizer technology.
If your situation does not involve executing queries on multiple domains, and I/O is
the only significant factor affecting performance, then the standard optimizing
techniques are probably sufficient. However, if there are other factors such as
CPU cost affecting performance, you may still use the extensible optimizer.

2-2
 Chapter 2
Installing and Using Data Cartridges

Figure 2-1 Cartridge Development Process

Inventory domain

Yes Have
implementation

No

Write SQL and PL/SQL

for object’s
type specification

Inventory access
methods

Yes Existing indexes No

access data

Build regular indexes Define index types

Yes Multi-domain No
queries

Implement extensible No I/O cost

optimizer only significant
factor
Yes

Package existing Use existing

code optimizer

Test

Develop installation
scripts and
documentation

2.2 Installing and Using Data Cartridges

Installation of a data cartridge is the process of assembling its components so that
the server can locate them and understand the user-defined type definitions. To
correctly place these components, you must:
1. Define tables and user-defined types in the server. This is usually accomplished
by running SQL scripts.
2. Place the dynamic link libraries in the location expected by the linkage
specification.
3. Copy online documentation, help files, and error message files to a managed
location.

2-3
 Chapter 2
Requirements and Guidelines for Data Cartridge Components

4. Register the user-defined types with the server by running SQL scripts that load
each new types defined for the cartridge. This step must be performed from a
privileged account.
5. Grant the necessary access privileges to the users of the cartridge.

2.3 Requirements and Guidelines for Data Cartridge

Components
The following requirements and guidelines apply to some database objects associated
with data cartridges.

2.3.1 Cartridge Schemas

The database components that form each cartridge must be installed in a schema that
has the same name as the cartridge. If a cartridge uses multiple schemes, the first 10
characters of each schema name must be identical to the cartridge name. Note that
the length of schema names in Oracle is limited to 30 bytes, or 30 characters in a
single-byte language.
The database components of a data cartridge that must be placed in the cartridge
schema include names for types, tables, views, directories, libraries and packages.
Because the schema name and username are always the same in Oracle, the choice
of a schema name determines the username.

2.3.2 Cartridge Globals

Some database-level cartridge components are in scope, and are therefore visible to
all users instead of being within the scope of a single user or schema. Examples of
such globals are roles, synonyms, and sequences. All global names should start with
the cartridge name, and be of the form:
C$CARTRIDGEGLOBAL

2.3.3 Cartridge Error Message Names or Error Codes

Currently, error code ORA-20000 is reserved for all errors generated by applications
that use Oracle products. The error message text is customizable. You should write
the cartridge-specific error messages in the form:
ORA-20000: C$CARTRIDGE-NNNN:%s

where
• C$CARTRIDGE is the name of the cartridge where the error originated
• NNNN is the number of the error message, unique to that cartridge
• %s is the description of the cartridge-specific error

2-4
 Chapter 2
Data Cartridge Deployment

Note:
Oracle Database Error Messages for information on writing and managing
error messages

2.3.4 Cartridge Installation Directory

Oracle recommends that you create a cartridge installation directory, specific to a
vendor or client organization. This installation directory should includes the operating
system-level components of the cartridge, such as shared libraries, configuration files,
directories, and similar components. This directory name should be identical to the
prefix chosen by the organization, and created under the root directory for the
platform.

2.3.5 Cartridge Files

Oracle recommends that you place error message files associated with each cartridge
into cartridge-specific subdirectories. It is also convenient to keep configuration files in
a cartridge-specific subdirectory.

2.3.6 Shared Library Names for External Procedures

Shared libraries (.so or .dll files) can be placed either into the cartridge installation
directory (all library names must be unique), or into a separate directory. If you are
using a separate directory, the file names should start with the cartridge name,
excluding the initial C$. If there are many such libraries, each name should start with
the first seven letters of the cartridge name, again excluding the C$.

2.4 Data Cartridge Deployment

At the deployment level, you face several common issues. The optimal approach to
these problems depends on the needs of your application. The following list includes
tasks that should form the basis of your checklist, and some proposed solutions.
• You need a way to install and uninstall your cartridge components. This includes
libraries, database objects, flat files, programs, configuration tools, administration
tools, and other objects.
• You should allow for installation of multiple versions of a cartridge to provide
backward compatibility and availability. Incorporate Oracle's migration facilities into
your strategy.
• You must track which data cartridges are installed to support other cartridges that
depend on them.
• You must track different versions of installed components.
• You must provide an upgrade path for migrating to newer versions of cartridges.
Again, Oracle's migration facilities can be helpful.
• To limit access to cartridge components to specific users and roles, combine
Oracle's security mechanisms with procedures that operate under invoker's and
definer's rights depending on the need.

2-5
 Chapter 2
Data Cartridge Deployment

• You must keep track of which users have access to a cartridge for administration
purposes. Consider making use of a table with appropriate triggers.
• Knowing where cartridges are installed is often a security and administration
concern. There is currently no easy way of knowing which cartridges are installed
in a particular database or what users have access to the cartridge or any of its
components. If this information is important in your situation, keep track of it by
any convenient method.

2.4.1 Data Cartridge Naming Conventions

Consider how the components of a data cartridge should be named; this discussion is
intended for independent software vendors (ISVs) and others who are creating
cartridges to be used by others.
Note that most examples that illustrate Data Cartridges do not follow the naming
conventions, primarily because they are intended to be as simple and as generic as
possible. As your familiarity with the technology increases and you consider building
data cartridges to be for use by others, you should understand and follow these
naming conventions.
The naming conventions assume a single-byte character set.

See Also:

• "Cartridge Internationalization" for information on using other character

sets
• "Globalization Support" for information on support for multiple languages
and locales

2.4.1.1 Need for Naming Conventions in Data Cartridges

In a production environment, an Oracle database might have multiple data cartridges
installed. These data cartridges could be from different development groups or
vendors, thus developed in isolation. Each data cartridge consists of various schema
objects inside the database, and other components visible at the operating system
level, such as external procedures in shared libraries. If multiple data cartridges tried
to use the same names for schema objects or operating system-level entities, the
result would be incorrect and inconsistent behavior.
Furthermore, because exception conditions during the run-time operation of data
cartridges can cause the Oracle server to return errors, it is important to prevent
conflicts between error or message codes of different data cartridges. These conflicts
can arise if, for example, two cartridges use the same error code for different error
conditions. Having unique error and message codes ensures that the origin of the
exception condition can be readily identified.

2.4.1.2 Unique Name Format

To prevent multiple data cartridge components from having the same name, Oracle
recommends the following convention to ensure unique naming of data cartridges.
This convention depends on each organization developing data cartridges choosing a

2-6
 Chapter 2
Data Cartridge Deployment

unique name. Oracle recommends that cartridge developers follow a unique name
format that starts with a C$.

Data cartridges and their components should have names of the following format:
C$pppptttm.ccccc

Table 2-1 describes the parts of this naming convention format.

Oracle recommends that all characters in the name except for the dollar sign, $, as the
second character be alphanumeric: letters, numbers, underscores, and hyphens.
For example, Acme Cartridge Company chooses and registers a prefix of ACME. It
provides an audio data cartridge and a video data cartridge, and chooses AUD and VID
as the type codes, respectively. It has no other information to include in the cartridge
name, and so it chooses an arbitrary number 1 for the miscellaneous information
indicator. As a result, the two cartridge names are:
• C$ACMEAUD1
• C$ACMEVID1
For each cartridge, a separate schema must be created, and Acme uses the cartridge
name as the schema name. Thus, all database components of the audio cartridge
must be created under the schema C$ACMEAUD1, and all database components of the
video cartridge must be created under the schema C$ACMEVID1. Examples of some
components might include:
• C$ACMEVID1.mf_rewind
• C$ACMEVID1.vid_ops_package
• C$ACMEVID1.vid_stream_lib
Each organization is responsible for specific naming requirements after the C$pppp
portion of the object name. For example, Acme Cartridge Company must ensure that
all of its cartridges have unique names and that all components within a cartridge have
unique names.

2.4.1.3 Data Cartridge Naming Conventions

The following table describes the parts of this naming convention format

Table 2-1 Data Cartridge Naming Conventions

Part Explanation Example

C$ Recommended by Oracle for all data cartridges.
pppp Prefix selected by the data cartridge creator. (Must be ACME
exactly four characters.)
ttt Type of cartridge, using an abbreviation meaningful to AUD (for audio)
the creator. Three characters.
m Miscellaneous information indicator, to allow a 1 (perhaps a version
designation meaningful to the creator. One character. number)
. (period) Period required if specifying an object in full
schema.object form.
ccccc Component name. Variable length. mf_set_volume

2-7
 Chapter 2
Data Cartridge Deployment

2.4.2 Cartridge Registration

A naming scheme requires a registration process to handle the administration of
names of components that form a data cartridge.

2.4.3 Cartridge Directory Structure and Standards

You need some directory standards that specify where to put your binaries, support
files, messages files, administration files, and libraries.
You must also define a database user who installs your cartridges. One possible
solution is to use EXDSYS, for External Data Cartridge System user.

Note:
The EXDSYS user must have special privileges necessary for running
cartridges. This user may be installed as part of cartridge installation, but a
better solution is to make the user part of the database installation by
including this process in the standard database creation script.

2.4.4 Cartridge Upgrades

Administrators need a safe way to upgrade a cartridge and its related metadata to a
newer version of the cartridge. You also require a process for upgrading data and
removing obsolete data. This may entail installation support and database support for
moving to newer database cartridge types
Administrators also require a means to update tables using cartridge types when a
cartridge changes.

2.4.5 Import and Export of Cartridge Objects

To import and export objects, you must understand how Oracle's import and export
facilities handle Oracle objects. In particular, you must know how types are handled
and whether the type methods are imported and exported, and also whether user-
defined methods are supported.

2.4.6 Cartridge Versioning

There are two types of cartridge versioning problems that must be addressed: internal
and external.

External Versioning
External versioning is the easier of the two versioning problems. You must be able to
track a cartridge version number and act accordingly upon installation or configuration
based on versioning information.

2-8
 Chapter 2
Data Cartridge Deployment

Internal Versioning
Internal versioning is the harder problem. Ideally, you would like a mechanism to
support multiple versions of a cartridge in the database. This would provide backward
compatibility and also make for high availability.

2.4.7 Cartridge Internationalization

You might want to internationalize your cartridges, so they can support multiple
languages and access Globalization Support facilities for messages and parsing.
Oracle recommends that data cartridge component names use the ASCII character
set.
If you must name the data cartridge components in a character set other than ASCII,
Oracle assigns you a unique four-character prefix. However, this increases the number
of bytes required to hold the prefix. The names of all Oracle schema objects must fit
into 30 bytes. In ASCII, this equals 30 characters. If you have, for example, a six-byte
character set and request a four-character prefix string, Oracle might truncate your
request to a smaller number of characters.

See Also:
Oracle Database Globalization Support Guide

2.4.8 Cartridge Administration

When planning and developing a data cartridge, you should consider the issues
involved in administering its use.

Administrating Data Cartridge Access

• How do administrators know who has access to a cartridge?
Administrators must administer access rights to internal and external components
such as programs and data files to specific users and roles.
• How do administrators restrict access to certain tables, types, views, and other
cartridge components to individual users and roles?
For security reasons, administrators must be allowed to restrict access to types on
an individual basis.
Each Cartridge design should have it's own security considerations. In building
complex data cartridges, you need a way to identify the various components of
your cartridge and instances of the cartridge, so administrators can grant and
revoke security roles on identifiable components.

Invoker’s Rights
Invoker's rights is a special privilege that allows the system to access database objects
to which it would not normally have access. The special user SYS has such rights.
Unless you grant privileges to public, the user you create to install and run your
cartridge needs this privilege.

2-9
 Chapter 2
Data Cartridge Deployment

Data Cartridge Configuration

Data cartridges need a front end to handle deployment issues, such as installation,
and configuration tools. While each data cartridge may have differing security needs, a
basic front end that allows a user to install, configure, and administer data cartridge
components is necessary.
This front end may just be some form of knowledge base or on-line documentation. In
any case, it should be online, easy to navigate, and contain templates exhibiting
standards and starting points.

2.4.9 Data Cartridge Development Approach

General Data cartridge development approach.
In developing a data cartridge, take a systematic approach. Start with small, easy
tasks and build incrementally toward a comprehensive solution.

2.4.9.1 Creating a Data Cartridge Plan

General overview of a plan for creating Data Cartridges.

1. Experiment with the examples in: Power Demand Cartridge Example , PSBTREE:
Extensible Indexing Example, and Pipelined Table Functions: Interface Approach
Example.

2. Create the prototype of your own data cartridge, starting with a single user-defined
type and a few data elements and methods. You can add user-defined types, data
elements, and methods, specific indextypes, and user-defined operators as you
expand the cartridge's capabilities.

3. Begin by implementing your methods entirely in SQL, and add callouts to 3GL
code later if you need them.
4. Test and debug your cartridge.

2.4.9.2 Developing Data Cartridges

A general development process for developing data cartridges
When you have a working prototype you may want to follow a development process
that includes the following steps.

1. Identify your areas of domain expertise.

2. Identify those areas of expertise that are relevant to persistent data.

3. Consider the feasibility of packaging one or more of these areas as a new data
cartridge or as an extension to an existing cartridge.
4. Use an object-oriented methodology to help decide what object types to include in
data cartridges.
5. Build and test the cartridges, one at a time.

2-10
Part II
Building Data Cartridges
This part describes the processes used to build data cartridge components.
3
Defining Object Types for Data Cartridges
Object types are crucial to building data cartridges in that they enable domain-level
abstractions to be captured in the database.

See Also:

• Oracle Database Object-Relational Developer's Guide

• Oracle Database Concepts
• Oracle Database Advanced Application Developer's Guide
• Oracle Database PL/SQL Language Reference

3.1 Objects and Object Types for Data Cartridges

In the Oracle Object-Relational Database Management System (ORDBMS), you use
object types to model real-world entities. An object type has attributes, which reflect
the entity's structure, and methods, which implement the operations on the entity.
Attributes are defined using built-in types or other object types. Methods are functions
or procedures written in PL/SQL or an external language, like C, and stored in the
database.
A typical use for an object type is to impose structure on some part of the data in the
database. For example, an object type named DataStream could be used by a
cartridge to store large amounts of data in a character LOB (a data type for large
objects). This object type has attributes such as an identifier, a name, a date, and so
on. The statement in Example 3-1 defines the DataStream data type:

A method is a procedure or function that is part of the object type definition and that
can operate on the object type data attributes. Such methods are called member
methods, and they take the keyword MEMBER when you specify them as a component
of the object type. The DataStream type definition declares three methods. The first
two, DataStreamMin and DataStreamMax, calculate the minimum and maximum values,
respectively, in the data stream stored inside the character LOB. The third method,
DataStreamToInt, a map method, governs comparisons between instances of data
stream type.
After declaring the type, define the type body. The body contains the code for type
methods. Example 3-2 shows the type body definition for the DataStream type. It
defines the member function methods, DataStreamMin and DataStreamMax, and the
map method DataStreamToInt.

DataStreamMin and DataStreamMax are call routines in a PL/SQL package named

DS_Package. Since these methods are likely to be compute-intensive (they process
numbers stored in the CLOB to determine minimum and maximum values), they are
defined as external procedures and implemented in C. The external dispatch is routed

3-1
 Chapter 3
Assigning an Object Identifier to an Object Type in Data Cartridges

through a PL/SQL package named DS_Package. Such packages are discussed in

Oracle Database PL/SQL Packages and Types Reference. The third method,
DataStreamToInt, is implemented in PL/SQL. Because we have a identifier, id,
attribute in DataStream, this method can return the value of the identifier attribute. Most
map methods, however, are more complex than DataStreamToInt.

See Also:

• "Object Comparison" for information about map methods

Working with Unstructured Data Types
• Oracle Database SecureFiles and Large Objects Developer's Guide for
general information about LOBs

3.1.1 Defining a DataStream Data Type

Example 3-1 Defining a DataStream data type
create or replace type DataStream as object (
id integer,
name varchar2(20),
createdOn date,
data clob,
MEMBER FUNCTION DataStreamMin return pls_integer,
MEMBER FUNCTION DataStreamMax return pls_integer,
MAP MEMBER FUNCTION DataStreamToInt return integer;

3.1.2 Defining the Type Body

Example 3-2 Defining the Type Body
CREATE OR REPLACE TYPE BODY DataStream IS
MEMBER FUNCTION DataStreamMin return pls_integer is
a pls_integer := DS_Package.ds_findmin(data);
begin return a; end;
MEMBER FUNCTION DataStreamMax return pls_integer is
b pls_integer := DS_Package.ds_findmax(data);
begin return b; end;
MAP MEMBER FUNCTION DataStreamToInt return integer is
c integer := id;
begin return c; end;
end;

3.2 Assigning an Object Identifier to an Object Type in Data

Cartridges
The CREATE TYPE statement has an optional keyword OID, which associates a user-
specified object identifier (OID) with the type definition. It necessary to anyone who
creates an object type used in several database.s
Each type has an OID. If you create an object type and do not specify an OID, Oracle
generates an OID and assigns it to the type. Oracle uses the OID internally for

3-2
 Chapter 3
Constructor Methods in Data Cartridges

operations pertaining to that type. Using the same OID for a type is important if you
plan to share instances of the type across databases for such operations as export/
import and distributed queries.
Note that in CREATE TYPE with OID, an OID is assigned to the type itself. Each row in a
table with a column of the specified type has a row-specific OID.

Consider creating a SpecialPerson type, and then instantiating this type in two
different databases in tables named SpecialPersonTable1 and SpecialPersonTable2.
The RDBMS must know that the SpecialPerson type is the same type in both
instances, and therefore the type must be defined using the same OID in both
databases. If you do not specify an OID with CREATE TYPE, a unique identifier is created
automatically by the RDBMS. The syntax for specifying an OID for an object type is in
Example 3-3.
In Example 3-4, the SELECT statement generates an OID, and the CREATE TYPE
statement uses the OID in creating an object type named mytype. Be sure to use the
SELECT statement to generate a different OID for each object type to be created,
because this is the only way to guarantee that each OID is valid and globally unique.

3.2.1 Specifying an ODI for an Object Type in Data Cartridges

Example 3-3 Specifying an ODI for an Object Type
CREATE OR REPLACE TYPE type_name OID 'oid' AS OBJECT (attribute datatype [,...]);

3.2.2 Assigning and Using OIDs in Data Cartridges

Example 3-4 Assigning and Using OIDs
SQLPLUS> SELECT SYS_OP_GUID() FROM DUAL;
SYS_OP_GUID()
--------------------------------
19A57209ECB73F91E03400400B40BBE3
1 row selected.

SQLPLUS> CREATE TYPE mytype OID '19A57209ECB73F91E03400400B40BBE3'

2> AS OBJECT (attrib1 NUMBER);
Statement processed.

3.3 Constructor Methods in Data Cartridges

Oracle implicitly defines a constructor method for each object type that you define.
The name of the constructor method is identical to the name of the object type. The
parameters of the constructor method are exactly the data attributes of the object type,
and they occur in the same order as the attribute definition for the object type. Only
one constructor method can be defined for each object type.
In Example 3-5, the system creates a type named rational_type and implicitly
creates a constructor method for this object type.
When you instantiate an object of rational_type, you invoke the constructor method,
as demonstrated in Example 3-6:

3-3
 Chapter 3
Object Comparison in Data Cartridges

3.3.1 Creating a Type

Example 3-5 Creating a Type
CREATE TYPE rational_type (
numerator integer,
denominator integer);

3.3.2 Instantiating a Type Object

Example 3-6 Instantiating a Type Object
CREATE TABLE some_table (
c1 integer, c2 rational_type);
INSERT INTO some_table
VALUES (42, rational_type(223, 71));

3.4 Object Comparison in Data Cartridges

SQL performs comparison operations on objects. Some comparisons are explicit,
using the comparison operators (=, <, >, <>, <=, >=, !=) and the BETWEEN and IN
predicates. Other comparisons are implicit, as in the GROUP BY, ORDER BY, DISTINCT,
and UNIQUE clauses.

Comparison of objects uses special member functions of the object type: map
methods and order methods. To perform object comparison, you must implement
either a map method or an order method in the CREATE TYPE and CREATE TYPE BODY
statements. In Example 3-7, the type body for the DataStream type implements the
map member function.
The definition of the map member function relies on the presence of the id attribute of
the DataStream type to map instances to integers. Whenever a comparison operation
is required between objects of type DataStream, the map function DataStreamToInt()
is called implicitly by the system.
The object type rational_type does not have a simple id attribute like DataStream.
Instead, its map member function is complicated, as demonstrated in Example 3-8.
Because a map function can return any of the built-in types, rational_type can return
a value or type REAL.

If you have not defined a map or order function for an object type, it can only support
equality comparisons. Oracle SQL performs the comparison by doing a field-by-field
comparison of the attributes of that type.

3.4.1 Implementing a Member Function

Example 3-7 Implementing a Member Function
MAP MEMBER FUNCTION DataStreamToInt return integer is
c integer := id;
begin return c; end;

3-4
 Chapter 3
Object Comparison in Data Cartridges

3.4.2 Implementing Functions for Types Without a Simple Id Attribute

Example 3-8 Implementing Functions for Types Without a Simple Id Attribute
MAP MEMBER FUNCTION RationalToReal RETURN REAL IS
BEGIN
RETURN numerator/denominator;
END;
...

3-5
4
Implementing Data Cartridges in PL/SQL
You can use PL/SQL to implement data cartridge methods. Methods are procedures
and functions that define the operations permitted on data defined using the data
cartridge.

4.1 Methods
A method is procedure or function that is part of the object type definition, and that
can operate on the attributes of the type. Such methods are also called member
methods, and they take the keyword MEMBER when you specify them as a component
of the object type.
Consider simple examples for implementing a method, invoking a method, and
referencing an attribute in a method.

See Also:

• Oracle Database Concepts for information about method specifications,

names, and overloading
• Oracle Database PL/SQL Language Reference for further explanation
and examples

4.1.1 Implementing Methods

To implement a method, create the PL/SQL code and specify it within a CREATE TYPE
BODY statement. If an object type has no methods, no CREATE TYPE BODY statement for
that object type is required.
Example 4-1demonstrates the definition of an object type rational_type.

The definition in Example 4-2 defines the function gcd, which is used in the definition
of the normalize method in the CREATE TYPE BODY statement.

The statements in Example 4-3 implement the methods rat_to_real, normalize, and
plus for the object type rational_type.

4.1.1.1 Defining an Object Type

Example 4-1 Defining an Object Type
CREATE TYPE rational_type AS OBJECT
( numerator INTEGER,
denominator INTEGER,
MAP MEMBER FUNCTION rat_to_real RETURN REAL,
MEMBER PROCEDURE normalize,

4-1
 Chapter 4
Methods

MEMBER FUNCTION plus (x rational_type)

RETURN rational_type);

4.1.1.2 Defining a “Greatest Common Divisor” Function

Example 4-2 Defining a "Greatest Common Divisor" Function
CREATE FUNCTION gcd (x INTEGER, y INTEGER) RETURN INTEGER AS
-- Find greatest common divisor of x and y. For example, if
-- (8,12) is input, the greatest common divisor is 4.
-- This normalizes (simplifies) fractions.
-- (You need not try to understand how this code works, unless
-- you are a math wizard. It does.)
--
ans INTEGER;
BEGIN
IF (y <= x) AND (x MOD y = 0) THEN
ans := y;
ELSIF x < y THEN
ans := gcd(y, x); -- Recursive call
ELSE
ans := gcd(y, x MOD y); -- Recursive call
END IF;
RETURN ans;
END;

4.1.1.3 Implementing Methods for an Object Type

Example 4-3 Implementing Methods for an Object Type
CREATE TYPE BODY rational_type
( MAP MEMBER FUNCTION rat_to_real RETURN REAL IS
-- The rat-to-real function converts a rational number to
-- a real number. For example, 6/8 = 0.75
BEGIN
RETURN numerator/denominator;
END;

-- The normalize procedure simplifies a fraction.

-- For example, 6/8 = 3/4
MEMBER PROCEDURE normalize IS
divisor INTEGER := gcd(numerator, denominator);
BEGIN
numerator := numerator/divisor;
denominator := denominator/divisor;
END;

-- The plus function adds a specified value to the

-- current value and returns a normalized result.
-- For example, 1/2 + 3/4 = 5/4
--
MEMBER FUNCTION plus(x rational_type)
RETURN rational_type IS
-- Return sum of SELF + x
BEGIN
r = rational_type(numerator*x.demonimator +
x.numerator*denominator,
denominator*x.denominator);
-- Example adding 1/2 to 3/4:
-- (3*2 + 1*4) / (4*2)

4-2
 Chapter 4
Methods

-- Now normalize (simplify). Here, 10/8 = 5/4

r.normalize;
RETURN r;
END;
END;

4.1.2 Invoking Methods

To invoke a method, use the syntax in Example 4-4.
In SQL statements only, you can use the syntax in Example 4-5.
Example 4-6 shows how to invoke a method named get_emp_sal in PL/SQL.

An alternative way to invoke a method is by using the SELF built-in parameter.

Because the implicit first parameter of each method is the name of the object on
whose behalf the method is invoked, Example 4-7 performs the same action as the
salary := employee.get_emp_sal(); line in Example 4-6.

In Example 4-7, employee is the name of the object on whose behalf the
get_emp_sal() method is invoked.

4.1.2.1 General Syntax for Invoking Methods

Example 4-4 Invoking Methods; General Syntax
object_name.method_name([parameter_list])

4.1.2.2 SQL Syntax for Invoking Methods

Example 4-5 Invoking Methods; SQL Syntax
correlation_variable.method_name([parameter_list])

4.1.2.3 PL/SQL Syntax for Invoking Methods

Example 4-6 Invoking Methods; PL/SQL Syntax
DECLARE
employee employee_type;
salary number;
...
BEGIN
salary := employee.get_emp_sal();
...
END;

4.1.2.4 Using the SELF Build-In Parameter

Example 4-7 Using the SELF Build-In Parameter
salary := get_emp_sal(SELF => employee);

4-3
 Chapter 4
Debugging PL/SQL Code

4.1.3 Referencing Attributes in a Method

Because member methods can reference the attributes and member methods of the
same object type without using a qualifier, a built-in reference, SELF, always identifies
the object on whose behalf the method is invoked.
Consider Example 4-8, where two statements set the value of variable var1 to 42.

The statements var1 := 42 and SELF.var1 := 42 have the same effect. Because
var1 is the name of an attribute of the object type a_type and because set_var1 is a
member method of this object type, no qualification is required to access var1 in the
method code. However, for code readability and maintainability, you can use the
keyword SELF in this context to make the reference to var1 more clear.

4.1.3.1 Setting Variable Values

Example 4-8 Setting Variable Values
CREATE TYPE a_type AS OBJECT (
var1 INTEGER,
MEMBER PROCEDURE set_var1);
CREATE TYPE BODY a_type (
MEMBER PROCEDURE set_var1 IS
BEGIN
var1 := 42;
SELF.var1 := 42;
END set_var1;
);

4.2 Debugging PL/SQL Code

One of the simplest ways to debug PL/SQL code is to try each method, block, or
statement interactively using SQL*Plus, and fix any problems before proceeding to the
next statement. If you need more information on an error message, enter the
statement SHOW ERRORS. Also. consider displaying statements for run-time debugging.
You can debug stored procedures and packages using the DBMS_OUTPUT package, by
inserting PUT and PUTLINE statements into the code to output the values of variables
and expressions to your terminal, as demonstrated in Example 4-9.
A PL/SQL tracing tool provides more information about exception conditions in
application code. You can use this tool to trace the execution of server-side PL/SQL
statements. Object type methods cannot be traced directly, but you can trace any
PL/SQL functions or procedures that a method calls. The tracing tool also provides
information about exception conditions in the application code. The trace output is
written to the Oracle server trace file. Note that only the database administrator has
access to the file.

4-4
 Chapter 4
Debugging PL/SQL Code

See Also:

• The Oracle Database SQL Tuning Guide describes the tracing tool
• The Oracle Database PL/SQL Packages and Types Reference and the
Oracle Database PL/SQL Language Reference describe the
DBMS_OUTPUT package

Example 4-9 Outputing Variable Values to the Terminal, for Debugging

Location in module: location
Parameter name: name
Parameter value: value

4.2.1 Notes for C and C++ Developers of Data Cartridges

If you are a C or C++ programmer, several PL/SQL conventions and requirements
may differ from your expectations.
• = means equal (not assign).
• := means assign (as in Algol).
• VARRAYs begin at index 1 (not 0).
• Comments begin with two hyphens (--), not with // or /*.
• The IF statement requires the THEN keyword.
• The IF statement must be concluded with the END IF keyword (which comes after
the ELSE clause, if there is one).
• There is no PRINTF statement. The comparable feature is the
DBMS_OUTPUT.PUT_LINE statement. In this statement, literal and variable text is
separated using the double vertical bar, ||.
• A function must have a return value, and a procedure cannot have a return value.
• If you call a function, it must be on the right side of an assignment operator.
• Many PL/SQL keywords cannot be used as variable names.

4.2.2 Common Potential Errors

Several kinds of errors that may occur while creating a data cartridge.

4.2.2.1 Signature Mismatches

13/19 PLS-00538: subprogram or cursor '<name>' is declared in an object
type specification and must be defined in the object type body
15/19 PLS-00539: subprogram '<name>' is declared in an object type body
and must be defined in the object type specification

If you see either or both of these messages, you have made an error with the
signature for a procedure or function. In other words, you have a mismatch between
the function or procedure prototype that you entered in the object specification, and
the definition in the object body.

4-5
 Chapter 4
Debugging PL/SQL Code

Ensure that parameter orders, parameter spelling (including case), and function
returns are identical. Use copy-and-paste to avoid errors in typing.

4.2.2.2 RPC Time Out

ORA-28576: lost RPC connection to external procedure agent
ORA-06512: at "<name>", line <number>
ORA-06512: at "<name>", line <number>
ORA-06512: at line 34

This error might occur after you exit the debugger for the DLL. Restart the program
outside the debugger.

4.2.2.3 Package Corruption

ERROR at line 1:
ORA-04068: existing state of packages has been discarded
ORA-04063: package body "<name>" has errors
ORA-06508: PL/SQL: could not find program unit being called
ORA-06512: at "<name>", line <number>
ORA-06512: at line <number>

This error might occur if you are extending an existing data cartridge; it indicates that
the package has been corrupted and must be recompiled.
Before you can perform the recompilation, you must delete all tables and object types
that depend upon the package that you are recompiling. To find the dependents on a
Windows NT system, use the Oracle Administrator toolbar. Click the Schema button,
log in as sys\<password>, and find packages and tables that you created. Drop these
packages and tables by entering SQL statements in the SQL*Plus interface, as shown
in Example 4-10:
Example 4-10 Dropping Packages and Tables
Drop type type_name;
Drop table table_name cascade constraints;

The recompilation can then be done using the SQL statements in Example 4-11:
Example 4-11 Recompiling Packages
Alter type type_name compile body;
Alter type type_name compile specification;

4-6
5
Implementing Data Cartridges in C, C++,
and Java
You can use C, C++, and Java to implement data cartridge methods. Methods are
procedures and functions that define the operations permitted on data defined using
the data cartridge. The focus is on issues related to developing and debugging
external procedures.

5.1 Shared Libraries

A shared library is an operating system file, such as a Windows DLL or a Solaris
shared object, which stores the coded implementation of external procedures. You can
access the shared library from Oracle by using an alias library, which is a schema
object that represents the library within PL/SQL. For security reasons, you need DBA
privileges to create an alias library.

See Also:

• Oracle Database PL/SQL Language Reference for more information on

using dedicated external procedure agents
• Oracle Database Security Guide for more information on Directory
Object support for libraries

5.1.1 Using Shared Libraries

To create the alias library, you must decide on the operating system location for the
library, log in as a DBA or as a user with the CREATE LIBRARY privilege, and then enter
the statement in Example 5-1. This creates the alias library schema object in the
database. After the alias library is created, you can refer to the shared library by the
name DS_Lib from PL/SQL. Example 5-1 specifies an absolute path for the library.

If you have copies of the library on multiple systems, to support distributed execution
of external procedures by designated or dedicated agents, use an environment
variable to specify the location of the libraries more generally, as in Example 5-2. This
statement uses the environment variable ${DS_LIB_HOME} to specify a common point
of reference or root directory from which the library can be found on all systems. The
string following the AGENT keyword specifies the agent (actually, a database link) that is
used to run any external procedure declared to be in library DS_Lib.

Oracle Database 12c Release 1 (12.1) introduces two extensions to the CREATE
LIBRARY syntax: an additional optional CREDENTIAL option and a DIRECTORY object
option. The credential option specifies the credentials of the operating system user
that the extproc agent impersonates when running an external subprogram that

5-1
 Chapter 5
External Procedures

specifies the library. The directory object option specifies the directory where the
shared library can be found.
Example 5-3 specifies a directory object option instead of a full path string; it also
specifies an optional credential argument. When an external procedure call through
the extproc process loads the PL/SQL library, extproc can authenticate and
'impersonate' on behalf of the defined smith_credential credential.

Note that if the directory or the credential does not exist when calling CREATE LIBRARY
statement, the library is created with errors.

5.1.1.1 Creating an Alias Library

Example 5-1 Creating an Alias Library
CREATE OR REPLACE LIBRARY DS_Lib AS
'/data_cartridge_dir/libdatastream.so';

5.1.1.2 Specifying the Location of the Library

Example 5-2 Specifying the Location of the Library Using an Environment
Variable
CREATE OR REPLACE LIBRARY DS_Lib AS
'${DS_LIB_HOME}/libdatastream.so' AGENT 'agent_link';

5.1.1.3 Creating an Alias Library through Directory Objects

Example 5-3 Creating an Alias Library Using the Directory Object Option
CREATE OR REPLACE LIBRARY DS_Lib AS
'libdatastream.so' IN data_cartridge_dir AGENT 'sales.hq.example.com'
CREDENTIAL smith_credential;

5.2 External Procedures

PL/SQL is a powerful language for database programming, but some methods are too
complex to code optimally in PL/SQL. For example, a routine to perform numeric
integration probably runs faster if it is implemented in C rather than PL/SQL.
To support such special-purpose processing, PL/SQL provides an interface for calling
routines written in other languages. This makes the strengths and capabilities of 3GLs,
like C, available through calls from a database server. Such a 3GL routine is called an
external procedure; it is stored in a shared library, registered with PL/SQL, and called
from PL/SQL at run time.
External procedures are an important tool for data cartridge developers. They can be
used not only to write fast, efficient, computation-intensive routines for cartridge types,
but also to integrate existing code with the database as data cartridges. Existing
shared libraries from other languages, such as a Windows NT DLL with C routines to
perform format conversions for audio files, can be called directly from a method in a
type implemented by an audio cartridge. Similarly, you can use external procedures to
process signals, drive devices, analyze data streams, render graphics, or process
numeric data.

5-2
 Chapter 5
External Procedures

See Also:
PL/SQL User's Guide and Reference for details on external procedures and
their use

5.2.1 Registering an External Procedure

To call an external procedure, you must not only instruct PL/SQL regarding the alias
library where the external procedure is defined, but also how to call this procedure and
what arguments to pass to it.
The DataStream type was defined in Example 3-1, and Example 3-2 defined methods
of DataStream by calling functions from the DS_Package package. Example 5-4 defines
the body of this package.
Note that in the PACKAGE BODY declaration clause, the package functions are tied to
external procedures in a shared library. The EXTERNAL clause in the function
declaration registers information about the external procedure, such as its name
(found after the NAME keyword), its location (which must be an alias library, following
the LIBRARY keyword), the language in which the external procedure is written
(following the LANGUAGE keyword), and so on.

The final part of the EXTERNAL clause in the example is the WITH CONTEXT specification.
Here, a context pointer is passed to the external procedure. The context pointer is
opaque to the external procedure, but is available so that the external procedure can
call back to the Oracle server, to potentially access more data in the same transaction
context.
Although the example describes external procedure calls from object type methods, a
data cartridge can use external procedures from a variety of other places in PL/SQL.
External procedure calls can appear in:
• Anonymous blocks
• Standalone and packaged subprograms
• Methods of an object type
• Database triggers
• SQL statements (calls to packaged functions only)

See Also:

• Description of the parameters that can accompany an EXTERNAL clause

in PL/SQL User's Guide and Reference.
Information about external procedures and formatting the call
specification when passing an object type to a C routine in Oracle
Database Advanced Application Developer's Guide
• The WITH CONTEXT clause in "Using the WITH CONTEXT Clause".

5-3
 Chapter 5
How PL/SQL Calls an External Procedure

5.2.1.1 Defining the Body of a Package

Example 5-4 Defining the Body of a Package
CREATE OR REPLACE PACKAGE BODY DS_Package AS
FUNCTION DS_Findmin(data CLOB) RETURN PLS_INTEGER IS EXTERNAL
NAME "c_findmin" LIBRARY DS_Lib LANGUAGE C WITH CONTEXT;
FUNCTION DS_Findmax(data CLOB) RETURN PLS_INTEGER IS EXTERNAL
NAME "c_findmax" LIBRARY DS_Lib LANGUAGE C WITH CONTEXT;
END;

5.3 How PL/SQL Calls an External Procedure

To call an external procedure, PL/SQL must know the DLL or shared library in which
the procedure resides. PL/SQL looks up the alias library in the EXTERNAL clause of the
subprogram that registered the external procedure. The data dictionary is used to
determine the actual path to the operating system shared library or DLL.
PL/SQL alerts a Listener process, which in turn starts a session-specific agent. Unless
some other particular agent has been designated either in the CREATE LIBRARY
statement for the procedure's specified library or in the agent argument of the CREATE
PROCEDURE statement, the default agent extproc is launched. The Listener hands over
the connection to the agent. PL/SQL passes the agent the name of the DLL, the name
of the external procedure, and any parameters passed in by the caller. The rest of this
account assumes that the agent launched is the default agent extproc.

After receiving the name of the DLL and the external procedure, extproc loads the
DLL and runs the external procedure. Also, extproc handles service calls, such as
raising an exception, and callbacks to the Oracle server. Finally, extproc passes to
PL/SQL any values returned by the external procedure. Figure 5-1 shows the flow of
control.

Figure 5-1 Calling an External Procedure

Oracle Server PL/SQL Listener

extproc

Oracle
Database /data_cartridge_dir/libdatastream.so

Oracle Address Space External Address Space

After the external procedure completes, extproc remains active throughout your
Oracle session. Thus, you incur the cost of spawning extproc only one time, no matter
how many calls you make. Still, you should call an external procedure only when the
computational benefits outweigh the cost. When you log off, extproc is killed.

5-4
 Chapter 5
Configuring Files for External Procedures

Note that the Listener must start extproc on the system that runs the Oracle server.
Starting extproc on a different system is not supported.

See Also:

• Oracle Database PL/SQL Language Reference for more information on

using dedicated external procedure agents to run an external procedure
• Oracle Database Administrator's Guide for information about
administering extproc and external procedure call

5.4 Configuring Files for External Procedures

The configuration files listener.ora and tnsnames.ora must have appropriate
entries, so that the Listener can dispatch the external procedures.
The Listener configuration file listener.ora must have a SID_DESC entry for the
external procedure, as demonstrated in Example 5-5.
Example 5-5 assumes the following:
• The Oracle instance is called o10.
• The system or node on which the Oracle server runs is named unix123.
• The installation directory for the Oracle server is /rdbms/u01.
• The port number for Oracle TCP/IP communication is the default Listener port
1521.
The tnsnames.ora file is the network substrate configuration file, and it must also be
updated to refer to the external procedure, as demonstrated in Example 5-6.
Example 5-6 assumes that IPC mechanisms are used to communicate with the
external procedure. You can also use, for example, TCP/IP for communication, in
which case the PROTOCOL parameter must be set to tcp.

See Also:
Oracle Database Administrator's Guide for more information about
configuring the listener.ora and tnsnames.ora files

5.4.1 Updating the Listener Configuration File

Example 5-5 Setting the SID_DESC Entry in the Listener Configuration FIle
# Listener configuration file
# This file is generated by stkconf.tsc

CONNECT_TIMEOUT_LISTENER = 0

LISTENER = (ADDRESS_LIST=
(ADDRESS=(PROTOCOL=ipc)(KEY=o10))

5-5
 Chapter 5
Configuring Files for External Procedures

(ADDRESS=(PROTOCOL=tcp)(HOST=unix123)(PORT=1521))
)

SID_LIST_LISTENER = (SID_LIST=
SID_DESC=(SID_NAME=o10)(ORACLE_HOME=/rdbms/u01/app/oracle/product/11.2.0.1.0)
SID_DESC=(SID_NAME=extproc)
(ORACLE_HOME=/rdbms/u01/app/oracle/product/11.2.0.1.0)
(PROGRAM=extproc))

5.4.2 Directing Network to Refer to External Procedures

Example 5-6 Updating the Network Substrate Configuration to Refer to
External Procedures
o10 = (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=unix123)(PORT=1521))
(CONNECT_DATA=(SID=o10)))
extproc_connection_data = (DESCRIPTION=(ADDRESS=(PROTOCOL=ipc)(KEY=o10))
CONNECT_DATA=(SID=extproc)))

5.4.3 Passing Parameters to an External Procedure

Passing parameters to an external procedure is complicated by several
circumstances:
• The set of PL/SQL data types does not correspond one-to-one with the set of C
data types.
• PL/SQL parameters can be null, whereas C parameters cannot. Unlike C,
PL/SQL includes the RDBMS concept of nullity.
• The external procedure might need the current length or maximum length of CHAR,
LONG RAW, RAW, and VARCHAR2 parameters.
• The external procedure might need character set information about CHAR,
VARCHAR2, and CLOB parameters.
• PL/SQL might need the current length, maximum length, or null status of values
returned by the external procedure.
Consider how to best specify a parameter list that deals with these circumstances. An
example of parameter passing is shown in Example 5-7, where the package function
DS_Findmin(data CLOB) calls the C routine c_findmin and the CLOB argument is
passed to the C routine as an OCILobLocator().

5.4.4 Specifying Data Types

You do not pass parameters to an external procedure directly. Instead, you pass them
to the PL/SQL subprogram that registered the external procedure. So, you must
specify PL/SQL data types for the parameters. Table 5-1 maps each PL/SQL data type
to a default external data type. The external data types map to C data type.
In some cases, you can use the PARAMETERS clause to override the default data type
mappings. For example, you can re-map the PL/SQL data type BOOLEAN from external
data type INT to external data type CHAR.

To avoid errors when declaring C prototype parameters, refer to Table 5-2, which
shows the C data type to specify for a given external data type and PL/SQL parameter

5-6
 Chapter 5
Configuring Files for External Procedures

mode. For example, if the external data type of an OUT parameter is CHAR, specify the
data type char* in your C prototype.

5.4.4.1 Conversion to External Datatypes

Table 5-1 Parameter Datatype Mappings

PL/SQL Type Supported External Types Default External

Type

BINARY_INTEGER, CHAR, UNSIGNED CHAR, SHORT, UNSIGNED INT

BOOLEAN, SHORT, INT, UNSIGNED INT, LONG,
PLS_INTEGER UNSIGNED LONG, SB1, UB1, SB2, UB2,
SB4, UB4, SIZE_T

NATURAL, NATURALN, CHAR, UNSIGNED CHAR, SHORT, UNSIGNED UNSIGNED INT

POSITIVE, SHORT, INT, UNSIGNED INT, LONG,
POSITIVEN, UNSIGNED LONG, SB1, UB1, SB2 ,UB2,
SIGNTYPE SB4, UB4, SIZE_T

FLOAT, REAL FLOAT FLOAT

DOUBLE PRECISION DOUBLE DOUBLE

CHAR, CHARACTER, STRING STRING

LONG, ROWID, VARCHAR,
VARCHAR2

LONG RAW, RAW RAW RAW

BFILE, BLOB, CLOB OCILOBLOCATOR OCILOBLOCATOR

5.4.4.2 Conversion from External Datatypes

Table 5-2 External Data Type Mappings

External Data Type IN, IN by Reference, IN OUT,

RETURN RETURN by Reference OUT

CHAR char char * char *

UNSIGNED CHAR unsigned char unsigned char * unsigned char *

SHORT short short * short *

UNSIGNED SHORT unsigned short unsigned short * unsigned short *

INT int int * int *

5-7
 Chapter 5
Configuring Files for External Procedures

Table 5-2 (Cont.) External Data Type Mappings

External Data Type IN, IN by Reference, IN OUT,

RETURN RETURN by Reference OUT

UNSIGNED INT unsigned int unsigned int * unsigned int *

LONG long long * long *

UNSIGNED LONG unsigned long unsigned long * unsigned long *

SIZE_T size_t size_t * size_t *

SB1 sb1 sb1 * sb1 *

UB1 ub1 ub1 * ub1 *

SB2 sb2 sb2 * sb2 *

UB2 ub2 ub2 * ub2 *

SB4 sb4 sb4 * sb4 *

UB4 ub4 ub4 * ub4 *

FLOAT float float * float *

DOUBLE double double * double *

STRING char * char * char *

RAW unsigned char * unsigned char * unsigned char *

OCILOBLOCATOR OCILobLocator * OCILobLocator * OCILobLocator **

5.4.5 Using the Parameters Clause

You can optionally use the PARAMETERS clause to pass additional information about
PL/SQL formal parameters and function return values to an external procedure. You
can also use this clause to reposition parameters.

See Also:
Oracle Database PL/SQL Language Reference.

5-8
 Chapter 5
Using Callbacks

5.4.6 Using the WITH CONTEXT Clause

When launched, an external procedure must access the database. For example,
DS_Findmin does not copy the entire CLOB data over to c_findmin, because doing so
would vastly increase the amount of stack that the C routine needs. Instead, the
PL/SQL function just passes a LOB locator to the C routine, with the intent that the
database is accessed again from C to read the actual LOB data.

When the C routine reads the data, it can use the OCI buffering and streaming
interfaces associated with LOBs, so that only incremental amounts of stack are
needed. Such re-access of the database from an external procedure is known as a
callback.
To be able to call back to a database, you must use the WITH CONTEXT clause to give
the external procedure access to the database environment, service, and error
handles. When an external procedure is called using WITH CONTEXT, the corresponding
C routine automatically gets an argument of type OCIExtProcContext* as its first
parameter. The order of the parameters can be changed using the PARAMETERS clause.
You can use this context pointer to fetch the handles using the OCIExtProcGetEnv call,
and then call back to the database. This procedure is shown in Example 5-7.

See Also:
Oracle Call Interface Programmer's Guide for details about OCI callbacks

5.5 Using Callbacks

An external procedure that runs on the Oracle server can call the access function
OCIExtProcGetEnv() to obtain the OCI environment and service handles. With the
OCI, you can use callbacks to execute SQL statements and PL/SQL subprograms,
fetch data, and manipulate LOBs. Moreover, callbacks and external procedures
operate in the same user session and transaction context, so they have the same user
privileges.
Example 5-7 is a version of c_findmin that is simplified to illustrate callbacks.

Example 5-7 Using Callbacks

Static OCIEnv *envhp;
Static OCISvcCtx *svchp;
Static OCIError *errhp;
Int c_findmin (OCIExtProcContext *ctx, OCILobLocator *lobl) {
sword retval;
retval = OCIExtProcGetEnv (ctx, &envhp, &svchp, &errhp);
if ((retval != OCI_SUCCESS) && (retval != OCI_SUCCESS_WITH_INFO))
exit(-1);
/* Use lobl to read the CLOB, compute the minimum, and store the value
in retval. */
return retval;
}

5-9
 Chapter 5
Common Potential Errors

5.5.1 Restrictions on Callbacks

With callbacks, the following SQL statements and OCI routines are not supported:
• Transaction control statements such as COMMIT
• Data definition statements such as CREATE
• Object-oriented OCI routines such as OCIRefClear
• Polling-mode OCI routines such as OCIGetPieceInfo
• The following OCI routines:
– OCIEnvInit()
– OCIInitialize()
– OCIPasswordChange()
– OCIServerAttach()
– OCIServerDetach()
– OCISessionBegin ()
– OCISessionEnd ()
– OCISvcCtxToLda()
– OCITransCommit()
– OCITransDetach()
– OCITransRollback()
– OCITransStart()
• Also, with OCI routine OCIHandleAlloc(), the following handle types are not
supported:
– OCI_HTYPE_SERVER
– OCI_HTYPE_SESSION
– OCI_HTYPE_SVCCTX
– OCI_HTYPE_TRANS

5.6 Common Potential Errors

Consider several common errors that may occur when running external procedures.

5.6.1 Calls to External Functions

You may have specified the wrong path or wrong name for the DLL file, or you may
have tried to use a DLL on a network mounted drive (a remote drive).
Can't Find DLL
ORA-06520: PL/SQL: Error loading external library
ORA-06522: Unable to load DLL
ORA-06512: at "<name>", line <number>

5-10
 Chapter 5
Debugging External Procedures

ORA-06512: at "<name>", line <number>

ORA-06512: at line <number>

5.6.2 RPC Time Out

This error might occur after you exit a debugger while debugging a shared library or
DLL. Simply disconnect your client and reconnect to the database.
ORA-28576: lost RPC connection to external procedure agent
ORA-06512: at "<name>", line <number>
ORA-06512: at "<name>", line <number>
ORA-06512: at line <number>

5.7 Debugging External Procedures

Usually, when an external procedure fails, its C prototype is faulty. That is, the
prototype does not match the one generated internally by PL/SQL. This can happen if
you specify an incompatible C data type. For example, to pass an OUT parameter of
type REAL, you must specify float *. Specifying float, double *, or any other C data
type, results in a mismatch.
In such cases, you might get a lost RPC connection to external procedure agent error,
which means that agent extproc terminated abnormally because the external
procedure caused a core dump. To avoid errors when declaring C prototype
parameters, refer to Table 5-2.

5.7.1 Using Package DEBUG_EXTPROC

To help you debug external procedures, PL/SQL provides the utility package
DEBUG_EXTPROC. To install the package, run the script dbgextp.sql, which you can find
in the PL/SQL demo directory.
To use the package, follow the instructions in dbgextp.sql. Your Oracle account must
have EXECUTE privileges on the package and CREATE LIBRARY privileges.

Note that DEBUG_EXTPROC works only on platforms with debugger utilities that can
attach to a running process.

5.7.2 Debugging C Code in DLLs on Windows NT Systems

If you are developing on a Windows NT system, you may perform the following
additional actions to debug external procedures:
1. Invoke the Windows NT Task Manager; press Ctrl+Alt+Del and select Task
Manager.
2. In the Processes display, select ExtProc.exe.
3. Right click, and select Debug.
4. Select OK in the message box.
At this point, if you have built your DLL in a debug fashion with Microsoft Visual C+
+, Visual C++ is activated.
5. In the Visual C++ window, select Edit > Breakpoints.
6. Use the breakpoint identified in dbgextp.sql in the PL/SQL demo directory.

5-11
 Chapter 5
Guidelines for Using External Procedures with Data Cartridges

5.8 Guidelines for Using External Procedures with Data

Cartridges
Make sure to write thread-safe external procedures. In particular, avoid using static
variables, which can be shared by routines running in separate threads.
For help in creating a dynamic link library, look in the RDBMS subdirectory /public,
where a template makefile can be found.

When calling external procedures, never write to IN parameters or overflow the

capacity of OUT parameters. PL/SQL does no run-time checks for these error
conditions. Likewise, never read an OUT parameter or a function result. Also, always
assign a value to IN OUT and OUT parameters and to function results. Otherwise, your
external procedure does not return successfully.
If you include the WITH CONTEXT and PARAMETERS clauses, you must specify the
parameter CONTEXT, which shows the position of the context pointer in the parameter
list. If you omit the PARAMETERS clause, the context pointer is the first parameter passed
to the external procedure.
If you include the PARAMETERS clause and the external procedure is a function, you
must specify the parameter RETURN (not RETURN property) in the last position.

For every formal parameter, there must be a corresponding parameter in the

PARAMETERS clause. Also, ensure that the data types of parameters in the PARAMETERS
clause are compatible with those in the C prototype, because no implicit conversions
are done.
A parameter for which you specify INDICATOR or LENGTH has the same parameter
mode as the corresponding formal parameter. However, a parameter for which you
specify MAXLEN, CHARSETID, or CHARSETFORM is always treated like an IN parameter,
even if you also specify BY REFERENCE.

With a parameter of type CHAR, LONG RAW, RAW, or VARCHAR2, you must use the property
LENGTH. Also, if that parameter is IN OUT or OUT and null, you must set the length of the
corresponding C parameter to zero.

See Also:
For more information about multithreading, see the Oracle Database
Heterogeneous Connectivity Administrator's Guide.

5.9 Java Methods

To use Java Data Cartridges, it is important that you know how to load Java class
definitions, about how to call stored procedures, and about context management.

5-12
 Chapter 5
Java Methods

See Also:
Cartridge Services Using C, C++ and Javafor information on ODCI classes.

5-13
6
Working with Unstructured Data Types
Some data types that are not native to the Oracle Database can be represented in the
Oracle Database as Large Objects (LOBs). You can implement them in PL/SQL and
OCI for Data Cartridges.

6.1 Overview of Cartridges and Unstructured Data Types

Some data cartridges must handle large amounts oMultif raw binary data, such as
graphic images or sound waveforms, or character data, such as text or streams of
numbers. Oracle supports large objects, LOBs, to handle these kinds of data.

• Internal LOBs are stored in the database tablespaces in a way that optimizes
space and provides efficient access. Internal LOBs participate in the transactional
model of the server.
Internal LOBs can store binary data (BLOBs), single-byte character data (CLOBs), or
fixed-width single-byte or multibyte character data (NCLOBs). An NCLOB consists of
character data that corresponds to the national character set defined for the
Oracle database. Varying width character data is not supported in Oracle.
• External LOBs are stored in operating system files outside the database
tablespaces as BFILEs, binary data. They cannot participate in transactions.
Both internal LOBs and in BFILEs provide considerable flexibility in handling large
amounts of data.
Data stored in a LOB is called the LOB's value. To the Oracle server, a LOB's value is
unstructured and cannot be queried. You must unpack and interpret a LOB's value in
cartridge-specific ways.
LOBs can be manipulated using the Oracle Call Interface, OCI, or the PL/SQL
DBMS_LOB package. You can write functions, including methods on object types that
can contain LOBs, to manipulate parts of LOBs.

See Also:
Oracle Database SecureFiles and Large Objects Developer's Guide for
details on LOBs

6.2 Using DDL for LOBs

LOB definition can involve the CREATE TYPE and the CREATE TABLE statements.
Example 6-1 specifies a CLOB within a data type named lob_type.

Example 6-2 creates an object table, lob_table, in which each row is an instance of
lob_type data:

6-1
 Chapter 6
LOB Locators

Example 6-3 shows how to store LOBs in a regular table, as opposed to an object table
as in Example 6-2.
When creating LOBs in tables, you can set the LOB storage, buffering, and caching
properties.

See Also:
Oracle Database SQL Language Reference manual and the Oracle
Database SecureFiles and Large Objects Developer's Guide for information
about using LOBs in CREATE TABLE, ALTER TABLE, CREATE TYPE and ALTER
TYPE statements

6.2.1 Creating a LOB Attribute of a Type

Example 6-1 Creating a CLOB Attribute of a Type
CREATE OR REPLACE TYPE lob_type AS OBJECT (
id INTEGER,
data CLOB );

6.2.2 Creating a LOB Object Table

Example 6-2 Creating a LOB Object Table
CREATE TABLE lob_table OF lob_type;

6.2.3 Creating LOB Columns

Example 6-3 Creating LOB Columns in a Table
CREATE TABLE lob_table1 (
id INTEGER,
b_lob BLOB,
c_lob CLOB,
nc_lob NCLOB,
b_file BFILE );

6.3 LOB Locators

LOBs can be stored with other row data or separate from row data. Regardless of the
storage location, each LOB has a locator, which can be viewed as a handle or pointer
to the actual location. Selecting a LOB returns the LOB locator instead of the LOB value.
Example 6-4 selects the LOB locator for b_lob and places it a PL/SQL local variable
named image1.

When you use an API function to manipulate the LOB value, you refer to the LOB using
the locator. The PL/SQL DBMS_LOB package contains useful routines to manipulate
LOBs, such as PUT_LINE() and GETLENGTH(), as in Example 6-5.

In the OCI, LOB locators are mapped to LOBLocatorPointers, such as OCILobLocator

*.

6-2
 Chapter 6
Emptying LOBs

For a BFILE, the LOB column has its own distinct locator, which refers to the LOB's
value that is stored in an external file in the server's file system. This implies that two
rows in a table with a BFILE column may refer to the same file or two distinct files. A
BFILE locator variable in a PL/SQL or OCI program behaves like any other automatic
variable. With respect to file operations, it behaves like a file descriptor available as
part of the standard I/O library of most conventional programming languages.

See Also:

• Oracle Database SecureFiles and Large Objects Developer's Guide

• Oracle Database SecureFiles and Large Objects Developer's Guide for
DBMS_LOB API

6.3.1 Selecting a LOBs and Assigning it to a Local Variable

Example 6-4 Selecting a LOB Locator and Assigning it to a Local Variable
DECLARE
image1 BLOB;
image_no INTEGER := 101;
BEGIN
SELECT b_lob INTO image1 FROM lob_table
WHERE key_value = image_no;
...
END;

6.3.2 Manipulating LOBs

Example 6-5 Manipulating LOBs with PUT_LINE() and GETLENGTH()
BEGIN
DBMS_OUTPUT.PUT_LINE('Size of the Image is: ',
DBMS_LOB.GETLENGTH(image1));
END;

6.4 Emptying LOBs

Use the special functions EMPTY_BLOB and EMPTY_CLOB in INSERT or UPDATE statements
of SQL DML to initialize a NULL or non-NULL internal LOB to empty. These are available
as special functions in Oracle SQL DML, and are not part of the DBMS_LOB package.

Before writing data to an internal LOB using OCI or the DBMS_LOB package, the LOB
column must be made non-null; it must contain a locator that points to an empty or
populated LOB value. You can initialize a BLOB column's value to empty by using the
function EMPTY_BLOB in the VALUES clause of an INSERT statement. Similarly, a CLOB or
NCLOB column's value can be initialized by using the function EMPTY_CLOB. Here is the
syntax of these functions:
FUNCTION EMPTY_BLOB() RETURN BLOB;
FUNCTION EMPTY_CLOB() RETURN CLOB;

6-3
 Chapter 6
Using the OCI to Manipulate LOBs

EMPTY_BLOB returns an empty locator of type BLOB and EMPTY_CLOB returns an empty
locator of type CLOB, which can also be used for NCLOBs. The functions don't have an
associated pragma.
An exception is raised if you use these functions anywhere but in the VALUES clause of
a SQL INSERT statement or as the source of the SET clause in a SQL UPDATE
statement.
Example 6-6 shows EMPTY_BLOB() used with SQL DML.

Example 6-7 shows how to use EMPTY_CLOB() in PL/SQL programs.

6.4.1 Using EMPTY_BLOB() in SQL

Example 6-6 Using EMPTY_BLOB() with SQL DML
INSERT INTO lob_table VALUES (1001, EMPTY_BLOB(), 'abcde', NULL);
UPDATE lob_table SET c_lob = EMPTY_CLOB() WHERE key_value = 1001;
INSERT INTO lob_table VALUES (1002, NULL, NULL, NULL);

6.4.2 Using EMPTY_CLOB() in PL/SQL

Example 6-7 Using EMPTY_CLOB() in PL/SQL Programs
DECLARE
lobb CLOB;
read_offset INTEGER;
read_amount INTEGER;
rawbuf RAW(20);
charbuf VARCHAR2(20);
BEGIN
read_amount := 10; read_offset := 1;
UPDATE lob_table SET c_lob = EMPTY_CLOB()
WHERE key_value = 1002 RETURNING c_lob INTO lobb;
dbms_lob.read(lobb, read_amount, read_offset, charbuf);
dbms_output.put_line('lobb value: ' || charbuf);
END

6.5 Using the OCI to Manipulate LOBs

The OCI includes functions that enable access to data stored in BLOBs, CLOBs, NCLOBs,
and BFILEs. These functions are introduced in Table 6-1.

Table 6-2 compares the OCI and PL/SQL (DBMS_LOB package) interfaces in terms of
LOB access.

Example 6-8 shows how to select a LOB from the database into a locator. It assumes
that the type lob_type has two attributes, id of type INTEGER and data of type CLOB,
and that a table, lob_table, of type lob_type, exists.

A sample program, populate.c, uses the OCI to populate a CLOB with the contents of
a file.

6-4
 Chapter 6
Using the OCI to Manipulate LOBs

See Also:
Oracle Call Interface Programmer's Guide for detailed documentation,
including parameters, parameter types, return values, and example code.

6.5.1 OCI Functions for Manipulating LOBs

Table 6-1 Summary of OCI Functions for Manipulating LOBs

Function Description
OCILobAppend() Appends LOB value to another LOB.
OCILobArrayRead() Reads LOB data for multiple locators in one round-trip.
OCILobArrayWrite() Writes LOB data for multiple locators in one round-trip.
OCILobAssign() Assigns one LOB locator to another.
OCILobCharSetForm() Returns the character set form of a LOB.
OCILobCharSetId() Returns the character set ID of a LOB.
OCILobClose() Closes a previously opened LOB or BFILE.
OCILobCopy2() Copies all or a portion of a LOB into another LOB;
replaces the deprecated method OCILobCopy().
OCILobCreateTemporary() Creates a temporary LOB.
OCILobErase2() Erases all or part of aLOB, starting at a specified offset;
replaces the deprecated method OCILobErase().
OCILobFileClose() Closes an open BFILE.
OCILobFileCloseAll() Closes all open BFILEs.
OCILobFileExists() Determines if a BFILE exists.
OCILobFileGetName() Returns the name of a BFILE.
OCILobFileIsOpen() Determines if the BFILE is open.
OCILobFileOpen() Opens a BFILE.
OCILobFileSetName() Sets the name of a BFILE in a locator.
OCILobFreeTemporary() Frees a temporary LOB.
OCILobGetChunkSize() Gets the chunk size of a LOB.
OCILobGetContentType() Gets the user-specified content type string for the data
in a SecureFile, if set.
OCILobGetLength2() Gets the length of a LOB; replaced the deprecated
method OCILobGetLength().
OCILobGetOptions() Gets the enabled settings that correspond to the
specified input option types for a specified SecureFile
LOB.
OCILobGetStorageLimit() Gets the maximum length of a LOB (BLOB, CLOB, or
NCLOB), in bytes.
OCILobIsEqual() Determines if two LOB locators refer to the same LOB.
OCILobIsOpen() Determines if a LOB or BFILE is open.

6-5
 Chapter 6
Using the OCI to Manipulate LOBs

Table 6-1 (Cont.) Summary of OCI Functions for Manipulating LOBs

Function Description
OCILobIsTemporary() Determines if a locator points to a temporary LOB.
OCILobLoadFromFile2() Loads BFILE data into an internal LOB; replaced the
deprecated method OCILobLoadFromFile().
OCILobLocatorAssign() Assigns one LOB or BFILE locator to another.
OCILobLocatorIsInit() Tests to see if a LOB locator is initialized.
OCILobOpen() Opens a LOB in the specified mode.
OCILobRead2() Reads a specified portion of a non-null LOB or a BFILE
into a buffer; replaces the deprecated method
OCILobRead().
OCILobSetContentType() Sets a content type string for the data in the
SecureFile LOB.
OCILobSetOptions() Enables option settings for a SecureFile LOB.
OCILobTrim2() Truncates a LOB; replaces the deprecated method
OCILobTrim().
OCILobWrite2() Writes data from a buffer into a LOB, writing over
existing data; replaces the deprecated method
OCILobWrite().
OCILobWriteAppend2() Writes data starting at the current end of a LOB;
replaces the deprecated method
OCILobWriteAppend().

6.5.2 Comparing OCI and PL/SQL Interfaces

Table 6-2 OCI and PL/SQL (DBMS_LOB) Interfaces Compared

OCI (ociap.h) PL/SQL DBMS_LOB (dbmslob.sql)

N/A
DBMS_LOB.COMPARE()

N/A
DBMS_LOB.INSTR()

N/A
DBMS_LOB.SUBSTR()

OCILobAppend() DBMS_LOB.APPEND()

N/A [use PL/SQL assign operator]

OCILobAssign()

N/A
OCILobCharSetForm()

N/A
OCILobCharSetId()

6-6
 Chapter 6
Using the OCI to Manipulate LOBs

Table 6-2 (Cont.) OCI and PL/SQL (DBMS_LOB) Interfaces Compared

OCI (ociap.h) PL/SQL DBMS_LOB (dbmslob.sql)

OCILobCopy() DBMS_LOB.COPY()

OCILobErase() DBMS_LOB.ERASE()

OCILobFileClose() DBMS_LOB.FILECLOSE()

OCILobFileCloseAll() DBMS_LOB.FILECLOSEALL()

OCILobFileExists() DBMS_LOB.FILEEXISTS()

OCILobFileGetName() DBMS_LOB.FILEGETNAME()

OCILobFileIsOpen() DBMS_LOB.FILEISOPEN()

OCILobFileOpen() DBMS_LOB.FILEOPEN()

OCILobFileSetName() N/A (use BFILENAME operator)

OCILobGetLength() DBMS_LOB.GETLENGTH()

N/A
OCILobIsEqual()

N/A [always initialize]

OCILobLocatorIsInit()

OCILobRead() DBMS_LOB.READ()

OCILobTrim() DBMS_LOB.TRIM()

OCILobWrite() DBMS_LOB.WRITE()

6.5.3 Selecting a Stored LOB into a Locator

Example 6-8 Selecting a LOB from the Database into a Locator
/*-----------------------------------------------------------------------*/
/* Select lob locators from a CLOB column */
/* Use the 'FOR UPDATE' clause for writing to the LOBs. */
/*-----------------------------------------------------------------------*/
static OCIEnv *envhp;
static OCIServer *srvhp;
static OCISvcCtx *svchp;
static OCIError *errhp;
static OCISession *authp;
static OCIStmt *stmthp;

6-7
 Chapter 6
Using the OCI to Manipulate LOBs

static OCIDefine *defnp1;

static OCIBind *bndhp;

sb4 select_locator(int rowind)

{
sword retval;
boolean flag;
int colc = rowind;
OCILobLocator *clob;
text *sqlstmt = (text *)"SELECT DATA FROM LOB_TABLE WHERE ID = :1 FOR UPDATE";

if (OCIStmtPrepare(stmthp, errhp, sqlstmt, (ub4) strlen((char *)sqlstmt),

(ub4) OCI_NTV_SYNTAX, (ub4) OCI_DEFAULT))
{
(void) printf("FAILED: OCIStmtPrepare() sqlstmt\n");
return OCI_ERROR;
}

if (OCIStmtBindByPos(stmthp, bndhp, errhp, (ub4) 1, (dvoid *) &colc,

(sb4) sizeof(colc), SQLT_INT, (dvoid *) 0, (ub2 *)0, (ub2 *)0, (ub4) 0,
(ub4 *) 0, (ub4) OCI_DEFAULT))
{
(void) printf("FAILED: OCIStmtBindByPos()\n");
return OCI_ERROR;
}

if (OCIDefineByPos(stmthp, &defnp1, errhp, (ub4) 1, (dvoid *) &clob, (sb4) -1,

(ub2) SQLT_CLOB, (dvoid *) 0, (ub2 *) 0, (ub2 *) 0, (ub4) OCI_DEFAULT))
{
(void) printf("FAILED: OCIDefineByPos()\n");
return OCI_ERROR;
}

/* Execute the select and fetch one row */

if (OCIStmtExecute(svchp, stmthp, errhp, (ub4) 1, (ub4) 0,
(CONST OCISnapshot*) 0, (OCISnapshot*) 0, (ub4) OCI_DEFAULT))
{
(void) printf("FAILED: OCIStmtExecute() sqlstmt\n");
report_error();
return OCI_ERROR;
}

/* Now test to see if the LOB locator is initialized */

retval = OCILobLocatorIsInit(envhp, errhp, clob, &flag);
if ((retval != OCI_SUCCESS) && (retval != OCI_SUCCESS_WITH_INFO))
{
(void) printf("Select_Locator --ERROR: OCILobLocatorIsInit(),
retval = %d\n", retval);
report_error();
checkerr(errhp, retval);
return OCI_ERROR;
}

if (!flag)
{
(void) printf("Select_Locator --ERROR: LOB Locator is not initialized.\n");
return OCI_ERROR;
}

return OCI_SUCCESS;
}

6-8
 Chapter 6
Using DBMS_LOB Package to Manipulate LOBs

6.6 Using DBMS_LOB Package to Manipulate LOBs

The DBMS_LOB package can be used to manipulate LOBs from PL/SQL. Table 6-3
introduces its routines.
Example 6-9 calls the TRIM procedure to trim a CLOB value to a smaller length. It
assumes that the type lob_type has two attributes, id of type INTEGER and data of
type CLOB, and that a table, lob_table, of type lob_type, exists. Because this example
deals with CLOB data, the second argument to DBMS_LOB.TRIM, the literal 834004,
specifies the number of characters. If the example dealt with BLOB data, this argument
would be interpreted as a number of bytes.

See Also:
Oracle Database PL/SQL Packages and Types Reference provides full
details on using the routines of the DBMS_LOB package.

6.6.1 DBMS_LOB Package Routines

Table 6-3 Summary of DBMS_LOB Package Routines

Routine Description

APPEND() Appends the contents of the source LOB to the destination LOB.

COPY() Copies all or part of the source LOB to the destination LOB.

ERASE() Erases all or part of a LOB.

LOADBLOBFROMFILE() Loads BFILE data into an internal BLOB.

LOADCLOBFROMFILE() Loads BFILE data into an internal CLOB.

TRIM() Trims the LOB value to the specified shorter length.

WRITE() Write data to the LOB from a specified offsets

GETLENGTH Gets the length of the LOB value.

Return the matching position of the nth occurrence of the pattern

INSTR()
in the LOB.

READ() Reads data from the LOB starting at the specified offset

SUBSTR() Returns part of the LOB value starting at the specified offset.

6-9
 Chapter 6
LOBs in External Procedures

Table 6-3 (Cont.) Summary of DBMS_LOB Package Routines

Routine Description
Closes the file.
FILECLOSE()

Closes all previously opened files.

FILECLOSEALL()

Tests if the file exists on the server.

FILEEXISTS()

Gets the directory alias and file name.

FILEGETNAME()

FILEISOPEN() Tests the file was opened using the input BFILE locators.

Opens a file.
FILEOPEN()

6.6.2 Trimming a CLOB

Example 6-9 Trimming a CLOB
PROCEDURE Trim_Clob IS
clob_loc CLOB;
BEGIN
-- get the LOB Locator
SELECT data into clob_loc FROM lob_table
WHERE id = 179 FOR UPDATE;
-- call the TRIM Routine
DBMS_LOB.TRIM(clob_loc, 834004);
COMMIT;
END;

6.7 LOBs in External Procedures

LOB locators can be passed as arguments to an external procedure, as defined in
Example 6-1.
The corresponding C function gets an argument of type OCILobLocator *. When the
function defined in Example 6-10 is called, it invokes a c routine, c_findmin(), with the
signature int c_findmin(OCILobLocator*).

The routine c_findmin is in a shared library associated with DS_Lib. To use the pointer
OCILobLocator* to get data from the LOB, you must reconnect to the database by
making a callback.

6.7.1 Defining an External Procedure (PL/SQL)

Example 6-10 Defining a PL/SQL External Procedure
FUNCTION DS_Findmin(data CLOB) RETURN PLS_INTEGER IS EXTERNAL
NAME "c_findmin" LIBRARY DS_Lib LANGUAGE C;

6-10
 Chapter 6
LOBs and Triggers

6.8 LOBs and Triggers

You cannot write to a LOB (:old or :new value) in any kind of trigger.

In regular triggers, you can read the :old value, but you cannot read the :new value. In
INSTEAD OF triggers, you can read both the :old and the :new values.

You cannot specify LOB type columns in an OF clause, because BFILE types can be
updated without updating the underlying table on which the trigger is defined.
Using OCI functions or the DBMS_LOB package to update LOB values or LOB attributes of
object columns does not fire triggers defined on the table that contains the columns or
attributes.

6.9 Using Open/Close as Bracketing Operations for Efficient

Performance
The Open/Close functions let you indicate the beginning and end of a series of LOB
operations, so that large-scale operations, such updating indexes, can be performed
when the Close function is called. This means that when the Open call is made, the
index would not be updated each time the LOB is modified, and that such updating
would not resume until the Close call.

You do not have to wrap all LOB operations inside the Open/Close operations, but code
block can be very valuable for the following reasons:
• If you do not wrap LOB operations inside an Open/Close call, then each
modification to the LOB implicitly opens and closes the LOB, thereby firing all
triggers. If you do wrap the LOB operations inside a pair of Open...Close
operations, then the triggers are not fired for each LOB modification. Instead, one
trigger is fired when the Close call is made. Likewise, extensible indexes are not
updated until the Close call. This means that any extensible indexes on the LOB
are not valid between the Open...Close calls.
• You must apply this technology carefully because state, which reflects the
changes to the LOB, is not saved between the Open and the Close operations.
When you have called Open, Oracle no longer keeps track of what portions of the
LOB value were modified, nor of the old and new values of the LOB that result from
any modifications. The LOB value is still updated directly for each OCILob* or
DBMS_LOB operation, and the usual read consistency mechanism is still in place.
You may also want extensible indexes on the LOB to be updated, as LOB
modifications are made because the extensible LOB indexes are always valid and
may be used at any time.
• The API enables you to determine if the LOB is open. In all cases, openness is
associated with the LOB, not the locator. The locator does not save any state
information.

6.9.1 Errors and Restrictions Regarding Open/Close Operations

It is an error to commit the transaction before closing all previously opened LOBs. At
transaction rollback time, all LOBs that are still open are discarded, which means that
they are not closed, which fires the triggers.

6-11
 Chapter 6
Using Open/Close as Bracketing Operations for Efficient Performance

It is an error to Open/Close the same LOB twice, either with different locators or with the
same locator. It is an error to close a LOB that has not been opened.

Example 6-11 assumes that loc1 is refers to an open LOB, and is assigned to loc2. If
loc2 is subsequently used to modify the LOB value, the modification is grouped with
loc1's modifications. This means that there is only one entry in the LOB manager's
state, not one for each locator. When the LOB is closed, either through loc1 or loc2,
the triggers are fired, so all updates made to the LOB through either locator are
committed. After the close of the LOB, if the user tries to use either locator to modify the
LOB, the operation performs an implicit Open() and Close(), as Open() ...
operation ... Close(). Note that consistent read is still maintained for each locator.
Remember that it is the LOB, not the locator, that is opened and closed. No matter how
many copies of the locator are made, the triggers for the LOB are fired only one time on
the first Close() call.

6.9.1.1 Working with Open() and Close() Code Blocks

Example 6-11 Using Open() and Close() Code Block
open (loc1);
loc2 := loc1;
write (loc1);
write (loc2);
open (loc2); /* error because the LOB is open */
close (loc1); /* triggers are fired and all LOB updates made before this
statement by any locator are incorporated in the extensible
index */
write (loc2); /* implicit open, write, implicit close */

6-12
7
Using Extensible Indexing
Extensible indexing allows you to implement modes of indexing in addition to those
that are built into Oracle. Consider when you should create domain indexes, which
are indexes created using the extensible indexing framework.

7.1 Overview of Extensible Indexing

Consider some terms and methods for building indexes; this material may be familiar
to developers who worked on database applications previously.

7.1.1 Purpose of Indexes

With large amounts of data such as that in databases, indexes make locating and
retrieving the data faster and more efficient. Whether they refer to records in a
database or text in a technical manual, entries in an index indicate three things about
the items they refer to:
• What the item is ("employee information on Mary Lee" or "the definition of
extensible indexing")
• Where the item is ("record number 1000" or "page 100")
• How the item is stored ("in a consecutive series of records" or "as text on a page")
Most sets of data can be indexed in several different ways. To provide the most useful
and efficient access to data, it is often critical to choose the right style of indexing. This
is because no indexing method is optimal for every application.
Database applications normally retrieve data with queries, which often use indexes in
selecting subsets of the available data. Queries can differ radically in the operators
used to express them, and thus in the methods of indexing that provide the best
access.
• To learn which sales people work in the San Francisco office, you need an
operator that checks for equality. Hash structures handle equality operators very
efficiently.
• To learn which sales people earn more than x but less than y, you need an
operator that checks ranges. B-tree structures are better at handling range-
oriented queries.

7.1.2 Purpose of Extensible Indexing

Databases are constantly incorporating new types of information that are more
complex and more specific to certain tasks, such as medical or multimedia
applications. As a result, queries are becoming more complex, and the amount of data
they must scan continues to grow. Oracle provides the extensible indexing framework
so you can tailor your indexing methods to your data and your applications, thus
improving performance and ease of use.

7-1
 Chapter 7
Overview of Extensible Indexing

With extensible indexing, your application

• Defines the structure of the index
• Stores the index data, either inside the Oracle database (for example, in the form
of index-organized tables) or outside the Oracle database
• Manages, retrieves, and uses the index data to evaluate user queries
Thus, your application controls the structure and semantic content of the index. The
database system cooperates with your application to build, maintain, and employ the
domain index. As a result, you can create indexes to perform tasks that are specific to
the domain in which you work, and your users compose normal-looking queries using
operators you define.

7.1.3 When to Use Extensible Indexing

Oracle's built-in indexing facilities are appropriate to a large number of situations.
However, as data becomes more complex and applications are tailored to specific
domains, situations arise that require other approaches. For example, extensible
indexing can help you solve problems like these:
• Implementing new search operators using specialized index structures
You can define operators to perform specialized searches using your index
structures.
• Indexing unstructured data
The built-in facilities cannot index a column that contains LOB values.
• Indexing attributes of column objects
The built-in facilities cannot index column objects or the elements of a collection
type.
• Indexing values derived from domain-specific operations
Oracle object types can be compared with map functions or order functions. If the
object uses a map function, then you can define a function-based index for use in
evaluating relational predicates. However, this only works for predicates with
parameters of finite range; it must be possible to precompute function values for all
rows. In addition, you cannot use order functions to construct an index.

7.1.4 Index Structures

Consider some frequently-used index structures that illustrate the choices available to
designers of domain indexes.

7.1.4.1 B-tree
No index structure can satisfy all needs, but the self-balancing B-tree index comes
closest to optimizing the performance of searches on large sets of data. Each B-tree
node holds multiple keys and pointers. The maximum number of keys in a node
supported by a specific B-tree is the order of that tree. Each node has a potential of
order+1 pointers to the level below it. For example, the order=2 B-tree illustrated in
Figure 7-1 has tree pointers: to child nodes whose value is less than the first key, to
the child nodes whose value is greater than the first key and less than the second key,
and to the child nodes whose value is greater than the second key. Thus, the B-tree
algorithm minimizes the number of reads and writes necessary to locate a record by

7-2
 Chapter 7
Overview of Extensible Indexing

passing through fewer nodes than in a binary tree algorithm, which has only one key
and at most two children for each decision node. Here we describe the Knuth
variation in which the index consists of two parts: a sequence set that provides fast
sequential access to the data, and an index set that provides direct access to the
sequence set.
Although the nodes of a B-tree generally do not contain the same number of data
values, and they usually contain a certain amount of unused space, the B-tree
algorithm ensures that the tree remains balanced and that the leaf nodes are at the
same level.

Figure 7-1 B-tree Index Structure

x x

x x x x x x
Index set

x x x x x x x x x x x x x x x x x x x x x x x x x x x Sequence set
(with pointers to
data records)

7.1.4.2 Hash
Hashing gives fast direct access to a specific stored record based on a given field
value. Each record is placed at a location whose address is computed as some
function of some field of that record. The same function is used to insert and retrieve.
The problem with hashing is that the physical ordering of records has little if any
relation to their logical ordering. Also, there can be large unused areas on the disk.

Figure 7-2 Hash Index Structure

0 1 2
S300 Blake 30 Paris

3 4 5
S200 Jones 10 Paris

6 7 8
S500 Adams 30 Athens

9 10 11
S100 Smith 30 London S400 Clark 20 London

12

7-3
 Chapter 7
Overview of Extensible Indexing

7.1.4.3 k-d tree

Data that has two dimensions, such as latitude and longitude, can be stored and
retrieved efficiently using a variation on the k-d tree known as the 2-d tree.
In this structure, each node is a data type with fields for information, the two co-
ordinates, and a left-link and right-link, which can point to two children.

Figure 7-3 2-d Index Structure

A (XX, XX)

A (XX, XX)

B (XX, XX)

A (XX, XX)

B (XX, XX)

C (XX, XX)

This structure is good at range queries. That is, if the user specifies a point (xx, xx)
and a distance, the query returns the set of all points within the specified distance of
the original point.
2-d trees are easy to implement. However, because a 2-d tree containing k nodes can
have a height of k, insertion and querying can be complex.

7.1.4.4 Point Quadtree

The point quadtree, in Figure 7-4, is also used to represent point data in a two
dimensional spaces, but these structures divide regions into four parts where 2-d trees
divide regions into two. The fields of the record type for this node comprise an attribute
for information, two co-ordinates, and four compass points (such as NW, SW, NE, SE)
that can point to four children.

7-4
 Chapter 7
Extensible Indexing Framework

Figure 7-4 Point Quadtree Index Structure

A B

A B C

Like 2-d trees, point quadtrees are easy to implement. However, a point quadtree
containing k nodes can have a height of k, so insertion and querying can be complex.
Each comparison requires comparisons on at least two co-ordinates. In practice,
though, the lengths from root to leaf tend to be shorter in point quadtrees.

7.2 Extensible Indexing Framework

The extensible indexing framework is a SQL-based interface that lets you define
domain-specific operators and indexing schemes, and integrate these into the Oracle
server.
The extensible indexing framework consists of the following components:
• Indextypes: An indextype schema object specifies the routines that manage
definition, maintenance, and scan operations for application-specific indexes. An

7-5
 Chapter 7
Using the Text Indextype

indextype tells the Oracle server how to establish a user-defined index on a

column of a table or attribute of an object.
• Domain Indexes: An application-specific index created using an indextype is
called a domain index because it indexes data in application-specific domains. A
domain index is an instance of an index that is created, managed, and accessed
by the routines specified by an indextype.
• Operators: Queries and data manipulation statements can use application-
specific operators, such as the Overlaps operator in the spatial domain. User-
defined operators are bound to functions. They can also be evaluated using
indexes. For instance, the equality operator can be evaluated using a hash index.
An indextype provides an index-based implementation for the operators it defines.
• Index-Organized Tables: With index-organized tables, your application can
define, build, maintain, and access indexes for complex objects using a table
metaphor. To the application, an index is modeled as a table, where each row is
an index entry. Index-organized tables handle duplicate index entries, which can
be important with complex types of data.
The extensible indexing framework lets you perform the following actions:
• Encapsulate application-specific index management routines as an indextype
schema object.
• Define a domain index on table columns.
• Process application-specific operators efficiently.
With the extensible indexing framework, you can build a domain index that operates
much like any other Oracle index. Users write standard queries using operators you
define. To create, drop, truncate, modify, and search a domain index, the Oracle
server invokes the application code you specify as part of the indextype.

See Also:

• Defining Operators for detailed information on user-defined operators

• Oracle Database Administrator's Guide for detailed information on index-
organized tables

7.3 Using the Text Indextype

Consider an extensible indexing framework with a skeletal example that both defines a
new text indexing scheme using the Text indextype, and uses the Text indextype to
index and operate on textual data.

7.3.1 Defining the Indextype

The order in which you create the components of an indextype depends on whether or
not you are creating an index-based functional implementation.

7.3.1.1 Non-Index-Based Functional Implementations

To define the Text indextype, the indextype designer must follow these steps:

7-6
 Chapter 7
Using the Text Indextype

1. Define and code the functional implementation for the supported operator
The Text indextype supports an operator called Contains, which accepts a text
value and a key, and returns a number indicating whether the text contains the
key. The functional implementation of this operator is a regular function defined as:
CREATE FUNCTION TextContains(Text IN VARCHAR2, Key IN VARCHAR2)
RETURN NUMBER AS
BEGIN
.......
END TextContains;
2. Create the new operator and bind it to the functional implementation
CREATE OPERATOR Contains
BINDING (VARCHAR2, VARCHAR2) RETURN NUMBER USING TextContains;
3. Define a type that implements the index interface ODCIIndex
This involves implementing routines for index definition, index maintenance, and
index scan operations. Oracle calls:
• The index definition routines ODCIIndexCreate(), ODCIIndexAlter(), and
ODCIIndexDrop() to perform the appropriate operations when the index is
created, altered, or dropped, or the base table is truncated
• The index maintenance routines ODCIIndexInsert(), ODCIIndexDelete(), and
ODCIIndexUpdate() to maintain the text index when table rows are inserted,
deleted, or updated
• The index scan routines ODCIIndexStart(), ODCIIndexFetch(), and
ODCIIndexClose() to scan the text index and retrieve rows of the base table
that satisfy the operator predicate
CREATE TYPE TextIndexMethods
(
STATIC FUNCTION ODCIIndexCreate(...)
...
);
CREATE TYPE BODY TextIndexMethods
(
...
);
4. Create the Text indextype schema object
The indextype definition specifies the operators supported by the new indextype
and the type that implements the index interface.
CREATE INDEXTYPE TextIndexType
FOR Contains(VARCHAR2, VARCHAR2)
USING TextIndexMethods
WITH SYSTEM MANAGED STORAGE TABLES;

7.3.1.2 Index-Based Functional Implementations

If you are creating an index-based functional implementation, you perform the same
operations as for non-index-based functional implementations, but in a different order.
This order is required because definition of an index-based functional implementation
requires the implementation type as a parameter.
1. Define the implementation type
2. Define and code the functional implementation

7-7
 Chapter 7
Using the Text Indextype

3. Create the operator

4. Create the indextype

7.3.2 Using the Indextype

When the Text indextype has been successfully defined, users can define text indexes
on text columns and use the Contains operator to query text data.

Suppose the MyEmployees table is defined by the statement in Example 7-1.

To build a text domain index on the resume column, a user issues the statement in
Example 7-2.
To query the text data in the resume column, users issue statements like the one in
Example 7-3. The query execution uses the text index on resume to evaluate the
Contains predicate.

7.3.2.1 Declaring a New Table

Example 7-1 Declaring a New Table
CREATE TABLE MyEmployees
(employee_id NUMBER(6),
first_name VARCHAR2(20),
last_name VARCHAR2(25),
salary NUMBER(8,2),
resume VARCHAR2(2000),
location VARCHAR2(200),
department_id NUMBER(4));

7.3.2.2 Building a Text Domain Index for the Table

Example 7-2 Building a Text Domain Index
CREATE INDEX ResumeIndex ON MyEmployees(resume) INDEXTYPE IS TextIndexType;

7.3.2.3 Querying a Table Using a Contains() Operator

Example 7-3 Using the Contains() Operator
SELECT * FROM MyEmployees WHERE Contains(resume, 'Oracle') =1;

7-8
8
Building Domain Indexes
Consider domain indexes and the ODCIIndex interface: how to use of domain indexes,
their partitioning, applicable restrictions, and migration procedures.
If you use user-managed domain indexes, the information specific to their
implementation is in User-Managed Local Domain Indexes

8.1 Overview of Indextypes and Domain Indexes

A domain index is an index designed for a specialized domain, such as spatial or
image processing. Users can build a domain index of a given type after the designer
creates the indextype. The behavior of domain indexes is specific to an industry, a
business function, or some other special purpose; you must specify it during cartridge
development.
The system-managed approach to domain indexes, new in the Oracle Database 11g
Release 1, requires less programmatic overhead and delivers better performance than
the earlier user-managed domain indexes. It addresses the limitations of the user-
managed approach, and has the following benefits:
• Because the kernel performs many more maintenance tasks on behalf of the user,
there is no need for programmatic support for table and partition maintenance
operations. These operations are implemented by taking actions in the server,
thus requiring a very minimal set of user-defined interface routines to be coded by
the user. The cartridge code can then be relatively unaware of partition issues.
• The number of objects that must be managed to support local partitioned domain
indexes is identical to identical to those for non-partitioned domain indexes. For
local partitioned indexes, the domain index storage tables are equipartitioned with
respect to the base tables (using system-partitioned tables); therefore, the number
of domain index storage tables does not increase with an increase in the number
of partitions.
• A single set of query and DML statements can now access and manipulate the
system-partitioned storage tables, facilitating cursor sharing and enhancing
performance.
Oracle recommends that you develop new applications with system-managed domain
indexes instead of user-managed domain indexes.
Indextypes encapsulate search and retrieval methods for complex domains such as
text, spatial, and image processing. An indextype is similar to the indexes that are
supplied with the Oracle Database. The difference is that you provide the application
software that implements the indextype.
An indextype has two major components:
• The methods that implement the behavior of the indextype, such as creating and
scanning the index
• The operators that the indextype supports, such as Contains() or Overlaps()

8-1
 Chapter 8
ODCIIndex Interface

To create an indextype, you must perform the following steps:

• Define the supported operators and create the functions that implement them
• Create the methods that implement the ODCIIndex interface, and define the type
that encapsulates them, called the implementation type
• Create the indextype, specifying the implementation type and listing the operators
with their bindings
In this context:
• Interface means a logical set of documented method specifications (not a separate
schema object)
• ODCIIndex interface means a set of index definition, maintenance, and scan
routine specifications

See Also:
Defining Operators

8.2 ODCIIndex Interface

The ODCIIndex interface specifies all the routines you must supply to implement an
indextype. The routines must be implemented as type methods.
The ODCIIndex interface comprises the following method classes:

• Index definition methods

• Index maintenance methods
• Index scan methods
• Index metadata method

See Also:
Extensible Indexing Interface for method signatures and parameter
descriptions

8.2.1 Index Definition Methods

The application calls index definition methods when a user issues a CREATE, ALTER,
DROP, or TRUNCATE statement on an index of indextype.

8.2.1.1 ODCIIndexCreate()
When a user issues a CREATE INDEX statement that references the indextype, Oracle
calls your ODCIIndexCreate() method, passing it any parameters specified as part of
the CREATE INDEX... PARAMETERS (...) statement, plus the description of the index.

8-2
 Chapter 8
ODCIIndex Interface

Typically, this method creates the tables or files in which you plan to store index data.
Unless the base table is empty, the method should also build the index.

8.2.1.2 ODCIIndexAlter()
When a user issues an ALTER INDEX statement referencing your indextype, Oracle
calls your ODCIIndexAlter() method, passing it the description of the domain index to
be altered along with any specified parameters. This method is also called to handle
an ALTER INDEX with the REBUILD or RENAME options. What your method must do
depends on the nature of your domain index, so the details are left to you as the
designer of the indextype.

8.2.1.3 ODCIIndexDrop()
When a user destroys an index of your indextype by issuing a DROP INDEX statement,
Oracle calls your ODCIIndexDrop() method.

8.2.2 Index Maintenance Methods

Your index maintenance methods are called when users issue INSERT, UPDATE, and
DELETE statements on tables with columns or object type attributes indexed by your
indextype.

8.2.2.1 ODCIIndexInsert()
When a user inserts a record, Oracle calls your ODCIIndexInsert() method, passing it
the new values in the indexed columns and the corresponding row identifier.

8.2.2.2 ODCIIndexDelete()
When a user deletes a record, Oracle calls your ODCIIndexDelete() method, passing it
the old values in the indexed columns and the corresponding row identifier.

8.2.2.3 ODCIIndexUpdate()
When a user updates a record, Oracle calls your ODCIIndexUpdate() method, passing
it the old and new values in the indexed columns and the corresponding row identifier.

8.2.3 Index Scan Methods

Your index scan methods specify the index-based implementation for evaluating
predicates containing the operators supported by your indextype. Index scans involve
methods for initialization, fetching rows or row identifiers, and cleaning up after all rows
are returned.
There are two modes of evaluating the operator predicate and returning the resulting
set of rows:
• Precompute All: Compute the entire result set in ODCIIndexStart(). Iterate over
the results returning a batch of rows from each call to ODCIIndexFetch(). This
mode is applicable to operators that must look at the entire result set to compute
ranking, relevance, and so on for each candidate row. It is also possible to return
one row at a time if your application requires that.

8-3
 Chapter 8
ODCIIndex Interface

• Incremental Computation: Compute a batch of result rows in each call to

ODCIIndexFetch(). This mode is applicable to operators that can determine the
candidate rows one at a time without having to look at the entire result set. It is
also possible to return one row at a time if your application requires that.

8.2.3.1 ODCIIndexStart()
Oracle calls your ODCIIndexStart() method at the beginning of an index scan,
passing it information on the index and the operator. Typically, this method:
• Initializes data structures used in the scan
• Parses and executes SQL statements that query the tables storing the index data
• Saves any state information required by the fetch and cleanup methods, and
returns the state or a handle to it
• Sometimes generates a set of result rows to be returned at the first invocation of
ODCIIndexFetch()
The information on the index and the operator is not passed to the fetch and cleanup
methods. Thus, ODCIIndexStart() must save state data that must be shared among
the index scan routines and return it through an output sctx parameter. To share large
amounts of state data, allocate cursor-duration memory and return a handle to the
memory in the sctx parameter.

As member methods, ODCIIndexFetch() and ODCIIndexClose() are passed the built-

in SELF parameter, through which they can access the state data.

See Also:

• ODCIIndexClose()
• ODCIIndexFetch()
• ODCIIndexStart()
• Oracle Call Interface Programmer's Guide for information on memory
services and maintaining context

8.2.3.2 ODCIIndexFetch()
Oracle calls your ODCIIndexFetch() method to return the row identifiers of the next
batch of rows that satisfies the operator predicate, passing it the state data returned by
ODCIIndexStart() or the previous ODCIIndexFetch() call. The operator predicate is
specified in terms of the operator expression (name and arguments) and a lower and
upper bound on the operator return values. Thus, ODCIIndexFetch() must return the
row identifiers of the rows for which the operator return value falls within the specified
bounds. To indicate the end of index scan, return a NULL.

8-4
 Chapter 8
ODCIIndex Interface

See Also:

• ODCIIndexFetch()
• ODCIIndexStart()

8.2.3.3 ODCIIndexClose()
Oracle calls your ODCIIndexClose() method when the cursor is closed or reused,
passing it the current state. ODCIIndexClose() should perform whatever cleanup or
closure operations your indextype requires.

8.2.4 Index Metadata Method

The ODCIIndexGetMetadata() method is optional. If you implement it, the Export utility
calls it to write implementation-specific metadata into the Export dump file. This
metadata might be policy information, version information, individual user settings, and
so on, which are not stored in the system catalogs. The metadata is written to the
dump files as anonymous PL/SQL blocks that are executed at import time immediately
before the creation of the associated index.

8.2.5 Transaction Semantics During Index Method Execution

The index interface methods (with the exception of the index definition methods,
ODCIIndexCreate(), ODCIIndexAlter(), and ODCIIndexDrop()) are invoked under the
same transaction that triggered these actions. Thus, the changes made by these
routines are atomic and are committed or aborted based on the parent transaction. To
achieve this, there are certain restrictions on the nature of the actions that you can
perform in the different indextype routines:
• Index definition routines have no restrictions.
• Index maintenance routines can only execute Data Manipulation Language
statements. These DML statements cannot update the base table on which the
domain index is created.
• Index scan routines can only execute SQL query statements.
For example, if an INSERT statement caused the ODCIIndexInsert() routine to be
invoked, ODCIIndexInsert() runs under the same transaction as INSERT. The
ODCIIndexInsert() routine can execute any number of DML statements (for example,
insert into index-organized tables). If the original transaction aborts, all the changes
made by the indextype routines are rolled back.
However, if the indextype routines cause changes external to the database (like writing
to external files), transaction semantics are not assured.

8-5
 Chapter 8
ODCIIndex Interface

See Also:

• ODCIIndexAlter()
• ODCIIndexCreate()
• ODCIIndexDrop()
• ODCIIndexInsert()

8.2.6 Transaction Semantics for Index Definition Routines

The index definition routines do not have any restrictions on the nature of actions
within them. Consider ODCIIndexCreate() to understand this difference. A typical set
of actions to be performed in ODCIIndexCreate() could be:

1. Create an index-organized table.

2. Insert data into the index-organized table.
3. Create a secondary index on a column of the index-organized table.
To allow ODCIIndexCreate() to execute an arbitrary sequence of DDL and DML
statements, each statement is considered to be an independent operation.
Consequently, the changes made by ODCIIndexCreate() are not guaranteed to be
atomic. The same is true for other index-definition routines.

See Also:
ODCIIndexCreate()

8.2.7 Consistency Semantics during Index Method Execution

The index maintenance (and scan routines) execute with the same snapshot as the
top level SQL statement performing the DML (or query) operation. This keeps the
index data processed by the index method consistent with the data in the base tables.

8.2.8 Privileges During Index Method Execution

Indextype routines always execute as the owner of the index. To support this, the
index access driver dynamically changes user mode to index owner before invoking
the indextype routines.
For certain operations, indextype routines store information in tables owned by the
indextype designer. The indextype implementation must perform those actions in a
separate routine, which is executed using the definer's privileges.

8-6
 Chapter 8
Creating, Dropping, and Commenting Indextypes

See Also:
Oracle Database SQL Language Reference for details on CREATE TYPE

8.3 Creating, Dropping, and Commenting Indextypes

Consider the SQL statements that manipulate indextypes.

See Also:
Oracle Database SQL Language Reference for complete descriptions of
these SQL statements

8.3.1 Creating Indextypes

When you have implemented the ODCIIndex interface and defined the implementation
type, you can create a new indextype by specifying the list of operators supported by
the indextype and referring to the type that implements the index interface. Using the
information retrieval example, the DDL statement for defining the new indextype
TextIndexType, which supports the Contains operator and whose implementation is
provided by the type TextIndexMethods, as demonstrated by Example 8-1.

In addition to the ODCIIndex interface routines, the implementation type must

implement the ODCIGetInterfaces() routine. This routine returns the version of the
interface implemented by the implementation type. Oracle invokes the
ODCIGetInterfaces() routine when executing CREATE INDEXTYPE.

Example 8-1 Creating an Indextype

CREATE INDEXTYPE TextIndexType
FOR Contains (VARCHAR2, VARCHAR2)
USING TextIndexMethods
WITH SYSTEM MANAGED STORAGE TABLES;

8.3.2 Dropping Indextypes

To remove the definition of an indextype, use the DROP statement, as in Example 8-2:

The default DROP behavior is DROP RESTRICT semantics, that is, if one or more domain
indexes exist that uses the indextype then the DROP operation is disallowed. Users can
override the default behavior with the FORCE option, which drops the indextype and
marks any dependent domain indexes invalid.

See Also:
"Object Dependencies_ Drop Semantics_ and Validation" for details on
object dependencies and drop semantics

8-7
 Chapter 8
Domain Indexes

Example 8-2 Dropping an IndexType

DROP INDEXTYPE TextIndexType;

8.3.3 Commenting Indextypes

Use the COMMENT statement to supply information about an indextype or operator, as
shown in Example 8-3.
To place a comment on an indextype, the indextype must be in your own schema, or
you must have the COMMENT ANY INDEXTYPE privilege.

Example 8-3 Commenting an INDEXTYPE

COMMENT ON INDEXTYPE
TextIndexType IS 'implemented by the type TextIndexMethods to support the Contains
operator';

8.3.3.1 INDEXTYPE Comments

Comments on indextypes can be viewed in these data dictionary views:
• ALL_INDEXTYPE_COMMENTS displays comments for the user-defined indextypes
accessible to the current user.
• DBA_INDEXTYPE_COMMENTS displays comments for all user-defined indextypes in the
database.
• USER_INDEXTYPE_COMMENTS displays comments for the user-defined indextypes
owned by the current user.

Table 8-1 Views *_INDEXTYPE_COMMENTS

Column Data Type Required Description

Owner of the user-defined
OWNER VARCHAR2(30) NOT NULL
indextype

Name of the user-defined

INDEXTYPE_NAME VARCHAR2(30) NOT NULL
indextype
Comment for the user-defined
COMMENT VARCHAR2(4000)
indextype

8.4 Domain Indexes

Consider domain index operations and how metadata associated with the domain
index may be obtained.

8.4.1 Domain Index Operations

Consider how to create, alter, truncate, and drop domain indexes.

8-8
 Chapter 8
Domain Indexes

8.4.1.1 Creating a Domain Index

A domain index can be created on a column of a table, just like a B-tree index.
However, an indextype must be explicitly specified. Example 8-4 shows how to specify
an indextype on the MyEmployees table that was declared in Example 7-1.

The INDEXTYPE clause specifies the indextype to be used. The PARAMETERS clause
identifies any parameters for the domain index, specified as a string. This string is
passed uninterpreted to the ODCIIndexCreate() routine for creating the domain index.
In the preceding example, the parameters string identifies the language of the text
document (thus identifying the lexical analyzer to use) and the list of stop words which
are to be ignored while creating the text index.
Example 8-4 Creating a Domain Index
CREATE INDEX ResumeTextIndex ON MyEmployees(resume)
INDEXTYPE IS TextIndexType
PARAMETERS (':Language English :Ignore the a an');

8.4.1.2 Changing a Domain Index

The parameter string is passed uninterpreted to ODCIIndexAlter() routine, which takes
appropriate actions to alter the domain index. This example specifies an additional
stop word to ignore in the text index.
Example 8-5 Changing a Domain Index
ALTER INDEX ResumeTextIndex PARAMETERS (':Ignore on');

8.4.1.3 Renaming a Domain Index

The ALTER statement can be used to rename a domain index, as shown in
Example 8-6.
A statement of this form causes Oracle to invoke the ODCIIndexAlter() method, which
takes appropriate actions to rename the domain index.
Example 8-6 Renaming a Domain Index
ALTER INDEX ResumeTextIndex RENAME TO ResumeTIdx;

8.4.1.4 Rebuilding a Domain Index

The ALTER statement can be used to rebuild a domain index.

The same ODCIIndexAlter() routine is called as before, but with additional information
about the ALTER option.

When the end user executes an ALTER INDEX domain_index UPDATE BLOCK
REFERENCES for a domain index on an index-organized table (IOT), ODCIIndexAlter() is
called with the AlterIndexUpdBlockRefs bit set. This gives you the opportunity to
update guesses as to the block locations of rows that are stored in the domain index in
logical rowids.
Example 8-7 Rebuilding a Domain Index
ALTER INDEX ResumeTextIndex REBUILD PARAMETERS (':Ignore off');

8-9
 Chapter 8
Domain Indexes

8.4.1.5 Truncating a Domain Index

There is no explicit statement for truncating a domain index. However, when the
corresponding base table is truncated, the underlying storage table for the domain
indexes are also truncated. Additionally, ODCIIndexAlter() is invoked by the command
in Example 8-8, and it truncates ResumeTextIndex because its alter_option is set to
AlterIndexRebuild:

Example 8-8 Truncating a Domain Index

TRUNCATE TABLE MyEmployees;

8.4.1.6 Dropping a Domain Index

To drop an instance of a domain index, use the DROP INDEX statement, shown in
Example 8-9:
This results in Oracle calling the ODCIIndexDrop() method, passing it information
about the index.
Example 8-9 Dropping a Domain Index
DROP INDEX ResumeTextIndex;

8.4.2 Domain Indexes on Index-Organized Tables

Consider some issues that arise when an indextype creates domain indexes on index-
organized tables. You can use the IndexOnIOT bit of IndexInfoFlags in the
ODCIIndexInfo structure to determine if the base table is an IOT.

8.4.2.1 About Rowid Storage in a UROWID Column

When the base table of a domain index is an index-organized table, and you want to
store rowids for the base table in a table of your own, you should store the rowids in a
UROWID (universal rowid) column if you are testing rowids for equality.

If the rowids are stored in a VARCHAR column instead, comparisons for textual equality
of a rowid from the base table and a rowid from your own table fail in some cases
where the rowids pick out the same row. This is because index-organized tables use
logical instead of physical rowids, and, unlike physical rowids, logical rowids for the
same row can have different textual representations. Two logical rowids are equivalent
when they have the same primary key, regardless of the guess data block addresses
stored with them.
A UROWID column can contain both physical and logical rowids. Storing rowids for an
IOT in a UROWID column ensures that the equality operator succeeds on two logical
rowids that have the same primary key information but different guess DBAs.
If you create an index storage table with a rowid column by performing a CREATE TABLE
AS SELECT from the IOT base table, then a UROWID column of the correct size is
created for you in your index table. If you create a table with a rowid column, then you
must explicitly declare your rowid column to be of type UROWID(x), where x is the size
of the UROWID column. The size chosen should be large enough to hold any rowid from
the base table; thus, it should be a function of the primary key from the base table.

8-10
 Chapter 8
Domain Indexes

Use the query demonstrated by Example 8-10 to determine a suitable size for the
UROWID column.

Doing an ALTER INDEX REBUILD on index storage tables raises the same issues as
doing a CREATE TABLE if you drop your storage tables and re-create them. If, on the
other hand, you reuse your storage tables, no additional work should be necessary if
your base table is an IOT.

8.4.2.2 Determining the Size of a UROWID Column

Example 8-10 Getting the Size of a UROWID Column
SELECT (SUM(column_length + 3) + 7)
FROM user_ind_columns ic, user_indexes i
WHERE ic.index_name = i.index_name
AND i.index_type = 'IOT - TOP'
AND ic.table_ name = base_table;

8.4.2.3 DML on Index Storage Tables

If you maintain a UROWID column in the index storage table, then you must change the
type of the rowid bind variable in DML INSERT, UPDATE, and DELETE statements so that
it works for all kinds of rowids. Converting the rowid argument passed in to a text string
and then binding it as a text string works well for both physical and universal rowids.
This strategy may help you to code your indextype to work with both regular tables and
IOTs.

8.4.2.4 Start, Fetch, and Close Operations on Index Storage Tables

If you use an index scan-context structure to pass context between Start, Fetch, and
Close, you must alter this structure. In particular, if you store the rowid define variable
for the query in a buffer in this structure, then you must allocate the maximum size for
a UROWID in this buffer (3800 bytes for universal rowids in byte format, 5072 for
universal rowids in character format) unless you know the size of the primary key of
the base table in advance or wish to determine it at run time. You must also store a bit
in the context to indicate if the base table is an IOT, since ODCIIndexInfo is not
available in Fetch.

As with DML operations, setting up the define variable as a text string works well for
both physical and universal rowids. When physical rowids are fetched from the index
table, you can be sure that their length is 18 characters. Universal rowids, however,
may be up to 5072 characters long, so a string length function must be used to
determine the actual length of a fetched universal rowid.

8.4.2.5 Indexes on Non-Unique Columns

All values of a primary key column must be unique, so a domain index defined upon a
non-unique column of a table cannot use this column as the primary key of an
underlying IOT used to store the index. To work around this, you can add a column in
the IOT, holding the index data, to hold a unique sequence number. When a column
value is inserted in the table, generate a unique sequence number to go with it; you
can then use the indexed column with the sequence number as the primary key of the
IOT. (Note that the sequence-number column cannot be a UROWID because UROWID
columns cannot be part of a primary key for an IOT.) This approach also preserves the
fast access to primary key column values that is a major benefit of IOTs.

8-11
 Chapter 8
Domain Indexes

8.4.3 Domain Index Metadata

For B-tree indexes, users can query the USER_INDEXES view to get index information.
To provide similar support for domain indexes, you can provide domain-specific
metadata in the following manner:
• Define one or more tables that contain this meta information. The key column of
this table must be a unique identifier for the index. This unique key could be the
index name (schema.index). The remainder of the columns can contain your
metadata.
• Create views that join the system-defined metadata tables with the index meta
tables to provide a comprehensive set of information for each instance of a domain
index. It is your responsibility as the indextype designer to provide the view
definitions.

8.4.4 Moving Domain Indexes Using Export/Import

Like B-tree and bitmap indexes, domain indexes are exported and subsequently
imported when their base tables are exported. However, domain indexes can have
implementation-specific metadata associated with them that is not stored in the system
catalogs. For example, a text domain index can have associated policy information, a
list of irrelevant words, and so on. The export/import mechanism moves this metadata
from the source platform to the target platform.
To move the domain index metadata, the indextype must implement the
ODCIIndexGetMetadata() interface method. When a domain index is being exported,
this method is invoked and passes the domain index information. It can return any
number of anonymous PL/SQL blocks that are written into the dump file and executed
on import. If present, these anonymous PL/SQL blocks are executed immediately
before the creation of the associated domain index.
By default, secondary objects of the domain are not imported or exported. However, if
the interfaces ODCIIndexUtilGetTableNames() and ODCIIndexUtilCleanup() are
present, the system invokes them to determine if the secondary objects associated
with the domain indexes are part of the export/import operation.

See Also:

• ODCIIndexGetMetadata()
• ODCIIndexUtilCleanup()
• ODCIIndexUtilGetTableNames()
• Oracle Database Utilities for information about using Export/Import

8.4.5 Moving Domain Indexes Using Transportable Tablespaces

The transportable tablespaces feature lets you move tablespaces from one Oracle
database into another. You can use transportable tablespaces to move domain index
data as an alternative to exporting and importing it.

8-12
 Chapter 8
Domain Indexes

Moving data using transportable tablespaces can be much faster than performing
either an export and import, or unload and load of the data because transporting a
tablespace only requires copying datafiles and integrating tablespace structural
information. Also, you do not have to rebuild the index afterward as you do when
loading or importing. You can check for the presence of the TransTblspc flag in
ODCIIndexInfo to determine whether the ODCIIndexCreate() call is the result of an
imported domain index.
To use transportable tablespace for the secondary tables of a domain index, you must
provide two additional ODCI interfaces, ODCIIndexUtilGetTableNames() and
ODCIIndexUtilCleanup(), in the implementation type.

See Also:
Oracle Database Administrator’s Guide for information about using
transportable tablespaces

8.4.6 Domain Index Views

The following views provide information about secondary objects associated with
domain indexes accessible to the user; they are only relevant for domain indexes.
• ALL_SECONDARY_OBJECTS provide information about secondary objects associated
with domain indexes accessible to the user.
• DBA_SECONDARY_OBJECTS provides information about all secondary objects that are
associated with domain indexes in the database.
• USER_SECONDARY_OBJECTS provides information about secondary objects
associated with domain indexes owned by the current user.

Table 8-2 Views *_SECONDARY_OBJECTS

Column Data Type Required Description

Name of the domain index owner
INDEX_OWNER VARCHAR2(30) NOT NULL

Name of the domain index

INDEX_NAME VARCHAR2(30) NOT NULL

Owner of the secondary object

SECONDARY_INDEX_OWNER VARCHAR2(30) NOT NULL
created by the domain index

Name of the secondary object

SECONDARY_INDEX_NAME VARCHAR2(30) NOT NULL
created by the domain index
Specifies if a secondary object is
SECONDARY_OBJDATA_TYPE VARCHAR2(20) NOT NULL
created by either indextype or
statistics type

Example 8-11 demonstrates how the USER_SECONDARY_OBJECTS view may be used to

obtain information on the ResumeTextIndex that was created in Example 8-4.

8-13
 Chapter 8
Object Dependencies, Drop Semantics, and Validation

Example 8-11 Using *_SECONDARY_OBJECTS Views

SELECT SECONDARY_OBJECT_OWNER, SECONDARY_OBJECT_NAME
FROM USER_SECONDARY_OBJECTS
WHERE INDEX_OWNER = USER and INDEX_NAME = 'ResumeTextIndex'

8.5 Object Dependencies, Drop Semantics, and Validation

Consider issues that affect objects used in domain indexes.

8.5.1 Object Dependencies

The dependencies among various objects are as follows:
• Functions, Packages, and Object Types: referenced by operators and
indextypes
• Operators: referenced by indextypes, DML, and query SQL Statements
• Indextypes: referenced by domain indexes
• Domain Indexes: referenced (used implicitly) by DML and query SQL statements
Thus, the order in which these objects must be created, or their definitions exported for
future import, is:
1. Functions, packages, and object types
2. Operators
3. Indextypes

8.5.2 Object Drop Semantics

The drop behavior for an object is as follows:
• RESTRICT semantics: if there are any dependent objects the drop operation is
disallowed.
• FORCE semantics: the object is dropped even in the presence of dependent
objects; any dependent objects are recursively marked invalid.
Table 8-3 shows the default and explicit drop options supported for operators and
indextypes. The other schema objects are included for completeness and context.

Table 8-3 Default and Explicit Drop Options for Operators and Index Types

Schema Object Default Drop Behavior Explicit Options Supported

Function FORCE None
Package FORCE None
Object Types RESTRICT FORCE
Operator RESTRICT FORCE
Indextype RESTRICT FORCE

8-14
 Chapter 8
Indextype, Domain Index, and Operator Privileges

8.5.3 Object Validation

Invalid objects are automatically validated, if possible, the next time they are
referenced.

8.6 Indextype, Domain Index, and Operator Privileges

• To create an operator and its bindings, you must have EXECUTE privilege on the
function, operator, package, or the type referenced in addition to CREATE OPERATOR
or CREATE ANY OPERATOR privilege.
• To create an indextype, you must have EXECUTE privilege on the type that
implements the indextype in addition to CREATE INDEXTYPE or CREATE ANY
INDEXTYPE privilege. Also, you must have EXECUTE privileges on the operators that
the indextype supports.
• To alter an indextype in your own schema, you must have CREATE INDEXTYPE
system privilege.
• To alter an indextype or operator in another user's schema, you must have the
ALTER ANY INDEXTYPE or ALTER ANY OPERATOR system privilege.
• To create a domain index, you must have EXECUTE privilege on the indextype in
addition to CREATE INDEX or CREATE ANY INDEX privileges.
• To alter a domain index, you must have EXECUTE privilege on the indextype.
• To use the operators in queries or DML statements, you must have EXECUTE
privilege on the operator and the associated function, package, and indextype.
• To change the implementation type, you must have EXECUTE privilege on the new
implementation type.

8.7 Partitioned Domain Indexes

A domain index can be built to have discrete index partitions that correspond to the
partitions of a range-, list-, hash-, or interval-partitioned table. Such an index is called a
local domain index, as opposed to a global domain index, which has no index
partitions. Local domain index refers to a partitioned index as a whole, not to the
partitions that compose a local domain index.

Note:
You cannot convert a global domain index into a local domain index by using
the ALTER TABLE MODIFY PARTITION BY statement.

You also cannot use ALTER TABLE MODIFY PARTITION BY statement to

modify the partitioning scheme of a table with any domain index defined on it.

8-15
 Chapter 8
Partitioned Domain Indexes

8.7.1 Using Local Domain Index Methods

A local domain index is equipartitioned with the underlying table. That is, all keys in a
local domain index refer to rows stored in its corresponding table partition, none refer
to rows in other partitions. You can provide for local domain indexes in the indextype
by calling the CREATE INDEXTYPE statement, as demonstrated in Example 8-12.

This statement specifies that the implementation type TextIndexMethods is capable of

creating and maintaining local domain indexes.
Example 8-12 Using Local Domain Index Methods Within an Indextype
CREATE INDEXTYPE TextIndexType
FOR Contains (VARCHAR2, VARCHAR2)
USING TextIndexMethods
WITH LOCAL PARTITION
WITH SYSTEM MANAGED STORAGE TABLES;

8.7.2 About Partitioned Indexes

The CREATE INDEX statement creates and partitions the index, as demonstrated by
Example 8-13.
The LOCAL [PARTITION] clause indicates that the index is a local index on a
partitioned table. You can specify partition names or allow Oracle to generate them.
The PARALLEL clause specifies that the index partitions are to be created in parallel.
The ODCIIndexAlter() routines, which correspond to index partition create, rebuild, or
populate, are called in parallel.
In the PARAMETERS clause, specify the parameter string that is passed uninterpreted to
the appropriate ODCI indextype routine. The maximum length of the parameter string
is 1000 characters.

When you specify this clause at the top level of the syntax, the parameters become
the default parameters for the index partitions. If you specify this clause as part of the
LOCAL [PARTITION] clause, you override any default parameters with parameters for
the individual partition. The LOCAL [PARTITION] clause can specify multiple partitions.

Example 8-13 Creating and Partitioning an Index

CREATE INDEX [schema.]index
ON [schema.]table [t.alias] (indexed_column)
INDEXTYPE IS indextype
[LOCAL [PARTITION [partition [PARAMETERS ('string')]]] [...] ]
[PARALLEL parallel_degree]
[PARAMETERS ('string')];

8.7.3 Creating a Local Domain Index

When the domain index is created, Oracle invokes the appropriate ODCI routine. If the
routine does not return successfully, the domain index is marked FAILED. The only
operations supported on an failed domain index are DROP INDEX and (for non-local
indexes) REBUILD INDEX. Example 8-14 creates a local domain index ResumeIndex,
shown in Example 8-14.

8-16
 Chapter 8
Partitioned Domain Indexes

Example 8-14 Creating a Local Domain Index

CREATE INDEX ResumeIndex ON MyEmployees(Resume)
INDEXTYPE IS TextIndexType LOCAL;

8.7.4 Dropping a Local Domain Index

A specified index partition cannot be dropped explicitly. To drop a local index partition,
you must drop the entire local domain index:
Example 8-15 Dropping a Local Index Partition
DROP INDEX ResumeIndex;

8.7.5 Altering a Local Domain Index

The ALTER INDEXTYPE statement lets you change properties and the implementation
type of an indextype without having to drop and re-create the indextype, then rebuild
all dependent indexes.
Use the ALTER INDEX statement to perform the following operations on a local domain
index:
• Rename the top level index.
• Modify the default parameter string for all the index partitions.
• Modify the parameter string associated with a specific partition.
• Rename an index partition.
• Rebuild an index partition.

See Also:
olink:SQLRF-GUID-BFA7E29C-7905-4811-9119-B20FD8EA18F2 for
complete syntax of SQL statements.

8.7.6 Summary of Index States

Like a domain index, a partition of a local domain index can be in one or more of
several states, listed in Table 8-4.

Table 8-4 Summary of Index States

State Description
IN_PROGRESS The index or the index partition is in this state before and during
the execution of the ODCIIndex DDL interface routines. The state
is generally transitional and temporary. However, if the routine
ends prematurely, the index could remain marked IN_PROGRESS.
FAILED If the ODCIIndex interface routine doing DDL operations on the
index returns an error, the index or index partition is marked
FAILED.

8-17
 Chapter 8
Partitioned Domain Indexes

Table 8-4 (Cont.) Summary of Index States

State Description
UNUSABLE Same as for regular indexes: An index on a partitioned table is
marked UNUSABLE as a result of certain partition maintenance
operations. Note that, for partitioned indexes, UNUSABLE is
associated only with an index partition, not with the index as a
whole.
VALID An index is marked VALID if an object that the index directly or
indirectly depends upon is exists and is valid. This property is
associated only with an index, never with an index partition.
INVALID An index is marked INVALID if an object that the index directly or
indirectly depends upon is dropped or invalidated. This property is
associated only with an index, never with an index partition.

8.7.7 DML Operations with Local Domain Indexes

DML operations cannot be performed on the underlying table if an index partition of a
local domain index is in any of these states: IN_PROGRESS, FAILED, or UNUSABLE.
However, if the index is marked UNUSABLE, and SKIP_UNUSABLE_INDEXES = true, then
index maintenance is not performed.

8.7.8 Table Operations that Affect Indexes

The following tables list operations that may be performed on the underlying table of
an inde,x and describe the effect, if any, on the index. Table 8-5 lists TABLE
operations, while Table 8-6 lists ALTER TABLE operations.

Table 8-5 Summary of Table Operations

Table Operation Description

DROP table Drops the table. Drops all the indexes and their corresponding partitions
TRUNCATE table Truncates the table. Truncates all the indexes and the index partitions

Table 8-6 Summary of ALTER TABLE Operations with Partition Maintenance

ALTER TABLE Operation Description

Modify Partition Unusable Marks the local index partition associated with the table partition
local indexes as UNUSABLE
Modify Partition Rebuild Rebuilds the local index partitions that are marked UNUSABLE and
Unusable local indexes are associated with this table partition
Add Partition Adds a new table partition. Also adds a new local index partition.
Coalesce Partition Applicable to only hash partitioned tables. Drops a base table
partition. Also drops the associated local index partition.
Drop Partition Drops a base table partition. Also drops the associated local index
partition

8-18
 Chapter 8
Using System Partitioning

Table 8-6 (Cont.) Summary of ALTER TABLE Operations with Partition

Maintenance

ALTER TABLE Operation Description

Truncate Partition Truncate the table partition. Also truncates the associated local
index partition
Move Partition Moves the base table partition to another tablespace.
Corresponding local index partitions are marked UNUSABLE.
Split Partition Splits a table partition into two partitions. Corresponding local
index partition is also split. If the resulting partitions are non-
empty, the index partitions are marked UNUSABLE.
Merge Partition Merges two table partitions into one partition. Corresponding local
index partitions should also merge. If the resulting partition
contains data, the index partition is marked UNUSABLE.
Exchange Partition Exchanges a table partition with a non-partitioned table. Local
Excluding Indexes index partitions and global indexes are marked UNUSABLE.
Exchange Partition Exchanges a table partition with a non-partitioned table. Local
Including Indexes index partition is exchanged with global index on the non-
partitioned table. Index partitions remain USABLE.

8.7.9 ODCIIndex Interfaces for Partitioning Domain Indexes

To support local domain indexes, you must implement the standard ODCIIndex
methods, plus two additional methods that are specific to local domain indexes:
• ODCIIndexExchangePartition()
• ODCIIndexUpdPartMetadata()

8.7.10 Using SQL*Loader for Domain Indexes

SQL*Loader conventional path loads and direct path loads are supported for tables on
which domain indexes are defined, with this limitation: The table must be heap-
organized.
To do a direct path load on a domain index defined on an IOT or on a LOB column,
perform these tasks:
1. Drop the domain index
2. Do the direct path load in SQL*Loader.
3. Re-create the domain indexes.

8.8 Using System Partitioning

System Partitioning enables you to create a single table consisting of multiple physical
partitions. System partitioning does not use partitioning keys. Instead, it creates the
number of partitions specified. Therefore, the resulting partitions have no bounds
(range), values (list), or a partitioning method.

8-19
 Chapter 8
Using System Partitioning

Because there are no partitioning keys, you must explicitly map the distributed table
rows to the destination partition. When inserting a row, for example, you must use the
partition extended syntax to specify the partition to which a row must be mapped.

See Also:
Supporting SQL syntax in the Oracle Database SQL Language Reference

8.8.1 Advantages of System Partitioned Tables

The main advantages of system-partitioned tables is that it can be used to create and
maintain tables that are equipartitioned with respect to another table. For example, this
means that a dependent table could be created as a system-partitioned table, with the
same number of partitions as the base table. It follows that such a system-partitioned
table can be used to store index data for a domain index, with the following
implications:
• Pruning follows the base table pruning rules: when a partition is accessed in the
base table, the corresponding partition can be accessed in the system-partitioned
table.
• DDLs of the base table can be duplicated on the system-partitioned table.
Therefore, if a partition is dropped on the base table, the corresponding partition
on the system-partitioned table is dropped automatically.

8.8.2 Implementing System Partitioning

Consider how to implement system partitioning.

8.8.2.1 Creating a System-Partitioned Table

Example 8-16 describes how to create a system-partitioned table with four partitions.
Each partition can have different physical attributes.
Example 8-16 Creating System-Partitioned Tables
CREATE TABLE SystemPartitionedTable (c1 integer, c2 integer)
PARTITION BY SYSTEM
(
PARTITION p1 TABLESPACE tbs_1,
PARTITION p2 TABLESPACE tbs_2,
PARTITION p3 TABLESPACE tbs_3,
PARTITION p4 TABLESPACE tbs_4
);

8.8.2.2 Inserting Data into a System-Partitioned Table

Example 8-17 demonstrates how to insert data into a system-partitioned table. Both
INSERT and MERGE statements (not shown here) must use the partition extended syntax
to identify the partition to which the row should be added. The tuple (4,5) could have
been inserted into any of the four partitions created in Example 8-16.
DATAOBJ_TO_PARTITION can also be used, as demonstrated by Example 8-18.

8-20
 Chapter 8
Using System Partitioning

Starting with Oracle Database 12c, Oracle recommends using

DATAOBJ_TO_MAT_PARTITION, as demonstrated in Example 8-19, instead of the
DATAOBJ_TO_PARTITION function. The DATAOBJ_TO_MAT_PARTITION function supports
local domain indexes on interval partitioned tables.
Note that the first line of code shows how to insert data into a named partition, while
the second line of code shows that data can also be inserted into a partition based on
the partition's order. The support for bind variables, illustrated on the third code line, is
important because it allows cursor sharing between INSERT statements.

The DATAOBJ_TO_PARTITION function shown in Example 8-18 determines the absolute

partition number, given the physical partition identifier. However, if the base table is
interval partitioned, then there might be holes in the partition numbers corresponding
to unmaterialized partitions. Because the system partitioned table only has
materialized partitions, DATAOBJ_TO_PARTITION numbers can cause a mis-match
between the partitions of the base table and the partitions of the underlying system
partitioned index storage tables.
The new function, DATAOBJ_TO_MAT_PARTITION, shown in Example 8-19, returns the
materialized partition number (as opposed to the absolute partition number) and helps
keep the two tables in sync. Indextypes planning to support local domain indexes on
interval partitioned tables should migrate to the use of this function.
Example 8-17 Inserting Data into System-Partitioned Tables
INSERT INTO SystemPartitionedTable PARTITION (p1) VALUES (4,5);

Example 8-18 Inserting Data into System-Partitioned Tables;

DATAOBJ_TO_PARTITION
INSERT INTO SystemPartitionedTable PARTITION
(DATAOBJ_TO_PARTITION (base_table, :physical_partid))
VALUES (...);

Example 8-19 Inserting Data into System-Partitioned Tables;

DATAOBJ_TO_MAT_PARTITION
INSERT INTO SystemPartitionedTable PARTITION
(DATAOBJ_TO_MAT_PARTITION (base_table, :physical_partid))
VALUES (...);

8.8.2.3 Deleting and Updating Data in a System-Partitioned Table

While delete and update operations do not require the partition extended syntax,
Oracle recommends that you use it if at all possible. Because there is no partition
pruning, the entire table is scanned to execute the operation if the partition-extended
syntax is omitted. This highlights the fact that there is no implicit mapping between the
rows and the partitions.

8.8.3 Supporting Operations with System-Partitioned Tables

The following operations continue to be supported by system partitioning:
• Partition maintenance operations and other DDLs, with the exception of:
– ALTER INDEX SPLIT PARTITION
– ALTER TABLE SPLIT PARTITION

8-21
 Chapter 8
Using System Partitioning

– CREATE TABLE (as SELECT)

• Creation of local indexes, with the exception of unique local indexes because they
require a partitioning key
• Creation of local bitmapped indexes
• Creation of global indexes
• All DML operations
• INSERT AS SELECT operations with partition extended syntax, as shown in the
following code example:
Inserting Data into a Particular Partition of a Table
INSERT INTO TableName
PARTITION (
PartitionName|
DATAOBJ_TO_MAT_PARTITION(base_table, :physical_partid))
AS SubQuery
The following operations are no longer supported by system partitioning because
system partitioning does not use a partitioning method, and therefore does not
distribute rows to partitions.
• CREATE TABLE AS SELECT An alternative approach is to first create the table, and
then insert rows into each partition.
• INSERT INTO TableName AS SubQuery

8.8.4 Running Partition Maintenance Operations

As an example, consider an ALTER TABLE SPLIT PARTITION routine issued for the
base table of a domain index.
1. The system invokes the ODCIIndexUpdPartMetadata() method using the
information about the partition being added or dropped; remember that a 1:2 split
involves dropping of one partition and adding two new partitions.
2. The system invokes the ODCIStatsUpdPartStatistics() on the affected partitions.
3. The system drops the partition that has been split from all system-partition index
and statistics storage tables.
4. The system adds two new partitions to the system-partitioned tables.
5. If the partition that is being split is empty, then one call to ODCIIndexAlter()
rebuilds the split partition, and a second call to ODCIIndexAlter() rebuilds the
newly added partition.

8.8.5 Altering Table Exchange Partitions with Indexes

The ALTER TABLE EXCHANGE PARTITION command is allowed for tables with domain
indexes only under the following circumstances:
• a domain index is defined on both the non-partitioned table, and the partitioned
table
• both the non-partitioned table and the partitioned table have the same associated
indextype

8-22
 Chapter 8
Using System-Managed Domain Indexes

Note:
The ALTER TABLE EXCHANGE PARTITION CASCADE command is not allowed if
there is a local domain index on the reference partitioned table.

The ALTER TABLE EXCHANGE PARTITION routine invokes the following user-
implemented methods:
1. ODCIIndexExchangePartition() for the affected partition and index
2. ODCIStatsExchangePartition() for the affected partition and index if statistics are
collected for them

8.9 Using System-Managed Domain Indexes

Consider how system-managed domain indexes work, how to collect and store
statistics for them, and restrictions on their use.
Let us examine how system-managed domain indexes work.
Figure 8-1 illustrates the initial setup of a base table T1. T1 has the following elements:

• three partitions
• a local domain index on one of its columns, IT1
• a table of corresponding metadata objects, MT1, which is the optional metadata
table created by the indextype to store information specific to each partition of the
local domain index
• a system-partitioned table, SPT1, created by the indextype to store index data
The structures shown in these tables (table T1, index IT1 and the system partitioned
table SPT1) have the same number of partitions, in a one-to-one relationship. The
metadata table MT1 has as many rows as the number of partitions in these tables.

8-23
 Chapter 8
Using System-Managed Domain Indexes

Figure 8-1 Three-Partition Table with a Local Domain Index, and Associated
Structures

Server Tables
Base Table (T1) Local Index (IT1)

PartName PartNum PartName PartNum

P1 1 IP1 1

P2 2 IP2 2

P3 3 IP3 3

Index Tables
Metadata Table (MT1) System Partitioned Table (SPT1)

PartName PartId Metadata PartName PartNum

IP1 101 Params1 IP1 1

IP2 102 Params2 IP2 2

IP3 103 Params3 IP3 3

Figure 8-2 illustrates what happens to T1 and its related structures after splitting one of
its partitions with the operation in Example 8-20:
• the partition P2 in the base table T1 splits into P21 and P22
• in the local domain index, partition IP2 is dropped and two new partitions, IP21
and IP22, are created
• the indextype invokes the ODCIIndexUpdPartMetadata() method that makes the
necessary updates to the metadata table MT1
• in the system partitioned table SPT1, the partition that corresponds to partition IP2
is dropped and two new partitions are created
• index partitions are marked UNUSABLE as a result of the split operation; they must
be rebuilt to make them USABLE

8-24
 Chapter 8
Designing System-Managed Domain Indexes

Figure 8-2 A Three-Partition Table after ALTER TABLE SPLIT PARTITION

Server Tables
Base Table (T1) Local Index (IT1)

PartName PartNum PartName PartNum

P1 1 IP1 1

P21 2 IP21 2

P22 3 IP22 3

P3 4 IP3 4

Index Tables
Metadata Table (MT1) System Partitioned Table (SPT1)

PartName PartId Metadata PartName PartNum

IP1 101 Params1 IP1 1

IP21 1021 Params2 IP21 2

IP22 1022 Params2 IP22 3

IP3 103 Params3 IP3 4

Example 8-20 Splitting an Existing Table Partition

ALTER TABLE T1 SPLIT PARTITION P2 INTO P21, P22

8.10 Designing System-Managed Domain Indexes

When a top-level DDL that affects a non-partitioned domain index is called, the system
invokes user-implemented ODCIIndexXXX() and ODCIStatsXXX() methods. Table 8-7
shows these methods.
When a top-level DDL that affects a local system managed domain index is called, the
system invokes user-implemented ODCIIndexXXX() and ODCIStatsXXX() methods.
Table 8-8 shows these methods. In summary, the following rules apply:
• For ODCIIndexXXX () DMLs and queries, both the index partition object identifier
(ODCIIndexInfo.IndexPartitionIden) and a base table partition physical
identifier (ODIIndexInfo.IndexCols(1).TablePartitionIden) are required. For
ODCIIndexXXX () DDL routines, both the index partition object identifier and the
index partition name are supplied.

8-25
 Chapter 8
Designing System-Managed Domain Indexes

• The CREATE INDEX routine uses two calls to ODCIIndexCreate() (one at the
beginning and one at the end), and as many calls to ODCIIndexAlter() with
alter_option=AlterIndexRebuild as there are partitions.
• The TRUNCATE TABLE routine uses as many calls to ODCIIndexAlter() with
alter_option=AlterIndexRebuild as there are partitions.
• All partition maintenance operations invoke ODCIIndexUpdPartMetadata() so that
the indextype correctly updates its partition metadata table. The list of index
partitions is specified by the index partition name and the index partition object
identifier, and is supplied with information regarding addition or dropping of the
partition. No DDLs are allowed in these calls.With each partition maintenance
operation, the system implicitly transforms the system-partitioned storage tables
that were created using domain indexes. The names of the newly generated
partitions correspond to the index partition names.
• If the system-partitioned tables are used to store partition-level statistics, then the
tables and indexes created by ODCIStatsCollect() and dropped by
ODCIStatsDelete() are tracked by the system to ensure that they remain
equipartitioned.
• If the application implements user-defined partition-level statistics, the system
invokes ODCIStatsUpdPartStatistics() with every partition maintenance operation.
This ensure that the statistics type updates its partition-level statistics, and
(optionally) its aggregate global statistics. No DDLs are allowed in these calls. If
ODCIStatsUpdPartStatistics() is not implemented, the system does not raise an
exception but proceeds to the next programmatic step.

8.10.1 Methods for Non-Partitioned Domain Indexes

Table 8-7 ODCIXXX() Methods for Non-Partitioned Domain Indexes

DDL ODCIXXX() Method Used in System-Managed Approach

Specify the system-managed approach
CREATE INDEXTYPE

ODCIIndexCreate()
CREATE INDEX

ODCIIndexAlter(),
TRUNCATE TABLE
with the alter_option=AlterIndexRebuild
ODCIIndexAlter()
ALTER INDEX

ODCIStatsCollect()
GATHER_INDEX_STATS()

in DBMS_STATS
ODCIStatsDelete()
DELETE_INDEX_STATS()

in DBMS_STATS
ODCIIndexDrop() and ODCIStatsDelete()
DROP INDEX

(Force)

8-26
 Chapter 8
Designing System-Managed Domain Indexes

Table 8-7 (Cont.) ODCIXXX() Methods for Non-Partitioned Domain Indexes

DDL ODCIXXX() Method Used in System-Managed Approach

ODCIIndexInsert()
INSERT

ODCIIndexDelete()
DELETE

ODCIIndexUpdate()
UPDATE

ODCIIndexStart(), ODCIIndexFetch() and ODCIIndexClose()

QUERY

8.10.2 Methods for Local System-Managed Domain Indexes

Table 8-8 ODCIXXX() Methods for Local System-Managed Domain Indexes

DDL ODCIXXX() Method Used in System-Managed Approach

Specify the system-managed approach
CREATE INDEXTYPE

One call to ODCIIndexCreate(), one ODCIIndexAlter() call for

CREATE INDEX
each partition, with alter_option=AlterIndexRebuild,
and then a final call to ODCIIndexCreate()
One call for each partition: ODCIIndexAlter(), with
TRUNCATE TABLE
alter_option=AlterIndexRebuild
ODCIIndexAlter()
ALTER INDEX

One call to ODCIStatsCollect()

GATHER_INDEX_STATS()

in DBMS_STATS
One call to ODCIStatsDelete()
DELETE_INDEX_STATS()

in DBMS_STATS
ODCIIndexDrop(), and if user-defined statistics have been
DROP INDEX
collected then ODCIStatsDelete()
(Force)
ODCIIndexUpdPartMetadata(), ODCIIndexAlter() with
ALTER TABLE ADD PARTITION
alter_option=AlterIndexRebuild
ALTER TABLE COALESCE ODCIIndexUpdPartMetadata();
PARTITION ODCIStatsUpdPartStatistics() if statistics are collected
ODCIIndexUpdPartMetadata(); ODCIStatsUpdPartStatistics() if
ALTER TABLE DROP PARTITION
statistics are collected

8-27
 Chapter 8
Creating Local Domain Indexes

Table 8-8 (Cont.) ODCIXXX() Methods for Local System-Managed Domain

Indexes

DDL ODCIXXX() Method Used in System-Managed Approach

ODCIIndexUpdPartMetadata(); ODCIIndexAlter() with
ALTER TABLE TRUNCATE
PARTITION alter_option=AlterIndexRebuild;
ODCIStatsUpdPartStatistics() if a statistics type is associated
with the indextype and if user-defined statistics have been
collected
ODCIIndexUpdPartMetadata(); ODCIIndexAlter() with
ALTER TABLE SPLIT
PARTITION alter_option=AlterIndexRebuild only if the result
partitions are empty; ODCIStatsUpdPartStatistics() if a
statistics type is associated with the indextype and if user-
defined statistics have been collected
ODCIIndexUpdPartMetadata(); ODCIIndexAlter() with
ALTER TABLE MERGE
PARTITION alter_option=AlterIndexRebuild only if the result
partitions are empty; ODCIStatsUpdPartStatistics() if a
statistics type is associated with the indextype and if user-
defined statistics have been collected
ODCIIndexExchangePartition(); if a statistics type is associated
ALTER TABLE EXCHANGE
with the indextype, and if user-defined statistics have been
PARTITION
collected, also ODCIStatsExchangePartition()
ODCIIndexUpdPartMetadata() if a partitioned table has a valid
ALTER TABLE MOVE PARTITION
system-managed local domain index that has been updated as
part of a partition MOVE and rename operation. If a partition is
moved without updating the system-managed indexes, the
index partition is marked UNUSABLE.
One call to ODCIStatsCollect()
GATHER_TABLE_STATS()

in DBMS_STATS
One call to ODCIStatsDelete(), if a statistics type is associated
DELETE_TABLE_STATS()
with the indextype, and if user-defined statistics have been
collected
in DBMS_STATS
ODCIIndexAlter()
ALTER INDEX PARTITION

ODCIIndexInsert()
INSERT

ODCIIndexDelete()
DELETE

ODCIIndexUpdate()
UPDATE

ODCIIndexStart(), ODCIIndexFetch() and ODCIIndexClose()

QUERY

8.11 Creating Local Domain Indexes

The CREATE INDEX routine implements the following steps:

8-28
 Chapter 8
Creating Local Domain Indexes

1. To create system-partitioned storage tables, the system calls ODCIIndexCreate()

with index information. The number of partitions is supplied in the
ODCIIndexInfo.IndexPartitionTotal attribute. Note that all the partitioned
storage tables should be system-partitioned.
The object-level CREATE routine passes in only the object-level parameter string.
To construct the storage attributes for all partitions, the indextype needs partition-
level parameter strings. To obtain these, the cartridge must programmatically
query the XX_IND_PARTITIONS or XXX_IND_SUBPARTITIONS views on the dictionary
tables.
Oracle recommends that the indextype assign names to the storage tables and its
partitions using the index partition name. Note that you must also obtain index
partition names programmatically, from the XXX_IND_PARTITIONS or
XXX_IND_SUBPARTITIONS views.
2. For each partition, the system calls the ODCIIndexAlter() method with
alter_option=AlterIndexRebuild.
You can verify if this ODCIIndexAlter() call has been made as part of a CREATE
INDEX call by checking whether the ODICEnv.IntermediateCall bit was set.
Programatically select the index column values for each partition from the base
table partition, transform them, and store the transformed data in the
corresponding system-partitioned table.
During DML or query operations, if the indextype must refer to the metadata table,
it should be programmed to insert the index partition object identifier into the
corresponding row of the metadata table.
To store values in non-partitioned tables, you can program the cartridge either at
the level of the initial ODCIIndexCreate() call, or at the level of the
ODCIIndexAlter() calls made for each partition.
3. The system makes a final call to the ODCIIndexCreate() method so that the
indextype can create any necessary indexes on the storage tables.
The CREATE routine may use temporary storage tables for intermediate data. As an
option, you can programmatically instruct your application to use external files; it is
then the application's responsibility to manage these files.
After this ODCIIndexCreate() call completes, all partitioned tables created and not
dropped by this call are managed by the system.
Note that creation of global indexes of any type on a system-partitioned index storage
table is flagged as an error.

See Also:

• ODCIIndexAlter()
• ODCIIndexCreate()

8-29
 Chapter 8
Maintaining Local Domain Indexes with INSERT, DELETE, and UPDATE

8.12 Maintaining Local Domain Indexes with INSERT,

DELETE, and UPDATE
DML operations should be implemented in the following manner:
1. One of ODCIIndexInsert(), ODCIIndexDelete(), or ODCIIndexUpdate() is invoked.
Both the index partition object identifier (for accessing the metadata table) and the
base table partition physical identifier (for performing DMLs in the corresponding
partition) are supplied as part of the ODICIndexInfo structure.
2. To implement DMLs on a system-partitioned table, the cartridge code must include
the syntax in the following code example. The DATAOBJ_TO_MAT_PARTITION()
function is provided by the system.
Calling DML Operations on System-Partitioned Tables
INSERT INTO SP PARTITION
(DATAOBJ_TO_MAT_PARTITION(base_table, :physical_partid)) VALUES(...)

8.13 Querying Local Domain Indexes

Follow these steps to query local domain indexes:
1. When the optimizer receives a query that has a user-defined operator, if it
determines to use a domain index scan for evaluation, ODCIIndexStart(),
ODCIIndexFetch(), or ODCIIndexClose() is invoked.
2. The index partition object identifier and the base table partition physical identifier
are passed in as part of ODCIIndexInfo structure.
3. The index partition object identifier can be used to look up the metadata table, if
necessary.
4. And the base table physical partition identifier can be used to query the
corresponding partition of the system partitioned table.
5. The cartridge code must use the syntax as shown in the following code example
and the provided function DATAOBJ_TO_MAT_PARTITION(), for querying the system
partitioned table.
Querying a System-Partitioned Table
SELECT FROM SP PARTITION
(DATAOBJ_TO_MAT_PARTITION(base_table, :physical_partid)) WHERE <..>;

8.14 System Managed Domain Index - Supported Schemes

The system-managed domain indexing approach supports the following:
• Non-partitioned system-managed domain indexes.
• Local indexes on range-partitioned, list-partitioned, hash-partitioned, interval-
partitioned, auto-list-partitioned, and composite-partitioned tables.
• Local indexes on reference partitioned tables when the root table of the reference
partitioned table is range-partitioned, list-partitioned, hash-partitioned, interval

8-30
 Chapter 8
Restrictions of System-Managed Domain Indexing

partitioned, range-composite partitioned, list-composite partitioned, or hash-

composite partitioned.

8.15 Restrictions of System-Managed Domain Indexing

• A system-managed domain index can index only a single column.
• Local domain indexes cannot be created for REF-partitioned tables or IOTs.
• Local domain indexes on reference partitioned tables are not supported when the
root table of the reference partitioned table is interval-composite partitioned, or
interval-subpartitioned.
• Local domain indexes are not supported on interval-composite, interval-
subpartitioned, auto-list-composite, or auto-list-subpartitioned tables.
• A bitmap or unique domain index cannot be specified.

8.16 Migrating Non-Partitioned Indexes

The following steps show how to migrate non-partitioned user-managed domain
indexes into system-managed domain indexes.
1. Modify metadata: issue an ALTER INDEXTYPE command to register the property of
the indextype with the system. This disassociates the statistics type.
2. The index is marked as INVALID. You must implicitly issue an ALTER INDEX ...
COMPILE command to validate the index again. This calls the ODCIIndexAlter()
method with alter_option=AlterIndexMigrate.
3. Issue an ASSOCIATE STATISTICS command to associate a system-managed
statistics type with the system-managed indextype.

8.17 Migrating Local Partitioned Indexes

The following steps show how to migrate local partitioned user-managed domain
indexes into system-managed equi-partitioned domain indexes.
1. Modify metadata: issue an ALTER INDEXTYPE command to register the new index
routines and the property of the indextype so it can be managed by the system. All
indexes of this indextype are marked INVALID, and cannot be used until after the
completion of the next step. This disassociates the statistics type and erases the
old statistics.
2. Modify index data: invoke the ALTER INDEX ... COMPILE command for the new
indextype of each index. This calls the ODCIIndexAlter() method with
alter_option=AlterIndexMigrate. You must implement this method to transform
groups on non-partitioned tables into system-partitioned tables. For each set of n
tables that represent a partitioned table, the cartridge code should perform the
following actions. Note that the migration does not require re-generation of index
data, but involves only exchange operations.
• Create a system-partitioned table.
• For each of the n non-partitioned tables, call the ALTER TABLE EXCHANGE
PARTITION [INCLUDING INDEXES] routine to exchange a non-partitioned table
for a partition of the system-partitioned table.

8-31
 Chapter 8
Migrating Local Partitioned Indexes

• Drop all n non-partitioned tables.

3. Issue an ASSOCIATE STATISTICS command to associate a system-managed
statistics type with the system-managed indextype.

8-32
9
Defining Operators
You can define operators and use them with and without indextypes.

9.1 User-Defined Operators

A user-defined operator is a top-level schema object. In many ways, user-defined
operators act like the built-in operators such as <, >, and =; for instance, they can be
invoked in all the same situations. They contribute to ease of use by simplifying SQL
statements, making them shorter and more readable.
User-defined operators are:
• Identified by names, which are in the same namespace as tables, views, types,
and standalone functions
• Bound to functions, which define operator behavior in specified contexts
• Controlled by privileges, which indicate the circumstances in which each operator
can be used
• Often associated with indextypes, which can be used to define indexes that are
not built into the database

See Also:
Oracle Database SQL Language Reference for detailed information on
syntax and privileges

9.1.1 Operator Bindings

An operator binding associates the operator with the signature of a function that
implements the operator. A signature consists of a list of the data types of the
arguments of the function, in order of occurrence, and the function's return type.
Operator bindings tell Oracle which function to execute when the operator is invoked.
An operator can be bound to several functions if each function has a different
signature. To be considered different, functions must have different argument lists.
Functions whose argument lists match, but whose return data types do not match, are
not considered different and cannot be bound to the same operator.
Operators can be bound to:
• Standalone functions
• Package functions
• User-defined type member methods

9-1
 Chapter 9
User-Defined Operators

Operators can be bound to functions and methods in any accessible schema. Each
operator must have at least one binding when you create it. If you attempt to specify
non-unique operator bindings, the Oracle server raises an error.

9.1.2 Operator Privileges

To create an operator and its bindings, you must have:
• CREATE OPERATOR or CREATE ANY OPERATOR privilege
• EXECUTE privilege on the function, operator, package, or type referenced
To drop a user-defined operator, you must own the operator or have the DROP ANY
OPERATOR privilege.

To invoke a user-defined operator in an expression, you must own the operator or

have EXECUTE privilege on it.

9.1.3 Creating Operators

To create an operator, specify its name and its bindings with the CREATE OPERATOR
statement. Example 9-1 creates the operator Contains(), binding it to functions that
provide implementations in the Text and Spatial domains.
Example 9-1 Creating an Operator
CREATE OPERATOR Contains
BINDING
(VARCHAR2, VARCHAR2) RETURN NUMBER USING text.contains,
(Spatial.Geo, Spatial.Geo) RETURN NUMBER USING Spatial.contains;

9.1.4 Dropping Operators

To drop an operator and all its bindings, specify its name with the DROP OPERATOR
statement. Example 9-2 drops the operator Contains().

The default DROP behavior is DROP RESTRICT: if there are dependent indextypes or
ancillary operators for any of the operator bindings, then the DROP operation is
disallowed.
To override the default behavior, use the FORCE option. Example 9-3 drops the
operator and all its bindings and marks any dependent indextype objects and
dependent ancillary operators invalid.
Example 9-2 Dropping an Operator; RESTRICT Option
DROP OPERATOR Contains;

Example 9-3 Dropping an Operator; FORCE Option

DROP OPERATOR Contains FORCE;

9.1.5 Altering Operators

You can add bindings to or drop bindings from an existing operator with the ALTER
OPERATOR statement. Example 9-4 adds a binding to the operator Contains().

9-2
 Chapter 9
User-Defined Operators

Example 9-4 Adding a Binding to an Operator

ALTER OPERATOR Contains
ADD BINDING (music.artist, music.artist) RETURN NUMBER
USING music.contains;

9.1.5.1 Necessary Privileges for ALTER OPERATOR

To alter an operator, the operator must be in your own schema, or you must have the
ALTER ANY OPERATOR privilege. You must also have EXECUTE privileges on the
operators and functions referenced.

9.1.5.2 Restrictions of ALTER OPERATOR

The following restrictions apply to the ALTER OPERATOR statement:

• You can only issue ALTER OPERATOR statements that relate to existing operators.
• You can only add or drop one binding in each ALTER OPERATOR statement.
• You cannot drop an operator's only binding with ALTER OPERATOR; use the DROP
OPERATOR statement to drop the operator. An operator cannot exist without any
bindings.
• If you add a binding to an operator associated with an indextype, the binding is not
associated to the indextype unless you also issue the ALTER INDEXTYPE ADD
OPERATOR statement

9.1.6 Commenting Operators

To add comment text to an operator, specify the name and text with the COMMENT
statement. Example 9-5 supplies information about the Contains() operator:

Comments on operators are available in the data dictionary through these views:
• USER_OPERATOR_COMMENTS
• ALL_OPERATOR_COMMENTS
• DBA_OPERATOR_COMMENTS
You can only comment operators in your own schema unless you have the COMMENT
ANY OPERATOR privilege.

Example 9-5 Adding COMMENTs to an Operator

COMMENT ON OPERATOR
Contains IS 'a number that indicates if the text contains the key';

9.1.7 About Invoking Operators

Like built-in operators, user-defined operators can be invoked wherever expressions
can occur. For example, user-defined operators can be used in:
• The select list of a SELECT command.
• The condition of a WHERE clause.
• The ORDER BY and GROUP BY clauses.

9-3
 Chapter 9
Operators and Indextypes

When an operator is invoked, Oracle evaluates the operator by executing a function

bound to it. When several functions are bound to the operator, Oracle executes the
function whose argument data types match those of the invocation (after any implicit
type conversions). Invoking an operator with an argument list that does not match the
signature of any function bound to that operator causes an error to be raised. Because
user-defined operators can have multiple bindings, they can be used as overloaded
functions.
Assume that Example 9-6 creates the operator Contains().

If Contains() is used in Example 9-7, the operator invocation Contains(resume,

'Oracle') causes Oracle to execute the function text.contains(resume, 'Oracle')
because the signature of the function matches the data types of the operator
arguments. Similarly, the operator invocation Contains(location, :bay_area)
executes the function spatial.contains(location, :bay_area).

Executing the statement in Example 9-8 raises an error because none of the operator
bindings satisfy the argument data types.

9.1.7.1 Creating Contains() Operator

Example 9-6 Creating the Contains() Operator
CREATE OPERATOR Contains
BINDING
(VARCHAR2, VARCHAR2) RETURN NUMBER
USING text.contains,
(spatial.geo, spatial.geo) RETURN NUMBER
USING spatial.contains;

9.1.7.2 Using Contains() Operator in a Query

Example 9-7 Using the Operator Contains() in a Query
SELECT * FROM MyEmployees
WHERE Contains(resume, 'Oracle')=1 AND Contains(location, :bay_area)=1;

9.1.7.3 Using Contains() Operator Incorrectly

Example 9-8 An Incorrect Use of the Operator Contains()
SELECT * FROM MyEmployees
WHERE Contains(address, employee_addr_type('123 Main Street', 'Anytown', 'CA',
'90001'))=1;

9.2 Operators and Indextypes

Operators are often defined in connection with indextypes. After creating the operators
with their functional implementations, you can create an indextype that supports
evaluations of these operators using an index scan.
Operators that occur outside WHERE clauses are essentially stand-ins for the functions
that implement them; the meaning of such an operator is determined by its functional
implementation. Operators that occur in WHERE clauses are sometimes evaluated using
functional implementations; at other times they are evaluated by index scans.

9-4
 Chapter 9
Operators and Indextypes

9.2.1 Operators in the WHERE Clause

Operators appearing in the WHERE clause can be evaluated efficiently by an index scan
using the scan methods provided by the indextype. This process involves the following
steps.
1. Creating an indextype that supports the evaluation of the operator
2. Recognizing operator predicates of a certain form
3. Selecting a domain index
4. Setting up an appropriate index scan
5. Executing the index scan methods

9.2.1.1 Using Operator Predicates

An indextype supports efficient evaluation of operator predicates that can be
represented by a range of lower and upper bounds on the operator return values.
Specifically, predicates of the forms listed in Example 9-9 are candidates for index
scan-based evaluation.
Operator predicates that Oracle can convert internally into one of the forms in
Example 9-9 can also make use of the index scan-based evaluation.
Using the operators in expressions, such as op(...) + 2 = 3, precludes index scan-
based evaluation.
Predicates of the form op() is NULL are evaluated using the functional
implementation.
Example 9-9 Operator Predicates
op(...) LIKE value_expression
op(...) relop value_expression

where value_expression must evaluated to a constant (not a column) that can be

used as a domain index key, and relop is one of <, <=, =, >=, or >

9.2.1.2 Resolving Query Results with the Contains() Operator

An index scan-based evaluation of an operator is only possible if the operator applies
to a column or object attribute indexed by an indextype. The optimizer makes the final
decision between the indexed implementation and the functional implementation,
taking into account the selectivity and cost while generating the query execution plan.
Consider the query in Example 9-10.
The optimizer can choose to use a domain index in evaluating the Contains()
operator if
• The resume column has a defined index.
• The index is of type TextIndexType.
• TextIndexType supports the appropriate Contains() operator.
If any of these conditions do not hold, Oracle performs a complete scan of the
MyEmployees table and applies the functional implementation of Contains() as a post-

9-5
 Chapter 9
Operators and Indextypes

filter. However, if all these conditions are met, the optimizer uses selectivity and cost
functions to compare the cost of index-based evaluation with the full table scan and
generates the appropriate execution plan.
Consider a slightly different query in Example 9-11.
Here, you can access the MyEmployees table through an index on the id column, one
on the resume column, or a bitmap merge of the two. The optimizer estimates the costs
of the three plans and picks the least expensive variant one, which could be to use the
index on id and apply the Contains() operator on the resulting rows. In that case,
Oracle would use the functional implementation of Contains() rather than the domain
index.
Example 9-10 Using the Contains() Operator in a Simple Query
SELECT * FROM MyEmployees WHERE Contains(resume, 'Oracle') = 1;

Example 9-11 Using the Contains() Operator in a Complex Query

SELECT * FROM MyEmployees WHERE Contains(resume, 'Oracle') =1 AND id =100;

9.2.1.3 Setting Up an Index Scan

If a domain index is selected for the evaluation of an operator predicate, an index scan
is set up. The index scan is performed by the scan methods ODCIIndexStart(),
ODCIIndexFetch(), and ODCIIndexClose(), specified as part of the corresponding
indextype implementation. The ODCIIndexStart() method is invoked with the
operator-related information, including name and arguments and the lower and upper
bounds describing the predicate. After the ODCIIndexStart() call, a series of fetches
are performed to obtain row identifiers of rows satisfying the predicate, and finally the
ODCIIndexClose() is called when the SQL cursor is destroyed.

See Also:

• ODCIIndexClose()
• ODCIIndexFetch()
• ODCIIndexStart()

9.2.1.4 Execution Model for Index Scan Methods

To implement index scan routines, you must understand how they are invoked and
how multiple sets of invocations may be combined.
As an example, consider the query in Example 9-12.
If the optimizer chooses to use the domain indexes on the resume columns of both
tables, the indextype routines might be invoked in the sequence demonstrated in
Example 9-13.
In this example, a single indextype routine is invoked several times for different
instances of the Contains() operator. It is possible that many operators are being
evaluated concurrently through the same indextype routines. A routine that gets all the
information it needs through its parameters, such as the CREATE routine, does not

9-6
 Chapter 9
Operators and Indextypes

maintain any state across calls, so evaluating multiple operators concurrently is not a
problem. Other routines that must maintain state across calls, like the FETCH routine,
must know which row to return next. These routines should maintain state information
in the SELF parameter that is passed in to each call. The SELF parameter, an instance
of the implementation type, can be used to store either the entire state or a handle to
the cursor-duration memory that stores the state (if the state information is large).

9.2.1.5 Filtering Multiple Table Queries with Contains() Operator

Example 9-12 Using the Contains() Operator in a Multiple Table Query
SELECT * FROM MyEmployees1, MyEmployees2
WHERE
Contains(MyEmployees1.resume, 'Oracle') =1 AND
Contains(MyEmployees2.resume, 'UNIX') =1 AND
MyEmployees1.employee_id = MyEmployees2.employee_id;

9.2.1.6 Invoking Indextrype Routines for the Contains() Operator

Example 9-13 Invoking Indextype Routines for the Contains() Operator Query
start(ctx1, ...); /* corr. to Contains(MyEmployees1.resume, 'Oracle') */
start(ctx2, ...); /* corr. to Contains(MyEmployees2.resume, 'UNIX');
fetch(ctx1, ...);
fetch(ctx2, ...);
fetch(ctx1, ...);
...
close(ctx1);
close(ctx2);

9.2.2 Using Operators Outside the WHERE Clause

Operators that are used outside the WHERE clause are evaluated using the functional
implementation. To execute the statement in Example 9-14, Oracle scans the
MyEmployees table and invokes the functional implementation for Contains() on each
instance of resume, passing it the actual value of the resume, the text data, in the
current row. Note that this function would not make use of any domain indexes built on
the resume column.

Because functional implementations can make use of domain indexes, consider how
to write functions that use domain indexes and how they are invoked by the system.
Example 9-14 Using Operators Outside the WHERE Clause
SELECT Contains(resume, 'Oracle') FROM MyEmployees;

9.2.2.1 Creating Index-based Functional Implementations

For many domain-specific operators, such as Contains(), the functional
implementation has two options:
• If the operator is operating on a column or OBJECT attribute that has a domain
index, the function can evaluate the operator by looking at the index data rather
than the actual argument value.
For example, when Contains(resume, 'Oracle') is invoked on a particular row of
the MyEmployees table, it is easier for the function to look up the text domain index

9-7
 Chapter 9
Operators and Indextypes

defined on the resume column and evaluate the operator based on the row
identifier for the row containing the resume than to work on the resume text data
argument.
• If the operator is operating on a column that does not have an appropriate domain
index defined on it or if the operator is invoked with literal values (non-columns),
the functional implementation evaluates the operator based on the argument
values. This is the default behavior for all operator bindings.
To make your operator handle both options, provide a functional implementation that
has three arguments in addition to the original arguments to the operator:
• Index context: domain index information and the row identifier of the row on which
the operator is being evaluated
• Scan context: a context value to share state with subsequent invocations of the
same operator operating on other rows of the table
• Scan flag: indicates whether the current call is the last invocation during which all
cleanup operations should be performed
The function TextContains() in Example 9-15 provides the index-based functional
implementation for the Contains() operator.

The Contains() operator is bound to the functional implementation, as demonstrated

in Example 9-16.
The WITH INDEX CONTEXT clause specifies that the functional implementation can
make use of any applicable domain indexes. The SCAN CONTEXT specifies the data type
of the scan context argument, which must be identical to the implementation type of
the indextype that supports this operator.

9.2.2.2 Implementing the Contains() Operator in Index-Based Functions

Example 9-15 Implementing the Contains() Operator in Index-Based Functions
CREATE FUNCTION TextContains (Text IN VARCHAR2, Key IN VARCHAR2,
indexctx IN ODCIIndexCtx, scanctx IN OUT TextIndexMethods, scanflg IN NUMBER)
RETURN NUMBER AS
BEGIN
.......
END TextContains;

9.2.2.3 Binding the Contains() Operator to the Functional Implementation

Example 9-16 Binding the Contains() Operator to the Functional
Implementation
CREATE OPERATOR Contains
BINDING (VARCHAR2, VARCHAR2) RETURN NUMBER
WITH INDEX CONTEXT, SCAN CONTEXT TextIndexMethods
USING TextContains;

9.2.2.4 Operator Resolution

Oracle invokes the functional implementation for the operator if the operator appears
outside the WHERE clause. If the functional implementation is index-based, or defined to
use an indextype, the additional index information is passed in as arguments, but only

9-8
 Chapter 9
Operators and Indextypes

if the operator's first argument is a column or object attribute with a domain index of
the appropriate indextype.
For example, in the query SELECT Contains(resume, 'Oracle & Unix') FROM
MyEmployees, Oracle evaluates the operator Contains() using the index-based
functional implementation, passing it the index information about the domain index on
the resume column instead of the resume data.

9.2.2.5 Operator Execution

To execute the index-based functional implementation, Oracle sets up the arguments
in the following manner:
• The initial set of arguments is identical to those specified by the user for the
operator.
• If the first argument is not a column, the ODCIIndexCtx attributes are set to NULL.
• If the first argument is a column, the ODCIIndexCtx attributes are set up as follows.
– If there is an applicable domain index, the ODCIIndexInfo attribute contains
information about it; otherwise the attribute is set to NULL.
– The rowid attribute holds the row identifier of the row being operated on.
• The scan context is set to NULL on the first invocation of the operator. Because it is
an IN/OUT parameter, the return value from the first invocation is passed in to the
second invocation and so on.
• The scan flag is set to RegularCall for all normal invocations of the operator. After
the last invocation, the functional implementation is invoked one more time, at
which time any cleanup actions can be performed. During this call, the scan flag is
set to CleanupCall and all other arguments except the scan context are set to
NULL.
When index information is passed in, the implementation can compute the operator
value with a domain index lookup using the row identifier as key. The index metadata
is used to identify the index structures associated with the domain index. The scan
context is typically used to share state with the subsequent invocations of the same
operator.
If there is no indextype that supports the operator, or if there is no domain index on the
column passed to the operator as its first argument, then the index context argument is
null. However, the scan context argument is still available, Thus, the operator can
maintain state between invocations even if no index is used by the query.

9.2.3 Operators that Return Ancillary Data

In addition to filtering rows, operators in WHERE clauses sometimes must return ancillary
data. Ancillary data is modeled as one or more operators, each of which has
• A single literal number argument, which ties it to the corresponding primary
operator
• A functional implementation with access to state generated by the index scan-
based implementation of the primary operator
In the query in Example 9-17, the primary operator, Contains(), can be evaluated
using an index scan that determines which rows satisfy the predicate, and computes a
score value for each row. The functional implementation for the Score operator

9-9
 Chapter 9
Operators and Indextypes

accesses the state generated by the index scan to obtain the score for a given row
identified by its row identifier. The literal argument 1 associates the ancillary operator
Score to the primary operator Contains(), which generates the ancillary data.

The functional implementation of an ancillary operator can use either the domain index
or the state generated by the primary operator. When invoked, the functional
implementation is passed three extra arguments:
• The index context, which contains the domain index information
• The scan context, which provides access to the state generated by the primary
operator
• A scan flag to indicate whether the functional implementation is being invoked for
the last time
Consider how to define and invoke operators that modeling ancillary data.
Example 9-17 Accessing Ancillary Data with the Contains() Operator
SELECT Score(1) FROM MyEmployees
WHERE Contains(resume, 'OCI & UNIX', 1) =1;

9.2.3.1 Operator Bindings that Compute Ancillary Data

An operator binding that computes ancillary data is called a primary binding.
Example 9-18 defines a primary binding for the operator Contains().

This definition registers two bindings for Contains():

• CONTAINS(VARCHAR2, VARCHAR2), used when ancillary data is not required

• CONTAINS(VARCHAR2, VARCHAR2, NUMBER), used when ancillary data is required (the
NUMBER argument associates this binding with the ancillary operator binding)
The two bindings have a single functional implementation, as shown in Example 9-19:
Example 9-18 Comparing Ancillary Data with the Contains() Operator
CREATE OPERATOR Contains
BINDING (VARCHAR2, VARCHAR2) RETURN NUMBER
WITH INDEX CONTEXT, SCAN CONTEXT TextIndexMethods COMPUTE ANCILLARY DATA
USING TextContains;

Example 9-19 Implementing Bindings for Computations

TextContains(VARCHAR2, VARCHAR2, ODCIIndexCtx, TextIndexMethods, NUMBER).

9.2.3.2 Operator Bindings That Model Ancillary Data

An operator binding that models ancillary data is called an ancillary binding.
Functional implementations for ancillary data operators are similar to index-based
functional implementations. When you have defined the function, you bind it to the
operator with an additional ANCILLARY TO attribute, indicating that the functional
implementation must share its state with the primary operator binding.
Note that the functional implementation for the ancillary operator binding must have
the same signature as the functional implementation for the primary operator binding.
Example 9-20 demonstrates how to evaluate the ancillary operator inside a
TextScore() function.

9-10
 Chapter 9
Operators and Indextypes

Using the TextScore() definition, you could create an ancillary binding, as in

Example 9-21.
The ANCILLARY TO clause specifies that Score shares state with the primary operator
binding CONTAINS(VARCHAR2, VARCHAR2).

The ancillary operator binding is invoked with a single literal number argument, such
as Score(1), Score(2), and so on.

Example 9-20 Evaluating an Ancillary Operator

CREATE FUNCTION TextScore (Text IN VARCHAR2, Key IN VARCHAR2,
indexctx IN ODCIIndexCtx, scanctx IN OUT TextIndexMethods, scanflg IN NUMBER)
RETURN NUMBER AS
BEGIN
.......
END TextScore;

Example 9-21 Creating an Ancillary Operator Binding

CREATE OPERATOR Score
BINDING (NUMBER) RETURN NUMBER
ANCILLARY TO Contains(VARCHAR2, VARCHAR2)
USING TextScore;

9.2.3.3 Operator Resolution

The operators corresponding to ancillary data are invoked by the user with a single
number argument. This number argument must be a literal in both the ancillary
operation, and in the primary operator invocation, so that the operator association can
be done at query compilation time.
To determine the corresponding primary operator, Oracle matches the number passed
to the ancillary operator with the number passed as the last argument to the primary
operator. It is an error to find zero or more than one matching primary operator
invocation. After the matching primary operator invocation is found,
• The arguments to the primary operator become operands of the ancillary operator.
• The ancillary and primary operator executions are passed the same scan context.
For example, in the Example 9-17 query, the invocation of Score is determined to be
ancillary to Contains() based on the number argument 1, and the functional
implementation for Score gets the operands (resume, 'Oracle&Unix', indexctx,
scanctx, scanflg), where scanctx is shared with the invocation of Contains().

9.2.3.4 Operator Execution

Operator execution uses an index scan to process the Contains() operator. For each
of the rows returned by the fetch() call of the index scan, the functional
implementation of Score is invoked by passing to it the ODCIIndexCtx argument, which
contains the index information, row identifier, and a handle to the index scan state.
The functional implementation can use the handle to the index scan state to compute
the score.

9-11
10
Using Extensible Optimizer
You can use the Oracle Database extensible optimizer to optimize SQL statement
execution. Optimization concepts, statistics, selectivity, cost analysis, the ordering of
predicates, and the dependency model of the optimizer are described.

10.1 Overview of Query Optimization

Query Optimization is the process of choosing the most efficient way to execute a
SQL statement. The Extensible Indexing feature discussed in Defining Operators
introduces user-defined access methods.
The extensible optimizer feature allows authors of user-defined functions and indexes
to create statistics collection, selectivity, and cost functions that are used by the
optimizer in choosing a query plan. The optimizer cost model is extended to integrate
information supplied by the user to assess CPU and the I/O cost, where CPU cost is
the number of machine instructions used, and I/O cost is the number of data blocks
fetched.
Specifically, you can perform the following computations.
• Associate cost functions and default costs with domain indexes (partitioned or
non-partitioned), indextypes, packages, and standalone functions. The optimizer
can obtain the cost of scanning a single partition of a domain index, multiple
domain index partitions, or an entire index.
• Associate selectivity functions and default selectivity with methods of object types,
package functions, and standalone functions. The optimizer can estimate user-
defined selectivity for a single partition, multiple partitions, or the entire table
involved in a query.
• Associate statistics collection functions with domain indexes and columns of
tables. The optimizer can collect user-defined statistics at both the partition level
and the object level for a domain index or a table.
• Order predicates with functions based on cost.
• Select a user-defined access method (domain index) for a table based on access
cost.
• Use the DBMS_STATS package to invoke user-defined statistics collection and
deletion functions.
• Use new data dictionary views to include information about the statistics collection,
cost, or selectivity functions associated with columns, domain indexes, indextypes
or functions.
• Add a hint to preserve the order of evaluation for function predicates.
Please note that only the cost-based optimizer has been enhanced; Oracle has not
altered the operation of the rule-based optimizer.
The optimizer generates an execution plan for SQL queries and DML statements
SELECT, INSERT, UPDATE, or DELETE. For simplicity, we describe the generation of an

10-1
 Chapter 10
Overview of Query Optimization

execution plan in terms of a SELECT statement, but the process for DML statements is
similar.
An execution plan includes an access method for each table in the FROM clause, and
an ordering, called the join order, of the tables in the FROM clause. System-defined
access methods include indexes, hash clusters, and table scans. The optimizer
chooses a plan by generating a set of join orders, or permutations, by computing the
cost of each, and then by selecting the process with the lowest cost. For each table in
the join order, the optimizer computes the cost of each possible access method and
join method and chooses the one with the lowest cost. The cost of the join order is the
sum of the access method and join method costs. The costs are calculated using
algorithms that comprise the cost model. The cost model includes varying level of
detail about the physical environment in which the query is executed.
The optimizer uses statistics about the objects referenced in the query to compute the
selectivity and costs. The statistics are gathered using the DBMS_STATS package. The
selectivity of a predicate is the fraction of rows in a table that is chosen by the
predicate, and it is a number between 0 and 1.

The Extensible Indexing feature allows users to define new operators, indextypes, and
domain indexes. For user-defined operators and domain indexes, the Extensible
Optimizer feature enables you to control the three main components used by the
optimizer to select an execution plan statistics, selectivity, and cost.

See Also:

• Oracle Database Concepts for an introduction to optimization

• Oracle Database 2 Day + Performance Tuning Guide for information
about using hints in SQL statements
• Oracle Database PL/SQL Packages and Types Reference for
information about DBMS_STATS

10.1.1 Statistics
Statistics for tables and indexes can be generated by using the DBMS_STATS package.
In general, the more accurate the statistics, the better the execution plan generated by
the optimizer.

10.1.1.1 User-Defined Statistics

The Extensible Optimizer feature lets you define statistics collection functions for
domain indexes, indextypes, data types, individual table columns, and partitions. This
means that whenever a domain index is analyzed, a call is made to the user-specified
statistics collection function. The database does not know the representation and
meaning of the user-collected statistics.
In addition to domain indexes, Oracle supports user-defined statistics collection
functions for individual columns of a table, and for user-defined data types. In the
former case, whenever a column is analyzed, the user-defined statistics collection
function is called to collect statistics in addition to any standard statistics that the

10-2
 Chapter 10
Overview of Query Optimization

database collects. If a statistics collection function exists for a data type, it is called for
each column of the table being analyzed that has the required type.
The cost of evaluating a user-defined function depends on the algorithm and the
statistical properties of its arguments. It is not practical to store statistics for all
possible combinations of columns that could be used as arguments for all functions.
Therefore, Oracle maintains only statistics on individual columns. It is also possible
that function costs depend on the different statistical properties of each argument.
Every column could require statistics for every argument position of every applicable
function. Oracle does not support such a proliferation of statistics and cost functions
because it would decrease performance.
A user-defined function to drop statistics is required whenever there is a user-defined
statistics collection function.

10.1.1.2 User-Defined Statistics for Partitioned Objects

When using system-managed local domain indexes, you must implement two methods
of the ODCIStats interface: ODCIStatsExchangePartition(), and
ODCIStatsUpdPartStatistics().

10.1.2 Selectivity
The optimizer uses statistics to calculate the selectivity of predicates. The selectivity is
the fraction of rows in a table or partition that is chosen by the predicate. It is a number
between 0 and 1. The selectivity of a predicate is used to estimate the cost of a
particular access method; it is also used to determine the optimal join order. A poor
choice of join order by the optimizer could result in a very expensive execution plan.
Currently, the optimizer uses a standard algorithm to estimate the selectivity of
selection and join predicates. However, the algorithm does not always work well in
cases in which predicates contain functions or type methods. In addition, predicates
can contain user-defined operators about which the optimizer does not have any
information. In that case the optimizer cannot compute an accurate selectivity.

10.1.2.1 User-Defined Selectivity

For greater control over the optimizer's selectivity estimation, this feature lets you
specify user-defined selectivity functions for predicates containing user-defined
operators, standalone functions, package functions, or type methods. The user-
defined selectivity function is called by the optimizer whenever it encounters a
predicate with one of the forms shown in Example 10-1:
For such cases, users can define selectivity functions associated with operator(...).
The arguments to operator can be columns, constants, bind variables, or attribute
references. When optimizer encounters such a predicate, it calls the user-defined
selectivity function and passes the entire predicate as an argument (including the
operator, function, or type method and its arguments, the relational operator
relational_operator, and the constant expression or bind variable). The return value
of the user-defined selectivity function must be expressed as a percent, and be
between 0 and 100 inclusive; the optimizer ignores values outside this range.
Wherever possible, the optimizer uses user-defined selectivity values. However, this is
not possible in the following cases:

10-3
 Chapter 10
Overview of Query Optimization

• The user-defined selectivity function returns an invalid value (less than 0 or greater
than 100).
• There is no user-defined selectivity function defined for the operator, function, or
method in the predicate.
• The predicate does not have one of the forms listed in Example 10-1; it may also
be of the form operator(...) + 3 relational_operator constant.
In each of these cases, the optimizer uses heuristics to estimate the selectivity.
Example 10-1 Three Predicate Forms that Trigger a Call to the Optimizer
operator(...) relational_operator constant
constant relational_operator operator(...)
operator(...) LIKE constant

where
• operator(...) is a user-defined operator, standalone function, package function,
or type method,
• relational_operator is one of {<, <=, =, >=, >}, and
• constant is a constant value expression or bind variable.

10.1.3 Cost
The optimizer estimates the cost of various access paths to choose an optimal plan.
For example, it computes the CPU and I/O cost of using an index and a full table scan
to choose between the two. However the optimizer does not know the internal storage
structure of domain indexes, and so it cannot compute a good estimate of the cost of a
domain index.

10.1.3.1 User-Defined Cost

For greater flexibility, the cost model has been extended to let you define costs for
domain indexes, index partitions, and user-defined standalone functions, package
functions, and type methods. The user-defined costs can be in the form of default
costs that the optimizer looks up, or they can be full-fledged cost functions which the
optimizer calls to compute the cost.
Like user-defined selectivity statistics, user-defined cost statistics are optional. If no
user-defined cost is available, the optimizer uses heuristics to compute an estimate.
However, in the absence of sufficient useful information about the storage structures in
user-defined domain indexes and functions, such estimates can be very inaccurate
and result in the choice of a sub-optimal execution plan.
User-defined cost functions for domain indexes are called by the optimizer only if a
domain index is a valid access path for a user-defined operator. User-defined cost
functions for functions, methods and domain indexes are only called when a predicate
has one of the forms outlined in Example 10-1, which is identical to the conditions for
user-defined selectivity functions.
User-defined cost functions can return three cost values, each value representing the
cost of a single execution of a function or domain index implementation:
• CPU — the number of machine cycles executed by the function or domain index
implementation. This does not include the overhead of invoking the function.

10-4
 Chapter 10
Defining Statistics, Selectivity, and Cost Functions

• I/O — the number of data blocks read by the function or domain index
implementation. For a domain index, this does not include accesses to the Oracle
table. The multiblock I/O factor is not passed to the user-defined cost functions.
• NETWORK — the number of data blocks transmitted. This is valid for distributed
queries, functions, and domain index implementations. For Oracle, this cost
component is not used and is therefore ignored; however, the user is required to
stipulate a value so ensure backward compatibility.
The optimizer computes a composite cost from these cost values.
The package DBMS_ODCI contains a function estimate_cpu_units to help get the CPU
and I/O cost from input consisting of the elapsed time of a user function.
estimate_cpu_units measures CPU units by multiplying the elapsed time by the
processor speed of the machine and returns the approximate number of CPU
instructions associated with the user function. For a multiprocessor machine,
estimate_cpu_units considers the speed of a single processor.

The cost of a query is a function of the cost values. The settings of optimizer
initialization parameters determine which cost to minimize. If optimizer_mode is
first_rows, the resource cost of returning a single row is minimized, and the optimizer
mode is passed to user-defined cost functions. Otherwise, the resource cost of
returning all rows is minimized.

10.2 Defining Statistics, Selectivity, and Cost Functions

You can compute and store user-defined statistics for domain indexes and columns.
User-defined selectivity and cost functions for functions and domain indexes can use
both standard and user-defined statistics in their computation. The internal
representation of these statistics need not be known to Oracle, but you must provide
methods for their collection. You are solely responsible for defining the representation
of such statistics and for maintaining them. Note that user-collected statistics are used
only by user-defined selectivity and cost functions; the optimizer uses only its standard
statistics.
User-defined statistics collection, selectivity, and cost functions must be defined in a
user-defined type. Depending on the functionality you want it to support, this type must
implement as methods some or all of the functions defined in the system interface
ODCIStats, Oracle Data Cartridge Interface Statistics, in Extensible Optimizer
Interface.
Example 10-2 shows a type definition (or the outline of one) that implements all the
functions in the ODCIStats interface.

The object type that you define, referred to as a statistics type, need not implement
all the functions from ODCIStats. User-defined statistics collection, selectivity, and cost
functions are optional, so a statistics type may contain only a subset of the functions in
ODCIStats. Table 10-1 summarizes the type methods and default statistics associated
with different kinds of schema objects.

10-5
 Chapter 10
Defining Statistics, Selectivity, and Cost Functions

Table 10-1 Statistics Methods and Default Statistics for Various Schema
Objects

ASSOCIATE Statistics Type Methods Used Default

STATISTICS Statistics
column ODCIStatsCollect(), ODCIStatsDelete()
object type ODCIStatsCollect(), ODCIStatsDelete(), cost, selectivity
ODCIStatsFunctionCost(), ODCIStatsSelectivity()
function ODCIStatsFunctionCost(), ODCIStatsSelectivity() cost, selectivity
package ODCIStatsFunctionCost(), ODCIStatsSelectivity() cost, selectivity
index ODCIStatsCollect(), ODCIStatsDelete(), cost
ODCIStatsIndexCost()
indextype ODCIStatsCollect(), ODCIStatsDelete(), cost
ODCIStatsIndexCost(),
ODCIStatsUpdPartStatistics(),
ODCIStatsExchangePartition()

The types of the parameters of statistics type methods are system-defined ODCI data
types. These are described in Extensible Optimizer Interface.
The selectivity and cost functions must not change any database or package state.
Consequently, no SQL DDL or DML operations are permitted in the selectivity and
cost functions. If such operations are present, the functions are not called by the
optimizer.

10.2.1 Defining a Statistics Type

Example 10-2 Defining a Statistics Type
CREATE TYPE my_statistics AS OBJECT (

-- Function to get current interface

FUNCTION ODCIGetInterfaces(ifclist OUT ODCIObjectList) RETURN NUMBER,

-- User-defined statistics functions

FUNCTION ODCIStatsCollect(col ODCIColInfo, options ODCIStatsOptions,
statistics OUT RAW, env ODCIEnv) RETURN NUMBER,
FUNCTION ODCIStatsCollect(ia ODCIIndexInfo, options ODCIStatsOptions,
statistics OUT RAW, env ODCIEnv) RETURN NUMBER,
FUNCTION ODCIStatsDelete(col ODCIColInfo, statistics OUT RAW, env ODCIEnv)
RETURN NUMBER,
FUNCTION ODCIStatsDelete(ia ODCIIndexInfo, statistics OUT RAW, env ODCIEnv)
RETURN NUMBER,

-- User-defined statistics functions for local domain index

FUNCTION ODCIStatsUpdPartStatistics(ia ODCIIndexInfo, palistODCIPartInfoList,
env ODCIEnv) RETURN NUMBER;
FUNCTION ODCIStatsExchangePartition(ia ODCIIndexInfo, ia1 ODCIIndexInfo,
env ODCIEnv) RETURN NUMBER;

-- User-defined selectivity function

FUNCTION ODCIStatsSelectivity(pred ODCIPredInfo, sel OUT NUMBER, args
ODCIArgDescList, start <function_return_type>,
stop <function_return_type>, <list of function arguments>,
env ODCIEnv) RETURN NUMBER,

10-6
 Chapter 10
Defining Statistics, Selectivity, and Cost Functions

-- User-defined cost function for functions and type methods

FUNCTION ODCIStatsFunctionCost(func ODCIFuncInfo, cost OUT ODCICost,
args ODCIArgDescList, <list of function arguments>) RETURN NUMBER,

-- User-defined cost function for domain indexes

FUNCTION ODCIStatsIndexCost(ia ODCIIndexInfo, sel NUMBER,
cost OUT ODCICost, qi ODCIQueryInfo, pred ODCIPredInfo,
args ODCIArgDescList, start <operator_return_type>,
stop <operator_return_type>, <list of operator value arguments>,
env ODCIEnv) RETURN NUMBER
)

10.2.2 User-Defined Statistics Functions

There are two user-defined statistics collection functions, one for collecting statistics
and the other for deleting them.
The first, ODCIStatsCollect(), is used to collect user-defined statistics; its interface
depends on whether a column or domain index is being analyzed. It is called when
analyzing a column of a table or a domain index and takes two parameters:
• col for the column being analyzed, or ia for the domain index being analyzed;
• options for options specified in the DBMS_STATS package.
As mentioned, the database does not interpret statistics collected by
ODCIStatsCollect(). For system-managed domain index statistics, you don't return
the statistics collected by ODCIStatsCollect(). You should store these statistics in a
user-managed format, as described in Generating Statistics for System-Managed
Domain Indexes, and illustrated in Figure 10-1, Figure 10-2, and Figure 10-3.
User-collected statistics are deleted by calling the ODCIStatsDelete() function whose
interface depends on whether the statistics for a column or domain index are being
dropped. It takes a single parameter: col, for the column whose user-defined statistics
must be deleted, or ia, for the domain index whose statistics are to be deleted.

If a user-defined ODCIStatsCollect() function is present in a statistics type, the

corresponding ODCIStatsDelete() function must also be present.

The return values of the ODCIStatsCollect() and ODCIStatsDelete() functions must

be Success, Error, or Warning; these return values are defined in a system package
ODCIConst.

See Also:

• ODCIStatsCollect()
• ODCIStatsDelete()
•

10.2.3 User-Defined Selectivity Functions

User-defined selectivity functions are used only for predicate forms listed in
Example 10-1.

10-7
 Chapter 10
Defining Statistics, Selectivity, and Cost Functions

A user-defined selectivity function ODCIStatsSelectivity() takes five sets of input

parameters that describe the predicate:
• The pred parameter describes the function operator and the relational operator
relational_operator.
• The args parameter describes the start and stop values (that is, <constant>) of
the function and the actual arguments to the function (operator()).
• The start parameter, whose data type is identical to that of the function's return
value, describes the start value of the function.
• The stop parameter, whose data type is identical to that of the function's return
value, describes the stop value of the function.
• A list of function arguments whose number, position, and type must match the
arguments of the function operator.
The computed selectivity is returned in the output parameter sel as a number between
0 and 100 (inclusive) that represents a percentage. The optimizer ignores numbers
less than 0 or greater than 100 as invalid values. If the computed selectivity is less
than 0.5%, a value of 0 may be returned in the output parameter sel. A selectivity of 0
does not mean that the predicate will be removed.
The return value of the ODCIStatsSelectivity() function must be one of Success,
Error, or Warning.

As an example, consider a function myFunction, as defined in Example 10-3.

A user-defined selectivity function ODCIStatsSelectivity() is detailed in Extensible

Optimizer Interface.
If myFunction() is called using literal arguments, such as myFunction(2, 'TEST') >
5, then the selectivity function is called as out lined in Example 10-4.

If, on the other hand, myFunction() is called with some non-literals arguments, such
as myFunction(Test_tab.col_a, 'TEST')> 5, where col_a is a column in table
Test_tab, then the selectivity function is called as outlined in Example 10-5.

In summary, the start, stop, and function argument values are passed to the selectivity
function only if they are literals; otherwise they are NULL. ODCIArgDescList describes
all the arguments that follow it.
Example 10-3 Defining a User-Defined Function
myFunction (a NUMBER, b VARCHAR2(10)) return NUMBER

Example 10-4 Calling a Selectivity Function Using Literal Arguments

ODCIStatsSelectivity(ODCIPredInfo_constructor, sel,
ODCIArgDescList_constructor, 5, NULL, 2, 'TEST', ODCIEnv_flag)

Example 10-5 Calling a Selectivity Function Using Non-Literal Arguments

ODCIStatsSelectivity(ODCIPredInfo_constructor, sel,
ODCIArgDescList_constructor, 5, NULL, NULL, 'TEST', ODCIEnv_flag)

10-8
 Chapter 10
Defining Statistics, Selectivity, and Cost Functions

See Also:

• ODCIStatsSelectivity()
• ODCIArgDescList

10.2.4 User-Defined Cost Functions for Functions

User-defined cost functions are only used for predicate forms listed in Example 10-1.
You can define a function, ODCIStatsFunctionCost(), for computing the cost of
standalone functions, package functions, or type methods. This function takes three
sets of input parameters describing the predicate:
• The func parameter describes the function operator.
• The args parameter describes the actual arguments to the function operator.
• A list of function arguments whose number, position, and type must match the
arguments of the function operator.
The ODCIStatsFunctionCost() function returns its computed cost in the cost
parameter. The returned cost can have two components, a CPU cost and an I/O cost,
which are combined by the optimizer to compute a composite cost. The costs returned
by user-defined cost functions must be positive whole numbers. Invalid values are
ignored by the optimizer.
The return value of the ODCIStatsFunctionCost() function must be one of Success,
Error, or Warning.

Consider a myFunction(), defined in Example 10-3.

A user-defined cost function ODCIStatsFunctionCost() is detailed in Extensible

Optimizer Interface.
If myFunction() is called using literal arguments, such as myFunction(2, 'TEST') >
5, where col_a is a column in table Test_tab, then the cost function is called as out
lined in Example 10-6.
If, on the other hand, myFunction() is called with non-literal arguments, such as
myFunction(Test_tab.col_a, 'TEST') > 5, where col_a is a column in table
Test_tab, then the cost function is called as out lined in Example 10-7.

In summary, function argument values are passed to the cost function only if they are
literals; otherwise, they are NULL. ODCIArgDescList describes all the arguments that
follow it.
Example 10-6 Calling a Cost Function Using Literal Arguments
ODCIStatsFunctionCost(ODCIFuncInfo_constructor, cost,
ODCIArgDescList_constructor, 2, 'TEST', ODCIEnv_flag)

Example 10-7 Calling a Cost Function Using Non-Literal Arguments

ODCIStatsFunctionCost(ODCIFuncInfo_constructor, cost,
ODCIArgDescList_constructor, NULL, 'TEST', ODCIEnv_flag)

10-9
 Chapter 10
Defining Statistics, Selectivity, and Cost Functions

See Also:

• ODCIStatsFunctionCost()

10.2.5 User-Defined Cost Functions for Domain Indexes

User-defined cost functions for domain indexes are used for the same type of
predicates mentioned previously, except that operator must be a user-defined
operator for which a valid domain index access path exists.
The ODCIStatsIndexCost() function takes these sets of parameters:

• ia describing the domain index

• sel representing the user-computed selectivity of the predicate
• cost giving the computed cost
• qi containing additional information about the query
• pred describing the predicate
• args describing the start and stop values (that is, <constant>) of the operator and
the actual arguments to the operator operator
• start, whose data type is identical to that of the operator's return value, describing
the start value of the operator
• stop whose data type is identical to that of the operator's return value, describing
the stop value of the operator
• a list of operator value arguments whose number, position, and type must match
the arguments of the operator operator. The value arguments of an operator are
the arguments excluding the first argument.
• env, an environment flag set by the server to indicate which call is being made in
cases where multiple calls are made to the same routine. The flag is reserved for
future use; currently it is always set to 0.
The computed cost of the domain index is returned in the output parameter, cost.

ODCIStatsIndexCost() returns Success, Error or Warning.

Consider an operator defined in Example 10-8, which returns 1 or 0 depending on

whether or not the string b_string is contained in the string a_string. Further,
assume that the operator is implemented by a domain index.
A user-defined index cost function ODCIStatsIndexCost() is detailed in Extensible
Optimizer Interface.
If contains() is called using non-literal arguments, such as
Contains(Test_tab.col_c,'TEST') <= 1, then the index cost function is called as out
lined in Example 10-9.
Note that the first argument, a_string, of Contains does not appear as a parameter of
ODCIStatsIndexCost(). This is because the first argument to an operator must be a
column for the domain index to be used, and this column information is passed in

10-10
 Chapter 10
Defining Statistics, Selectivity, and Cost Functions

through the ODCIIndexInfo parameter. Only the operator arguments after the first (the
value arguments) must appear as parameters to the ODCIStatsIndexCost() function.

In summary, the start, stop, and operator argument values are passed to the index
cost function only if they are literals; otherwise they are NULL. ODCIArgDescList
describes all the arguments that follow it.
Example 10-8 Defining an Operator
Contains(a_string VARCHAR2(2000), b_string VARCHAR2(10))

Example 10-9 Calling an Index Cost Function Using Non-Literal Arguments

ODCIStatsIndexCost(ODCIIndexInfo_constructor, sel, cost,
ODCIQueryInfo_constructor, ODCIPredInfo_constructor,
ODCIArgDescList_constructor, NULL, 1, 'TEST', ODCIEnv_flag)

See Also:
ODCIStatsIndexCost()

10.2.6 Generating Statistics for System-Managed Domain Indexes

If you choose the system-managed approach to maintain domain indexes and must
associate a statistics type with the domain index or the indextype, then the statistics
type must also be managed by the system.
Statistics may be collected when issuing an ODCIStatsCollect() call for a system-
managed domain index. For a non-partitioned index, the statistics may be stored with
the index storage table, as a separate table, or in a data cartridge metadata table with
index name qualified rows.
For local partitioned domain indexes, there are three options for storing statistics. All
use the ODCIStatsUpdPartStatistics() method during a partition maintenance
operation in the following ways. Please note that in all the following examples, no
DDLs are executed inside the ODCIStatsUpdPartStatistics() call, and only DML and
query instructions are allowed in the implementation of
ODCIStatsUpdPartStatistics().

See Also:

• ODCIStatsCollect()
• ODCIStatsUpdPartStatistics()

10.2.6.1 Index-Partition Statistics Storage in an Index Table

The system calls the ODCIStatsUpdPartStatistics() method If the statistics are stored
with the indexed data in the index storage (system-partitioned) tables, as illustrated in
Figure 10-1. The method can optionally maintain any statistics-related partition

10-11
 Chapter 10
Defining Statistics, Selectivity, and Cost Functions

metadata, or be a null operation. The server deletes or drops the statistics for the
affected partitions along with the index data specific to these partitions.

Figure 10-1 Storing Index-Specific Statistics with Index Tables

Partition IndexData Statistics

(value) (frequency)

P1 25 10

35 5
P2 57 22
76 10
99 5
P3 120 15

150 5

10.2.6.2 Index-Partition Statistics Storage in a Separate Table

If the statistics are stored in separate system-partitioned tables, as illustrated in
Figure 10-2, the server tracks the creation of these system partitioned tables of store
statistics during an ODCIStatsCollect() call. These tables are maintained by the server
in the same manner as for index storage tables.

Figure 10-2 Storing Index-Specific Statistics in a Separate Table

Partition Statistics

P1 StatsP1_1

StatsP1_2
P2 StatsP2_1
StatsP2_2
StatsP2_3
P3 StatsP3_1

StatsP3_2

10.2.6.3 Index-Partition Statistics Storage in a Common Table

If the statistics are stored in a non-partitioned table as either schema-name, index-
name, or partition-name qualified rows, as illustrated in Figure 10-3, then you have to
maintain the partition-level statistics with a call to ODCIStatsUpdPartStatistics(). The
server does not perform any operation on these tables.

10-12
 Chapter 10
Using User-Defined Statistics, Selectivity, and Cost

Figure 10-3 Storing Index-Partition Statistics in a Common Table

Schema Index IndexPartition UserdefinedStatistics

U1 I1 IP1 Statistics1

U1 I1 IP2 Statistics2

U2 I2 IP1 Statistics3

U2 I2 IP2 Statistics4

U2 I2 IP3 Statistics5

10.3 Using User-Defined Statistics, Selectivity, and Cost

Statistics types act as interfaces for user-defined functions that influence the choice of
an execution plan by the optimizer. However, for the optimizer to be able to use a
statistics type, it requires a mechanism to bind the statistics type to a database object
such as a column, a standalone function, an object type, an index, an indextype or a
package. You cannot associate a statistics type with a partition of a table or a partition
of a domain index. The ASSOCIATE STATISTICS command creates this association.

10.3.1 User-Defined Statistics

User-defined statistics functions are relevant for columns that use both standard SQL
data types and object types, and for domain indexes. The functions
ODCIStatsSelectivity(), ODCIStatsFunctionCost(), and ODCIStatsIndexCost() are not
used for user-defined statistics, so statistics types used only to collect user-defined
statistics need not implement these functions.
Users could create their own tables. This approach requires that privileges on these
tables be administered properly, backup and restoration of these tables be done along
with other dictionary tables, and point-in-time recovery considerations be resolved.

10.3.1.1 Column Statistics

Consider a table Test_tab, defined as in Example 10-10, where typ1 is an object type.

Suppose that stat is a statistics type that implements ODCIStatsCollect() and

ODCIStatsDelete() functions.User-defined statistics are collected by the DBMS_STATS
package for the column col_b if we bind a statistics type with the column, as
demonstrated in Example 10-11:
A list of columns can be associated with the statistics type stat. Note that Oracle
supports only associations with top-level columns, not attributes of object types; if you
wish, the ODCIStatsCollect() function can collect individual attribute statistics by
traversing the column.
Another way to collect user-defined statistics is to declare an association with a data
type, as in Example 10-12, which declares stat_typ1 as the statistics type for the type
typ1. When the table Test_tab is analyzed with this association, user-defined
statistics are collected for the column col_b using the ODCIStatsCollect() function of
statistics type stat_typ1.

10-13
 Chapter 10
Using User-Defined Statistics, Selectivity, and Cost

Individual column associations always have precedence over associations with types.
Thus, in the preceding example, if both ASSOCIATE STATISTICS commands are issued,
DBMS_STATS would use the statistics type stat (and not stat_typ1) to collect user-
defined statistics for column col_b. It is also important to note that standard statistics,
if possible, are collected along with user-defined statistics.
User-defined statistics are deleted using the ODCIStatsDelete() function from the
same statistics type that was used to collect the statistics.
Associations defined by the ASSOCIATE STATISTICS command are stored in a
dictionary table called ASSOCIATION$.

Only user-defined data types can have statistics types associated with them; you
cannot declare associations for standard SQL data types.
Example 10-10 Creating a Table with an Object Type Column
CREATE TABLE Test_tab (
col_a NUMBER,
col_b typ1,
col_c VARCHAR2(2000)
)

Example 10-11 Associating Statistics with Columns for User-Defined Statistics

ASSOCIATE STATISTICS WITH COLUMNS Test_tab.col_b USING stat

Example 10-12 Associating Statistics with Data Types for User-Defined

Statistics
ASSOCIATE STATISTICS WITH TYPES typ1 USING stat_typ1

10.3.1.2 Implementing Domain Index Statistics

A domain index has an indextype. A statistics type for a system-managed domain
index is defined by associating it only with its indextype. Example 10-13 demonstrates
how to create an indextype, an index, and an operator on the table Test_tab from
Example 10-10:
Here, indtype is the indextype, userOp is a user-defined operator supported by
indtype, userOp_func is the functional implementation of userOp, and imptype is the
implementation type of the indextype indtype.

A statistics type stat_indtype can be associated with the system-managed indextype,

as demonstrated in Example 10-14. When the domain index Test_indx that has an
indextype indtype is analyzed, user-defined statistics for the index are collected by
calling the ODCIStatsCollect() function of stat_indtype.

To drop index statistics, use the ODCIStatsDelete() method which is defined for the
same statistics type that defined the earlier ODCIStatsCollect() method.
Example 10-13 Creating an Indextype, an Index and an Operator for User-
Defined Statistics
CREATE INDEXTYPE indtype
FOR userOp(NUMBER)
USING imptype WITH SYSTEM MANAGED STORAGE TABLES;

CREATE INDEX Test_indx ON Test_tab(col_a)

INDEXTYPE IS indtype PARAMETERS('example');

10-14
 Chapter 10
Using User-Defined Statistics, Selectivity, and Cost

CREATE OPERATOR userOp BINDING (NUMBER) RETURN NUMBER

USING userOp_func;

Example 10-14 Associating Statistics with System-Managed Indextypes

ASSOCIATE STATISTICS WITH INDEXTYPES indtype USING stat_indtype
WITH SYSTEM MANAGED STORAGE TABLES

10.3.2 User-Defined Selectivity

The optimizer uses selectivity functions to compute the selectivity of predicates in a
query. The predicates must have one of the appropriate forms and can contain user-
defined operators, standalone functions, package functions, or type methods.

10.3.2.1 User-Defined Operators

Suppose that the association in Example 10-15 is declared. If the optimizer encounters
the userOp(Test_tab.col_a) = 1 predicate, it calls the ODCIStatsSelectivity()
function (if present) in the statistics type stat_userOp_func that is associated with the
functional implementation of the userOp_func of the userOp operator.

Example 10-15 Associating Statistics with User-Defined Operators

ASSOCIATE STATISTICS WITH FUNCTIONS userOp_func USING stat_userOp_func

10.3.2.2 Standalone Functions

If the association in Example 10-16 is declared for a standalone function myFunction,
then the optimizer calls the ODCIStatsSelectivity() function (if present) in the statistics
type stat_myFunction for the myFunction(Test_tab.col_a, 'TEST') = 1 predicate.

Example 10-16 Associating Statistics with Standalone Functions

ASSOCIATE STATISTICS WITH FUNCTIONS myFunction USING stat_MyFunction

10.3.2.3 Package Functions

If the association in Example 10-17 is declared for a package Demo_pack, then the
optimizer calls the ODCIStatsSelectivity() function (if present) in the statistics type
stat_Demo_pack for the Demo_pack.myDemoPackFunction(Test_tab.col_a, 'TEST')
= 1 predicate, where myDemoPackFunction is a function in Demo_pack.

Example 10-17 Associating Statistics with Package Functions

ASSOCIATE STATISTICS WITH PACKAGES Demo_pack USING stat_Demo_pack

10.3.2.4 Type Methods

If the association in Example 10-18 is declared for a type Example_typ, then the
optimizer calls the ODCIStatsSelectivity() function (if present) in the statistics type
stat_Example_typ for the myExampleTypMethod(Test_tab.col_b) = 1 predicate,
where myExampleTypMethod is a method in Example_typ.

Example 10-18 Associating Statistics with Type Methods

ASSOCIATE STATISTICS WITH TYPES Example_typ USING stat_Example_typ

10-15
 Chapter 10
Using User-Defined Statistics, Selectivity, and Cost

10.3.2.5 Default Selectivity

An alternative to selectivity functions is user-defined default selectivity. The default
selectivity is a value between 0 and 100%; the optimizer looks it up instead of calling a
selectivity function. Default selectivities can be used for predicates with user-defined
operators, standalone functions, package functions, or type methods.
The association in Example 10-19 declares that the myFunction(Test_tab.col_a) =
1 predicate always has a selectivity of 20% (or 0.2), regardless of the parameters of
myFunction, the comparison operator =, or the constant 1. The optimizer uses this
default selectivity instead of calling a selectivity function.
An association can be declared using either a statistics type or a default selectivity, but
not both. Thus, the following statement is illegal:
ASSOCIATE STATISTICS WITH FUNCTIONS myFunction USING stat_myFunction
DEFAULT SELECTIVITY 20

Other examples of default selectivity declarations include:

ASSOCIATE STATISTICS WITH PACKAGES Demo_pack DEFAULT SELECTIVITY 20
ASSOCIATE STATISTICS WITH TYPES Example_typ DEFAULT SELECTIVITY 20

Example 10-19 Associating Statistics with Default Selectivity

ASSOCIATE STATISTICS WITH FUNCTIONS myFunction DEFAULT SELECTIVITY 20

10.3.3 User-Defined Cost

The optimizer uses user-defined cost functions to compute the cost of predicates in a
query. The predicates must have one of the forms listed earlier and can contain user-
defined operators, standalone functions, package functions, or type methods. In
addition, user-defined cost functions are also used to compute the cost of domain
indexes.

10.3.3.1 User-Defined Operators

If the association in Example 10-20 is declared, consider the userOp(Test_tab.col_a)
= 1 predicate. If the optimizer evaluates the domain index Test_indx with an indtype
indextype that implements userOp, it calls the ODCIStatsIndexCost() method (if
present) in the statistics type stat_indtype. If the domain index is not used, however,
the optimizer calls the ODCIStatsFunctionCost() (if present) in the statistics type
stat_userOp to compute the cost of the functional implementation of the operator
userOp.

Example 10-20 Associating Statistics with User-Defined Operators

ASSOCIATE STATISTICS WITH INDEXTYPES indtype USING stat_indtype
WITH SYSTEM MANAGED STORAGE TABLES
ASSOCIATE STATISTICS WITH FUNCTIONS userOp USING stat_userOp_func

10.3.3.2 Standalone Functions

If the association in Example 10-21 is declared for a standalone function myFunction,
then the optimizer calls the ODCIStatsFunctionCost() function (if present) in the

10-16
 Chapter 10
Using User-Defined Statistics, Selectivity, and Cost

statistics type stat_myFunction for the myFunction(Test_tab.col_a, 'TEST') = 1

predicate.
User-defined function costs do not influence the choice of access methods; they are
only used for ordering predicates, described in Extensible Optimizer Interface.
Example 10-21 Associating Statistics with Standalone Functions
ASSOCIATE STATISTICS WITH FUNCTIONS myFunction USING stat_myFunction;

10.3.3.3 Package Functions

If the association in Example 10-22 is declared for a package Demo_pack, then the
optimizer calls the ODCIStatsFunctionCost() function, if present, in the statistics type
stat_Demo_pack for the Demo_pack.myDemoPackFunction(Test_tab.col_a) = 1
predicate, where myDemoPackFunction is a function in Demo_pack.

Example 10-22 Associating Statistics with Package Functions

ASSOCIATE STATISTICS WITH PACKAGES Demo_pack USING stat_Demo_pack;

10.3.3.4 Type Methods

If the association is declared, as in Example 10-23, for a type Example_typ, then the
optimizer calls the ODCIStatsFunctionCost() function, if present, in the statistics type
stat_Example_typ for the myExampleTypMethod(Test_tab.col_b) = 1 predicate,
where myExampleTypMethod is a method in Example_typ.

Example 10-23 Associating Statistics with Type Methods

ASSOCIATE STATISTICS WITH TYPES Example_typ USING stat_Example_typ;

10.3.3.5 Default Cost

Like default selectivity, default costs can be used for predicates with user-defined
operators, standalone functions, package functions, or type methods. The command in
Example 10-24 declares that using the domain index Test_indx to implement the
userOp(Test_tab.col_a) = 1 predicate always has a CPU cost of 100, an I/O cost of
5, and a network cost of 0 (the network cost is ignored in Oracle), regardless of the
parameters of userOp, the comparison operator "=", or the constant "1". The optimizer
uses this default cost instead of calling the ODCIStatsIndexCost() function.
You can declare an association using either a statistics type or a default cost but not
both. Thus, the following statement is illegal:
ASSOCIATE STATISTICS WITH INDEXES Test_indx USING stat_Test_indx
DEFAULT COST (100, 5, 0)

The following are some more examples of default cost declarations:

ASSOCIATE STATISTICS WITH FUNCTIONS myFunction DEFAULT COST (100, 5, 0)
ASSOCIATE STATISTICS WITH PACKAGES Demo_pack DEFAULT COST (100, 5, 0)
ASSOCIATE STATISTICS WITH TYPES Example_typ DEFAULT COST (100, 5, 0)
ASSOCIATE STATISTICS WITH INDEXTYPES indtype DEFAULT COST (100, 5, 0)

Example 10-24 Associating Statistics with Default Cost

ASSOCIATE STATISTICS WITH INDEXES Test_indx DEFAULT COST (100, 5, 0);

10-17
 Chapter 10
Using User-Defined Statistics, Selectivity, and Cost

10.3.4 Declaring a NULL Association for an Index or Column

An association of a statistics type defined for an indextype or object type is inherited
by index instances of that indextype and by columns of that object type. An inherited
association can be overridden by explicitly defining a different association for an index
instance or column, but there may be occasions when you would prefer an index or
column not to have any association at all. For example, for a particular query the
benefit of a better plan may not outweigh the additional compilation time incurred by
invoking the cost or selectivity functions. For cases like this, you can use the
ASSOCIATE command to declare a NULL association for a column or index, as in
Example 10-25.
If the NULL association is specified, the schema object does not inherit any statistics
type from the column type or the indextype. A NULL association also precludes default
values.
Example 10-25 Declaring NULL Statistics Associations for Columns and
Indexes
ASSOCIATE STATISTICS WITH COLUMNS columns NULL;
ASSOCIATE STATISTICS WITH INDEXES indexes NULL;

10.3.5 How DDL Operations Affect Statistics

Partition-level and schema object-level aggregate statistics are affected by DDL
operations in the same way as standard statistics. Table 10-2 summarizes the effects.

Table 10-2 Effects of DDL on Partition and Global Statistics

Operation Effect on Partition Effect on Global Statistics

Statistics
None No Action
ADD PARTITION

COALESCE PARTITION Statistics deleted Statistics recalculated (if

_minimal_stats_aggregation
is FALSE, otherwise, no effect)
Statistics deleted Statistics recalculated (if
DROP PARTITION
_minimal_stats_aggregation
is FALSE, otherwise no effect)
Statistics deleted None
SPLIT PARTITION

Statistics deleted None

MERGE PARTITION

Statistics deleted None

TRUNCATE PARTITION

Statistics deleted Statistics recalculated (if

EXCHANGE PARTITION
_minimal_stats_aggregation
is FALSE, otherwise no effect)
None None
REBUILD PARTITION

10-18
 Chapter 10
Predicate Ordering

Table 10-2 (Cont.) Effects of DDL on Partition and Global Statistics

Operation Effect on Partition Effect on Global Statistics

Statistics
None None
MOVE PARTITION

None None
RENAME PARTITION

If an existing partition is exchanged, or dropped with an ALTER TABLE DROP PARTITION

statement, and the _minimal_stats_aggregation parameter is set to FALSE, the
statistics for that partition are deleted, and the aggregate statistics of the table or index
are recalculated.

10.4 Predicate Ordering

In the absence of an ORDERED_PREDICATES hint, predicates (except those used for
index keys) are evaluated in the order specified by the following rules:
• Predicates without any user-defined functions, type methods, or subqueries are
evaluated first, in the order specified in the WHERE clause.
• Predicates with user-defined functions and type methods which have user-
computed costs are evaluated in increasing order of their cost.
• Predicates with user-defined functions and type methods that have no user-
computed cost are evaluated next, in the order specified in the WHERE clause.
• Predicates not specified in the WHERE clause (for example, predicates transitively
generated by the optimizer) are evaluated next.
• Predicates with subqueries are evaluated last in the order specified in the WHERE
clause.

10.5 Dependency Model

The dependency model reflects the actions that are taken when you issue any of the
SQL commands described in Table 10-3.

Table 10-3 Dependency Model for DDLs

Command Action

DROP statistics_type If an association is defined with statistics_type, the

command fails, otherwise the type is dropped.

DROP statistics_type FORCE Calls DISASSOCIATE FORCE for all objects associated
with the statistics_type; drops statistics_type.

DROP object Calls DISASSOCIATE, drops object_type if

DISASSOCIATE succeeds.

10-19
 Chapter 10
Restrictions and Suggestions

Table 10-3 (Cont.) Dependency Model for DDLs

Command Action
If association is present for the column, this calls
ALTER TABLE DROP COLUMN
DISASSOCIATE FORCE with column; if no entry in
ASSOCIATION$ but there are entries in type USATS$,
then ODCIStatsDelete() for the columns is invoked.
If user-defined statistics collected with the
DISASSOCIATE
statistics_type are present, the command fails.
Deletes the entry in ASSOCIATION$ and calls
DISASSOCIATE FORCE
ODCIStatsDelete().
Delete index statistics using the The ODCIStatsDelete() function is invoked; if any
DBMS_STATISTICS package errors are raised, statistics deletion fails and an error is
reported.
If an association or user-defined statistics are present
ASSOCIATE
for the associated object, the command fails.

See Also:
ODCIStatsDelete()

10.6 Restrictions and Suggestions

A statistics type is an ordinary object type. Since an object type must have at least one
attribute, so must a statistics type. However, because it is never be accessed or set,
this is a dummy attribute.

10.6.1 Distributed Execution

Oracle's distributed implementation does not support adding functions to the remote
capabilities list. All functions referencing remote tables are executed as filters. The
placement of the filters occurs outside the optimizer. The cost model reflects this
implementation and does not attempt to optimize placement of these predicates.
Since predicates are not shipped to the remote site, you cannot use domain indexes
on remote tables. Therefore, the DESCRIBE protocol is unchanged, and remote domain
indexes are not visible from the local site.

10.6.2 System-Managed Storage Tables and ASSOCIATE

STATISTICS
If you are creating an indextype WITH SYSTEM MANAGED STORAGE TABLES, you should
also create its associated statistics type WITH SYSTEM MANAGED STORAGE TABLES. If you
are collecting statistics on the local indexed column using system partitioned tables,
then the Oracle server maintains the system-partitioned statistics tables for them
during partition maintenance operations. You can only use the WITH SYSTEM MANAGED

10-20
 Chapter 10
Restrictions and Suggestions

STORAGE TABLES option when an indextype is associated with the statistics type;
otherwise the system raises an error.

10.6.3 Aggregate Object-Level Statistics

When using local indexes, it may be useful to maintain both partition-level and
aggregate object-level statistics. During partition maintenance operations, the partition
level statistics are deleted, while the aggregate object-level statistics are either
adjusted to reflect the operation or left "as is" for later recomputation.
The decision to adjust or recompute the aggregate statistics is made based on
_minimal_stats_aggregation parameter in the server. If the parameter is FALSE, the
aggregate statistics are recomputed. If the parameter is TRUE, the statistics are not
recomputed.

10.6.4 System-Managed Domain Indexing

The system-managed domain indexing approach supports system-managed statistics
that are associated with indextypes; indextype itself should also be system-managed.

10.6.5 Collecting and Deleting User-Defined Statistics for System-

Managed Indexes
User-defined statistics collection for system-managed domain indexes does not
happen when analyzing the domain index itself. Instead, the call to
ODCIStatsCollect() will be invoked when index statistics are gathered during calls to
GATHER_*_STATS procedures defined inside the DBMS_STATS package. Likewise, the call
to ODCIStatsDelete() to delete domain index statistics will be invoked during calls to
DELETE_*_STATS procedures defined inside the DBMS_STATS package.

Example 10-26 demonstrates how to collect statistics for the TXT_IDX domain index on
the SCOTT schema by issuing a call to DBMS_STATS.GATHER_INDEX_STATS().

Example 10-27 shows how to delete statistics for the same index.
Example 10-28 demonstrates how to use the granularity argument to the
DBMS_STATS.GATHER_INDEX_STATS() to collect partition statistics from the PT_TXT_IDX
local domain index on the SCOTT schema.

Example 10-29 shows how to delete partition statistics for the same index. Example
10-30 and Example 10-31 demonstrate how to collect and delete statistics for a single
partition of a local domain index respectively.

See Also:
Oracle Database PL/SQL Packages and Types Reference

10.6.5.1 Collecting statistics for a system-managed domain index

This topics shows how to collect statistics for a system-managed domain index.

10-21
 Chapter 10
Restrictions and Suggestions

Example 10-26 Collecting statistics for a system-managed domain index

BEGIN
SYS.DBMS_STATS.GATHER_INDEX_STATS('SCOTT', 'TXT_IDX');
END;
/

10.6.5.2 Deleting statistics for a system-managed domain index

This topic shows how to delete statistics for a system-managed domain index.
Example 10-27 Deleting statistics for a system-managed domain index
BEGIN
SYS.DBMS_STATS.DELETE_INDEX_STATS('SCOTT', 'TXT_IDX');
END;
/

10.6.5.3 Collecting statistics for all partitions of a local system-managed

domain index
This topic shows how to collect statistics for all partitions of a local system-managed
domain index.
Example 10-28 Collecting statistics for all partitions of a local system-
managed domain index
BEGIN
SYS.DBMS_STATS.GATHER_INDEX_STATS('SCOTT', 'PT_TXT_IDX',
granularity=>'PARTITION');
END;
/

10.6.5.4 Deleting statistics for all partitions of a local system-managed domain

index
This topic shows how to delete statistics for all partitions of a local system-managed
domain index.
Example 10-29 Deleting statistics for all partitions of a local system-managed
domain index
BEGIN
SYS.DBMS_STATS.DELETE_INDEX_STATS('SCOTT', 'PT_TXT_IDX', cascade_parts=>TRUE);
END;
/

10.6.5.5 Collecting statistics for partition P2 of a local system-managed domain

index
This topic shows how to collect statistics for partition P2 of a local system-managed
domain index.

10-22
 Chapter 10
Restrictions and Suggestions

Example 10-30 Collecting statistics for partition P2 of a local system-managed

domain index
BEGIN
SYS.DBMS_STATS.GATHER_INDEX_STATS('SCOTT', 'PT_TXT_IDX', P2,
granularity=>'PARTITION');
END;
/

10.6.5.6 Deleting statistics for partition P2 of a local system-managed domain

index
This topics shows how to delete statistics for partition P2 of a local system-managed
domain index.
Example 10-31 Deleting statistics for partition P2 of a local system-managed
domain index
BEGIN
SYS.DBMS_STATS.DELETE_INDEX_STATS('SCOTT', 'PT_TXT_IDX', P2,
cascade_parts=>FALSE);
END;
/

When the local domain index is composite partitioned, it is possible to collect and
delete statistics for all subpartitions of a composite partition as shown in Example
10-32 and Example 10-33.
Example 10-34 and Example 10-35 demonstrate how to collect and delete statistics for
a single subpartition.
Additionally, if collection or deletion of statistics in all subpartitions is required, the
same calls shown in Example 10-28 and Example 10-29 must be used.

10.6.5.7 Collecting statistics for all subpartitions of a composite partition of a

local system-managed domain index
This topic shows how to collect statistics for all subpartitions of a composite partition of
a local system-managed domain index.
Example 10-32 Collecting statistics for all subpartitions of a composite
partition of a local system-managed domain index
BEGIN
SYS.DBMS_STATS.GATHER_INDEX_STATS('SCOTT', 'CPT_TXT_IDX', CP2,
granularity=>'PARTITION');
END;
/

10.6.5.8 Deleting statistics for all subpartitions of a composite partition of a

local system-managed domain index
This topic shows how to delete statistics for all subpartitions of a composite partition of
a local system-managed domain index

10-23
 Chapter 10
Restrictions and Suggestions

Example 10-33 Deleting statistics for all subpartitions of a composite partition

of a local system-managed domain index
BEGIN
SYS.DBMS_STATS.DELETE_INDEX_STATS('SCOTT', 'CPT_TXT_IDX', CP2,
cascade_parts=>FALSE);
END;
/

10.6.5.9 Collecting statistics for a subpartition of a local system-managed

domain index
This topic shows how to collect statistics for a subpartition of a local system-managed
domain index
Example 10-34 Collecting statistics for a subpartition of a local system-
managed domain index
BEGIN
SYS.DBMS_STATS.GATHER_INDEX_STATS('SCOTT', 'CPT_TXT_IDX', CP2_S1,
granularity=>'SUBPARTITION');
END;
/

10.6.5.10 Deleting statistics for a subpartition of a local system-managed

domain index
This topic shows how to delete statistics for a subpartition of a local system-managed
domain index
Example 10-35 Deleting statistics for a subpartition of a local system-managed
domain index
BEGIN
SYS.DBMS_STATS.DELETE_INDEX_STATS('SCOTT', 'CPT_TXT_IDX', CP2_S1,
cascade_parts=>FALSE);
END;
/

10.6.6 Performance
The cost of execution of the queries remains the same with the extensible optimizer if
the same plan is chosen. If a different plan is chosen, the execution time should be
better assuming that the user-defined cost, selectivity, and statistics collection
functions are accurate. In light of this, you are strongly encouraged to provide statistics
collection, selectivity, and cost functions for user-defined structures because the
optimizer defaults can be inaccurate and lead to an expensive execution plan.

10-24
11
Using Cartridge Services
Consider how to use cartridge services.

11.1 Introduction to Cartridge Services

Cartridge services are a set of services that help you create data cartridges in the
Oracle Extensibility framework.
Using Oracle Cartridge Services offers you the following advantages:

Portability
Oracle Cartridge Services offers you the flexibility to work across different machine
architectures

Flexibility Within Oracle Environments

Another type of flexibility is offered to you in terms of the fact that all cartridge services
work with your Oracle Database, irrespective of the configuration of operations that
has been purchased by your client.

Language Independence
The use of the Globalization Support services lets you internationalize your cartridge.
Language independence means that you can have different instances of your cartridge
operating in different language environments.

Tight Integration with the Server

Various cartridge services have been designed to facilitate access with Oracle
ORDBMS. This offers far superior performance to client -side programs attempting to
perform the same operations.

Guaranteed Compatibility
Oracle Database is a rapidly evolving technology and it is likely that your clients might
be operating with different releases of Oracle. The cartridge services operate with all
versions of Oracle Database.

Integration of Different Cartridges

The integration of cartridge services lets you produce a uniform integration of different
data cartridges.
Consider the set of services that you can use as part of your data cartridge. The APIs
that describe these interfaces are in Cartridge Services Using C_ C++ and Java

11-1
 Chapter 11
Cartridge Handle

11.2 Cartridge Handle

Cartridge services require various handles that are encapsulated inside two types of
OCI handles:

Environment Handle
The environment handle is either OCIEnv or OCI_HTYPE_ENV. Various cartridge services
are required at the process level when no session is available. The OCIInitialize()
should use the OCI_OBJECT option for cartridge service.

User Session handle

The user session handle is either OCISession or OCI_HTYPE_SESSION. In a callout, the
services can be used when the handle is allocated even without opening a connection
back to the database.
All cartridge service calls take a void * OCI handle as one of the arguments that may
be either an environment or a session handle. While most service calls are allowed
with either of the handles, certain calls may not be valid with one of the handles. For
example, it may be an error to allocate OCI_DURATION_SESSION with an environment
handle. An error is typically returned in an error handle.

11.2.1 Client Side Usage

Most of the cartridge service can also be used on the client side code. Refer to
individual services for restrictions. To use cartridge service on the client side, the OCI
environment has to be initialized with OCI_OBJECT option. This is automatically effected
in a cartridge.

11.2.2 Cartridge Side Usage

Most of the services listed in this document can be used in developing a database
cartridge, but please refer to documentation of each individual service for restrictions.
New service calls are available to obtain the session handle in a callout. The session
handle is available without opening a connection back to the server.

11.2.3 Making Service Calls

Before using any service, the OCI environment handle must be initialized. All the
services take an OCI environment (or user_session) handle as an argument. Errors
are returned in an OCI error handle. The sub handles required for various service calls
are not allocated along with the OCI environment handle. Services which must
initialize an environment provide methods to initialize it. Example 11-1 demonstrates
the initialization of these handles.
Example 11-1 Initializing OCI Handles
{
OCIEnv *envhp;
OCIError *errhp;
(void) OCIInitialize(OCI_OBJECT, (dvoid *)0, 0, 0, 0);
(void) OCIEnvInit(&envhp, OCI_OBJECT, (size_t)0, (dvoid **)0);
(void) OCIHandleAlloc((dvoid *)envhp, (dvoid **)errhp, OCI_HTYPE_ERROR,

11-2
 Chapter 11
Memory Services

(size_t)0, (dvoid **)0);

/* ... use the handles ... */
(void) OCIHandleFree((dvoid *)errhp, OCI_HTYPE_ERROR);
}

11.2.4 Handling Errors

Routines that return errors generally return OCI_SUCCESS or OCI_ERROR. Some routines
may return OCI_SUCCESS_WITH_INFO, OCI_INVALID_HANDLE, or OCI_NO_DATA. If
OCI_ERROR or OCI_SUCCESS_WITH_INFO is returned, then an error code, an error facility,
and possibly an error message can be retrieved by calling OCIErrorGet, as
demonstrated in Example 11-2.
Example 11-2 Retrieving Error Information Using OCIErrorGet()
{
OCIError *errhp;
ub4 errcode;
text buffer[512];
(void) OCIErrorGet((dvoid *)errhp, 1, (text *)NULL, &errcode, buffer,
sizeof(buffer), OCI_HTYPE_ERROR);
}

11.3 Memory Services

The memory service allows the client to allocate or free memory chunks. Each
memory chunk is associated with a duration. This allows clients to automatically free
all memory associated with a duration (at the end of the duration). The duration
determines the heap that is used to allocate the memory. The memory service
predefines three kinds of durations: call (OCI_DURATION_CALL), statement
(OCI_DURATION_STATEMENT) and session (OCI_DURATION_SESSION).

The client can also create a user duration. The client has to explicitly start and
terminate a user duration. Thus, the client can control the length of a user duration.
Like the predefined durations, a user duration can be used to specify the allocation
duration (for example, memory chunks are freed at the end of the user duration).
Each user duration has a parent duration. A user duration terminates implicitly when
its parent duration terminates. A parent duration can be call, statement, transaction,
session or any other user duration. Memory allocated in the user duration comes from
the heap of its parent duration.
The Oracle RDBMS memory manager supports a variety of memory models. Currently
callouts support memory for the duration of that callout. With the extension of row
sources to support external indexing, there is a need for memory of durations greater
than a callout.
The following functionality is supported:
• Allocate (permanent and friable) memory of following durations
– call to agent process
– statement
– session
– shared attributes (metadata) for cartridges
• Ability to re-allocate memory

11-3
 Chapter 11
Maintaining Context

• Ability to create a subduration memory, a sub heap which gets freed up when the
parent heap gets freed up. Memory for this sub heap can be allocated and freed.
• Ability to specify zeroed memory
• Ability to allocate large contiguous memory

11.4 Maintaining Context

Context management allows the clients to store values across calls. Cartridge services
provide a mechanism for saving and restoring context.
Most operating systems that support threads have the concept of thread context.
Threads can store thread specific data in this context (or state) and retrieve it at any
point. This provides a notion of thread global variable. Typically a pointer which points
to the root of a structure is stored in the context.
When the row source mechanism is externalized, you must have a mechanism to
maintain state between multiple calls to the same row source.
You must maintain session, statement and process states. Session state includes
information about multiple statements that are open, message files based on sessions'
Globalization Support settings, and so on. Process state includes shared metadata
(including systemwide metadata), message files, and so on. Depending on whether
the cartridge application is truly multi threaded, information sharing can be at a
process level or system level.
Since a user can be using multiple cartridges at any time, the state must be
maintained for each cartridge. This is done by requiring the user to supply a key for
each duration.

11.4.1 Durations
There are various predefined types of durations supported on memory and context
management calls. An additional parameter in all these calls is a context.
• OCI_DURATION_CALL. The duration of this operation is that of a callout.
• OCI_DURATION_STATEMENT. The duration of this operation is the external row
source.
• OCI_DURATION_SESSION. The duration of this operation is the user session.
• OCI_DURATION_PROCESS. The duration of this is agent process.

11.5 Globalization Support

To support multilingual application, Globalization Support functionality is required for
cartridges and callouts. NLSRTL is a multiplatform, multilingual library current used in
RDBMS and provides consistent Globalization Support behavior to all Oracle products.
Globalization Support basic services provide the following language and cultural
sensitive functionality:
• Locale information retrieval.
• String manipulation in the format of multibyte and wide-char.
• Character set conversion including Unicode support.

11-4
 Chapter 11
Parameter Manager Interface

• Messaging mechanism.

11.5.1 Globalization Support Language Information Retrieval

An Oracle locale consists of language, territory and character set definitions. The
locale determines conventions such as native day and month names; and date, time,
number, and currency formats. An internationalized application obeys a user's locale
setting and cultural convention. For example, in a German locale setting, users expect
to see day and month names in German spelling. The following interface provides a
simple way to retrieve local sensitive information.

11.5.2 String Manipulation

Two types of data structure are supported for string manipulation: multibyte string and
wide char string. Multibyte string is in native Oracle character set encoding, and
functions operated on it take the string as a whole unit. Wide char string function
provides more flexibility in string manipulation and supports character-based and
string-based operations.
The wide char data type we use here is Oracle-specific and not to be confused with
the wchar_t defined by the ANSI/ISO C standard. The Oracle wide char is always 4
bytes in all the platforms, while wchar_t is dependent on the implementation and
platform. The idea of Oracle wide char is to normalize multibyte characters to have a
fixed-width for easy processing. Round-trip conversion between Oracle wide char and
native character set is guaranteed.
The string manipulation can be classified into the following categories:
• Conversion of string between multibyte and wide char.
• Character classifications.
• Case conversion.
• Display length calculation.
• General string manipulation, such as compare, concatenation and searching.

11.6 Parameter Manager Interface

The parameter manager provides a set of routines to process parameters from a file or
a string. Routines are provided to process the input and to obtain key and value pairs.
These key and value pairs are stored in memory and routines are provided which can
access the values of the stored parameters.
The input processing routines match the contents of the file or the string against an
existing grammar and compare the key names found in the input against the list of
known keys that the user has registered. The behavior of the input processing routines
can be configured depending on the bits that are set in the flag argument.
The parameters can be retrieved either one at a time, or all at the same time by calling
a function that iterates over the stored parameters.

11.6.1 Input Processing and Support for Special Characters

Parameters consist of a key, or parameter name, type, and a value and must be
specified by the format key = value.

11-5
 Chapter 11
Parameter Manager Interface

Parameters can optionally accept lists of values which may be surrounded by

parentheses, either as key = (value1, ..., valuen) or as key = value1, ...,
valuen.

A value can be a string, integer, OCINumber, or Boolean. A boolean value starting with
'y' or 't' maps to TRUE and a boolean value starting with 'n' or 'f' maps to FALSE. The
matching for boolean values is case insensitive.
The parameter manager views certain characters as special characters which are not
parsed literally. The special characters and their meanings are indicated in Table 11-1.

Table 11-1 Special Characters

Character Description
# Comment (only for files)
( Start a list of values
) End a list of values
" Start or end of quoted string
' Start or end of quoted string
= Separator of keyword and value
\ Escape character

If a special character must be treated literally, then it must either be prefaced by the
escape character or the entire string must be surrounded by single or double quotes.
A key string can contain alphanumeric characters only. A value can contain any
characters. However, the value cannot contain special characters unless they are
quoted or escaped.

11.6.2 Parameter Manager Behavior Flag

The routines to process a file or a string use a behavior flag that alters default
characteristics of the parameter manager. These bits can be set in the flag:
• OCI_EXTRACT_CASE_SENSITIVE. All comparisons are case sensitive. The default is
to use case insensitive comparisons.
• OCI_EXTRACT_UNIQUE_ABBREVS. Unique abbreviations are allowed for keys. The
default is that unique abbreviations are not allowed.
• OCI_EXTRACT_APPEND_VALUES. If a value or values are stored for a particular key,
then any new values for this key should be appended. The default is to return an
error.

11.6.3 Key Registration

Before invoking the input processing routines (OCIExtractFromFile() or
OCIExtractFromString(), all of the keys must be registered by calling
OCIExtractSetNumKeys() followed by OCIExtractSetKey(), which requires:

• Name of the key

• Type of the key (integer, string, boolean, OCINumber)

11-6
 Chapter 11
File I/O

• OCI_EXTRACT_MULTIPLE is set for the flag value if multiple values are allowed
(default: only one value allowed)
• Default value to be used for the key (may be NULL)
• Range of allowable integer values specified by starting and ending values,
inclusive (may be NULL)
• List of allowable string values (may be NULL)

11.6.4 Parameter Storage and Retrieval

The results of processing the input into a set of keys and values are stored. The
validity of the parameters is checked before storing the parameters in memory. The
values are checked to see if they are of the proper type. In addition, if you wish, the
values can be checked to see if they fall within a certain range of integer values or are
members of a list of enumerated string values. Also, if you do not specify that a key
can accept multiple values, then an error is returned if a key is specified more than
one time in a particular input source. Also, an error is returned if the key is unknown.
Values of keys can be retrieved when processing is completed, using specific routines
for retrieving string, integer, OCINumber, or boolean values.

It is possible to retrieve all parameters at the same time. The function

OCIExtractToList() must first be called to generate a list of parameters that is
created from the parameter structures stored in memory. OCIExtractToList() returns
the number of unique keys stored in memory, and then OCIExtractFromList() can be
called to return the list of values associated with each key.

11.6.5 Parameter Manager Context

The parameter manager maintains its own context within the OCI environment handle.
This context stores all the processed parameter information and some internal
information. It must be initialized with a call to OCIExtractInit() and cleaned up with
a call to OCIExtractTerm().

11.7 File I/O

The OCI file I/O package is designed to make it easier for you to write portable code
that interacts with the file system by providing a consistent view of file I/O across
multiple platforms.
You must be aware of two issues when using this package in a data cartridge
environment. The first issue is that this package does not provide any security when
opening files for writing or when creating new files in a directory other than the security
provided by the operating system protections on the file and directory. The second
issue is that this package does not support the use of file descriptors across calls in a
multithreaded server environment.

11.8 String Formatting

The OCI string formatting package facilitates writing portable code that handles string
manipulation by means of the OCIFormatString() routine. This is an improved and
portable version of sprintf that incorporates additional functionality and error
checking that the standard sprintf does not. This additional functionality includes:

11-7
 Chapter 11
String Formatting

• Arbitrary argument selection.

• Variable width and precision specification.
• Length checking of the buffer.
• Oracle Globalization Support for internationalization.

11-8
12
Using User-Defined Aggregate Functions
User-defined aggregate functions may be used both singly and in parallel; consider
large aggregation contexts and materialized views.

See Also:
User-Defined Aggregate Functions Interface for a detailed description of the
ODCIAggregate interface.

12.1 Overview of User-Defined Aggregate Functions

Oracle provides several pre-defined aggregate functions such as MAX, MIN, and SUM for
performing operations on a set of rows. These pre-defined aggregate functions can be
used only with scalar data, not with complex data types such as multimedia data
stored using object types, opaque types, and LOBs. You can, however, define custom
implementations of these functions for complex data types. You can also define
entirely new aggregate functions to use with complex data. User-defined aggregate
functions can be used in SQL DML statements just like Oracle's built-in aggregates.
When functions are registered with the server, Oracle simply invokes the user-defined
aggregation routines supplied by you instead of the native routines. User-defined
aggregates can also be used with scalar data, such as complex statistical data
necessary for scientific applications.
User-defined aggregates are a feature of the Extensibility Framework, and you can
implement them using ODCIAggregate interface routines.

You can create a user-defined aggregate function by implementing a set of routines

collectively known as the ODCIAggregate routines. You can implement these routines
as methods within an object type, so the implementation can be in any language that
Oracle supports, PL/SQL, C, C++ or Java. When the object type is defined and the
routines are implemented in the type body, use the CREATE FUNCTION statement to
create the aggregate function.
Each user-defined aggregate function uses up to four ODCIAggregate routines, or
steps, to define internal operations that any aggregate function performs, namely:
initialization, iteration, merging, and termination.
• Initialization is accomplished by the ODCIAggregateInitialize() routine, which is
invoked by Oracle to initialize the computation of the user-defined aggregate. The
initialized aggregation context is passed back to Oracle as an object type instance.
• Iteration is performed through the ODCIAggregateIterate() routine, which is
repeatedly invoked by Oracle. On each invocation, a new value or a set of new
values and the current aggregation context are passed in. The routine processes
the new values and returns the updated aggregation context. This routine is
invoked for every non-NULL value in the underlying group. NULL values are ignored
during aggregation and are not passed to the routine.

12-1
 Chapter 12
Creating a User-Defined Aggregate

• Merging is performed by ODCIAggregateMerge(), a routine invoked by Oracle to

combine two aggregation contexts. This routine takes the two contexts as inputs,
combines them, and returns a single aggregation context.
• Termination takes place when the ODCIAggregateTerminate() routine is invoked
by Oracle as the final step of aggregation. The routine takes the aggregation
context as input and returns the resulting aggregate value.
The process is illustrated in the Using User-Defined Aggregate Functions section.

12.1.1 Using User-Defined Aggregate Functions

Consider the aggregate function AVG() in the following statement:
SELECT AVG(T.Sales)
FROM AnnualSales T
GROUP BY T.State;

To perform this computation, the aggregate function AVG() goes through these steps:

1. Initializes the computation by initializing the aggregation context, or the rows over
which aggregation is performed:
runningSum = 0; runningCount = 0;
2. Iteratively processes each successive input value and updates the context:
runningSum += inputval; runningCount++;
3. [Optional] Merge by combining the two aggregation contexts and return a single
context. This operation combines the results of aggregation over subsets to obtain
the aggregate over the entire set. This extra step can be required during either
serial or parallel evaluation of an aggregate. If needed, it is performed before step
4:
runningSum = runningSum1 + runningSum2;
runningCount = runningCount1 + runningCount2

Evaluating User-Defined Aggregates in Parallel describes this step in greater

detail.
4. Terminates by computing the result; uses the context to return the resultant
aggregate value:
return (runningSum/runningCount);
If AVG() were a user-defined function, the object type that embodies it would
implement a method for a corresponding ODCIAggregate routine for each of these
steps. The variables runningSum and runningCount, which determine the state of the
aggregation in the example, would be attributes of that object type.

12.2 Creating a User-Defined Aggregate

The process of creating a user-defined aggregate function has two steps, illustrated in
Example 12-1 and Example 12-2. Both examples use the SpatialUnion() aggregate
function defined by Oracle Spatial. The function computes the bounding geometry over
a set of input geometries.

12-2
 Chapter 12
Using a User-Defined Aggregate

Example 12-1 Implementing the ODCIAggregate Interface

The ODCIAggregate routines are implemented as methods within an object type
SpatialUnionRoutines. The actual implementation could be in any Oracle-supported
language for type methods, such as PL/SQL, C, C++ or Java.
CREATE TYPE SpatialUnionRoutines(
STATIC FUNCTION ODCIAggregateInitialize( ... ) ...,
MEMBER FUNCTION ODCIAggregateIterate(...) ... ,
MEMBER FUNCTION ODCIAggregateMerge(...) ...,
MEMBER FUNCTION ODCIAggregateTerminate(...)
);

CREATE TYPE BODY SpatialUnionRoutines IS

...
END;

Example 12-2 Defining a User-Defined Aggregate Function

This function definition creates the SpatialUnion() aggregate function by specifying
its signature and the object type that implements the ODCIAggregate interface:
CREATE FUNCTION SpatialUnion(x Geometry) RETURN Geometry
AGGREGATE USING SpatialUnionRoutines;

12.3 Using a User-Defined Aggregate

User-defined aggregates can be used just like built-in aggregate functions in SQL DML
and query statements. They can appear in the SELECT list, ORDER BY clause, or as part
of the predicate in the HAVING clause. The following Example 12-3, Example 12-4 and
Example 12-5 illustrate some options.

See Also:
Oracle Database Data Warehousing Guide for information about GROUP BY
extensions such as ROLLUP, CUBE and grouping sets

12.3.1 Using the SELECT Statement with User-Defined Aggregate

Functions
Example 12-3 Using the SELECT Statement with User-Defined Aggregate
Functions
The following query can be used to compute state boundaries by aggregating the
geometries of all counties belonging to the same state:
SELECT SpatialUnion(geometry)
FROM counties
GROUP BY state

12-3
 Chapter 12
Evaluating User-Defined Aggregates in Parallel

12.3.2 Using the HAVING Clause with User-Defined Aggregate

Functions
Example 12-4 Using the HAVING Clause with User-Defined Aggregate
Functions
User-defined aggregates can be used in the HAVING clause to eliminate groups from
the output based on the results of the aggregate function. Here, MyUDAG() is a user-
defined aggregate:
SELECT groupcol, MyUDAG(col)
FROM tab
GROUP BY groupcol
HAVING MyUDAG(col) > 100
ORDER BY MyUDAG(col);

12.3.3 Using Query Options with User-Defined Aggregate Functions

Example 12-5 Using other Query Options with User-Defined Aggregate
Functions
User-defined aggregates can take DISTINCT or ALL (default) options on the input
parameter. DISTINCT causes duplicate values to be ignored while computing an
aggregate. The SELECT statement that contains a user-defined aggregate can also
include GROUP BY extensions such as ROLLUP, CUBE and grouping sets:
SELECT ..., MyUDAG(col)
FROM tab
GROUP BY ROLLUP(gcol1, gcol2);

The ODCIAggregateMerge() interface is invoked to compute super aggregate values

in such roll-up operations.

12.4 Evaluating User-Defined Aggregates in Parallel

Like built-in aggregate functions, user-defined aggregates can be evaluated in parallel.
The aggregation contexts generated by aggregating subsets of the rows within the
parallel slaves are sent back to the next parallel step, either the query coordinator or
the next slave set. It then merges the aggregation contexts, and then invokes the
Terminate routine to obtain the aggregate value. This behaviour is illustrated in
Figure 12-1.

12-4
 Chapter 12
Handling Large Aggregation Contexts

Figure 12-1 Sequence of Calls for Parallel Evaluation of User-Defined

Aggregates

Initialize Iterate

Merge

Initialize Iterate Terminate

You should note that the aggregate function must be declared to be parallel-enabled,
as shown in Example 12-6:
Example 12-6 Parallel-Enabling a User-Defined Aggregate Function
CREATE FUNCTION MyUDAG(...) RETURN ...
PARALLEL_ENABLE AGGREGATE USING MyAggrRoutines;

12.5 Handling Large Aggregation Contexts

When the implementation type methods are implemented in an external language,
such as C++ or Java, the aggregation context must be passed back and forth between
the Oracle server process and the external function's language environment each time
an implementation type method is called. This can have an adverse effect on
performance as the size of the aggregation context increases.
To enhance performance, you can store the aggregation context in external memory,
allocated in the external function's execution environment. You can then pass the
reference or key between the Oracle server and the external function. The key itself
should be stored in the implementation type instance, the self. This approach keeps
the implementation type instance small so that it can be transferred quickly. Another
advantage of this strategy is that the memory used to hold the aggregation context is
allocated in the function's execution environment, such as extproc, and not in the
Oracle server.
Usually you should use ODCIAggregateInitialize() to allocate the memory to hold
the aggregation context and store the reference to it in the implementation type
instance. In subsequent calls, the external memory and the aggregation context that it
contains can be accessed using the reference. The external memory should usually be
freed in ODCIAggregateTerminate(). ODCIAggregateMerge() should free the external
memory used to store the merged context (the second argument of
ODCIAggregateMerge() after the merge is finished.

12-5
 Chapter 12
Handling Large Aggregation Contexts

See Also:

• ODCIAggregateInitialize()
• ODCIAggregateTerminate()
• ODCIAggregateMerge()

12.5.1 External Context and Parallel Aggregation

With parallel execution of queries with user-defined aggregates, the entire aggregation
context, which comprises all partial aggregates computed by slave processes, must
sometimes be transmitted to another slave or to the master process. You can
implement the optional routine ODCIAggregateWrapContext() to collect all the partial
aggregates. If a user-defined aggregate is being evaluated in parallel, and
ODCIAggregateWrapContext() is defined, Oracle invokes the routine to copy all
external context references into the implementation type instance and then frees the
external memory. To support ODCIAggregateWrapContext(), the implementation type
must contain attributes to hold the aggregation context and another attribute to hold
the key that identifies the external memory.
When the aggregation context is stored externally, the key attribute of the
implementation type should contain the reference identifying the external memory, and
the remaining attributes of the implementation type should be NULL. After a
ODCIAggregateWrapContext() call runs successfully, the key attribute should be NULL,
and the other attributes should hold the actual aggregation context.
Each of the implementation type's member methods should begin by checking whether
the context is inline (contained in the implementation type instance) or in external
memory. If the context is inline, as it would be if it was sent from another parallel slave,
it should be copied to external memory so that it can be passed by reference.
Implementation of the ODCIAggregateWrapContext() routine is optional. It is necessary
only when external memory holds the aggregation context, and the user-defined
aggregate is evaluated in parallel. If the user-defined aggregate is never evaluated in
parallel, ODCIAggregateWrapContext() is not needed. If the
ODCIAggregateWrapContext() method is not defined, Oracle assumes that the
aggregation context is not stored externally and does not try to call the method.

Note:
ODCIAggregateWrapContext()

12.5.1.1 Using External Memory to Store Aggregate Context

This example shows how an aggregation context type that contains references to
external memory can also store the entire context, when needed.
The 4 byte key parameter is used to look up the external context. When NULL, it implies
that the entire context value is held by the rest of the attributes in the object. The other
attributes, such as GeometrySet, correspond to the actual aggregation context. If the

12-6
 Chapter 12
Handling Large Aggregation Contexts

key value is not NULL, these attributes must have a NULL value. However, when the
context object is self-contained, as after a call to ODCIAggregateWrapContext(), these
attributes hold the current context values.
CREATE TYPE MyAggrRoutines AS OBJECT
(
key RAW(4),
ctxval GeometrySet,
ctxval2 ...
);

12.5.2 User-Defined Aggregates and Analytic Functions

Analytic functions enable you to compute various cumulative, moving, and centered
aggregates over a set of rows called a window. For each row in a table, analytic
functions return a value computed on the other rows contained in the given row's
window. These functions provide access to several rows of a table without a self-join.
User-defined aggregates can be used as analytic functions.

12.5.2.1 Using User-Defined Aggregates and Analytic Functions

SELECT Account_number, Trans_date, Trans_amount,
MyAVG (Trans_amount) OVER(
PARTITION BY Account_number ORDER BY Trans_date
RANGE INTERVAL '7' DAY PRECEDING) AS mavg_7day
FROM Ledger;

12.5.3 Reuse of Aggregation Context for Analytic Functions

When a user-defined aggregate is used as an analytic function, the aggregate is
calculated for each row's corresponding window. Generally, each successive window
contains largely the same set of rows, such that the new aggregation context, the new
window, differs by only a few rows from the old aggregation context, the previous
window. To reuse the aggregation context, any new rows that were not in the old
context must be iterated over to add them, and any rows from the old context that do
not belong in the new context must be removed. If the aggregation context cannot be
reused, all the rows it contains must be reiterated to rebuild it.
You can implement an optional routine, ODCIAggregateDelete(), to allow Oracle to
reuse the aggregation context more efficiently. ODCIAggregateDelete() removes from
the aggregation context rows from the previous context that are not in the new
(current) window. Oracle calls this routine for each row that must be removed. For
each row that must be added, Oracle calls ODCIAggregateIterate().

If the new aggregation context is a superset of the old one, then it contains all the rows
from the old context and no rows must be deleted. Oracle then reuses the old context
even if ODCIAggregateDelete() is not implemented.

12-7
 Chapter 12
Using Materialized Views with User-Defined Aggregates

See Also:

• Oracle Database Data Warehousing Guide for information about analytic

functions
• ODCIAggregateDelete()
• ODCIAggregateIterate()

12.5.4 External Context and User-Defined Analytic Functions

When user-defined aggregates are used as analytic functions, the aggregation context
can be reused from one window to the next. In these cases, the flag argument of the
ODCIAggregateTerminate() function has its ODCI_AGGREGATE_REUSE_CTX bit set to
indicate that the external memory holding the aggregation context should not be freed.
Also, the ODCIAggregateInitialize() method is passed the implementation type
instance of the previous window, so instead of having to allocate memory again, you
can access and re-initialize the external memory previously allocated. To support
external context for user-defined analytic functions, you should follow these steps:
1. ODCIAggregateInitialize() - If the implementation type instance passed is not
NULL, use the previously allocated external memory instead of allocating new
external memory, and reinitialize the aggregation context.
2. ODCIAggregateTerminate() - Free external memory only if the bit
ODCI_AGGREGATE_REUSE_CTX of the flag argument is not set.
3. ODCIAggregateMerge() - Free external memory associated with the merged
aggregation context.
4. ODCIAggregateTerminate() - Copy the aggregation context from the external
memory into the implementation type instance, and free the external memory.
5. All member methods - First determine if the context is stored externally or inline. If
the context is inline, allocate external memory and copy the context there.

See Also:

• ODCIAggregateInitialize()
• ODCIAggregateMerge()
• ODCIAggregateTerminate()

12.6 Using Materialized Views with User-Defined

Aggregates
A materialized view definition can contain user-defined aggregates and built-in
aggregate operators, as demonstrated in Example 12-7.

12-8
 Chapter 12
Creating and Using a User-Defined Aggregate Function

To enable the materialized view for query rewrite, the user-defined aggregates in the
materialized view must be declared as DETERMINISTIC, as demonstrated in
Example 12-8.
When a user-defined aggregate is dropped or re-created, all of its dependent
materialized views are marked invalid.
Example 12-7 Creating Materialized Views
CREATE MATERIALIZED VIEW MyMV AS
SELECT gcols, MyUDAG(c1) FROM tab GROUP BY (gcols);

Example 12-8 Enabling Materialized Views for Query Rewrite

CREATE FUNCTION MyUDAG(x NUMBER) RETURN NUMBER
DETERMINISTIC
AGGREGATE USING MyImplType;

CREATE MATERIALIZED VIEW MyMV

ENABLE QUERY REWRITE AS
SELECT gcols, MyUDAG(c1) FROM tab GROUP BY (gcols);

See Also:
Oracle Database Data Warehousing Guide for information about materialized
views

12.7 Creating and Using a User-Defined Aggregate Function

Example 12-9 illustrates how to create and use a simple user-defined aggregate
function, SecondMax().

Example 12-9 Creating and Using a User-Defined Aggregate Function

SecondMax() returns the second-largest value in a set of numbers.

1. Implement the type SecondMaxImpl to contain the ODCIAggregate routines:

create type SecondMaxImpl as object
(
max NUMBER, -- highest value seen so far
secmax NUMBER, -- second highest value seen so far
static function ODCIAggregateInitialize(sctx IN OUT SecondMaxImpl)
return number,
member function ODCIAggregateIterate(self IN OUT SecondMaxImpl,
value IN number) return number,
member function ODCIAggregateTerminate(self IN SecondMaxImpl,
returnValue OUT number, flags IN number) return number,
member function ODCIAggregateMerge(self IN OUT SecondMaxImpl,
ctx2 IN SecondMaxImpl) return number
);
/
2. Implement the type body for SecondMaxImpl:
create or replace type body SecondMaxImpl is
static function ODCIAggregateInitialize(sctx IN OUT SecondMaxImpl)
return number is

12-9
 Chapter 12
Creating and Using a User-Defined Aggregate Function

begin
sctx := SecondMaxImpl(0, 0);
return ODCIConst.Success;
end;

member function ODCIAggregateIterate(self IN OUT SecondMaxImpl, value IN number)

return number is
begin
if value > self.max then
self.secmax := self.max;
self.max := value;
elsif value > self.secmax then
self.secmax := value;
end if;
return ODCIConst.Success;
end;

member function ODCIAggregateTerminate(self IN SecondMaxImpl,

returnValue OUT number, flags IN number) return number is
begin
returnValue := self.secmax;
return ODCIConst.Success;
end;

member function ODCIAggregateMerge(self IN OUT SecondMaxImpl, ctx2 IN

SecondMaxImpl) return number is
begin
if ctx2.max > self.max then
if ctx2.secmax > self.secmax then
self.secmax := ctx2.secmax;
else
self.secmax := self.max;
end if;
self.max := ctx2.max;
elsif ctx2.max > self.secmax then
self.secmax := ctx2.max;
end if;
return ODCIConst.Success;
end;
end;
/
3. Create the user-defined aggregate:
CREATE FUNCTION SecondMax (input NUMBER) RETURN NUMBER
PARALLEL_ENABLE AGGREGATE USING SecondMaxImpl;
4. Use SecondMax():
SELECT SecondMax(salary), department_id
FROM MyEmployees
GROUP BY department_id
HAVING SecondMax(salary) > 9000;

12-10
13
Using Pipelined and Parallel Table
Functions
The table functions and the generic datatypes ANYTYPE, ANYDATA, and ANYDATASET are
often used with table functions.

13.1 Overview of Table Functions

Table functions are functions that produce a collection of rows (either a nested table or
a varray) that can be queried like a physical database table. You use a table function
like the name of a database table, in the FROM clause of a query.

A table function can take a collection of rows as input. An input collection parameter
can be either a collection type or a REF CURSOR.

Table function may be executed in parallel, and returned rows can be streamed
directly to the next process without intermediate staging. Rows from a collection
returned by a table function can also be pipelined; this means that they are iteratively
returned as they are produced, instead of being returned in a single batch after all
processing of the table function's input is completed.
Streaming, pipelining, and parallel execution of table functions can improve
performance in the following manner:
• By enabling multithreaded, concurrent execution of table functions
• By eliminating intermediate staging between processes
• By improving query response time: With non-pipelined table functions, the entire
collection returned by a table function must be constructed and returned to the
server before the query can return a single result row. Pipelining enables rows to
be returned iteratively, as they are produced. This also reduces the memory that a
table function requires, as the object cache does not have to materialize the entire
collection.
• By iteratively providing result rows from the collection returned by a table function
as the rows are produced instead of waiting until the entire collection is staged in
tables or memory and then returning the entire collection
Figure 13-1 shows a typical data-processing scenario in which data goes through
several (in this case, three) transformations, implemented by table functions, before
finally being loaded into a database. In this scenario, the table functions are not run in
parallel, and the entire result collection must be staged after each transformation.

13-1
 Chapter 13
Table Function Concepts

Figure 13-1 Typical Data Processing with Non-Parallel, Non-Pipelined Table

Functions

OLTP T1 Stage 1 T2 Stage 2 T3 DSS

By contrast, Figure 13-2 shows how streaming and parallel execution can streamline
the same scenario.

Figure 13-2 Data Processing Using Pipelining and Parallel Execution

T1 T2 T3

OLTP T1 T2 T3 Data
Warehouse
T1 T2 T3

13.2 Table Function Concepts

Consider table functions and some concepts related to pipelining and parallel
execution of these table functions.

13.2.1 Table Functions

Table functions return a collection type instance and can be queried like a table by
calling the function in the FROM clause of a query. Table functions use the TABLE
keyword.
The following example shows a table function GetBooks that takes a CLOB as input and
returns an instance of the collection type BookSet_t. The CLOB column stores a catalog
listing of books in some format (either proprietary or following a standard such as
XML). The table function returns all the catalogs and their corresponding book listings.
The collection type BookSet_t is defined in Example 13-1.

The CLOBs are stored in a table Catalogs, as demonstrated in Example 13-2.

Function GetBooks() is defined in Example 13-3.

The query in Example 13-4 returns all the catalogs and their corresponding book
listings.
Example 13-1 Creating a Collection Type
CREATE TYPE Book_t AS OBJECT
( name VARCHAR2(100),
author VARCHAR2(30),
abstract VARCHAR2(1000));

CREATE TYPE BookSet_t AS TABLE OF Book_t;

13-2
 Chapter 13
Table Function Concepts

Example 13-2 Storing a Clob in a Table

CREATE TABLE Catalogs
( name VARCHAR2(30),
cat CLOB);

Example 13-3 Creating a Function that Returns a Collection Type

CREATE FUNCTION GetBooks(a CLOB) RETURN BookSet_t;

Example 13-4 Using a Collection Type in a Query

SELECT c.name, Book.name, Book.author, Book.abstract
FROM Catalogs c, TABLE(GetBooks(c.cat)) Book;

13.2.2 Pipelined Table Functions

Data is said to be pipelined if it is consumed by a consumer (transformation) as soon
as the producer (transformation) produces it, without being staged in tables or a cache
before being input to the next transformation.
Pipelining enables a table function to return rows faster and can reduce the memory
required to cache a table function's results.
A pipelined table function can return the table function's result collection in subsets.
The returned collection behaves like a stream that can be fetched from on demand.
This makes it possible to use a table function like a virtual table.
Pipelined table functions can be implemented in two ways:
• In the native PL/SQL approach, the consumer and producers can run on separate
execution threads (either in the same or different process context) and
communicate through a pipe or queuing mechanism. This approach is similar to
co-routine execution.
• In the interface approach, the consumer and producers run on the same execution
thread. Producer explicitly returns the control back to the consumer after
producing a set of results. In addition, the producer caches the current state so
that it can resume where it left off when the consumer invokes it again.
The interface approach requires you to implement a set of well-defined interfaces
in a procedural language.
The co-routine execution model provides a simpler, native PL/SQL mechanism for
implementing pipelined table functions, but this model cannot be used for table
functions written in C or Java. The interface approach, on the other hand, can. The
interface approach requires the producer to save the current state information in a
context object before returning so that this state can be restored on the next
invocation.
In this discussion, the term table function refers to a pipelined table function— a table
function that returns a collection in an iterative, pipelined way.

13.2.3 Pipelined Table Functions with REF CURSOR Arguments

A pipelined table function can accept any argument that regular functions accept. A
table function that accepts a REF CURSOR as an argument can serve as a
transformation function. That is, it can use the REF CURSOR to fetch the input rows,

13-3
 Chapter 13
Table Function Concepts

perform some transformation on them, and then pipeline the results out (using either
the interface approach or the native PL/SQL approach).
For example, the following code sketches the declarations that define a StockPivot
function. This function converts a row of the type (Ticker, OpenPrice, ClosePrice)
into two rows of the form (Ticker, PriceType, Price). Calling StockPivot for the row
("ORCL", 41, 42) generates two rows: ("ORCL", "O", 41) and ("ORCL", "C", 42).

Input data for the table function might come from a source such as table StockTable:
CREATE TABLE StockTable (
ticker VARCHAR(4),
openprice NUMBER,
closeprice NUMBER
);

The declarations are in Example 13-5.

Example 13-6 uses the StockPivot table function.

In the preceding query, the pipelined table function StockPivot fetches rows from the
CURSOR subquery SELECT * FROM StockTable, performs the transformation, and
pipelines the results back to the user as a table. The function produces two output
rows (collection elements) for each input row.
Note that when a CURSOR subquery is passed from SQL to a REF CURSOR function
argument as in the preceding example, the referenced cursor is open when the
function begins executing.
Note also that cursor operations are not allowed for REF CURSOR variables based on
table functions: SELECT FOR UPDATE, and WHERE CURRENT OF.

See Also:
Pipelined Table Functions: Interface Approach Example for a complete
implementation of this table function using the interface approach, in both C
and Java.

Example 13-5 Declaring a Pipelined Table Function with REF CURSOR

Arguments
-- Create the types for the table function's output collection
-- and collection elements

CREATE TYPE TickerType AS OBJECT

(
ticker VARCHAR2(4),
PriceType VARCHAR2(1),
price NUMBER
);

CREATE TYPE TickerTypeSet AS TABLE OF TickerType;

-- Define the ref cursor type

CREATE PACKAGE refcur_pkg IS

TYPE refcur_t IS REF CURSOR RETURN StockTable%ROWTYPE;

13-4
 Chapter 13
Pipelined Table Functions

END refcur_pkg;
/

-- Create the table function

CREATE FUNCTION StockPivot(p refcur_pkg.refcur_t) RETURN TickerTypeSet

PIPELINED ... ;
/

Example 13-6 Using a Pipelined Table Function with REF CURSOR Arguments
SELECT * FROM TABLE(StockPivot(CURSOR(SELECT * FROM StockTable)));

13.2.4 Parallel Execution of Table Functions

With parallel execution of a function that appears in the SELECT list, execution of the
function is pushed down to and conducted by multiple slave scan processes. These
each execute the function on a segment of the function's input data.
For example, the query
SELECT f(col1) FROM tab;

is parallelized if f is a pure function. The SQL executed by a slave scan process is

similar to:
SELECT f(col1) FROM tab WHERE ROWID BETWEEN :b1 AND :b2;

Each slave scan operates on a range of rowids and applies function f to each
contained row. Function f is then executed by the scan processes; it does not run
independently of them.
Unlike a function that appears in the SELECT list, a table function is called in the FROM
clause and returns a collection. This affects the way that table function input data is
partitioned among slave scans because the partitioning approach must be appropriate
for the operation that the table function performs. (For example, an ORDER BY operation
requires input to be range-partitioned, whereas a GROUP BY operation requires input to
be hash partitioned.)
A table function itself specifies in its declaration the partitioning approach that is
appropriate for it, as described in "Input Data Partitioning". The function is then
executed in a two-stage operation. First, one set of slave processes partitions the data
as directed in the function's declaration; then a second set of slave scans executes the
table function in parallel on the partitioned data. The table function in the following
query has a REF CURSOR parameter:
SELECT * FROM TABLE(f(CURSOR(SELECT * FROM tab)));

The scan is performed by one set of slave processes, which redistributes the rows
(based on the partitioning method specified in the function declaration) to a second set
of slave processes that actually executes function f in parallel.

13.3 Pipelined Table Functions

Consider issues involved in implementing pipelined table functions.

13-5
 Chapter 13
Pipelined Table Functions

13.3.1 Implementation Choices for Pipelined Table Functions

As noted previously, two approaches are supported for implementing pipelined table
functions: the interface approach and the PL/SQL approach.
The interface approach requires the user to supply a type that implements a
predefined Oracle interface consisting of start, fetch, and close operations. The type is
associated with the table function when the table function is created. During query
execution, the fetch method is invoked repeatedly to iteratively retrieve the results.
With the interface approach, the methods of the implementation type associated with
the table function can be implemented in any of the supported internal or external
languages (including PL/SQL, C/C++, and Java).
With the PL/SQL approach, a single PL/SQL function includes a special instruction to
pipeline results (single elements of the collection) out of the function instead of
returning the whole collection as a single value. The native PL/SQL approach is
simpler to implement because it requires writing only one PL/SQL function.
The approach used to implement pipelined table functions does not affect the way they
are used. Pipelined table functions are used in SQL statements in exactly the same
way regardless of the approach used to implement them.

13.3.2 Declaring Pipelined Table Functions

You declare a pipelined table function by specifying the PIPELINED keyword. This
keyword indicates that the function returns rows iteratively. The return type of the
pipelined table function must be a collection type (a nested table or a varray).

Example 13-7 shows declarations of pipelined table functions implemented using the
interface approach. The interface routines for functions GetBooks and StockPivot
have been implemented in the types BookMethods and StockPivotImpl, respectively.

Example 13-8 shows declarations of the same table functions implemented using the
native PL/SQL approach.
Example 13-7 Declaring Pipelined Table Functions for the Interface Approach
CREATE FUNCTION GetBooks(cat CLOB) RETURN BookSet_t PIPELINED USING BookMethods;

CREATE FUNCTION StockPivot(p refcur_pkg.refcur_t)

RETURN TickerTypeSet PIPELINED USING StockPivotImpl;

Example 13-8 Declaring Pipelined Table Functions for the Native PL/SQL
Approach
CREATE FUNCTION GetBooks(cat CLOB) RETURN BookSet_t PIPELINED IS ...;

CREATE FUNCTION StockPivot(p refcur_pkg.refcur_t) RETURN TickerTypeSet

PIPELINED IS...;

13.3.3 Implementing the Native PL/SQL Approach

In PL/SQL, the PIPE ROW statement causes a table function to pipe a row and continue
processing. The statement enables a PL/SQL table function to return rows as soon as
they are produced. This is demonstrated in Example 13-9. For performance reasons,
the PL/SQL run-time system provides the rows to the consumer in batches.

13-6
 Chapter 13
Pipelined Table Functions

In Example 13-9, the PIPE ROW(out_rec) statement pipelines data out of the PL/SQL
table function.
The PIPE ROW statement may be used only in the body of pipelined table functions; an
error is raised if it is used anywhere else. The PIPE ROW statement can be omitted for a
pipelined table function that returns no rows.
A pipelined table function must have a RETURN statement that does not return a value.
The RETURN statement transfers the control back to the consumer and ensures that the
next fetch gets a NO_DATA_FOUND exception.

Example 13-9 Implementing a Pipelined Table Function for the Native PL/SQL
Approach
CREATE FUNCTION StockPivot(p refcur_pkg.refcur_t) RETURN TickerTypeSet
PIPELINED IS
out_rec TickerType := TickerType(NULL,NULL,NULL);
in_rec p%ROWTYPE;
BEGIN
LOOP
FETCH p INTO in_rec;
EXIT WHEN p%NOTFOUND;
-- first row
out_rec.ticker := in_rec.Ticker;
out_rec.PriceType := 'O';
out_rec.price := in_rec.OpenPrice;
PIPE ROW(out_rec);
-- second row
out_rec.PriceType := 'C';
out_rec.Price := in_rec.ClosePrice;
PIPE ROW(out_rec);
END LOOP;
CLOSE p;
RETURN;
END;
/

13.3.4 Pipelining Between PL/SQL Table Functions

With serial execution, results are pipelined from one PL/SQL table function to another
using an approach similar to co-routine execution. Example 13-10 pipelines results
from function g to function f.

Parallel execution works similarly, except that each function executes in a different
process or set of processes.
Example 13-10 Pipelining Function Results from One Function to Another
SELECT * FROM TABLE(f(CURSOR(SELECT * FROM TABLE(g()))));

13.3.5 Combining PIPE ROW with AUTONOMOUS_TRANSACTION

Because table functions pass control back and forth to a calling routine as rows are
produced, there is a restriction on combining table functions and PRAGMA
AUTONOMOUS_TRANSACTIONs. If a table function is part of an autonomous transaction, it
must COMMIT or ROLLBACK before each PIPE ROW statement, to avoid an error in the
calling subprogram.

13-7
 Chapter 13
Pipelined Table Functions

13.3.6 Implementing the Interface Approach

To use the interface approach, you must define an implementation type that
implements the ODCITable interface. This interface consists of start, fetch, and close
routines whose signatures are specified by Oracle and which you implement as
methods of the type.
Oracle invokes the methods to perform the following steps in the execution of a query
that contains a table function:
1. Start by initializing the scan context parameter, using the ODCITableStart()
function.
2. Fetch to produce a subset of the rows in the result collection. The
ODCITableFetch() method is invoked as many times as necessary to return the
entire collection.
3. Close and clean up (release memory and so on) using ODCITableClose() after the
last ODCITableFetch().
The ODCITable interface also defines two optional routines, ODCITablePrepare() and
ODCITableDescribe(), that are invoked at compilation time:

• ODCITableDescribe() determines the structure of the data type the table function
returns, in situations where this cannot be defined in a static manner.
• ODCITablePrepare() initializes the scan context parameter. If this method is
implemented, the scan context it prepares is passed to the ODCITableStart()
routine, and the context is maintained between restarts of the table function. It also
provides projection information and supports the return of transient anonymous
types.

See Also:

• ODCITableClose()
• ODCITableDescribe()
• ODCITableFetch()
• ODCITablePrepare()
• ODCITableStart()

13.3.6.1 Scan Context

For the fetch method to produce the next set of rows, a table function must be able to
maintain context between successive invocations of the interface routines to fetch
another set of rows. This context, called the scan context, is defined by the attributes
of the implementation type. A table function preserves the scan context by modeling it
in an object instance of the implementation type.

13-8
 Chapter 13
Pipelined Table Functions

13.3.6.2 Start Routine

The start routine ODCITableStart() is the first routine that is invoked to begin
retrieving rows from a table function. This routine typically performs the setup needed
for the scan, creating the scan context (as an object instance sctx) and returning it to
Oracle. However, if ODCITablePrepare() is implemented, it creates the scan context,
which is then passed to the ODCITableStart() routine. The arguments to the table
function, specified by the user in the SELECT statement, are passed in as parameters to
this routine.
Note that any REF CURSOR arguments of a table function must be declared as
SYS_REFCURSOR type in the declaration of the ODCITableStart(). Ordinary REF CURSOR
types cannot be used as formal argument types in ODCITableStart(). Ordinary REF
CURSOR types can only be declared in a package, and types defined in a package
cannot be used as formal argument types in a type method. To use a REF CURSOR type
in ODCITableStart(), you must use the system-defined SYS_REFCURSOR type.

See Also:

• ODCITablePrepare()
• ODCITableStart()

13.3.6.3 Fetch Routine

The fetch routine ODCITableFetch() is invoked one or more times by Oracle to retrieve
all the rows in the table function's result set. The scan context is passed in as a
parameter. This routine returns the next subset of one or more rows.
The fetch routine is called by Oracle repeatedly until all the rows have been returned
by the table function. Returning more rows in each invocation of ODCITableFetch()
reduces the number of fetch calls that must be made and thus improves performance.
The table function should return a null collection to indicate that all rows have been
returned.
The nrows parameter indicates the number of rows that are required to satisfy the
current OCI call. For example, if the current OCI call is an ODCITableFetch() that
requested 100 rows, and 20 rows have been returned, then the nrows parameter is
equal to 80. The fetch function is allowed to return a different number of rows. The
main purpose of this parameter is to prevent ODCITableFetch() from returning more
rows than actually required. If ODCITableFetch() returns more rows than the value of
this parameter, the rows are cached and returned in subsequent ODCITableFetch()
calls, or they are discarded if the OCI statement handle is closed before they are all
fetched.

See Also:
ODCITableFetch()

13-9
 Chapter 13
Pipelined Table Functions

13.3.6.4 Close Routine

The close routine ODCITableClose() is invoked by Oracle after the last fetch
invocation. The scan context is passed in as a parameter. This routine performs the
necessary cleanup operations.

Figure 13-3 Flowchart of Table Function Row Source Execution

ODCITableStart

ODCITableFetch

Yes Is result
null?

No

Process Result

ODCITableClose

13.3.6.5 Describing Returned Data Sructures; Describe Method

Sometimes it is not possible to define the structure of the return type from the table
function statically. If the shape of the rows is different in different queries, it may
depend on the actual arguments with which the table function is invoked. Such table
functions can be declared to return AnyDataSet. AnyDataSet is a generic collection
type. It can be used to model any collection (of any element type) and has an
associated set of APIs (both PL/SQL and C) that enable you to construct AnyDataSet
instances and access the elements.
The following example shows a table function declared to return an AnyDataSet
collection whose structure is not fixed at function creation time:
CREATE FUNCTION AnyDocuments(VARCHAR2) RETURN ANYDATASET
PIPELINED USING DocumentMethods;

You can implement a ODCITableDescribe() routine to determine the format of the

elements in the result collection when the format depends on the actual parameters to
the table function. ODCITableDescribe() is invoked by Oracle at query compilation
time to retrieve the specific type information. Typically, the routine uses the user
arguments to determine the shape of the return rows. The format of elements in the
returned collection is conveyed to Oracle by returning an instance of AnyType.

13-10
 Chapter 13
Pipelined Table Functions

The AnyType instance specifies the actual structure of the returned rows of the specific
query. Like AnyDataSet, AnyType has an associated set of PL/SQL and C interfaces
with which to construct and access the metadata information.
The query in Example 13-11, for an AnyDocuments function, returns information on
either books or magazines.
Example 13-12 is an implementation of the ODCITableDescribe() method, which
consults the DTD of the XML documents at the specified location to return the
appropriate AnyType value, either a book or a magazine. The AnyType instance is
constructed by invoking the constructor APIs with the field name and data type
information.
When Oracle invokes ODCITableDescribe(), it uses the type information that is
returned in the AnyType OUT argument to resolve references in the command line, such
as the reference to the x.Abstract attribute in Example 13-12. This functionality is
applicable only when the returned type is a named type, and therefore has named
attributes.
Another feature of ODCITableDescribe() is its ability to describe SELECT list
parameters, such as using OCI interfaces, when executing a SELECT * query. The
information retrieved reflects one SELECT list item for each top-level attribute of the type
returned by ODCITableDescribe().

Because the ODCITableDescribe() method is called at compile time, the table function
should have at least one argument that has a value at compile time, like a constant. By
using the table function with different arguments, you can get different return types
from the function, as demonstrated in Example 13-13.
The ODCITableDescribe() functionality is available only if the table function is
implemented using the interface approach. A native PL/SQL implementation of a table
function that returns ANYDATASET returns rows whose structure is opaque to the server.

Example 13-11 Querying for AnyType Data

SELECT * FROM
TABLE(AnyDocuments('http://.../documents.xml')) x
WHERE x.Abstract like '%internet%';

Example 13-12 Implementing the ODCITableDescribe() Method

CREATE TYPE Mag_t AS OBJECT
( name VARCHAR2(100),
publisher VARCHAR2(30),
abstract VARCHAR2(1000)
);

STATIC FUNCTION ODCITableDescribe(rtype OUT ANYTYPE,

url VARCHAR2)
IS BEGIN
Contact specified web server and retrieve document...
Check XML doc schema to determine if books or mags...
IF books THEN
rtype=AnyType.AnyTypeGetPersistent('SYS','BOOK_T');
ELSE
rtype=AnyType.AnyTypeGetPersistent('SYS','MAG_T');
END IF;
END;

13-11
 Chapter 13
Pipelined Table Functions

Example 13-13 Using Functions that Return AnyType

-- Issue a query for books
SELECT x.Name, x.Author
FROM TABLE(AnyDocuments('Books.xml')) x;

-- Issue a query for magazines

SELECT x.Name, x.Publisher
FROM TABLE(AnyDocuments('Magazines.xml')) x;

See Also:

• ODCITableDescribe()
• "Transient and Generic Types" for information on AnyDataSet and
AnyType

13.3.6.6 Preparing a Query for Execution; Prepare Method

ODCITablePrepare() is invoked at query compilation time. It generates and saves
information to decrease the execution time of the query.
If you do not implement ODCITablePrepare(), ODCITableStart() initializes the context
each time it is called. However, if you do implement ODCITablePrepare(), it initializes
the scan context, which is passed to the ODCITableStart() when the query is
executed, reducing startup time. In addition, when ODCITablePrepare() is
implemented, ODCITableClose() is called only one time during the query, rather than
each time the table function is restarted. This has the following benefits:
• It decreases execution time by reducing the number of calls to ODCITableClose().
• It allows the scan context to be maintained between table function restarts.
ODCITablePrepare() also provides projection information to the table function. If you
do not implement ODCITablePrepare() for table functions that return collections of
user-defined types (UDTs), your table function must set every attribute of the UDT of
each element, because it has no way of knowing which attributes are used. In
contrast, selecting from a regular table fetches only the required columns, which is
naturally faster in most cases. However, if you do implement ODCITablePrepare(), it
can build an array of attribute positions, record the return type information in an
argument of type ODCITabFuncInfo, and save this information in the scan context, as
described in Example 13-14.
Implementing ODCITablePrepare() also allows your table function to return transient
anonymous types. ODCITablePrepare() is called at the end of query compilation, so it
can be passed the table descriptor object (TDO) built by the describe method. The
describe method can build and return a transient anonymous TDO. Oracle transforms
this TDO so that it can be used during query execution, and passes the transformed
TDO to the prepare method in the RetType attribute. If the describe method returns a
TDO for a type that is not anonymous, that TDO is identical to the transformed TDO.
Thus, if a table function returns:
• A named collection type, the RetType attribute contains the TDO of this type.
• AnyDataSet, and the describe method returns a named type, the RetType attribute
contains the TDO of the named type.

13-12
 Chapter 13
Pipelined Table Functions

• AnyDataSet, and the describe method returns an anonymous type, Oracle

transforms this type, and RetType contains the transformed TDO.
Example 13-14 Building an Array of Attribute Positions and Save it in a Scan
Context
CREATE TYPE SYS.ODCITabFuncInto AS OBJECT (
Attrs SYS.ODCINumberList,
RetType SYS.AnyType
);

See Also:

• ODCITableClose()
• ODCITablePrepare()
• ODCITableStart()

13.3.7 Querying Table Functions

Pipelined table functions are used in the FROM clause of SELECT statements
independently from implementation, either in native PL/SQL or through the interface
approach. The result rows are retrieved by Oracle iteratively from the table function
implementation, as demonstrated in Example 13-15.
However, if the output of a table function is determined solely by the values passed
into it as arguments, such that the function always produces exactly the same result
value for each respective combination of values passed in, you can declare the
function DETERMINISTIC, and Oracle automatically buffers rows for it. Note, though,
that the database has no way of knowing whether a function marked DETERMINISTIC
really is DETERMINISTIC, and if one is not, results are unpredictable.

Example 13-15 Using a Table Function to Iteratively Retrieve Rows

SELECT x.Ticker, x.Price
FROM TABLE(StockPivot(CURSOR(SELECT * FROM StockTable))) x
WHERE x.PriceType='C';

13.3.7.1 Implementing Multiple Calls to Table Functions

Multiple invocations of a table function, either within the same query or in separate
queries result in multiple executions of the underlying implementation. That is, in
general, there is no buffering or reuse of rows, as demonstrated in Example 13-16.
Example 13-16 Using Multiple Invocations of a Table Function
SELECT * FROM TABLE(f(...)) t1, TABLE(f(...)) t2
WHERE t1.id = t2.id;

SELECT * FROM TABLE(f());

SELECT * FROM TABLE(f());

13-13
 Chapter 13
Pipelined Table Functions

13.3.7.2 Using PL/SQL REF CURSOR Variables

PL/SQL REF CURSOR variables can be defined for queries over table functions, as
demonstrated in Example 13-17.
Cursors over table functions have the same fetch semantics as ordinary cursors. REF
CURSOR assignments based on table functions do not have special semantics.

However, the SQL optimizer does not optimize across PL/SQL statements; therefore,
Example 13-19 runs better than Example 13-18.
Additionally, Example 13-18 is slower because of the overhead associated with
executing two SQL statements, and because it does not take advantage of efficiencies
realized by pipelining results between two functions, as Example 13-19 does.
Example 13-17 Defining REF CURSOR Variables for Table Function Queries
OPEN c FOR SELECT * FROM TABLE(f(...));

Example 13-18 Using a REF CURSOR Variable

BEGIN
OPEN r FOR SELECT * FROM TABLE(f(CURSOR(SELECT * FROM tab)));
SELECT * BULK COLLECT INTO rec_tab FROM TABLE(g(r));
END;

Example 13-19 Using a REF CURSOR Variable More Effectively

SELECT * FROM TABLE(g(CURSOR(SELECT * FROM
TABLE(f(CURSOR(SELECT * FROM tab))))));

13.3.8 Performing DML Operations Inside Table Functions

A table function must be declared with the autonomous transaction pragma in order for
the function to execute DML statements. This pragma causes the function to execute
in an autonomous transaction not shared by other processes, as demonstrated in
Example 13-20.
During parallel execution, each instance of the table function creates an independent
transaction.
Example 13-20 Declaring a Table Function with Autonomous Transaction
Pragma
CREATE FUNCTION f(p SYS_REFCURSOR) return CollType PIPELINED IS
PRAGMA AUTONOMOUS_TRANSACTION;
BEGIN ... END;

13.3.9 Performing DML Operations on Table Functions

Table functions cannot be the target table in UPDATE, INSERT, or DELETE statements.
For example, the following statements raise an error:
UPDATE F(CURSOR(SELECT * FROM tab)) SET col = value;
INSERT INTO f(...) VALUES ('any', 'thing');

However, you can create a view over a table function and use INSTEAD OF triggers to
update it, as in Example 13-21.

13-14
 Chapter 13
Parallel Table Functions

Example 13-22 demonstrates how an INSTEAD OF trigger is fired when the user inserts
a row into the BookTable view:.

INSTEAD OF triggers can be defined for all DML operations on a view built on a table
function.
Example 13-21 Creating a View over a Table
CREATE VIEW BookTable AS
SELECT x.Name, x.Author
FROM TABLE(GetBooks('data.txt')) x;

Example 13-22 How an INSTEAD OF Trigger is Fired when a Row is Inserted

into a View
CREATE TRIGGER BookTable_insert
INSTEAD OF INSERT ON BookTable
REFERENCING NEW AS n
FOR EACH ROW
BEGIN
...
END;
INSERT INTO BookTable VALUES (...);

13.3.10 Handling Exceptions in Table Functions

Exception handling in table functions works just as it does with ordinary user-defined
functions.
Some languages, such as C and Java, provide a mechanism for user-supplied
exception handling. If an exception raised within a table function is handled, the table
function executes the exception handler and continues processing. Exiting the
exception handler takes control to the enclosing scope. If the exception is cleared,
execution proceeds normally.
An unhandled exception in a table function causes the parent transaction to roll back.

13.4 Parallel Table Functions

For a table function to be executed in parallel, it must have a partitioned input
parameter. Parallelism is turned on for a table function if, and only if, both the following
conditions are met:
• The function has a PARALLEL_ENABLE clause in its declaration.
• Exactly one REF CURSOR is specified with a PARTITION BY clause.
If the PARTITION BY clause is not specified for any input REF CURSOR as part of the
PARALLEL_ENABLE clause, the SQL compiler cannot determine how to partition the
data correctly.

13.4.1 Inputting Data with Cursor Variables

You can pass a set of rows to a PL/SQL function in a REF CURSOR parameter, as
demonstrated in Example 13-23.

13-15
 Chapter 13
Parallel Table Functions

Results of a subquery can be passed to a function directly, as demonstrated in

Example 13-24. The CURSOR keyword is required to indicate that the results of a
subquery should be passed as a REF CURSOR parameter.

Example 13-23 Passing a Set of Rows to a PL/SQL Function in a REF CURSOR

FUNCTION f(p1 IN SYS_REFCURSOR) RETURN ... ;

Example 13-24 Directly Passing Results from a Subquery to a Function

SELECT * FROM TABLE(f(CURSOR(SELECT empno FROM tab)));

13.4.1.1 Using Multiple REF CURSOR Input Variables

PL/SQL functions can accept multiple REF CURSOR input variables, as demonstrated in
Example 13-25.
Function g can be invoked as demonstrated in Example 13-26.

You can pass table function return values to other table functions by creating a REF
CURSOR that iterates over the returned data, as demonstrated in Example 13-27.

Example 13-25 Passing a Set of Rows to a PL/SQL Function Through REF

CURSOR
CREATE FUNCTION g(p1 pkg.refcur_t1, p2 pkg.refcur_t2) RETURN...
PIPELINED ... ;

Example 13-26 Invoking a Function that Uses Several REF CURSOR

Parameters
SELECT * FROM TABLE(g(CURSOR(SELECT empno FROM tab),
CURSOR(SELECT * FROM emp));

Example 13-27 Using REF CURSOR to Pass Return Values Between Table
Functions
SELECT * FROM TABLE(f(CURSOR(SELECT * FROM TABLE(g(...)))));

13.4.1.2 Explicitly Opening a REF CURSOR for a Query

You can explicitly open a REF CURSOR for a query and pass it as a parameter to a table
function, as demonstrated in Example 13-28.
Example 13-28 Explicitly Using a Query REF CURSOR as Table Function
Parameter
BEGIN
OPEN r FOR SELECT * FROM TABLE(f(...));
-- Must return a single row result set.
SELECT * INTO rec FROM TABLE(g(r));
END;

13.4.1.3 PL/SQL REF CURSOR Arguments to Java and C/C++ Functions

Parallel and pipelined table functions may be written in C/C++, Java, or PL/SQL.
Unlike PL/SQL, C/C++ and Java do not support the REF CURSOR type, but you can still
pass a REF CURSOR argument to C/C++ and Java functions.

13-16
 Chapter 13
Parallel Table Functions

If a table function is implemented as a C callout, then an IN REF CURSOR argument

passed to the callout is automatically available as an executed OCI statement handle.
You can use this handle like any other executed statement handle.
A REF CURSOR argument to a callout passed as an IN OUT parameter is converted to an
executed statement handle on the way in to the callout, and the statement handle is
converted back to a REF CURSOR on the way out. (The inbound and outbound
statement handles may be different.)
If a REF CURSOR type is used as an OUT argument, or a return type to a call, then the
call must return the statement handle, which are converted to a REF CURSOR for the
caller, as demonstrated in Example 13-28.
If the function is written as a Java call, the IN REF CURSOR argument is automatically
converted to an instance of the Java ResultSet class. The IN REF CURSOR to
ResultSet mapping is available only if you use a fat JDBC driver based on OCI. This
mapping is not available for a thin JDBC driver. As with an executed statement handle
in a C call, when a REF CURSOR is either an IN OUT argument, an OUT argument, or a
return type for the function, a Java ResultSet is converted back to a PL/SQL REF
CURSOR on its way out to the caller.

A predefined weak REF CURSOR type, SYS_REFCURSOR, is also supported. With

SYS_REFCURSOR, you do not have to first create a REF CURSOR type in a package before
you can use it. This weak REF CURSOR type can be used in the ODCITableStart()
method, which, as a type method, cannot accept a package type.
To use a strong REF CURSOR type, you still must create a PL/SQL package and declare
a strong REF CURSOR type in it. Also, if you are using a strong REF CURSOR type as an
argument to a table function, then the actual type of the REF CURSOR argument must
match the column type, or an error is generated.
To partition a weak REF CURSOR argument, you must partition by ANY, because a weak
REF CURSOR argument cannot be partitioned by RANGE or HASH. Oracle recommends
that you not use weak REF CURSOR arguments to table functions.

Example 13-29 Using a REF CURSOR in a Call

CREATE OR replace PACKAGE p1 AS
TYPE rc IS REF cursor;
END;

CREATE OR REPLACE LIBRARY MYLIB AS 'mylib.so';

CREATE OR REPLACE FUNCTION MyCallout (stmthp p1.rc)

RETURN binary_integer AS LANGUAGE C LIBRARY MYLIB
WITH CONTEXT
PARAMETERS (context, stmthp ocirefcursor, RETURN sb4);

sb4 MyCallout (OCIExtProcContext *ctx, OCIStmt ** stmthp)

OCIEnv *envhp; /* env. handle */
OCISvcCtx *svchp; /* service handle */
OCIError *errhp; /* error handle */
OCISession *usrhp; /* user handle */

int errnum = 29400; /* choose some oracle error number */

char errmsg[512]; /* error message buffer */
size_t errmsglen; /* Length of error message */
OCIDefine *defn1p = (OCIDefine *) 0;
OCINumber *val=(OCINumber *)0;

13-17
 Chapter 13
Parallel Table Functions

OCINumber *rval = (OCINumber *)0;

sword status = 0;
double num=0;
val = (OCINumber*) OCIExtProcAllocCallMemory(ctx, sizeof(OCINumber));
/* Get OCI handles */
if (GetHandles(ctx, &envhp, &svchp, &errhp, &usrhp,&rval))
return -1;
/* Define the fetch buffer */
psdro_checkerr(NULL, errhp, OCIDefineByPos(*stmthp, &defn1p, errhp, (ub4) 1,
(dvoid *) &num, (sb4) sizeof(num),
SQLT_FLT, (dvoid *) 0, (ub2 *)0,
(ub2 *)0, (ub4) OCI_DEFAULT));

/* Fetch loop */
while ((status = OCIStmtFetch(*stmthp, errhp, (ub4) 1, (ub4) OCI_FETCH_NEXT,
(ub4) OCI_DEFAULT)) == OCI_SUCCESS ||
status == OCI_SUCCESS_WITH_INFO)
{
printf("val=%lf\n",num);
}
return 0;
}

13.4.2 Input Data Partitioning

The table function declaration can specify data partitioning for exactly one REF CURSOR
parameter, as demonstrated in Example 13-30. The PARTITION BY phrase in the
PARALLEL_ENABLE clause specifies which one of the input cursors to partition, and what
columns to use for partitioning.
When explicit column names are specified in the column list, the partitioning method
may be either RANGE or HASH. The input rows are hash- or range-partitioned on the
specified columns.
The ANY keyword enables you to indicate that the function behavior is independent of
the partitioning of the input data. When this keyword is used, the run-time system
randomly partitions the data among the slaves. This keyword is appropriate for use
with functions that take in one row, manipulate its columns, and generate output row(s)
based on the columns of this row only.
To demonstrate, the pivot-like function StockPivot() in Example 13-31 takes as input
a row of the type (Ticker varchar(4), OpenPrice number, ClosePrice number),
and generates rows of the type (Ticker varchar(4), PriceType varchar(1), Price
number). Thus, the row ("ORCL", 41, 42) generates two rows: ("ORCL", "O", 41) and
("ORCL", "C", 42).

You can use directly TickerTypeSet, as created in Example 13-5, instead of

rec_tab_type or define it the same way.

The function f() may be used to generate another table from Stocks table, as shown
in Example 13-32.
If StockTable is scanned in parallel and partitioned on OpenPrice, then the function
StockPivot() is combined with the data-flow operator that scans StockTable and
therefore sees the same partitioning.

13-18
 Chapter 13
Parallel Table Functions

If StockTable is not partitioned, and the scan on it does not execute in parallel, the
insert into AlternateStockTable also runs sequentially, as demonstrated in
Example 13-33.
If function g() runs in parallel and is partitioned by ANY, then the parallel insert can
belong in the same data-flow operator as g().

Whenever the ANY keyword is specified, the data is partitioned randomly among the
slaves. This effectively means that the function is executed in the same slave set
which does the scan associated with the input parameter.
No redistribution or repartitioning of the data is required here. In the case, when the
cursor p itself is not parallel-enabled, the incoming data is randomly partitioned on the
columns in the column list. The round-robin table queue is used for this partitioning.
If you create an XMLIndex index on a partitioned XMLType table, or a partitioned table
with an XMLType column using range, list, or hash partitioning, that index is
equipartitioned with the base table.
Example 13-30 Specifying Data Partitioning for a REF CURSOR Parameter
CREATE FUNCTION f(p ref_cursor_type) RETURN rec_tab_type PIPELINED
PARALLEL_ENABLE(PARTITION p BY [{HASH | RANGE} (column_list) | ANY ]) IS
BEGIN ... END;

Example 13-31 Implementing the StockPivot() Function

CREATE FUNCTION StockPivot(p refcur_pkg.refcur_t) RETURN rec_tab_type PIPELINED
PARALLEL_ENABLE(PARTITION p BY ANY) IS
ret_rec rec_type;

BEGIN
FOR rec IN p LOOP
ret_rec.Ticker := rec.Ticker;
ret_rec.PriceType := "O";
ret_rec.Price := rec.OpenPrice;
PIPE ROW(ret_rec);

ret_rec.Ticker := rec.Ticker; -- Redundant; not required

ret_rec.PriceType := "C";
ret_rec.Price := rec.ClosePrice;
PIPE ROW ret_rec;
END LOOP;
RETURN;

END;

Example 13-32 Using a REF CURSOR to Generate a Table from Another Table
INSERT INTO AlternateStockTable
SELECT * FROM
TABLE(StockPivot(CURSOR(SELECT * FROM StockTable)));

Example 13-33 Using a REF CURSOR to Scan and Insert

CREATE FUNCTION g(p refcur_pkg.refcur_t) RETURN ... PIPELINED
PARALLEL_ENABLE (PARTITION p BY ANY)
BEGIN
...
END;

13-19
 Chapter 13
Input Data Streaming for Table Functions

INSERT INTO AlternateStockTable

SELECT * FROM TABLE(f(CURSOR(SELECT * FROM Stocks))), TABLE(g(CURSOR( ... )))
WHERE join_condition;

13.4.3 Parallel Execution of Leaf-Level Table Functions

Parallel execution of a table function requires a REF CURSOR as an input parameter. If
the table function does not inherently require a set of rows as input, then you must
redesign the function to require a REF CURSOR as input.

For example, the following function reads a set of external files in parallel and returns
the records they contain. To provide work for a REF CURSOR, you might first create a
table and populate it with the filenames. A REF CURSOR over this table can then be
passed as a parameter to the table function readfiles(), as demonstrated by
Example 13-34.
Example 13-34 Using a REF CURSOR to Read a Set of External FIles
CREATE TABLE filetab(filename VARCHAR(20));

INSERT INTO filetab VALUES('file0');

INSERT INTO filetab VALUES('file1');
...
INSERT INTO filetab VALUES('fileN');

SELECT * FROM TABLE(readfiles(CURSOR(SELECT filename FROM filetab)));

CREATE FUNCTION readfiles(p pkg.rc_t) RETURN coll_type

PARALLEL_ENABLE(PARTITION p BY ANY) IS
ret_rec rec_type;
BEGIN
FOR rec IN p LOOP
done := FALSE;
WHILE (done = FALSE) LOOP
done := readfilerecord(rec.filename, ret_rec);
PIPE ROW(ret_rec);
END LOOP;
END LOOP;
RETURN;
END;

13.5 Input Data Streaming for Table Functions

Data streaming is the manner in which a table function orders or clusters rows that it
fetches from cursor arguments. A function can stream its input data in any of the
following ways:
• Place no restriction on the ordering of the incoming rows.
• Order them on a particular key column or columns.
• Cluster them on a particular key.
To control the behavior of the input stream, use the syntax in Example 13-36.
Input streaming may be specified for either sequential or parallel execution of a
function. If an ORDER BY or CLUSTER BY clause is not specified, rows are input in a
random order. The semantics of ORDER BY are different for parallel execution from the
semantics of the ORDER BY clause in a SQL statement. In a SQL statement, the ORDER

13-20
 Chapter 13
Input Data Streaming for Table Functions

BY clause globally orders the entire data set. In a table function, the ORDER BY clause
orders the respective rows local to each instance of the table function running on a
slave.
Clustering causes rows that have the same key values to appear next to one another,
but it does not otherwise do any ordering of rows.
expr should be the REFCURSOR received in the function.

Example 13-35 illustrates the syntax for ordering the input stream. In the example,
function f() takes in rows of the kind (Region, Sales) and returns rows of the form
(Region, AvgSales), showing average sales for each region.

13.5.1 Setting up the Input Stream

Example 13-35 Ordering the Input Stream
CREATE FUNCTION f(p ref_cursor_type) RETURN tab_rec_type PIPELINED
CLUSTER p BY (Region)
PARALLEL_ENABLE(PARTITION p BY HASH(Region)) IS
ret_rec rec_type;
cnt number;
sum number;
BEGIN
LOOP
FETCH p INTO rec;
EXIT WHEN p%NOTFOUND;
IF (first rec in the group) THEN
cnt := 1;
sum := rec.Sales;
ELSIF (last rec in the group) THEN
IF (cnt <> 0) THEN
ret_rec.Region := rec.Region;
ret_rec.AvgSales := sum/cnt;
PIPE ROW(ret_rec);
END IF;
ELSE
cnt := cnt + 1;
sum := sum + rec.Sales;
END IF;
END LOOP;
RETURN;
END

Example 13-36 Controlling Input Data Streaming

FUNCTION f(p ref_cursor_type) RETURN tab_rec_type [PIPELINED]
{[ORDER | CLUSTER] expr BY column_list}
PARALLEL_ENABLE({PARTITION p BY
[ANY | {HASH | RANGE} column_list]} )
IS
BEGIN
...
END;

13.5.2 Parallel Execution: Partitioning and Clustering

Partitioning and clustering are easily confused, but they do different things. Sometimes
partitioning can be sufficient without clustering in parallel execution.

13-21
 Chapter 13
Creating Domain Indexes in Parallel

Consider a function SmallAggr that performs in-memory aggregation of salary for each
department_id, where department_id can be either 1, 2, or 3. The input rows to the
function can be partitioned by HASH on department_id so that all rows with
department_id equal to 1 go to one slave, all rows with department_id equal to 2 go
to another slave, and so on.
The input rows do not have to be clustered on department_id to perform the
aggregation in the function. Each slave could have a 1 by 3 array SmallSum[1..3], in
which the aggregate sum for each department_id is added in memory into
SmallSum[department_id]. On the other hand, if the number of unique values of
department_id were very large, you would want to use clustering to compute
department aggregates and write them to disk one department_id at a time.

13.6 Creating Domain Indexes in Parallel

Creating a domain index can be a lengthy process because of the large amount of
data that a domain index typically handles. You can exploit the parallel-processing
capabilities of table functions to alleviate this bottleneck by using table functions to
create domain indexes in parallel.
Typically, the ODCIIndexCreate() routine performs the following steps:

1. Creates tables for storing the index data

2. Fetches the relevant data, such as keycols and rowid, from the base table,
transforms it, and inserts relevant transformed data into the table created for
storing the index data.
3. Builds secondary indexes on the tables that store the index data, for faster access
at query time.
Step 2 is the bottleneck in creating domain indexes. You can speed up this step by
encapsulating these operations in a parallel table function and invoking the function
from the ODCIIndexCreate() function. In Example 13-37, a table function IndexLoad()
is defined to do just that.
where p is a cursor of the form:
SELECT /*+ PARALLEL (base_table, par_degree) */ keycols ,rowid
FROM base_table

The par_degree value can be explicitly specified; otherwise, it is derived from the
parallel degree of the base table.
The function IndexMerge(), defined in Example 13-38, is needed to merge the results
from the several instances of IndexLoad().

The new steps in ODCIIndexCreate() would be:

1. Create metadata structures for the index (tables to store the index data).
2. Explicitly commit the transaction so that the IndexLoad() function can access the
committed data.
3. Invoke IndexLoad() in parallel, as shown in the following code example.
Invoking the Merging of Parallel Domain Index Loads

13-22
 Chapter 13
Transient and Generic Types

status := ODCIIndexMerge(CURSOR(
SELECT * FROM TABLE(ODCIIndexLoad(ia, parms, CURSOR(
SELECT key_cols, ROWID FROM basetable)))))
4. Create secondary index structures.

See Also:
ODCIIndexCreate()

13.6.1 Loading Domain Indexes

The formation of secondary indexes is very time consuming, and Oracle recommends
that you implement parallel loading of domain index, as illustrated in the “Loading a
Domain Index in Parallel” section, and subsequently recombine them as shown in
“Merging the Results from Parallel Domain Index Loads” example.
Example 13-37 Loading a Domain Index in Parallel
CREATE FUNCTION IndexLoad(ia ODCIIndexInfo, parms VARCHAR2,
p refcur-type)
RETURN status_code_type
PARALLEL_ENABLE(PARTITION p BY ANY)
PRAGMA AUTONOMOUS_TRANSACTION
IS
BEGIN
FOR rec IN p LOOP
- process each rec and determine the index entry
- derive name of index storage table from parameter ia
- insert into table created in ODCIIndexCreate
END LOOP;
COMMIT; -- explicitly commit the autonomous txn
RETURN ODCIConst.Success;
END;

Example 13-38 Merging the Results from Parallel Domain Index Loads
CREATE FUNCTION IndexMerge(p refcur-type)
RETURN NUMBER
IS
BEGIN
FOR rec IN p LOOP
IF (rec != ODCIConst.Success)
RETURN Error;
END LOOP;
RETURN Success;
END;

13.7 Transient and Generic Types

Oracle has three special SQL data types that enable you to dynamically encapsulate
and access type descriptions, data instances, and sets of data instances of any other
SQL type, including object and collection types. You can also use these three special
types to create anonymous, or unnamed, types, including anonymous collection types.
See Table 13-1.

13-23
 Chapter 13
Transient and Generic Types

The three SQL types are implemented as opaque types; the internal structure of these
types is not known to the database: their data can be queried only by implementing
functions, typically 3GL routines. Oracle provides both an OCI and a PL/SQL API for
implementing such functions.

Table 13-1 Generic SQL Types

Type Description
SYS.ANYTYPE A type description type. A SYS.ANYTYPE can contain a type
description of any SQL type, named or unnamed, including object
types and collection types.
An ANYTYPE can contain a type description of a persistent type,
but an ANYTYPE itself is transient: the value in an ANYTYPE itself is
not automatically stored in the database. To create a persistent
type, use a CREATE TYPE statement from SQL.
SYS.ANYDATA A self-describing data instance type. A SYS.ANYDATA contains an
instance of a given type, with data, plus a description of the type.
In this sense, a SYS.ANYDATA is self-describing. An ANYDATA can
be persistently stored in the database.
SYS.ANYDATASET A self-describing data set type. A SYS.ANYDATASET type contains
a description of a given type plus a set of data instances of that
type. An ANYDATASET can be persistently stored in the database.

Each of these three types can be used with any built-in type native to the database
with object types and collection types, both named and unnamed. The types provide a
generic way to work dynamically with type descriptions, lone instances, and sets of
instances of other types. Using the APIs, you can create a transient ANYTYPE
description of any kind of type. Similarly, you can create or convert (cast) a data value
of any SQL type to an ANYDATA and can convert an ANYDATA (back) to a SQL type. And
similarly again with sets of values and ANYDATASET.

The generic types simplify working with stored procedures. You can use the generic
types to encapsulate descriptions and data of standard types and pass the
encapsulated information into parameters of the generic types. In the body of the
procedure, you can detail how to handle the encapsulated data and type descriptions
of whatever type.
You can also store encapsulated data of a variety of underlying types in one table
column of type ANYDATA or ANYDATASET. For example, you can use ANYDATA with
advanced queuing to model queues of heterogeneous types of data. You can query
the data of the underlying data types like any other data.
Note, however, that ANYDATA and ANYDATASET objects that were created on a transient
type, such as unnamed type constructed through an ANYTYPE API, cannot be stored
persistently in an ANYDATA or ANYDATASET table column.

Corresponding to the three generic SQL types are three OCI types that model them.
Each has a set of functions for creating and accessing the respective type:
• OCIType, corresponding to SYS.ANYTYPE
• OCIAnyData, corresponding to SYS.ANYDATA
• OCIAnyDataSet, corresponding to SYS.ANYDATASET

13-24
 Chapter 13
Transient and Generic Types

See Also:

• Oracle Call Interface Programmer's Guide for the OCIType, OCIAnyData,

and OCIAnyDataSet APIs and details on how to use them
• Oracle Database PL/SQL Packages and Types Reference for
information about the interfaces to the ANYTYPE, ANYDATA, and
ANYDATASET types and about the DBMS_TYPES package, which defines
constants for built-in and user-defined types, for use with ANYTYPE,
ANYDATA, and ANYDATASET
• Oracle Database Object-Relational Developer's Guide for definition of
generic, transient, and opaque types.

13-25
14
Designing Data Cartridges
There are various items for you to consider when designing data cartridges.

14.1 Choosing the Programming Language

You can implement methods for object types in PL/SQL, C/C++, or Java. PL/SQL and
Java methods run in the address space of the server. C/C++ methods are dispatched
as external procedures and run outside the address space of the server.
The best implementation choice depends on the situation. Here are some guidelines:
• A callout involving C or C++ is generally fastest if the processing is substantially
CPU-bound. However, callouts incur the cost of dispatch, which might be
important for small amounts of processing in C/C++.
• PL/SQL is most efficient for methods that are not computation-intensive. The other
implementation options are typically favored over PL/SQL if you have a large body
of code implemented in another language, and it can be used by the data
cartridge, or if you must perform extensive computations.
• Java is a relatively open implementation choice. Although Java is usually
interpreted, high-performance applications might benefit from pre-compilation of
methods or just-in-time compilers.

14.2 Invoker's Rights

The invoker's rights mechanism lets a function execute with the privileges of the
invoker. Thus, a cartridge can live within a schema dedicated to it, which can be used
by other schemas without privileges for operating on objects in the schema where the
cartridge resides.

14.3 Callouts and LOBs

When using LOBs with callouts, consider the following:

• It can be to your advantage to code your callout so that it is independent of LOB

types (BFILE/BLOB).
• The PL/SQL layer of your cartridge can open your BFILE so that no BFILE-specific
logic is required in your callout (other than error recovery from OCILob calls that do
not operate on BFILEs).
• With the advent of temporary LOBs, you must be aware of the deep copy that can
occur when assignments and calls are done with temporary LOBs. Use NOCOPY (BY
REFERENCE) on BLOB parameters as appropriate.

14-1
 Chapter 14
Saving and Passing State

14.4 Saving and Passing State

Traditionally, external procedures have a state-less model. All statement handles
opened during the invocation of an external procedure are closed implicitly at the end
of the call.
Oracle Database allows state information, such as OCI statement handles and
associated state in the database, to be saved and used across invocations of external
procedures in a session. By default, cartridges are stateless; however, you can use
OCIMemory services and OCIContext services with OCI_DURATION_SESSION or other
appropriate duration to save state. Statement handles created in one external
procedure invocation can be re-used in another. As the data cartridge developer, you
must explicitly free these handles. Oracle recommends that you do this as soon as the
statement handle is no longer needed. All state maintained for the statement in the
OCI handles and in the database is freed as a result. This helps to improve the
scalability of your data cartridge.

See Also:
Oracle Database PL/SQL Language Reference

14.5 Designing Indexes

Consider some factors that guide optimal design of indexes for your data cartridge.

14.5.1 Domain Index Performance

Creating a domain index is not always the best course. If you decide to create a
domain index, keep the following factors in mind:
• For complex domain indexes, the functional implementation works better with
small data size and when results are a large percentage of the total data size.
• Judicious use of the extensible optimizer can improve performance.

14.5.2 Domain Index Component Names

Naming internal components for a domain index implementation can be an issue.
Names of internal data objects are typically based on names you provide for table and
indexes. The problem is that the derived names for the internal objects must not
conflict with any other user-defined object or system object. To avoid this problem,
develop some policy that restricts names, or implement some metadata management
scheme to avoid errors during DROP, CREATE, and so on.

14.5.3 When to Use Index-Organized Tables

You can create secondary indexes on IOT because using them is more efficient than
storing data in a table and a separate index, particularly if most of your data is in the
index. This offers a big advantage if you are accessing the data in multiple ways. Note
that before the Oracle9i release, you could create only one index on IOTs.

14-2
 Chapter 14
Designing Indexes

14.5.4 Storing Index Structures in LOBs

Index structures can be stored in LOBs, but take care to tune the LOB for best
performance. If you are accessing a particular LOB frequently, create your table with
the CACHE option and place the LOB index in a separate tablespace. If you are updating
a LOB frequently, TURN OFF LOGGING and read/write in multiples of CHUNK size. If you are
accessing a particular portion of a LOB frequently, buffer your reads/writes using LOB
buffering or your own buffering scheme.

14.5.5 External Index Structures

With the extensible indexing framework, the meaning and representation of a user-
defined index is left to the cartridge developer. Oracle provides basic index
implementations such as IOTs. In certain cases, binary or character LOBs can also be
used to store complex index structures. IOTs, BLOBs and CLOBs all live within the
database. In addition to them, you may also store a user-defined index as a structure
external to the database, for example in a BFILE.

The external index structure gives you the most flexibility in representing your index.
An external index structure is particularly useful if you have invested in the
development of in-memory indexing structures. For example, an operating system file
may store index data, which is read into a memory mapped file at run time. Such a
case can be handled as a BFILE in the external index routines.

External index structures may also provide superior performance, although this gain
comes at some cost. Index structures external to the database do not participate in the
transaction semantics of the database, which, in the case of index structures inside the
database, make data and concomitant index updates atomic. This means that if an
update to the data causes an update for the external index to be invoked through the
extensible indexing interface, failures can cause the data updates to be rolled back but
not the index updates. The database can only roll back what is internal to it: external
index structures cannot be rolled back in synchronization with a database rollback.
External index structures are perhaps most useful for read-only access. Their
semantics become complex if updates to data are involved.

14.5.6 Multi-Row Fetch

When the ODCIIndexFetch() routine is called, the rowids of all the rows that satisfy the
operator predicate are returned. The maximum number of rows that can be returned
by the ODCIIndexFetch() routine is nrows (nrows being an argument to the
ODCIIndexFetch() routine). The value of nrows is decided by Oracle based on some
internal factors. If you have a better idea of the number of rows that ought to be
returned to achieve optimal query performance, you can determine that this number of
rows is returned in the ODCIRidList VARRAY instead of nrows. Note that the number of
values in the ODCIRidList must be less than or equal to nrows.

As the cartridge designer, you are in the best position to make a judgement regarding
the number of rows to be returned. For example, if in the index 1500 rowids are stored
together, and nrows = 2000, then it may be optimal to return 1500 rows instead of 2000
rows. Otherwise, the user would have to retrieve 3000 rowids, return 2000 of them, and
note which 1000 rowids were not returned.

14-3
 Chapter 14
Designing Operators

If you do not have any specific optimization in mind, you can use the value of nrows to
determine the number of rows to be returned. Currently the value of nrows has been
set to 2000.

If you implement indexes that use callouts, use multirow fetch to fetch the largest
number of rows back to the server. This offsets the cost of making the callout.

See Also:
ODCIIndexFetch()

14.6 Designing Operators

All domain indexes should contain both indexed and functional implementations of
operators, in case the optimizer chooses not to use the indexed implementation. You
can, however, use the indexing structures to produce the functional result.

14.7 Designing for the Extensible Optimizer

Data cartridges can be more efficient if they are designed with the extensible optimizer
in mind.

14.7.1 Weighing Cost and Selectivity

When estimating cost, Oracle considers the costs associated with CPU, I/O, and
Network.

14.7.2 Cost for functions

You can determine the cost of executing a C function using common profilers or tools.
For SQL queries, an explain plan of the query gives a rough estimate of the cost of the
query. In addition, the tkprof utility helps you gather information about the CPU and
the I/O cost involved in the operation. You can also determine the cost of executing a
callout by using it in a SQL query which "selects from dual" and then estimating its cost
using tkprof.

14.7.2.1 Selectivity for Functions

The selectivity of a predicate is the number of rows returned by the predicate divided
by the total number of rows in the tables. Selectivity refers to the fraction of rows of the
table returned by the predicate.
The selectivity function should use the statistics collected for the table to determine
what percentage of rows of the table the predicate returns with the given list of
arguments. For example, to compute the selectivity of a predicate
IMAGE_GREATER_THAN (Image SelectedImage) which determines the images that are
greater than the Image SelectedImage, you might use a histogram of the sizes of the
images in the database to compute the selectivity.
Statistics can affect the calculation of selectivity for predicates and the cost of domain
indexes.

14-4
 Chapter 14
Designing for Maintenance

14.7.2.2 Statistics for Tables

The statistics collected for a table can affect the computation of selectivity of a
predicate. Thus, statistics that help the user make a better judgement about the
selectivity of a predicate should be collected for tables and columns. Knowing the
predicates that can operate on the data is helpful in determining what statistics to
collect.
For example, in a spatial domain the average, minimum, and maximum number of
elements in a VARRAY that contains the nodes of the spatial objects is a useful statistic
to collect.

14.7.2.3 Statistics for Indexes

When a domain index is analyzed, statistics for the underlying objects that constitute
the domain index should be analyzed. For example, if the domain index is composed
of tables, the statistics collection function should analyze the tables when the domain
index is analyzed. The cost of accessing the domain index can be influenced by the
statistics that have been collected for the index. For instance, the cost of accessing a
domain index could be approximated as the selectivity times the total number of data
blocks in the various tables being accessed when the domain index is accessed.
To define cost, selectivity and statistics functions accurately requires a good
understanding of the domain. The preceding guidelines are meant to help you
understand some issues you must take into account while working on the cost,
selectivity and statistics functions. In general it may be a good idea to start by using
the default cost and selectivity, and observing how queries of interest behave.

14.8 Designing for Maintenance

When you design a data cartridge, keep in mind the issues regarding maintenance.
In particular, if your cartridge maintains a large number of objects, views, tables, and
so on, consider making a metadata table to maintain the relationships among the
objects for the user. This reduces the complexity of developing and maintaining the
cartridge when it is in use.

14.9 Enabling Cartridge Installation

• Include a README with your cartridge to tell users how to install the cartridge.
• Make the cartridge installable in one step in the database, if possible, such as in
sqlplus @imginst.
• Tell users how to start the listener if you are using callouts.
• Tell users how to setup extproc. Most users have never heard of extproc and
many users have never set up a listener. This is the primary problem when
deploying cartridges.
• With the Oracle Software Packager, you can easily create custom SQL install
scripts using the instantiate_file action. This feature lets you substitute
variables in your files when they are installed and it leaves your user with scripts
and files that are customized for their installation.

14-5
 Chapter 14
Designing for Portability

See Also:
Oracle Database Advanced Application Developer's Guide for information on
setting up the listener and extproc

14.10 Designing for Portability

To make your data cartridge more portable, consider the following:
• Use the data types in oratypes.h.
• Use OCI calls where ever possible.
• Use the switches that enforce ANSI C conformance when possible.
• Use ANSI C function prototypes.
• Build and test on your target platforms as early in your development cycle as
possible. This helps you locate platform-specific code and provides the maximum
amount of time to redesign.
Portability is reduced by:
• Storing endian (big/little) specific data
• Storing floating point data (IEEE/VAX/other)
• Operating system-specific calls (if you must use them, isolate them in a layer
specific to the operating system; however, if the calls you require are not in the
OCI, and also are not in POSIX, then you are likely to encounter intractable
problems)
• Implicitly casting int as size_t on a 64-bit platform

14-6
Part III
Scenarios and Examples
Consider the following examples that illustrate the techniques described in Building
Data Cartridges:
• Power Demand Cartridge Example
• PSBTREE: Extensible Indexing Example
• Pipelined Table Functions: Interface Approach Example
15
Power Demand Cartridge Example
The power demand cartridge in this example includes a user-defined object type,
extensible indexing, and optimization. The entire cartridge definition is available online
in file extdemo1.sql in the Oracle demo directory.

See Also:

• Building Domain Indexes for information about extensible query

optimization
• Using Extensible Optimizer for information about extensible indexing
• Using Cartridge Services for information about cartridge services

15.1 Feature Requirements

A power utility, Power-To-The-People, develops a sophisticated model to decide how
to deploy its resources. The region served by the utility is represented by a grid laid
over a geographic area. This grid is illustrated in Figure 15-1.

Figure 15-1 Region Served by the Power Utility

Vicksburg Clinton

15-1
 Chapter 15
Feature Requirements

This region may be surrounded by other regions some of whose power needs are
supplied by other utilities. As pictured, every region is composed of geographic
quadrants, called cells, on a 10x10 grid. There are several ways of identifying cells —
by spatial coordinates (longitude/latitude), by a matrix numbering (1,1; 1,2;...), and by
numbering them sequentially, as illustrated in Figure 15-2.

Figure 15-2 Regional Grid Cells in Numbered Sequence

1 2 3 4 5 6 7 8 9 10

11 12 13 14 15 16 17 18 19 20

21 22 23 24 25 26 27 28 29 30

31 32 33 34 35 36 37 38 39 40

41 42 43 44 45 46 47 48 49 50

51 52 53 54 55 56 57 58 59 60

61 62 63 64 65 66 67 68 69 70

71 72 73 74 75 76 77 78 79 80

81 82 83 84 85 86 87 88 89 90

91 92 93 94 95 96 97 98 99 100

Within the area represented by each cell, the power used by consumers in that area is
recorded each hour. For example, the power demand readings for a particular hour
might be represented by Table 15-1 (cells here represented on a matrix).

Table 15-1 Sample Power Demand Readings for an Hour

- 1 2 3 4 5 6 7 8 9 10
1 23 21 25 23 24 25 27 32 31 30
2 33 32 31 33 34 32 23 22 21 34
3 45 44 43 33 44 43 42 41 45 46
4 44 45 45 43 42 26 19 44 33 43
5 45 44 43 42 41 44 45 46 47 44
6 43 45 98 55 54 43 44 33 34 44
7 33 45 44 43 33 44 34 55 46 34
8 87 34 33 32 31 34 35 38 33 39
9 30 40 43 42 33 43 34 32 34 46

15-2
 Chapter 15
Feature Requirements

Table 15-1 (Cont.) Sample Power Demand Readings for an Hour

- 1 2 3 4 5 6 7 8 9 10
10 43 42 34 12 43 45 48 45 43 32

The power stations also receives reports from two other sources:
• Sensors on the ground provide temperature readings for every cell
By analyzing the correlation between historical power demand from cells and the
temperature readings for those regions, the utility is able to determine with a close
approximation the expected demand, given specific temperatures.
• Satellite cameras provide images regarding current conditions that are converted
into grayscale images that match the grid illustrated in Figure 15-3.

Figure 15-3 Grayscale Representation of Satellite Image

These images are designed so that lighter is colder. Thus, the image shows a cold
front moving into the region from the south-west. By correlating the data provided by
the grayscale images with temperature readings taken at the same time, the utility has
been able to determine what the power demand is given weather conditions viewed
from the stratosphere.
The reason that this is important is that a crucial part of this modeling has to do with
noting the rapidity and degree of change in the incoming reports as weather changes
and power is deployed. The following diagram shows same cold front at a second
recording, illustrated in Figure 15-4.

15-3
 Chapter 15
Feature Requirements

Figure 15-4 Grayscale Representation of Weather Conditions at Second

Recording

By analyzing the extent and speed of the cold front, the utility is able to project what
the conditions are likely to be in the short and medium term, as in Figure 15-5.

15-4
 Chapter 15
Feature Requirements

Figure 15-5 Grayscale Representation of Conditions as Projected

By combing the data about these conditions and other anomalous situations (such as
the failure of a substation), the utility must be able to organize the most optimal
deployment of its resources. Figure 15-6 reflects the distribution of substations across
the region.

15-5
 Chapter 15
Feature Requirements

Figure 15-6 Distribution of Power Stations Across the Region

The distribution of power stations means that the utility can redirect its deployment of
electricity to the areas of greatest need. Figure 15-7 gives a pictorial representation of
the overlap between three stations.

15-6
 Chapter 15
Modeling the Application

Figure 15-7 Areas Served by Three Power Stations

Depending on fluctuating requirements, the utility must be able to decide how to

deploy its resources, and even whether to purchase power from another utility in the
event of shortfall.

15.2 Modeling the Application

Consider a technical and business scenario for modeling an application. The Class
Diagram in Figure 15-8 describes the application objects using the Unified Modelling
Language (UML) notation.

15-7
 Chapter 15
Modeling the Application

Figure 15-8 Application Object Model of the Power Demand Cartridge

retrieves, stores Hourly Demand Status

Power Cartridge User
1
* Date
Time
TotalGridDemand
MaxCellDemand
Regional Grid is associated with MinCellDemand
Grid No 1
* calcTotalGridDemand
getMaxCellDemand
getMinCellDemand
isEqualToSpecificCell
isEqualToAnyCell
Locator
1
NW 1

has
NE
SW
SE 100
Cell Demand Reading
CellNo
Demand

has
1

for
demand
reads
1
1
Sensor 1 senses 1 Cell Temperature Reading Cell Meter
has
Temperature 1 1 CellNo
1 1
reads
temperature 1 1
for

has
2
Camera photographs Satellite Image 1
Grid Coordinate
GreyScaleValue provides
1 matching x
* image y

15.2.1 Sample Queries

Modelling the application in this way makes it possible the following specific queries:
• Find the cell (geographic quadrant) with the highest demand for a specified time-
period.
• Find the time-period with the highest total demand.
• Find all cells where demand is greater than some specified value.
• Find any cell at any time where the demand equals some specified value.
• Find any time period for which 3 or more cells have a demand greater than some
specified.
• Find the time-period for which there was the greatest disparity (difference)
between the cell with the minimum demand and the cell with the maximum
demand.
• Find the times for which 10 or more cells had demand not less than some
specified value.
• Find the times for which the average cell demand was greater than some specified
value.
Note that it is assumed that the average is easily computable through
TotalPowerDemand/100.

15-8
 Chapter 15
Modeling the Application

• Find the time-periods for which the median cell demand was greater than some
specified value.
Note that we assume that the median value is not easily computable.
• Find all time-periods for which the total demand rose 10 percent or more over the
preceding time's total demand.
These queries are, of course, only a short list of the possible information that could be
gleaned from the system. For instance, it is obvious that the developer of such an
application would want to build queries that are based on the information derived from
prior queries:
• What is the percentage change in demand for a particular cell as compared to a
previous time-period?
• Which cells demonstrate rapid increase or decrease in demand measured as
percentages that are greater or less than specified values?
Figure 15-9 describes and illustrates the Power Demand cartridge, as implemented.

Figure 15-9 Implementation Model of the Power Demand Cartridge

retrieves, stores Hourly Demand Status

Power Cartridge User
1
* Date
Time
TotalGridDemand
MaxCellDemand
Regional Grid is associated with MinCellDemand
Grid No 1
* calcTotalGridDemand
getMaxCellDemand
getMinCellDemand
isEqualToSpecificCell
isEqualToAnyCell
Locator
1
NW 1
has

NE
SW
SE 100
Cell Demand Reading
CellNo
Demand
has

1
for
demand
reads

Cell Meter
1 has
CellNo

1
has

2
Grid Coordinate
x
y

The utility receives ongoing reports from weather centers about current conditions and
from power stations about ongoing power utilization for specific geographical areas

15-9
 Chapter 15
Queries and Extensible Indexing

(represented by cells on a 10x10 grid). It then compares this information to historical

data so it may predict demand for power in the different geographic areas for given
time periods.
Each service area for the utility is considered as a 10x10 grid of cells, where each
cell's boundaries are associated with spatial coordinates (longitude/latitude). The
geographical areas represented by the cells can be uniform or can have different
shapes and sizes. Within the area represented by each cell, the power used by
consumers in that area is recorded each hour. For example, the power demand
readings for a particular hour might be represented by Table 15-2.

Table 15-2 Sample Power Demand Readings for an Hour

- 1 2 3 4 5 6 7 8 9 10
1 23 21 25 23 24 25 27 32 31 30
2 33 32 31 33 34 32 23 22 21 34
3 45 44 43 33 44 43 42 41 45 46
4 44 45 45 43 42 26 19 44 33 43
5 45 44 43 42 41 44 45 46 47 44
6 43 45 98 55 54 43 44 33 34 44
7 33 45 44 43 33 44 34 55 46 34
8 87 34 33 32 31 34 35 38 33 39
9 30 40 43 42 33 43 34 32 34 46
10 43 42 34 12 43 45 48 45 43 32

The numbers in each cell reflect power demand (in some unit of measurement
determined by the electric utility) for the hour for that area. For example, the demand
for the first cell (1,1) was 23, the demand for the second cell (1,2) was 21, and so on.
The demand for the last cell (10, 10) was 32.
The utility uses this data for many monitoring and analytical applications. Readings for
individual cells are monitored for unusual surges or decreases in demand. For
example, the readings of 98 for (6,3) and 87 for (8,1) might be unusually high, and the
readings of 19 for (4,7) and 12 for (10,4) might be unusually low. Trends are also
analyzed, such as significant increases or decreases in demand for each
neighborhood, for each station, and overall, over time.

15.3 Queries and Extensible Indexing

Consider the kinds of queries that benefit from domain indexes. The choice to use
extensible indexing depends on whether queries run as efficiently with a standard
Oracle index, or with no index at all.

15.3.1 Queries Not Benefiting from Extensible Indexing

A query does not require a domain index if both of the following are true:
• The desired information can be made an attribute (column) of the table and a
standard index can be defined on that column.

15-10
 Chapter 15
Queries and Extensible Indexing

• The operations in queries on the data are limited to those operations supported by
the standard index, such as equals, lessthan, greaterthan, max, and min for a b-
tree index.
In the PowerDemand_Typ object type cartridge example, the values for three columns
(TotGridDemand, MaxCellDemand, and MinCellDemand) are set by functions, after which
the values do not change. (For example, the total grid power demand for 13:00 on 01-
Jan-1998 does not change after it has been computed.) For queries that use these
columns, a standard b-tree index on each column is sufficient and recommended for
operations like equals, lessthan, greaterthan, max, and min.

Examples of queries that would not benefit from extensible indexing (using the power
demand cartridge) include:
• Find the cell with the highest power demand for a specific time.
• Find the time when the total grid power demand was highest.
• Find all cells where the power demand is greater than a specified value.
• Find the times for which the average cell demand or the median cell demand was
greater than a specified value.
To make this query run efficiently, define two additional columns in the
PowerDemand_Typ object type (AverageCellDemand and MedianCellDemand), and
create functions to set the values of these columns. (For example,
AverageCellDemand is TotGridDemand divided by 100.) Then, create b-tree
indexes on the AverageCellDemand and MedianCellDemand columns.

15.3.2 Queries Benefiting from Extensible Indexing

A query benefits from a domain index if the data being queried against cannot be
made a simple attribute of a table or if the operation to be performed on the data is not
one of the standard operations supported by Oracle indexes.
Examples of queries that would benefit from extensible indexing (using the power
demand cartridge) include:
• Find the first cell for a specified time where the power demand was equal to a
specified value.
By asking for the first cell, the query goes beyond a simple true-false check (such
as finding out whether any cell for a specified time had a demand equal to a
specified value), and thus benefits from a domain index.
• Find the time for which there was the greatest disparity, or difference between the
cell with the minimum demand and the cell with the maximum demand.
• Find all times for which 3 or more cells had a demand greater than a specified
value.
• Find all times for which 10 or more cells had a demand not less than a specified
value.
• Find all times for which the total grid demand rose 10 percent or more over the
preceding time's total grid demand.

15-11
 Chapter 15
Creating the Domain Index

15.4 Creating the Domain Index

Consider the parts of the power demand cartridge as they relate to extensible
indexing. Explanatory text and code segments are mixed.
The entire cartridge definition is available online as extdemo1.sql in the standard
Oracle demo directory (location is platform-dependent).

15.4.1 Creating the Schema to Own the Index

Before you create a domain index, create a database user, or schema. to own the
index. In the power demand example, the user PowerCartUser is created and granted
the appropriate privileges. All database structures related to the cartridge are created
under this user (that is, while the cartridge developer or DBA is connected to the
database as PowerCartUser), as demonstrated in Example 15-1.

Example 15-1 Creating a Database User for the Power Demand Cartridge
set echo on
connect sys/knl_test7 as sysdba;
drop user PowerCartUser cascade;
create user PowerCartUser identified by PowerCartUser;

-------------------------------------------------------------------
-- INITIAL SET-UP
-------------------------------------------------------------------
-- grant privileges --
grant connect, resource to PowerCartUser;
grant create operator to PowerCartUser;
grant create indextype to PowerCartUser;
grant create table to PowerCartUser;

15.4.2 Creating the Object Types

The object type PowerDemand_Typ stores the hourly power grid readings. This type is
used to define a column in the table in which the readings are stored.
First, three types are defined for later use, as demonstrated in Example 15-2.
• PowerGrid_Typ defines the cells in PowerDemand_Typ.
• NumTab_Typ is used in the table where the index entries are stored.
• The PowerDemand_Typ type includes the following:
– Three attributes (TotGridDemand, MaxCellDemand, MinCellDemand) that are set
by three member procedures
– Power demand readings (100 cells in a grid)
– The date/time of the power demand readings. (Every hour, 100 areas transmit
their power demand readings.)
Example 15-2 Creating the Types of Power Demand Cartridge
CREATE OR REPLACE TYPE PowerGrid_Typ as VARRAY(100) of NUMBER;

CREATE OR REPLACE TYPE NumTab_Typ as TABLE of NUMBER;

15-12
 Chapter 15
Creating the Domain Index

CREATE OR REPLACE TYPE PowerDemand_Typ AS OBJECT (

-- Total power demand for the grid
TotGridDemand NUMBER,
-- Cell with maximum/minimum power demand for the grid
MaxCellDemand NUMBER,
MinCellDemand NUMBER,
-- Power grid: 10X10 array represented as Varray(100)
-- using previously defined PowerGrid_Typ
CellDemandValues PowerGrid_Typ,
-- Date/time for power-demand samplings: Every hour,
-- 100 areas transmit their power demand readings.
SampleTime DATE,
--
-- Methods (Set...) for this type:
-- Total demand for the entire power grid for a
-- SampleTime: sets the value of TotGridDemand.
Member Procedure SetTotalDemand,
-- Maximum demand for the entire power grid for a
-- SampleTime: sets the value of MaxCellDemand.
Member Procedure SetMaxDemand,
-- Minimum demand for the entire power grid for a
-- SampleTime: sets the value of MinCellDemand.
Member Procedure SetMinDemand
);
/

15.4.3 Defining the Object Type Methods

The PowerDemand_Typ object type has methods that set the first three attributes in the
type definition:
• TotGridDemand, the total demand for the entire power grid for the hour in question
(identified by SampleTime)
• MaxCellDemand, the highest power demand value for all cells for the SampleTime
• MinCellDemand, the lowest power demand value for all cells for the SampleTime
The logic for each procedure is not complicated. SetTotDemand loops through the cell
values and creates a running total. SetMaxDemand compares the first two cell values
and saves the higher as the current highest value; it then examines each successive
cell, comparing it against the current highest value and saving the higher of the two as
the current highest value, until it reaches the end of the cell values. SetMinDemand
uses the same approach as SetMaxDemand, but it continually saves the lower value in
comparisons to derive the lowest value overall, as demonstrated in Example 15-3.
Example 15-3 Implementing an Object Type for Power Demand Cartridge
CREATE OR REPLACE TYPE BODY PowerDemand_Typ
IS
--
-- Methods (Set...) for this type:
-- Total demand for the entire power grid for a
-- SampleTime: sets the value of TotGridDemand.
Member Procedure SetTotalDemand
IS
I BINARY_INTEGER;
Total NUMBER;
BEGIN
Total :=0;
I := CellDemandValues.FIRST;

15-13
 Chapter 15
Creating the Domain Index

WHILE I IS NOT NULL LOOP

Total := Total + CellDemandValues(I);
I := CellDemandValues.NEXT(I);
END LOOP;
TotGridDemand := Total;
END;

-- Maximum demand for the entire power grid for a

-- SampleTime: sets the value of MaxCellDemand.
Member Procedure SetMaxDemand
IS
I BINARY_INTEGER;
Temp NUMBER;
BEGIN
I := CellDemandValues.FIRST;
Temp := CellDemandValues(I);
WHILE I IS NOT NULL LOOP
IF Temp < CellDemandValues(I) THEN
Temp := CellDemandValues(I);
END IF;
I := CellDemandValues.NEXT(I);
END LOOP;
MaxCellDemand := Temp;
END;

-- Minimum demand for the entire power grid for a

-- SampleTime: sets the value of MinCellDemand.
Member Procedure SetMinDemand
IS
I BINARY_INTEGER;
Temp NUMBER;
BEGIN
I := CellDemandValues.FIRST;
Temp := CellDemandValues(I);
WHILE I IS NOT NULL LOOP
IF Temp > CellDemandValues(I) THEN
Temp := CellDemandValues(I);
END IF;
I := CellDemandValues.NEXT(I);
END LOOP;
MinCellDemand := Temp;
END;
END;
/

15.4.4 Understanding Functions and Operators

The power demand cartridge is designed so that users can query the power grid for
relationships of equality, greaterthan, or lessthan. However, because of the way
the cell demand data is stored, the standard operators (=, >, <) cannot be used.
Instead, new operators must be created, and a function must be created to define the
implementation for each new operator (that is, how the operator is to be interpreted by
Oracle).
For this cartridge, each of the three relationships can be checked in two ways:
• Whether a specific cell in the grid satisfies the relationship. For example, are there
grids where cell (3,7) has demand equal to 25?

15-14
 Chapter 15
Creating the Domain Index

These operators have names in the form Power_XxxxxSpecific(), such as

Power_EqualsSpecific(), and the implementing functions have names in the form
Power_XxxxxSpecific_Func().
• Whether any cell in the grid satisfies the relationship. For example, are there grids
where any cell has demand equal to 25?
These operators have names in the form Power_XxxxxAny(), such as
Power_EqualsAny(), and the implementing functions have names in the form
Power_XxxxxAny_Func().
For each operator-function pair, the function is defined first and then the operator as
using the function. The function is the implementation that would be used if there were
no index defined. This implementation must be specified so that the Oracle optimizer
can determine costs, decide whether the index should be used, and create an
execution plan.
Table 15-3 shows the operators and implementing functions:

Table 15-3 Operators and Implementing Functions

Operator Implementing Function

Power_EqualsSpecific() Power_EqualsSpecific_Func()

Power_EqualsAny() Power_EqualsAny_Func()

Power_LessThanSpecific() Power_LessThanSpecific_Func()

Power_LessThanAny() Power_LessThanAny_Func()

Power_GreaterThanSpecific() Power_GreaterThanSpecific_Func()

Power_GreaterThanAny() Power_GreaterThanAny_Func()

Each function and operator returns a numeric value of 1 if the condition is true (for
example, if the specified cell is equal to the specified value), 0 if the condition is not
true, or null if the specified cell number is invalid.

The statements in Example 15-4 create the implementing functions,

Power_xxx_Func(), first the specific and then the any implementations.

The statements in Example 15-5 create the operators (Power_xxx). Each statement
specifies an implementing function.

15.4.4.1 Creating Functions and Operators

Examples Example 15-4 and Example 15-5 demonstrate how to implement functions
and operators for the previously defined object types of the Power Demand cartridge.
Example 15-4 Implementing Power_XXX_Func() Functions for Power Demand
Cartridge
CREATE FUNCTION Power_EqualsSpecific_Func(
object PowerDemand_Typ, cell NUMBER, value NUMBER)

15-15
 Chapter 15
Creating the Domain Index

RETURN NUMBER AS
BEGIN
IF cell <= object.CellDemandValues.LAST
THEN
IF (object.CellDemandValues(cell) = value) THEN
RETURN 1;
ELSE
RETURN 0;
END IF;
ELSE
RETURN NULL;
END IF;
END;
/
CREATE FUNCTION Power_GreaterThanSpecific_Func(
object PowerDemand_Typ, cell NUMBER, value NUMBER)
RETURN NUMBER AS
BEGIN
IF cell <= object.CellDemandValues.LAST
THEN
IF (object.CellDemandValues(cell) > value) THEN
RETURN 1;
ELSE
RETURN 0;
END IF;
ELSE
RETURN NULL;
END IF;
END;
/
CREATE FUNCTION Power_LessThanSpecific_Func(
object PowerDemand_Typ, cell NUMBER, value NUMBER)
RETURN NUMBER AS
BEGIN
IF cell <= object.CellDemandValues.LAST
THEN
IF (object.CellDemandValues(cell) < value) THEN
RETURN 1;
ELSE
RETURN 0;
END IF;
ELSE
RETURN NULL;
END IF;
END;
/
CREATE FUNCTION Power_EqualsAny_Func(
object PowerDemand_Typ, value NUMBER)
RETURN NUMBER AS
idx NUMBER;
BEGIN
FOR idx IN object.CellDemandValues.FIRST..object.CellDemandValues.LAST LOOP
IF (object.CellDemandValues(idx) = value) THEN
RETURN 1;
END IF;
END LOOP;
RETURN 0;
END;
/
CREATE FUNCTION Power_GreaterThanAny_Func(
object PowerDemand_Typ, value NUMBER)

15-16
 Chapter 15
Creating the Domain Index

RETURN NUMBER AS
idx NUMBER;
BEGIN
FOR idx IN object.CellDemandValues.FIRST..object.CellDemandValues.LAST LOOP
IF (object.CellDemandValues(idx) > value) THEN
RETURN 1;
END IF;
END LOOP;
RETURN 0;
END;
/
CREATE FUNCTION Power_LessThanAny_Func(
object PowerDemand_Typ, value NUMBER)
RETURN NUMBER AS
idx NUMBER;
BEGIN
FOR idx IN object.CellDemandValues.FIRST..object.CellDemandValues.LAST LOOP
IF (object.CellDemandValues(idx) < value) THEN
RETURN 1;
END IF;
END LOOP;
RETURN 0;
END;
/

Example 15-5 Implementing Power_XXX() Functions for Power Demand

Cartridge
CREATE OPERATOR Power_Equals BINDING(PowerDemand_Typ, NUMBER, NUMBER)
RETURN NUMBER USING Power_EqualsSpecific_Func;
CREATE OPERATOR Power_GreaterThan BINDING(PowerDemand_Typ, NUMBER, NUMBER)
RETURN NUMBER USING Power_GreaterThanSpecific_Func;
CREATE OPERATOR Power_LessThan BINDING(PowerDemand_Typ, NUMBER, NUMBER)
RETURN NUMBER USING Power_LessThanSpecific_Func;

CREATE OPERATOR Power_EqualsAny BINDING(PowerDemand_Typ, NUMBER)

RETURN NUMBER USING Power_EqualsAny_Func;
CREATE OPERATOR Power_GreaterThanAny BINDING(PowerDemand_Typ, NUMBER)
RETURN NUMBER USING Power_GreaterThanAny_Func;
CREATE OPERATOR Power_LessThanAny BINDING(PowerDemand_Typ, NUMBER)
RETURN NUMBER USING Power_LessThanAny_Func;

15.4.5 Creating the Indextype Implementation Methods

The power demand cartridge creates an object type for the indextype that specifies
methods for the domain index. These methods are part of the ODCIIndex (Oracle Data
Cartridge Interface Index) interface, and they collectively define the behavior of the
index in terms of the methods for defining, manipulating, scanning, and exporting the
index.
Table 15-4 shows the method functions (all but one starting with ODCIIndex) created
for the power demand cartridge.

Table 15-4 Indextype Methods

Method Description
ODCIGetInterfaces() Returns the list interface names implemented by the type.

15-17
 Chapter 15
Creating the Domain Index

Table 15-4 (Cont.) Indextype Methods

Method Description
ODCIIndexCreate() Creates a table to store index data. If the base table containing
data to be indexed is not empty, this method builds the index for
existing data.
This method is called when a CREATE INDEX statement is
issued that refers to the indextype. Upon invocation, any
parameters specified in the PARAMETERS clause are passed in
along with a description of the index.
ODCIIndexDrop() Drops the table that stores the index data. This method is called
when a DROP INDEX statement specifies the index.
ODCIIndexStart() Initializes the scan of the index for the operator predicate. This
method is invoked when a query is submitted involving an
operator that can be executed using the domain index.
ODCIIndexFetch() Returns the ROWID of each row that satisfies the operator
predicate.
ODCIIndexClose() Ends the current use of the index. This method can perform any
necessary clean-up.
ODCIIndexInsert() Maintains the index structure when a record is inserted in a
table that contains columns or object attributes indexed by the
indextype.
ODCIIndexDelete() Maintains the index structure when a record is deleted from a
table that contains columns or object attributes indexed by the
indextype.
ODCIIndexUpdate() Maintains the index structure when a record is updated
(modified) in a table that contains columns or object attributes
indexed by the indextype.
ODCIIndexGetMetadata() Allows the export and import of implementation-specific
metadata associated with the index.

15.4.6 Defining theType

Example 15-6 creates the power_idxtype_im object type. The methods of this type are
the ODCI methods to define, manipulate, and scan the domain index. The curnum
attribute is the cursor number used as context for the scan routines ODCIIndexStart(),
ODCIIndexFetch(), and ODCIIndexClose().
The CREATE TYPE statement is followed by a CREATE TYPE BODY statement that specifies
the implementation for each member function:
CREATE OR REPLACE TYPE BODY power_idxtype_im
IS
...

All the method definitions (except for ODCIIndexGetMetadata(), which returns a

VARCHAR2 string) have the following general form:
STATIC FUNCTION function-name (...)
RETURN NUMBER
IS
...
END;

15-18
 Chapter 15
Creating the Domain Index

Example 15-6 Creating power_idxtype_im Object Type for Power Demand

Cartridge
CREATE OR REPLACE TYPE power_idxtype_im AS OBJECT
(
curnum NUMBER,
STATIC FUNCTION ODCIGetInterfaces(ifclist OUT sys.ODCIObjectList)
RETURN NUMBER,
STATIC FUNCTION ODCIIndexCreate (ia sys.ODCIIndexInfo, parms VARCHAR2,
env sys.ODCIEnv) RETURN NUMBER,
STATIC FUNCTION ODCIIndexDrop(ia sys.ODCIIndexInfo, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIIndexStart(sctx IN OUT power_idxtype_im,
ia sys.ODCIIndexInfo,
op sys.ODCIPredInfo, qi sys.ODCIQueryInfo,
strt NUMBER, stop NUMBER,
cmppos NUMBER, cmpval NUMBER, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIIndexStart(sctx IN OUT power_idxtype_im,
ia sys.ODCIIndexInfo,
op sys.ODCIPredInfo, qi sys.ODCIQueryInfo,
strt NUMBER, stop NUMBER,
cmpval NUMBER, env sys.ODCIEnv)
RETURN NUMBER,
MEMBER FUNCTION ODCIIndexFetch(nrows NUMBER, rids OUT sys.ODCIRidList,
env sys.ODCIEnv) RETURN NUMBER,
MEMBER FUNCTION ODCIIndexClose (env sys.ODCIEnv) RETURN NUMBER,
STATIC FUNCTION ODCIIndexInsert(ia sys.ODCIIndexInfo, rid VARCHAR2,
newval PowerDemand_Typ, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIIndexDelete(ia sys.ODCIIndexInfo, rid VARCHAR2,
oldval PowerDemand_Typ, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIIndexUpdate(ia sys.ODCIIndexInfo, rid VARCHAR2,
oldval PowerDemand_Typ,
newval PowerDemand_Typ, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIIndexGetMetadata(ia sys.ODCIIndexInfo,
expversion VARCHAR2,
newblock OUT PLS_INTEGER,
env sys.ODCIEnv)
RETURN VARCHAR2
);
/

15.4.6.1 ODCIGetInterfaces()
The ODCIGetInterfaces() function returns the list of names of the interfaces
implemented by the type. To specify the current version of these interfaces, the
ODCIGetInterfaces() routine must return'SYS.ODCIINDEX2' in the OUT parameter, as
demonstrated in Example 15-7.
Example 15-7 Registering Interface and Index Functions in Power Demand
Cartridge
STATIC FUNCTION ODCIGetInterfaces(
ifclist OUT sys.ODCIObjectList)
RETURN NUMBER IS
BEGIN
ifclist := sys.ODCIObjectList(sys.ODCIObject('SYS','ODCIINDEX2'));

15-19
 Chapter 15
Creating the Domain Index

return ODCIConst.Success;
END ODCIGetInterfaces;

15.4.6.2 ODCIIndexCreate()
The ODCIIndexCreate() function creates the table to store index data. If the base table
containing data to be indexed is not empty, this method inserts the index data entries
for existing data.
The function takes the index information as an object parameter whose type is
SYS.ODCIINDEXINFO. The type attributes include the index name, owner name, and so
forth. The PARAMETERS string specified in the CREATE INDEX statement is also passed in
as a parameter to the function, as demonstrated in Example 15-8.
Example 15-8 Registering ODCIIndexCreate() for Power Demand Cartridge
STATIC FUNCTION ODCIIndexCreate (
ia sys.ODCIIndexInfo,
parms VARCHAR2,
env sys.ODCIEnv)
RETURN NUMBER IS
i INTEGER;
r ROWID;
p NUMBER;
v NUMBER;
stmt1 VARCHAR2(1000);
stmt2 VARCHAR2(1000);
stmt3 VARCHAR2(1000);
cnum1 INTEGER;
cnum2 INTEGER;
cnum3 INTEGER;
junk NUMBER;

The SQL statement to create the table for the index data is constructed and executed.
The table includes the ROWID of the base table, r, the cell position number (cpos) in the
grid from 1 to 100, and the power demand value in that cell (cval).
BEGIN
-- Construct the SQL statement.
stmt1 := 'CREATE TABLE ' || ia.IndexSchema || '.' || ia.IndexName ||'_pidx' ||
'( r ROWID, cpos NUMBER, cval NUMBER)';

-- Dump the SQL statement.

dbms_output.put_line('ODCIIndexCreate>>>>>');
sys.ODCIIndexInfoDump(ia);
dbms_output.put_line('ODCIIndexCreate>>>>>'||stmt1);

-- Execute the statement.

cnum1 := dbms_sql.open_cursor;
dbms_sql.parse(cnum1, stmt1, dbms_sql.native);
junk := dbms_sql.execute(cnum1);
dbms_sql.close_cursor(cnum1);

The function populates the index by inserting rows into the table. The function
"unnests" the VARRAY attribute and inserts a row for each cell into the table. Thus, each
10 X 10 grid (10 rows, 10 values for each row) becomes 100 rows in the table (one
row for each cell).
-- Now populate the table.
stmt2 := ' INSERT INTO '|| ia.IndexSchema || '.' || ia.IndexName || '_pidx' ||

15-20
 Chapter 15
Creating the Domain Index

' SELECT :rr, ROWNUM, column_value FROM THE' || ' (SELECT CAST (P.'||
ia.IndexCols(1).ColName||'.CellDemandValues AS NumTab_Typ)'|| ' FROM ' ||
ia.IndexCols(1).TableSchema || '.' || ia.IndexCols(1).TableName || ' P' ||
' WHERE P.ROWID = :rr)';

-- Execute the statement.

dbms_output.put_line('ODCIIndexCreate>>>>>'||stmt2);

-- Parse the statement.

cnum2 := dbms_sql.open_cursor;
dbms_sql.parse(cnum2, stmt2, dbms_sql.native);

stmt3 := 'SELECT ROWID FROM '|| ia.IndexCols(1).TableSchema || '.' ||

ia.IndexCols(1).TableName;
dbms_output.put_line('ODCIIndexCreate>>>>>'||stmt3);
cnum3 := dbms_sql.open_cursor;
dbms_sql.parse(cnum3, stmt3, dbms_sql.native);
dbms_sql.define_column_rowid(cnum3, 1, r);
junk := dbms_sql.execute(cnum3);

WHILE dbms_sql.fetch_rows(cnum3) > 0 LOOP

-- Get column values of the row. --
dbms_sql.column_value_rowid(cnum3, 1, r);
-- Bind the row into the cursor for the next insert. --
dbms_sql.bind_variable_rowid(cnum2, ':rr', r);
junk := dbms_sql.execute(cnum2);
END LOOP;

The function concludes by closing the cursors and returning a success status.
dbms_sql.close_cursor(cnum2);
dbms_sql.close_cursor(cnum3);
RETURN ODCICONST.SUCCESS;
END ODCIInexCreate;

15.4.6.3 ODCIIndexDrop()
The ODCIIndexDrop() function drops the table that stores the index data, as
demonstrated in Example 15-9. This method is called when a DROP INDEX statement is
issued.
Example 15-9 Registering ODCIIndexDrop() for Power Demand Cartridge
STATIC FUNCTION ODCIIndexDrop(ia sys.ODCIIndexInfo, env sys.ODCIEnv)
RETURN NUMBER IS
stmt VARCHAR2(1000);
cnum INTEGER;
junk INTEGER;
BEGIN
-- Construct the SQL statement.
stmt := 'drop table ' || ia.IndexSchema || '.' || ia.IndexName || '_pidx';

dbms_output.put_line('ODCIIndexDrop>>>>>');
sys.ODCIIndexInfoDump(ia);
dbms_output.put_line('ODCIIndexDrop>>>>>'||stmt);

-- Execute the statement.

cnum := dbms_sql.open_cursor;
dbms_sql.parse(cnum, stmt, dbms_sql.native);
junk := dbms_sql.execute(cnum);
dbms_sql.close_cursor(cnum);

15-21
 Chapter 15
Creating the Domain Index

RETURN ODCICONST.SUCCESS;
END ODCIIndexDrop;

15.4.6.4 ODCIIndexStart(); Specific Queries

The first definition of the ODCIIndexStart() function initializes the scan of the index to
return all rows that satisfy the operator predicate. For example, if a query asks for all
instances where cell (3,7) has a value equal to 25, the function initializes the scan to
return all rows in the index-organized table for which that cell has that value. This
definition of ODCIIndexStart() differs from the definition in ODCIIndexStart(); Any
Queries in that it includes the cmppos parameter for the position of the cell.

The self parameter is the context that is shared with the ODCIIndexFetch() and
ODCIIndexClose() functions. The ia parameter contains the index information as an
object instance of type SYS.ODCIINDEXINFO, and the op parameter contains the
operator information as an object instance of type SYS.ODCIOPERINFO. The strt and
stop parameters are the lower and upper boundary points for the operator return
value. The cmppos parameter is the cell position and cmpval is the value in the cell
specified by the operator Power_XxxxxSpecific(). This is demonstrated in
Example 15-10.
Example 15-10 Registering ODCIIndexStart() for Power Demand Cartridge
STATIC FUNCTION ODCIIndexStart(
sctx IN OUT power_idxtype_im,
ia sys.ODCIIndexInfo,
op sys.ODCIPredInfo,
qi sys.ODCIQueryInfo,
strt NUMBER, stop NUMBER,
cmppos NUMBER,
cmpval NUMBER,
env sys.ODCIEnv )
RETURN NUMBER IS
cnum INTEGER;
rid ROWID;
nrows INTEGER;
relop VARCHAR2(2);
stmt VARCHAR2(1000);
BEGIN
dbms_output.put_line('ODCIIndexStart>>>>>');
sys.ODCIIndexInfoDump(ia);
sys.ODCIPredInfoDump(op);
dbms_output.put_line('start key : '||strt);
dbms_output.put_line('stop key : '||stop);
dbms_output.put_line('compare position : '||cmppos);
dbms_output.put_line('compare value : '||cmpval);

The function checks for errors in the predicate.

-- Take care of some error cases.
-- The only predicates in which btree operators can appear are
-- op() = 1 OR op() = 0
if (strt != 1) and (strt != 0) then
raise_application_error(-20101, 'Incorrect predicate for operator');
END if;

if (stop != 1) and (stop != 0) then

15-22
 Chapter 15
Creating the Domain Index

raise_application_error(-20101, 'Incorrect predicate for operator');

END if;

The function generates the SQL statement to be executed. It determines the operator
name and the lower and upper index value bounds (the start and stop keys). The start
and stop keys can both be 1 (= TRUE) or both be 0 (= FALSE).
-- Generate the SQL statement to be executed.
-- First, figure out the relational operator needed for the statement.
-- Take into account the operator name and the start and stop keys. For now,
-- the start and stop keys can both be 1 (= TRUE) or both be 0 (= FALSE).
if op.ObjectName = 'POWER_EQUALS' then
if strt = 1 then
relop := '=';
else
relop := '!=';
end if;
elsif op.ObjectName = 'POWER_LESSTHAN' then
if strt = 1 then
relop := '<';
else
relop := '>=';
end if;
elsif op.ObjectName = 'POWER_GREATERTHAN' then
if strt = 1 then
relop := '>';
else
relop := '<=';
end if;
else
raise_application_error(-20101, 'Unsupported operator');
end if;

stmt := 'select r from '||ia.IndexSchema||'.'||ia.IndexName||'_pidx'||

' where cpos '|| '=' ||''''||cmppos||''''|| ' and cval ' ||relop||''''||
cmpval||'''';

dbms_output.put_line('ODCIIndexStart>>>>>' || stmt);
cnum := dbms_sql.open_cursor;
dbms_sql.parse(cnum, stmt, dbms_sql.native);
dbms_sql.define_column_rowid(cnum, 1, rid);
nrows := dbms_sql.execute(cnum);

The function stores the cursor number in the context, which is used by the
ODCIIndexFetch function, and sets a success return status.
-- Set context as the cursor number.
stcx := power_idxtype_im(cnum);

-- Return success.
RETURN ODCICONST.SUCCESS;
END ODCIIndexStart;

15-23
 Chapter 15
Creating the Domain Index

See Also:

• ODCIIndexClose()
• ODCIIndexFetch()
• ODCIIndexStart()
• ODCIIndexStart(); Any Queries

15.4.6.5 ODCIIndexStart(); Any Queries

This definition of the ODCIIndexStart() function initializes the scan of the index to
return all rows that satisfy the operator predicate. For example, if a query asks for all
instances where any cell has a value equal to 25, the function initializes the scan to
return all rows in the index-organized table for which that cell has that value. This
definition of differs from the definition in ODCIIndexStart(); Specific Queries in that
it does not include the cmppos parameter.

The self parameter is the context that is shared with the ODCIIndexFetch() and
ODCIIndexClose() functions. The ia parameter contains the index information as an
object instance of type SYS.ODCIINDEXINFO, and the op parameter contains the
operator information as an object instance of type SYS.ODCIOPERINFO. The strt and
stop parameters are the lower and upper boundary points for the operator return
value. The cmpval parameter is the value in the cell specified by the operator
Power_Xxxx().

Example 15-11 Registering ODCIIndexStart() for Any Queries for Power

Demand Cartridge
STATIC FUNCTION ODCIIndexStart(
sctx IN OUT power_idxtype_im,
ia sys.ODCIIndexInfo,
op sys.ODCIPredInfo,
qi sys.ODCIQueryInfo,
strt NUMBER,
stop NUMBER,
cmpval NUMBER,
env sys.ODCIEnv )
RETURN NUMBER IS
cnum INTEGER;
rid ROWID;
nrows INTEGER;
relop VARCHAR2(2);
stmt VARCHAR2(1000);
BEGIN
dbms_output.put_line('ODCIIndexStart>>>>>');
sys.ODCIIndexInfoDump(ia);
sys.ODCIPredInfoDump(op);
dbms_output.put_line('start key : '||strt);
dbms_output.put_line('stop key : '||stop);
dbms_output.put_line('compare value : '||cmpval);

The function checks for errors in the predicate.

-- Take care of some error cases.
-- The only predicates in which btree operators can appear are

15-24
 Chapter 15
Creating the Domain Index

-- op() = 1 OR op() = 0
if (strt != 1) and (strt != 0) then
raise_application_error(-20101, 'Incorrect predicate for operator');
END if;

if (stop != 1) and (stop != 0) then

raise_application_error(-20101, 'Incorrect predicate for operator');
END if;

The function generates the SQL statement to be executed. It determines the operator
name and the lower and upper index value bounds (the start and stop keys). The start
and stop keys can both be 1 (= TRUE) or both be 0 (= FALSE).
-- Generate the SQL statement to be executed.
-- First, figure out the relational operator needed for the statement.
-- Take into account the operator name and the start and stop keys. For now,
-- the start and stop keys can both be 1 (= TRUE) or both be 0 (= FALSE).
if op.ObjectName = 'POWER_EQUALSANY' then
relop := '=';
elsif op.ObjectName = 'POWER_LESSTHANANY' then
relop := '<';
elsif op.ObjectName = 'POWER_GREATERTHANANY' then
relop := '>';
else
raise_application_error(-20101, 'Unsupported operator');
end if;

-- This statement returns the qualifying rows for the TRUE case.
stmt := 'select distinct r from '||ia.IndexSchema||'.'||ia.IndexName||'_pidx'||'
where cval '||relop||''''||cmpval||'''';
-- In the FALSE case, we must find the complement of the rows.
if (strt = 0) then
stmt := 'select distinct r from '||ia.IndexSchema||'.'||ia.IndexName||
'_pidx'||' minus '||stmt;
end if;

dbms_output.put_line('ODCIIndexStart>>>>>' || stmt);
cnum := dbms_sql.open_cursor;
dbms_sql.parse(cnum, stmt, dbms_sql.native);
dbms_sql.define_column_rowid(cnum, 1, rid);
nrows := dbms_sql.execute(cnum);

The function stores the cursor number in the context, which is used by the
ODCIIndexFetch() function, and sets a success return status.
-- Set context as the cursor number.
self := power_idxtype_im(cnum);

-- Return success.
RETURN ODCICONST.SUCCESS;
END ODCIIndexStart;

15-25
 Chapter 15
Creating the Domain Index

See Also:

• ODCIIndexClose()
• ODCIIndexFetch()
• ODCIIndexStart()
• ODCIIndexStart(); Specific Queries

15.4.6.6 ODCIIndexFetch()
The ODCIIndexFetch() function, demonstrated in Example 15-12 returns a batch of
ROWIDs for the rows that satisfy the operator predicate. Each time ODCIIndexFetch() is
invoked, it returns the next batch of rows (rids parameter, a collection of type
SYS.ODCIRIDLIST) that satisfy the operator predicate. The maximum number of rows
that can be returned on each invocation is specified by the nrows parameter.

Oracle invokes ODCIIndexFetch() repeatedly until all rows that satisfy the operator
predicate have been returned.
Example 15-12 Registering ODCIIndexFetch() for Power Demand Cartridge
MEMBER FUNCTION ODCIIndexFetch(
nrows NUMBER,
rids OUT sys.ODCIRidList,
env sys.ODCIEnv)
RETURN NUMBER IS
cnum INTEGER;
idx INTEGER := 1;
rlist sys.ODCIRidList := sys.ODCIRidList();
done boolean := FALSE;

The function loops through the collection of rows selected by the ODCIIndexStart()
function, using the same cursor number, cnum, as in the ODCIIndexStart() function,
and returns the ROWIDs.
BEGIN
dbms_output.put_line('ODCIIndexFetch>>>>>');
dbms_output.put_line('Nrows : '||round(nrows));

cnum := self.curnum;

WHILE not done LOOP

if idx > nrows then
done := TRUE;
else
rlist.extEND;
if dbms_sql.fetch_rows(cnum) > 0 then
dbms_sql.column_value_rowid(cnum, 1, rlist(idx));
idx := idx + 1;
else
rlist(idx) := null;
done := TRUE;
END if;
END if;
END LOOP;

15-26
 Chapter 15
Creating the Domain Index

rids := rlist;
RETURN ODCICONST.SUCCESS;
END ODCIIndexFetch;

See Also:

• ODCIIndexFetch()
• ODCIIndexStart()

15.4.6.7 ODCIIndexClose()
The ODCIIndexClose() function, demonstrated in Example 15-13, closes the cursor
used by the ODCIIndexStart() and ODCIIndexFetch() functions.
Example 15-13 Registering ODCIIndexStart() for Power Demand Cartridge
MEMBER FUNCTION ODCIIndexClose (env sys.ODCIEnv)
RETURN NUMBER IS
cnum INTEGER;
BEGIN
dbms_output.put_line('ODCIIndexClose>>>>>');

cnum := self.curnum;
dbms_sql.close_cursor(cnum);
RETURN ODCICONST.SUCCESS;
END ODCIIndexClose;

15.4.6.8 ODCIIndexInsert()
The ODCIIndexInsert() function, demonstrated in Example 15-14, is called when a
record is inserted in a table that contains columns or OBJECT attributes indexed by the
indextype. The new values in the indexed columns are passed in as arguments along
with the corresponding row identifier.
Example 15-14 Registering ODCIIndexInsert() for Power Demand Cartridge
STATIC FUNCTION ODCIIndexInsert(
ia sys.ODCIIndexInfo,
rid VARCHAR2,
newval PowerDemand_Typ,
env sys.ODCIEnv)
RETURN NUMBER AS
cid INTEGER;
i BINARY_INTEGER;
nrows INTEGER;
stmt VARCHAR2(1000);
BEGIN
dbms_output.put_line(' ');
dbms_output.put_line('ODCIIndexInsert>>>>>'||' TotGridDemand= '||
newval.TotGridDemand ||' MaxCellDemand= '||newval.MaxCellDemand ||
' MinCellDemand= '||newval.MinCellDemand) ;
sys.ODCIIndexInfoDump(ia);

-- Construct the statement.

stmt := ' INSERT INTO '|| ia.IndexSchema || '.' || ia.IndexName || '_pidx' ||
' VALUES (:rr, :pos, :val)';

15-27
 Chapter 15
Creating the Domain Index

-- Execute the statement.

dbms_output.put_line('ODCIIndexInsert>>>>>'||stmt);
-- Parse the statement.
cid := dbms_sql.open_cursor;
dbms_sql.parse(cid, stmt, dbms_sql.native);
dbms_sql.bind_variable_rowid(cid, ':rr', rid);

-- Iterate over the rows of the Varray and insert them.

i := newval.CellDemandValues.FIRST;
WHILE i IS NOT NULL LOOP
-- Bind the row into the cursor for insert.
dbms_sql.bind_variable(cid, ':pos', i);
dbms_sql.bind_variable(cid, ':val', newval.CellDemandValues(i));
-- Execute.
nrows := dbms_sql.execute(cid);
dbms_output.put_line('ODCIIndexInsert>>>>>('||'RID'||' , '||i|| ' , '||
newval.CellDemandValues(i)|| ')');
i := newval.CellDemandValues.NEXT(i);
END LOOP;

dbms_sql.close_cursor(cid);
RETURN ODCICONST.SUCCESS;
END ODCIIndexInsert;

15.4.6.9 ODCIIndexDelete()
The ODCIIndexDelete() function, demonstrated in Example 15-15, is called when a
record is deleted from a table that contains columns or object attributes indexed by the
indextype. The old values in the indexed columns are passed in as arguments along
with the corresponding row identifier.
Example 15-15 Registering ODCIIndexDelete() for Power Demand Cartridge
STATIC FUNCTION ODCIIndexDelete(
ia sys.ODCIIndexInfo,
rid VARCHAR2,
oldval PowerDemand_Typ,
env sys.ODCIEnv)
RETURN NUMBER AS
cid INTEGER;
stmt VARCHAR2(1000);
nrows INTEGER;
BEGIN
dbms_output.put_line(' ');
dbms_output.put_line('ODCIIndexDelete>>>>>'||' TotGridDemand= '||
oldval.TotGridDemand ||' MaxCellDemand= '||oldval.MaxCellDemand ||
' MinCellDemand= '||oldval.MinCellDemand) ;
sys.ODCIIndexInfoDump(ia);

-- Construct the statement.

stmt := ' DELETE FROM '|| ia.IndexSchema || '.' ||ia.IndexName|| '_pidx' ||
' WHERE r=:rr';
dbms_output.put_line('ODCIIndexDelete>>>>>'||stmt);

-- Parse and execute the statement.

cid := dbms_sql.open_cursor;
dbms_sql.parse(cid, stmt, dbms_sql.native);
dbms_sql.bind_variable_rowid(cid, ':rr', rid);
nrows := dbms_sql.execute(cid);
dbms_sql.close_cursor(cid);

15-28
 Chapter 15
Creating the Domain Index

RETURN ODCICONST.SUCCESS;
END ODCIIndexDelete;

15.4.6.10 ODCIIndexUpdate()
The ODCIIndexUpdate() function, demonstrated in Example 15-16, is called when a
record is updated in a table that contains columns or object attributes indexed by the
indextype. The old and new values in the indexed columns are passed in as
arguments along with the row identifier.
Example 15-16 Registering ODCIIndexUpdate() for Power Demand Cartridge
STATIC FUNCTION ODCIIndexUpdate(
ia sys.ODCIIndexInfo,
rid VARCHAR2,
oldval PowerDemand_Typ,
newval PowerDemand_Typ,
env sys.ODCIEnv)
RETURN NUMBER AS
cid INTEGER;
cid2 INTEGER;
stmt VARCHAR2(1000);
stmt2 VARCHAR2(1000);
nrows INTEGER;
i NUMBER;
BEGIN
dbms_output.put_line(' ');
dbms_output.put_line('ODCIIndexUpdate>>>>> Old'||' TotGridDemand= '||
oldval.TotGridDemand||' MaxCellDemand= '||oldval.MaxCellDemand ||
' MinCellDemand= '||oldval.MinCellDemand) ;
dbms_output.put_line('ODCIIndexUpdate>>>>> New'||' TotGridDemand= '||
newval.TotGridDemand ||' MaxCellDemand= '||newval.MaxCellDemand ||
' MinCellDemand= '||newval.MinCellDemand) ;
sys.ODCIIndexInfoDump(ia);

-- Delete old entries.

stmt := ' DELETE FROM '||ia.IndexSchema ||'.'||ia.IndexName||'_pidx'||
' WHERE r=:rr';
dbms_output.put_line('ODCIIndexUpdate>>>>>'||stmt);

-- Parse and execute the statement.

cid := dbms_sql.open_cursor;
dbms_sql.parse(cid, stmt, dbms_sql.native);
dbms_sql.bind_variable_rowid(cid, ':rr', rid);
nrows := dbms_sql.execute(cid);
dbms_sql.close_cursor(cid);

-- Insert new entries.

stmt2 := ' INSERT INTO '||ia.IndexSchema||'.'||ia.IndexName||'_pidx'||
' VALUES (:rr, :pos, :val)';
dbms_output.put_line('ODCIIndexUpdate>>>>>'||stmt2);

-- Parse and execute the statement.

cid2 := dbms_sql.open_cursor;
dbms_sql.parse(cid2, stmt2, dbms_sql.native);
dbms_sql.bind_variable_rowid(cid2, ':rr', rid);

-- Iterate over the rows of the Varray and insert them.

i := newval.CellDemandValues.FIRST;
WHILE i IS NOT NULL LOOP

15-29
 Chapter 15
Creating the Domain Index

-- Bind the row into the cursor for insert.

dbms_sql.bind_variable(cid2, ':pos', i);
dbms_sql.bind_variable(cid2, ':val', newval.CellDemandValues(i));
nrows := dbms_sql.execute(cid2);
dbms_output.put_line('ODCIIndexUpdate>>>>>('||'RID'||' , '||i ||' , '||
newval.CellDemandValues(i)|| ')');
i := newval.CellDemandValues.NEXT(i);
END LOOP;
dbms_sql.close_cursor(cid2);

RETURN ODCICONST.SUCCESS;
END ODCIIndexUpdate;

ODCIIndexUpdate is the last method defined in the CREATE TYPE BODY statement, which
ends as follows:
END;
/

15.4.6.11 ODCIIndexGetMetadata()
The optional ODCIIndexGetMetadata() function, as demonstrated in Example 15-17, if
present, is called by the Export utility to write implementation-specific metadata (which
is not stored in the system catalogs) into the export dump file. This metadata might be
policy information, version information, user settings, and so on. This metadata is
written to the dump file as anonymous PL/SQL blocks that are executed at import time,
immediately before the associated index is created.
This method returns strings to the Export utility that comprise the code of the PL/SQL
blocks. The Export utility repeatedly calls this method until a zero-length string is
returned, thus allowing the creation of any number of PL/SQL blocks of arbitrary
complexity. Normally, this method calls functions within a PL/SQL package to make
use of package-level variables, such as cursors and iteration counters, that maintain
state across multiple calls by Export.
In the power demand cartridge, the only metadata that is passed is a version string of
V1.0, identifying the current format of the index-organized table that underlies the
domain index. The power_pkg.getversion function generates a call to the
power_pkg.checkversion procedure, to be executed at import time to check that the
version string is V1.0.

Example 15-17 Registering ODCIIndexGetMetadata() for Power Demand

Cartridge
STATIC FUNCTION ODCIIndexGetMetadata(
ia sys.ODCIIndexInfo,
expversion VARCHAR2,
newblock OUT PLS_INTEGER,
env sys.ODCIEnv)
RETURN VARCHAR2 IS

BEGIN
-- Let getversion do all the work since it has to maintain state across calls.

RETURN power_pkg.getversion (ia.IndexSchema, ia.IndexName, newblock);

EXCEPTION
WHEN OTHERS THEN
RAISE;

15-30
 Chapter 15
Creating the Domain Index

END ODCIIndexGetMetaData;

The power_pkg package is defined as follows:

Example 15-18 Creating Package power_pkg for the Power Demand Cartridge
CREATE OR REPLACE PACKAGE power_pkg AS
FUNCTION getversion(
idxschema IN VARCHAR2,
idxname IN VARCHAR2,
newblock OUT PLS_INTEGER)
RETURN VARCHAR2;

PROCEDURE checkversion (
version IN VARCHAR2);
END power_pkg;
/
SHOW ERRORS;

CREATE OR REPLACE PACKAGE BODY power_pkg AS

-- iterate is a package-level variable used to maintain state across calls
-- by Export in this session.

iterate NUMBER := 0;

FUNCTION getversion(
idxschema IN VARCHAR2,
idxname IN VARCHAR2,
newblock OUT PLS_INTEGER)
RETURN VARCHAR2 IS

BEGIN

-- We are generating only one PL/SQL block consisting of one line of code.
newblock := 1;

IF iterate = 0 THEN
-- Increment iterate so we'll know we're done next time we're called.
iterate := iterate + 1;

-- Return a string that calls checkversion with a version 'V1.0'

-- Note that export adds the surrounding BEGIN/END pair to form the anon.
-- block... we don't have to.

RETURN 'power_pkg.checkversion(''V1.0'');';
ELSE
-- reset iterate for next index
iterate := 0;
-- Return a 0-length string; we won't be called again for this index.
RETURN '';
END IF;
END getversion;

PROCEDURE checkversion (version IN VARCHAR2)

IS
wrong_version EXCEPTION;

BEGIN
IF version != 'V1.0' THEN

15-31
 Chapter 15
Defining Types and Methods for Extensible Optimizing

RAISE wrong_version;
END IF;
END checkversion;

END power_pkg;

See Also:
Oracle Database Utilities for information about the Export and Import utilities

15.4.7 Creating the Indextype

The power demand cartridge creates the indextype for the domain index. The
specification, in Example 15-19, includes the list of operators supported by the
indextype. It also identifies the implementation type containing the OCDI index
routines.
Example 15-19 Creating Indextype power_idxtype for Power Demand Cartridge
CREATE OR REPLACE INDEXTYPE power_idxtype
FOR
Power_Equals(PowerDemand_Typ, NUMBER, NUMBER),
Power_GreaterThan(PowerDemand_Typ, NUMBER, NUMBER),
Power_LessThan(PowerDemand_Typ, NUMBER, NUMBER),
Power_EqualsAny(PowerDemand_Typ, NUMBER),
Power_GreaterThanAny(PowerDemand_Typ, NUMBER),
Power_LessThanAny(PowerDemand_Typ, NUMBER)
USING power_idxtype_im;

15.5 Defining Types and Methods for Extensible Optimizing

Consider the parts of the power demand cartridge as they relate to extensible
optimization.

15.5.1 Creating the Statistics Table, PowerCartUserStats

The table PowerCartUserStats, demonstrated in Example 15-20, stores statistics
about the hourly power grid readings. The method ODCIStatsSelectivity() uses these
statistics to estimate the selectivity of operator predicates. Because of the types of
statistics collected, it is more convenient to use a separate table instead of letting
Oracle store the statistics.
The PowerCartUserStats table contains the following columns:

• The table and column for which statistics are collected

• The cell for which the statistics are collected
• The minimum and maximum power demand for the given cell over all power grid
readings
• The number of non-null readings for the given cell over all power grid reading

15-32
 Chapter 15
Defining Types and Methods for Extensible Optimizing

Example 15-20 Creating Statistics Table PowerCartUserStats for Power

Demand Cartridge
CREATE TABLE PowerCartUserStats (
-- Table for which statistics are collected
tab VARCHAR2(30),
-- Column for which statistics are collected
col VARCHAR2(30),
-- Cell position
cpos NUMBER,
-- Minimum power demand for the given cell
lo NUMBER,
-- Maximum power demand for the given cell
hi NUMBER,
-- Number of (non-null) power demands for the given cell
nrows NUMBER
);
/

15.5.2 Creating the Extensible Optimizer Methods

The power demand cartridge creates an object type that specifies methods used by
the extensible optimizer. These methods are part of the ODCIStats interface and they
collectively define the methods that are called by the methods of DBMS_STATS package,
or when the optimizer is deciding on the best execution plan for a query.
Table 15-5 shows the method functions created for the power demand cartridge.
Names of all but one of the functions begin with the string ODCIStats.

Table 15-5 Extensible Optimizer Methods

Method Description
ODCIGetInterfaces() Returns the list of names of the interfaces implemented by the
type.
ODCIStatsCollect() Collects statistics for columns of type PowerDemand_Typ or
domain indexes of indextype power_idxtype.
This method is called when a statement that refers either to a
column of the PowerDemand_Typ type or to an index of the
power_idxtype indextype is issued. Upon invocation, any
specified options are passed in along with a description of the
column or index.
ODCIStatsDelete() Deletes statistics for columns of type PowerDemand_Typ or
domain indexes of indextype power_idxtype.
This method is called when a statement to delete statistics for a
column of the appropriate type or an index of the appropriate
indextype is issued.
ODCIStatsSelectivity() Computes the selectivity of a predicate involving an operator or
its functional implementation.
Called by the optimizer when a predicate of the appropriate
type appears in the WHERE clause of a query.
ODCIStatsIndexCost() Computes the cost of a domain index access path.
Called by the optimizer to get the cost of a domain index access
path, assuming the index can be used for the query.

15-33
 Chapter 15
Defining Types and Methods for Extensible Optimizing

Table 15-5 (Cont.) Extensible Optimizer Methods

Method Description
ODCIStatsFunctionCost() Computes the cost of a function.
Called by the optimizer to get the cost of executing a function.
The function need not necessarily be an implementation of an
operator.

15.5.2.1 Creating the Type Definition

Example 15-21 creates the power_statistics object type. This object type's ODCI
methods are used to collect and delete statistics about columns and indexes, compute
selectivities of predicates with operators or functions, and to compute costs of domain
indexes and functions. The curnum attribute is not used.

The CREATE TYPE statement is followed by a CREATE TYPE BODY statement that
specifies the implementation for each member function:
CREATE OR REPLACE TYPE BODY power_statistics
IS
...

All the function definitions have the following general form:

STATIC FUNCTION function-name (...)
BEGIN
RETURN NUMBER IS
END;

Example 15-21 Creating power_statistics Object Type Definition for Power

Demand Cartridge
CREATE OR REPLACE TYPE power_statistics AS OBJECT
(
curnum NUMBER,
STATIC FUNCTION ODCIGetInterfaces(ifclist OUT sys.ODCIObjectList)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsCollect(col sys.ODCIColInfo,
options sys.ODCIStatsOptions, rawstats OUT RAW, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsDelete(col sys.ODCIColInfo, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsCollect(ia sys.ODCIIndexInfo,
options sys.ODCIStatsOptions, rawstats OUT RAW, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsDelete(ia sys.ODCIIndexInfo, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsSelectivity(pred sys.ODCIPredInfo,
sel OUT NUMBER, args sys.ODCIArgDescList, strt NUMBER, stop NUMBER,
object PowerDemand_Typ, cell NUMBER, value NUMBER, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsSelectivity(pred sys.ODCIPredInfo, sel OUT NUMBER,
args sys.ODCIArgDescList, strt NUMBER, stop NUMBER, object PowerDemand_Typ,
value NUMBER, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsIndexCost(ia sys.ODCIIndexInfo, sel NUMBER,
cost OUT sys.ODCICost, qi sys.ODCIQueryInfo, pred sys.ODCIPredInfo,

15-34
 Chapter 15
Defining Types and Methods for Extensible Optimizing

args sys.ODCIArgDescList, strt NUMBER, stop NUMBER, cmppos NUMBER,

cmpval NUMBER, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsIndexCost(ia sys.ODCIIndexInfo, sel NUMBER,
cost OUT sys.ODCICost, qi sys.ODCIQueryInfo, pred sys.ODCIPredInfo,
args sys.ODCIArgDescList, strt NUMBER, stop NUMBER, cmpval NUMBER,
env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsFunctionCost(func sys.ODCIFuncInfo,
cost OUT sys.ODCICost, args sys.ODCIArgDescList, object PowerDemand_Typ,
cell NUMBER, value NUMBER, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsFunctionCost(func sys.ODCIFuncInfo,
cost OUT sys.ODCICost, args sys.ODCIArgDescList, object PowerDemand_Typ,
value NUMBER, env sys.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIStatsFunctionCost(func sys.ODCIFuncInfo,
cost OUT sys.ODCICost, args sys.ODCIArgDescList, object PowerDemand_Typ,
cell NUMBER, value NUMBER, env sys.ODCIEnv)
RETURN NUMBER
);
/

15.5.2.2 ODCIGetInterfaces()
The ODCIGetInterfaces() function, demonstrated in Example 15-22, returns the list of
names of the interfaces implemented by the type. There is only one set of the
extensible optimizer interface routines, called SYS.ODCISTATS, but the server supports
multiple versions of them for backward compatibility. To specify the current version of
the routines, function ODCIGetInterfaces() must specify SYS.ODCISTATS2 in the OUT,
ODCIObjectList parameter.

Example 15-22 Registering interfaces and Statistics Functions for Power

Demand Cartridge
STATIC FUNCTION ODCIGetInterfaces(
ifclist OUT sys.ODCIObjectList)
RETURN NUMBER IS
BEGIN
ifclist := sys.ODCIObjectList(sys.ODCIObject('SYS','ODCISTATS2'));
RETURN ODCIConst.Success;
END ODCIGetInterfaces;

15.5.2.3 ODCIStatsCollect() Method for PowerDemand_Typ Columns

The ODCIStatsCollect() function, demonstrated in Example 15-23, collects statistics
for columns whose data type is the PowerDemand_Typ object type. The statistics are
collected for each cell in the column over all power grid readings. For a given cell, the
statistics collected are the minimum and maximum power grid readings, and the
number of non-null readings.
The function takes the column information as an object parameter whose type is
SYS.ODCICOLINFO. The type attributes include the table name, column name, and so
on. Options specified in the DBMS_STATS package command used to collect the column
statistics are also passed in as parameters. Since the power demand cartridge uses a
table to store the statistics, the output parameter rawstats is not used in this cartridge.

15-35
 Chapter 15
Defining Types and Methods for Extensible Optimizing

Example 15-23 Registering ODCIStatsCollect() for Power Demand Cartridge

STATIC FUNCTION ODCIStatsCollect(
col sys.ODCIColInfo,
options sys.ODCIStatsOptions,
rawstats OUT RAW,
env sys.ODCIEnv)
RETURN NUMBER IS
cnum INTEGER;
stmt VARCHAR2(1000);
junk INTEGER;
cval NUMBER;
colname VARCHAR2(30) := rtrim(ltrim(col.colName, '"'), '"');
statsexists BOOLEAN := FALSE;
pdemands PowerDemand_Tab%ROWTYPE;
user_defined_stats PowerCartUserStats%ROWTYPE;

CURSOR c1(tname VARCHAR2, cname VARCHAR2) IS

SELECT * FROM PowerCartUserStats
WHERE tab = tname AND col = cname;
CURSOR c2 IS
SELECT * FROM PowerDemand_Tab;

BEGIN
sys.ODCIColInfoDump(col);
sys.ODCIStatsOptionsDump(options);

IF (col.TableSchema IS NULL OR col.TableName IS NULL OR col.ColName IS NULL)

THEN
RETURN ODCIConst.Error;
END IF;

dbms_output.put_line('ODCIStatsCollect>>>>>');
dbms_output.put_line('**** Analyzing column '||col.TableSchema|| '.' ||
col.TableName|| '.' || col.ColName);

-- Check if statistics exist for this column

FOR user_defined_stats IN c1(col.TableName, colname) LOOP
statsexists := TRUE;
EXIT;
END LOOP;

The function checks whether statistics for this column exist. If so, it initializes them to
NULL; otherwise, it creates statistics for each of the 100 cells and initializes them to
NULL.
IF not statsexists THEN
-- column statistics don't exist; create entries for each of the 100 cells
cnum := dbms_sql.open_cursor;
FOR i in 1..100 LOOP
stmt := 'INSERT INTO PowerCartUserStats VALUES( '||''''|| col.TableName ||
''', '||''''||colname||''', '||to_char(i)||', '||'NULL, NULL, NULL)';
dbms_sql.parse(cnum, stmt, dbms_sql.native);
junk := dbms_sql.execute(cnum);
END LOOP;
dbms_sql.close_cursor(cnum);
ELSE
-- column statistics exist; initialize to NULL
cnum := dbms_sql.open_cursor;
stmt := 'UPDATE PowerCartUserStats'||
' SET lo = NULL, hi = NULL, nrows = NULL'||' WHERE tab = '||

15-36
 Chapter 15
Defining Types and Methods for Extensible Optimizing

col.TableName||' AND col = '||colname;

dbms_sql.parse(cnum, stmt, dbms_sql.native);
junk := dbms_sql.execute(cnum);
dbms_sql.close_cursor(cnum);
END IF;

The function collects statistics for the column by reading rows from the table that is
being analyzed. This is done by constructing and executing a SQL statement.
-- For each cell position, the following statistics are collected:
-- maximum value
-- minimum value
-- number of rows (excluding NULLs)
cnum := dbms_sql.open_cursor;
FOR i in 1..100 LOOP
FOR pdemands IN c2 LOOP
IF i BETWEEN pdemands.sample.CellDemandValues.FIRST AND
pdemands.sample.CellDemandValues.LAST THEN
cval := pdemands.sample.CellDemandValues(i);
stmt := 'UPDATE PowerCartUserStats SET '|| 'lo = least(' || 'NVL(' ||
to_char(cval)||', lo), '||'NVL('||'lo, '||to_char(cval)||')), '||
'hi = greatest('||'NVL('||to_char(cval)||', hi), '||'NVL('||
'hi, '||to_char(cval)||')), '||
'nrows = decode(nrows, NULL, decode('||to_char(cval)||
', NULL, NULL, 1), decode('||to_char(cval)||
', NULL, nrows, nrows+1)) '||'WHERE cpos = '||to_char(i)||
' AND tab = '''||col.TableName||''''||' AND col = '''||colname||
'''';
dbms_sql.parse(cnum, stmt, dbms_sql.native);
junk := dbms_sql.execute(cnum);
END IF;
END LOOP;
END LOOP;

The function concludes by closing the cursor and returning a success status.
dbms_sql.close_cursor(cnum);
rawstats := NULL;
return ODCIConst.Success;

END ODCIStatsCollect;

15.5.2.4 ODCIStatsDelete() Method for PowerDemand_Typ Columns

The ODCIStatsDelete() function, demonstrated in Example 15-24, deletes statistics of
columns whose data type is the PowerDemand_Typ object type. The function takes the
column information as an object parameter whose type is SYS.ODCICOLINFO. The type
attributes include the table name, column name, and so on.
Example 15-24 Registering ODCIStatsDelete() for Power Demand Cartridge
STATIC FUNCTION ODCIStatsDelete(
col sys.ODCIColInfo,
env sys.ODCIEnv)
RETURN NUMBER IS
cnum INTEGER;
stmt VARCHAR2(1000);
junk INTEGER;
colname VARCHAR2(30) := rtrim(ltrim(col.colName, '"'), '"');
statsexists BOOLEAN := FALSE;
user_defined_stats PowerCartUserStats%ROWTYPE;

15-37
 Chapter 15
Defining Types and Methods for Extensible Optimizing

CURSOR c1(tname VARCHAR2, cname VARCHAR2) IS

SELECT * FROM PowerCartUserStats
WHERE tab = tname AND col = cname;
BEGIN
sys.ODCIColInfoDump(col);

IF (col.TableSchema IS NULL OR col.TableName IS NULL OR col.ColName IS NULL)

THEN
RETURN ODCIConst.Error;
END IF;

dbms_output.put_line('ODCIStatsDelete>>>>>');
dbms_output.put_line('**** Analyzing (delete) column '|| col.TableSchema||
'.' ||col.TableName||'.'||col.ColName);

The function verifies that statistics for the column exist by checking the statistics table.
If statistics were not collected, then there is nothing to be done. If, however, statistics
are present, it constructs and executes a SQL statement to delete the relevant rows
from the statistics table.
-- Check if statistics exist for this column
FOR user_defined_stats IN c1(col.TableName, colname) LOOP
statsexists := TRUE;
EXIT;
END LOOP;

-- If user-defined statistics exist, delete them

IF statsexists THEN
stmt := 'DELETE FROM PowerCartUserStats'||' WHERE tab = '''||col.TableName||
''''|| ' AND col = ''' || colname || '''';
cnum := dbms_sql.open_cursor;
dbms_output.put_line('ODCIStatsDelete>>>>>');
dbms_output.put_line('ODCIStatsDelete>>>>>' || stmt);
dbms_sql.parse(cnum, stmt, dbms_sql.native);
junk := dbms_sql.execute(cnum);
dbms_sql.close_cursor(cnum);
END IF;

RETURN ODCIConst.Success;
END ODCStatsDelete;

15.5.2.5 ODCIStatsCollect() Method for power_idxtype Domain Indexes

The ODCIStatsCollect() function, demonstrated in Example 15-25, collects statistics
for domain indexes whose indextype is power_idxtype. In the power demand
cartridge, this function simply analyzes the index-organized table that stores the index
data.
The function takes the index information as an object parameter whose type is
SYS.ODCIINDEXINFO. The type attributes include the index name, owner name, and so
on. Options specified by the DBMS_STATS package are used to collect the index
statistics are also passed in as parameters. The output parameter rawstats is not
used.
Example 15-25 Registering ODCIStatsCollect() for Power Demand Cartridge
STATIC FUNCTION ODCIStatsCollect (
ia sys.ODCIIndexInfo,
options sys.ODCIStatsOptions,

15-38
 Chapter 15
Defining Types and Methods for Extensible Optimizing

rawstats OUT RAW,

env sys.ODCIEnv)
RETURN NUMBER IS
stmt VARCHAR2(1000);

BEGIN
-- To analyze a domain index, analyze the table that implements the index
sys.ODCIIndexInfoDump(ia);
sys.ODCIStatsOptionsDump(options);

stmt := 'dbms_stats.gather_table_stats('
|| '''' || ia.IndexSchema || ''', '
|| '''' || ia.IndexName || '_pidx' || ''');';
dbms_output.put_line('**** Analyzing index '
|| ia.IndexSchema || '.' || ia.IndexName);
dbms_output.put_line('SQL Statement: ' || stmt);
EXECUTE IMMEDIATE 'BEGIN ' || stmt || ' END;';
rawstats := NULL;

RETURN ODCIConst.Success;
END ODCIStatsCollect;

15.5.2.6 ODCIStatsDelete() Method for power_idxtype Domain Indexes

The ODCIStatsDelete() function, demonstrated in Example 15-26, deletes statistics for
domain indexes whose indextype is power_idxtype. In the power demand cartridge,
this function simply deletes the statistics of the index-organized table that stores the
index data.
The function takes the index information as an object parameter whose type is
SYS.ODCIINDEXINFO. The type attributes include the index name, owner name, and so
on.
Example 15-26 Registering ODCIStatsDelete() for Domain Indexes in Power
Demand Cartridge
STATIC FUNCTION ODCIStatsDelete(
ia sys.ODCIIndexInfo,
env sys.ODCIEnv)
RETURN NUMBER IS
stmt VARCHAR2(1000);
BEGIN
-- To delete statistics for a domain index, delete the statistics for the
-- table implementing the index
sys.ODCIIndexInfoDump(ia);
stmt := 'dbms_stats.delete_table_stats('|| '''' || ia.IndexSchema || ''', '
|| '''' || ia.IndexName || '_pidx' || ''');';
dbms_output.put_line('**** Analyzing (delete) index '||ia.IndexSchema||'.'||
ia.IndexName);
dbms_output.put_line('SQL Statement: ' || stmt);

EXECUTE IMMEDIATE 'BEGIN ' || stmt || ' END;';

RETURN ODCIConst.Success;
END ODCIStatsDelete;

15.5.2.7 ODCIStatsSelectivity() Method for Specific Queries

The first definition of the ODCIStatsSelectivity() function estimates the selectivity of
operator or function predicates for Specific queries. For example, if a query asks for

15-39
 Chapter 15
Defining Types and Methods for Extensible Optimizing

all instances where cell (3,7) has a value equal to 25, the function estimates the
percentage of rows in which the given cell has the specified value.
The pred parameter contains the function information (the functional implementation of
an operator in an operator predicate); this parameter is an object instance of type
SYS.ODCIPREDINFO. The selectivity is returned as a percentage in the sel output
parameter. The args parameter (an object instance of type SYS.ODCIARGDESCLIST)
contains a descriptor for each argument of the function, and the start and stop values
of the function. For example, if an argument is a column, the argument descriptor
contains the table name, column name, and so on. The strt and stop parameters are
the lower and upper boundary points for the function return value. If the function in a
predicate contains a literal of type PowerDemand_Typ, the object parameter contains
the value in the form of an object constructor. The cell parameter is the cell position
and the value parameter is the value in the cell specified by the function
(PowerXxxxxSpecific_Func).

The selectivity is estimated by using a technique similar to that used for simple range
predicates. For example, a simple estimate for the selectivity of a predicate like
c > v

is (M-v)/(M-m) where m and M are the minimum and maximum values, respectively, for
the column c (as determined from the column statistics), provided the value v lies
between m and M.

The get_selectivity function, demonstrated in Example 15-27, computes the

selectivity of a simple range predicate given the minimum and maximum values of the
column in the predicate. It assumes that the column values in the table are uniformly
distributed between the minimum and maximum values.
Example 15-27 Implementing Selectivity Function for Power Demand Cartridge
CREATE FUNCTION get_selectivity(relop VARCHAR2, value NUMBER,
lo NUMBER, hi NUMBER, ndv NUMBER)
RETURN NUMBER AS
sel NUMBER := NULL;
ndv NUMBER;
BEGIN
-- This function computes the selectivity (as a percentage)
-- of a predicate
-- col <relop> <value>
-- where <relop> is one of: =, !=, <, <=, >, >=
-- <value> is one of: 0, 1
-- lo and hi are the minimum and maximum values of the column in
-- the table. This function performs a simplistic estimation of the
-- selectivity by assuming that the range of distinct values of
-- the column is distributed uniformly in the range lo..hi and that
-- each distinct value occurs nrows/(hi-lo+1) times (where nrows is
-- the number of rows).

IF ndv IS NULL OR ndv <= 0 THEN

RETURN 0;
END IF;

-- col != <value>
IF relop = '!=' THEN
IF value between lo and hi THEN
sel := 1 - 1/ndv;
ELSE
sel := 1;

15-40
 Chapter 15
Defining Types and Methods for Extensible Optimizing

END IF;

-- col = <value>
ELSIF relop = '=' THEN
IF value between lo and hi THEN
sel := 1/ndv;
ELSE
sel := 0;
END IF;

-- col >= <value>

ELSIF relop = '>=' THEN
IF lo = hi THEN
IF value <= lo THEN
sel := 1;
ELSE
sel := 0;
END IF;
ELSIF value between lo and hi THEN
sel := (hi-value)/(hi-lo) + 1/ndv;
ELSIF value < lo THEN
sel := 1;
ELSE
sel := 0;
END IF;

-- col < <value>

ELSIF relop = '<' THEN
IF lo = hi THEN
IF value > lo THEN
sel := 1;
ELSE
sel := 0;
END IF;
ELSIF value between lo and hi THEN
sel := (value-lo)/(hi-lo);
ELSIF value < lo THEN
sel := 0;
ELSE
sel := 1;
END IF;

-- col <= <value>

ELSIF relop = '<=' THEN
IF lo = hi THEN
IF value >= lo THEN
sel := 1;
ELSE
sel := 0;
END IF;
ELSIF value between lo and hi THEN
sel := (value-lo)/(hi-lo) + 1/ndv;
ELSIF value < lo THEN
sel := 0;
ELSE
sel := 1;
END IF;

-- col > <value>

ELSIF relop = '>' THEN
IF lo = hi THEN

15-41
 Chapter 15
Defining Types and Methods for Extensible Optimizing

IF value < lo THEN

sel := 1;
ELSE
sel := 0;
END IF;
ELSIF value between lo and hi THEN
sel := (hi-value)/(hi-lo);
ELSIF value < lo THEN
sel := 1;
ELSE
sel := 0;
END IF;

END IF;

RETURN least(100, ceil(100*sel));

END;
/

The ODCIStatsSelectivity() function, demonstrated in Example 15-28, estimates the

selectivity for function predicates which have constant start and stop values. Further,
the first argument of the function in the predicate must be a column of type
PowerDemand_Typ and the remaining arguments must be constants.

Example 15-28 Registering ODCIStatsSelectivity() for Queries for Power

Demand Cartridge
STATIC FUNCTION ODCIStatsSelectivity(pred sys.ODCIPredInfo,
sel OUT NUMBER, args sys.ODCIArgDescList, strt NUMBER, stop NUMBER,
object PowerDemand_Typ, cell NUMBER, value NUMBER, env sys.ODCIEnv)
RETURN NUMBER IS
fname varchar2(30);
relop varchar2(2);
lo NUMBER;
hi NUMBER;
nrows NUMBER;
colname VARCHAR2(30);
statsexists BOOLEAN := FALSE;
stats PowerCartUserStats%ROWTYPE;
CURSOR c1(cell NUMBER, tname VARCHAR2, cname VARCHAR2) IS
SELECT * FROM PowerCartUserStats
WHERE cpos = cell
AND tab = tname
AND col = cname;
BEGIN
-- compute selectivity only when predicate is of the form:
-- fn(col, <cell>, <value>) <relop> <val>
-- In all other cases, return an error and let the optimizer
-- make a guess. We also assume that the function "fn" has
-- a return value of 0, 1, or NULL.

-- start value
IF (args(1).ArgType != ODCIConst.ArgLit AND
args(1).ArgType != ODCIConst.ArgNull) THEN
RETURN ODCIConst.Error;
END IF;

-- stop value
IF (args(2).ArgType != ODCIConst.ArgLit AND
args(2).ArgType != ODCIConst.ArgNull) THEN

15-42
 Chapter 15
Defining Types and Methods for Extensible Optimizing

RETURN ODCIConst.Error;
END IF;

-- first argument of function

IF (args(3).ArgType != ODCIConst.ArgCol) THEN
RETURN ODCIConst.Error;
END IF;

-- second argument of function

IF (args(4).ArgType != ODCIConst.ArgLit AND
args(4).ArgType != ODCIConst.ArgNull) THEN
RETURN ODCIConst.Error;
END IF;

-- third argument of function

IF (args(5).ArgType != ODCIConst.ArgLit AND
args(5).ArgType != ODCIConst.ArgNull) THEN
RETURN ODCIConst.Error;
END IF;

colname := rtrim(ltrim(args(3).colName, '"'), '"');

The first (column) argument of the function in the predicate must have statistics
collected for it. If statistics have not been collected, ODCIStatsSelectivity() returns an
error status.
-- Check if the statistics table exists (we are using a
-- user-defined table to store the user-defined statistics).
-- Get user-defined statistics: MIN, MAX, NROWS
FOR stats IN c1(cell, args(3).TableName, colname) LOOP
-- Get user-defined statistics: MIN, MAX, NROWS
lo := stats.lo;
hi := stats.hi;
nrows := stats.nrows;
statsexists := TRUE;
EXIT;
END LOOP;

-- If no user-defined statistics were collected, return error

IF not statsexists THEN
RETURN ODCIConst.Error;
END IF;

Each Specific function predicate corresponds to an equivalent range predicate. For

example, the predicate Power_EqualsSpecific_Func(col, 21, 25) = 0, which
checks that the reading in cell 21 is not equal to 25, corresponds to the equivalent
range predicate col[21] != 25.

The ODCIStatsSelectivity() function finds the corresponding range predicates for each
Specific function predicate. There are several boundary cases where the selectivity
can be immediately determined.
-- selectivity is 0 for "fn(col, <cell>, <value>) < 0"
IF (stop = 0 AND
bitand(pred.Flags, ODCIConst.PredIncludeStop) = 0) THEN
sel := 0;
RETURN ODCIConst.Success;
END IF;

-- selectivity is 0 for "fn(col, <cell>, <value>) > 1"

IF (strt = 1 AND

15-43
 Chapter 15
Defining Types and Methods for Extensible Optimizing

bitand(pred.Flags, ODCIConst.PredIncludeStart) = 0) THEN

sel := 0;
RETURN ODCIConst.Success;
END IF;

-- selectivity is 100% for "fn(col, <cell>, <value>) >= 0"

IF (strt = 0 AND
bitand(pred.Flags, ODCIConst.PredExactMatch) = 0 AND
bitand(pred.Flags, ODCIConst.PredIncludeStart) > 0) THEN
sel := 100;
RETURN ODCIConst.Success;
END IF;

-- selectivity is 100% for "fn(col, <cell>, <value>) <= 1"

IF (stop = 1 AND
bitand(pred.Flags, ODCIConst.PredExactMatch) = 0 AND
bitand(pred.Flags, ODCIConst.PredIncludeStop) > 0) THEN
sel := 100;
RETURN ODCIConst.Success;
END IF;

-- get function name

IF bitand(pred.Flags, ODCIConst.PredObjectFunc) > 0 THEN
fname := pred.ObjectName;
ELSE
fname := pred.MethodName;
END IF;

-- convert prefix relational operator to infix:

-- "Power_EqualsSpecific_Func(col, <cell>, <value>) = 1"
-- becomes "col[<cell>] = <value>"

-- Power_EqualsSpecific_Func(col, <cell>, <value>) = 0

-- Power_EqualsSpecific_Func(col, <cell>, <value>) <= 0
-- Power_EqualsSpecific_Func(col, <cell>, <value>) < 1
-- can be transformed to
-- col[<cell>] != <value>
IF (fname LIKE upper('Power_Equals%') AND
(stop = 0 OR
(stop = 1 AND
bitand(pred.Flags, ODCIConst.PredIncludeStop) = 0))) THEN
relop := '!=';

-- Power_LessThanSpecific_Func(col, <cell>, <value>) = 0

-- Power_LessThanSpecific_Func(col, <cell>, <value>) <= 0
-- Power_LessThanSpecific_Func(col, <cell>, <value>) < 1
-- can be transformed to
-- col[<cell>] >= <value>
ELSIF (fname LIKE upper('Power_LessThan%') AND
(stop = 0 OR
(stop = 1 AND
bitand(pred.Flags, ODCIConst.PredIncludeStop) = 0))) THEN
relop := '>=';

-- Power_GreaterThanSpecific_Func(col, <cell>, <value>) = 0

-- Power_GreaterThanSpecific_Func(col, <cell>, <value>) <= 0
-- Power_GreaterThanSpecific_Func(col, <cell>, <value>) < 1
-- can be transformed to
-- col[<cell>] <= <value>
ELSIF (fname LIKE upper('Power_GreaterThan%') AND
(stop = 0 OR

15-44
 Chapter 15
Defining Types and Methods for Extensible Optimizing

(stop = 1 AND
bitand(pred.Flags, ODCIConst.PredIncludeStop) = 0))) THEN
relop := '<=';

-- Power_EqualsSpecific_Func(col, <cell>, <value>) = 1

-- Power_EqualsSpecific_Func(col, <cell>, <value>) >= 1
-- Power_EqualsSpecific_Func(col, <cell>, <value>) > 0
-- can be transformed to
-- col[<cell>] = <value>
ELSIF (fname LIKE upper('Power_Equals%') AND
(strt = 1 OR
(strt = 0 AND
bitand(pred.Flags, ODCIConst.PredIncludeStart) = 0))) THEN
relop := '=';

-- Power_LessThanSpecific_Func(col, <cell>, <value>) = 1

-- Power_LessThanSpecific_Func(col, <cell>, <value>) >= 1
-- Power_LessThanSpecific_Func(col, <cell>, <value>) > 0
-- can be transformed to
-- col[<cell>] < <value>
ELSIF (fname LIKE upper('Power_LessThan%') AND
(strt = 1 OR
(strt = 0 AND
bitand(pred.Flags, ODCIConst.PredIncludeStart) = 0))) THEN
relop := '<';

-- Power_GreaterThanSpecific_Func(col, <cell>, <value>) = 1

-- Power_GreaterThanSpecific_Func(col, <cell>, <value>) >= 1
-- Power_GreaterThanSpecific_Func(col, <cell>, <value>) > 0
-- can be transformed to
-- col[<cell>] > <value>
ELSIF (fname LIKE upper('Power_GreaterThan%') AND
(strt = 1 OR
(strt = 0 AND
bitand(pred.Flags, ODCIConst.PredIncludeStart) = 0))) THEN
relop := '>';

ELSE
RETURN ODCIConst.Error;

END IF;

After the Specific function predicate is transformed into a simple range predicate,
ODCIStatsSelectivity() calls get_selectivity to compute the selectivity for the range
predicate (and thus, equivalently, for the Specific function predicate). It returns with a
success status.
sel := get_selectivity(relop, value, lo, hi, nrows);
RETURN ODCIConst.Success;
END;

15.5.2.8 ODCIStatsIndexCost() Method for Specific Queries

The first definition of the ODCIStatsIndexCost() function, demonstrated in
Example 15-29, estimates the cost of the domain index for Specific queries. For
example, if a query asks for all instances where cell (3,7) has a value equal to 25, the
function estimates the cost of the domain index access path to evaluate this query.
This definition of ODCIStatsIndexCost() differs from the definition in

15-45
 Chapter 15
Defining Types and Methods for Extensible Optimizing

ODCIStatsIndexCost() Method for Any Queries in that it includes the cmppos

parameter for the position of the cell.
The ia parameter contains the index information as an object instance of type
SYS.ODCIINDEXINFO. The sel parameter is the selectivity of the operator predicate as
estimated by the ODCIStatsSelectivity() function for Specific queries. The
estimated cost is returned in the cost output parameter. The qi parameter contains
some information about the query and its environment, such as whether the ALL_ROWS
or FIRST_ROWS optimizer mode is being used. The pred parameter contains the
operator information as an object instance of type SYS.ODCIPREDINFO. The args
parameter contains descriptors of the value arguments of the operator, and the start
and stop values of the operator. The strt and stop parameters are the lower and
upper boundary points for the operator return value. The cmppos parameter is the cell
position and cmpval is the value in the cell specified by the operator
Power_XxxxxSpecific().

In the power demand cartridge, the domain index cost for Specific queries is identical
to the domain index cost for Any queries, so this version of the ODCIStatsIndexCost()
function simply calls the second definition of the function, described in
ODCIStatsIndexCost() Method for Any Queries.

Example 15-29 Registering ODCISIndexCost() for Queries for Power Demand

Cartridge
STATIC FUNCTION ODCIStatsIndexCost(ia sys.ODCIIndexInfo,
sel NUMBER, cost OUT sys.ODCICost, qi sys.ODCIQueryInfo,
pred sys.ODCIPredInfo, args sys.ODCIArgDescList,
strt NUMBER, stop NUMBER, cmppos NUMBER, cmpval NUMBER, env sys.ODCIEnv)
RETURN NUMBER IS
BEGIN
-- This is the cost for queries on a specific cell; simply
-- use the cost for queries on any cell.
RETURN ODCIStatsIndexCost(ia, sel, cost, qi, pred, args,
strt, stop, cmpval, env);
END;

See Also:

• ODCIStatsIndexCost()
• ODCIStatsIndexCost() Method for Any Queries
• ODCIStatsSelectivity()

15.5.2.9 ODCIStatsIndexCost() Method for Any Queries

The second definition of the ODCIStatsIndexCost() function, demonstrated in
Example 15-30, estimates the cost of the domain index for Any queries. For example, if
a query asks for all instances where any cell has a value equal to 25, the function
estimates the cost of the domain index access path to evaluate this query. This
definition of ODCIStatsIndexCost() differs from the definition in
ODCIStatsIndexCost() Method for Specific Queries in that it does not include the
cmppos parameter.

15-46
 Chapter 15
Defining Types and Methods for Extensible Optimizing

The ia parameter contains the index information as an object instance of type

SYS.ODCIINDEXINFO. The sel parameter is the selectivity of the operator predicate as
estimated by the ODCIStatsSelectivity() function for Any queries. The estimated
cost is returned in the cost output parameter. The qi parameter contains some
information about the query and its environment, such as whether the ALL_ROWS or
FIRST_ROWS optimizer mode is being used. The pred parameter contains the operator
information as an object instance of type SYS.ODCIPREDINFO. The args parameter
contains descriptors of the value arguments of the operator, and the start and stop
values of the operator. The strt and stop parameters are the lower and upper
boundary points for the operator return value. The cmpval parameter is the value in the
cell specified by the operator Power_XxxxxAny().

The index cost is estimated as the number of blocks in the index-organized table
implementing the index multiplied by the selectivity of the operator predicate times a
constant factor.
Example 15-30 Registering ODCIStatsIndexCost() for Any Queries for Power
Demand Cartridge
STATIC FUNCTION ODCIStatsIndexCost(ia sys.ODCIIndexInfo,
sel NUMBER, cost OUT sys.ODCICost, qi sys.ODCIQueryInfo,
pred sys.ODCIPredInfo, args sys.ODCIArgDescList,
strt NUMBER, stop NUMBER, cmpval NUMBER, env sys.ODCIEnv)
RETURN NUMBER IS
ixtable VARCHAR2(40);
numblocks NUMBER := NULL;
get_table user_tables%ROWTYPE;
CURSOR c1(tab VARCHAR2) IS
SELECT * FROM user_tables WHERE table_name = tab;
BEGIN
-- This is the cost for queries on any cell.

-- To compute the cost of a domain index, multiply the

-- number of blocks in the table implementing the index
-- with the selectivity

-- Return if we don't have predicate selectivity

IF sel IS NULL THEN
RETURN ODCIConst.Error;
END IF;

cost := sys.ODCICost(NULL, NULL, NULL, NULL);

-- Get name of table implementing the domain index

ixtable := ia.IndexName || '_pidx';

-- Get number of blocks in domain index

FOR get_table IN c1(upper(ixtable)) LOOP
numblocks := get_table.blocks;
EXIT;
END LOOP;

IF numblocks IS NULL THEN

-- Exit if there are no user-defined statistics for the index
RETURN ODCIConst.Error;
END IF;

cost.CPUCost := ceil(400*(sel/100)*numblocks);
cost.IOCost := ceil(1.5*(sel/100)*numblocks);

15-47
 Chapter 15
Defining Types and Methods for Extensible Optimizing

RETURN ODCIConst.Success;
END;

See Also:

• ODCIStatsIndexCost()
• ODCIStatsIndexCost() Method for Specific Queries
• ODCIStatsSelectivity()

15.5.2.10 ODCIStatsFunctionCost() Method

The ODCIStatsFunctionCost() function, demonstrated in Example 15-31, estimates the
cost of evaluating a function Power_XxxxxSpecific_Func() or
Power_XxxxxAny_Func().

The func parameter contains the function information; this parameter is an object
instance of type SYS.ODCIFUNCINFO. The estimated cost is returned in the output cost
parameter. The args parameter as an object instance of type SYS.ODCIARGDESCLIST
contains a descriptor for each argument of the function. If the function contains a literal
of type PowerDemand_Typ as its first argument, the object parameter contains the
value in the form of an object constructor. The value parameter is the value in the cell
specified by the function PowerXxxxxSpecific_Func() or Power_XxxxxAny_Func().

The function cost is simply estimated as some default value depending on the function
name. Since the functions do not read any data from disk, the I/O cost is set to zero.
Example 15-31 Registering ODCIStatsFunctionCost() for Power Demand
Cartridge
STATIC FUNCTION ODCIStatsFunctionCost(func sys.ODCIFuncInfo,
cost OUT sys.ODCICost, args sys.ODCIArgDescList,
object PowerDemand_Typ, value NUMBER, env sys.ODCIEnv)
RETURN NUMBER IS
fname VARCHAR2(30);
BEGIN
cost := sys.ODCICost(NULL, NULL, NULL, NULL);

-- Get function name

IF bitand(func.Flags, ODCIConst.ObjectFunc) > 0 THEN
fname := func.ObjectName;
ELSE
fname := func.MethodName;
END IF;

IF fname LIKE upper('Power_LessThan%') THEN

cost.CPUCost := 5000;
cost.IOCost := 0;
RETURN ODCIConst.Success;
ELSIF fname LIKE upper('Power_Equals%') THEN
cost.CPUCost := 7000;
cost.IOCost := 0;
RETURN ODCIConst.Success;
ELSIF fname LIKE upper('Power_GreaterThan%') THEN
cost.CPUCost := 5000;
cost.IOCost := 0;

15-48
 Chapter 15
Defining Types and Methods for Extensible Optimizing

RETURN ODCIConst.Success;
ELSE
RETURN ODCIConst.Error;
END IF;
END;

15.5.3 Associating the Extensible Optimizer Methods with Database

Objects
In order for the optimizer to use the methods defined in the power_statistics object
type, they have to be associated with the appropriate database objects, as
demonstrated in Example 15-32.
Example 15-32 Using Statistics Methods with Database Objects for Power
Demand Cartridge
Associate statistics type with types, indextypes, and functions
ASSOCIATE STATISTICS WITH TYPES PowerDemand_Typ USING power_statistics;
ASSOCIATE STATISTICS WITH INDEXTYPES power_idxtype USING power_statistics
WITH SYSTEM MANAGED STORAGE TABLES;
ASSOCIATE STATISTICS WITH FUNCTIONS
Power_EqualsSpecific_Func,
Power_GreaterThanSpecific_Func,
Power_LessThanSpecific_Func,
Power_EqualsAny_Func,
Power_GreaterThanAny_Func,
Power_LessThanAny_Func
USING power_statistics;

15.5.4 Analyzing the Database Objects

Analyzing tables, columns, and indexes ensures that the optimizer has the relevant
statistics to estimate accurate costs for various access paths and choose a good plan.
Further, the selectivity and cost functions defined in the power_statistics object type
rely on the presence of statistics. Example 15-33 demonstrates statements that
analyze the database objects and verify that statistics were indeed collected.
Example 15-33 Analyzing Database Objects for the Power Demand Cartridge
-- Analyze the table
EXECUTE dbms_stats.gather_table_stats(
'POWERCARTUSER', 'POWERDEMAND_TAB', cascade => TRUE);

-- Verify that user-defined statistics were collected

SELECT tab tablename, col colname, cpos, lo, hi, nrows
FROM PowerCartUserStats
WHERE nrows IS NOT NULL
ORDER BY cpos;

-- Delete the statistics

EXECUTE dbms_stats.delete_table_stats('POWERCARTUSER', 'POWERDEMAND_TAB');

-- Verify that user-defined statistics were deleted

SELECT tab tablename, col colname, cpos, lo, hi, nrows
FROM PowerCartUserStats
WHERE nrows IS NOT NULL
ORDER BY cpos;

15-49
 Chapter 15
Testing the Domain Index

-- Re-analyze the table

EXECUTE dbms_stats.gather_table_stats(
'POWERCARTUSER', 'POWERDEMAND_TAB',cascade => TRUE);

-- Verify that user-defined statistics were re-collected

SELECT tab tablename, col colname, cpos, lo, hi, nrows
FROM PowerCartUserStats
WHERE nrows IS NOT NULL
ORDER BY cpos;

15.6 Testing the Domain Index

Consider the parts of the power demand example that perform some simple tests of
the domain index, and how to test the domain index and see if it is causing more
efficient execution of queries than would occur without an index. These tests consist
of:
• Creating the power demand table (PowerDemand_Tab) and populating it with a small
amount of data
• Executing some queries before the index is created (and showing the execution
plans without an index being used)
The execution plans show that a full table scan is performed in each case.
• Creating the index on the grid
• Executing the same queries after the index is created (and showing the execution
plans with the index being used)
The execution plans show that Oracle is using the index and not performing full
table scans, thus resulting in more efficient execution.
The statements described here are available online in the example file (tkqxpwr.sql).

15.6.1 Creating and Populating the Power Demand Table

The power demand table, as demonstrated in Example 15-34, is created with two
columns:
• region allows the electric utility to use the grid scheme in multiple areas or states.
Each region, such as New York, New Jersey, Pennsylvania, and so on, is
represented by a 10x10 grid.
• sample is a collection of samplings, or power demand readings from each cell in
the grid, defined using the PowerDemand_Typ object type.
Several rows are inserted, representing power demand data for two regions, 1 and 2,
for several hourly timestamps. For simplicity, values are inserted only into the first 5
positions of each grid; the remaining 95 values are set to null, as demonstrated in
Example 15-35.
Finally, the values for TotGridDemand, MaxCellDemand, and MinCellDemand are
computed and set for each of the newly inserted rows, and these values are displayed,
as demonstrated in Example 15-36.
Example 15-34 Creating PowerDemand_Tab Table for Power Demand Cartridge
CREATE TABLE PowerDemand_Tab (
-- Region for which these power demand readings apply

15-50
 Chapter 15
Testing the Domain Index

region NUMBER,
-- Values for each "sampling" time (for a given hour)
sample PowerDemand_Typ
);

Example 15-35 Populating PowerDemand_Tab Table for Power Demand

Cartridge
-- The next INSERT statements "cheats" by supplying only 5 grid values

-- First 5 INSERT statements are for region 1 (1 AM to 5 AM on 01-Feb-1998).

INSERT INTO PowerDemand_Tab VALUES(1,

PowerDemand_Typ(NULL, NULL, NULL, PowerGrid_Typ(55,8,13,9,5),
to_date('02-01-1998 01','MM-DD-YYYY HH'))
);

INSERT INTO PowerDemand_Tab VALUES(1,

PowerDemand_Typ(NULL, NULL, NULL, PowerGrid_Typ(56,8,13,9,3),
to_date('02-01-1998 02','MM-DD-YYYY HH'))
);

INSERT INTO PowerDemand_Tab VALUES(1,

PowerDemand_Typ(NULL, NULL, NULL, PowerGrid_Typ(55,8,13,9,3),
to_date('02-01-1998 03','MM-DD-YYYY HH'))
);

INSERT INTO PowerDemand_Tab VALUES(1,

PowerDemand_Typ(NULL, NULL, NULL, PowerGrid_Typ(54,8,13,9,3),
to_date('02-01-1998 04','MM-DD-YYYY HH'))
);

INSERT INTO PowerDemand_Tab VALUES(1,

PowerDemand_Typ(NULL, NULL, NULL, PowerGrid_Typ(54,8,12,9,3),
to_date('02-01-1998 05','MM-DD-YYYY HH'))
);

-- Also insert some rows for region 2.

INSERT INTO PowerDemand_Tab VALUES(2,

PowerDemand_Typ(NULL, NULL, NULL, PowerGrid_Typ(9,8,11,16,5),
to_date('02-01-1998 01','MM-DD-YYYY HH'))
);

INSERT INTO PowerDemand_Tab VALUES(2,

PowerDemand_Typ(NULL, NULL, NULL, PowerGrid_Typ(9,8,11,20,5),
to_date('02-01-1998 02','MM-DD-YYYY HH'))
);

Example 15-36 Computing Grid and Cell Demands for Power Demand
Cartridge
DECLARE
CURSOR c1 IS SELECT Sample, Region FROM PowerDemand_Tab FOR UPDATE;
s PowerDemand_Typ;
r NUMBER;
BEGIN
OPEN c1;
LOOP
FETCH c1 INTO s,r;
EXIT WHEN c1%NOTFOUND;

15-51
 Chapter 15
Testing the Domain Index

s.SetTotalDemand;
s.SetMaxDemand;
s.SetMinDemand;
dbms_output.put_line(s.TotGridDemand);
dbms_output.put_line(s.MaxCellDemand);
dbms_output.put_line(s.MinCellDemand);
UPDATE PowerDemand_Tab SET Sample = s WHERE CURRENT OF c1;
END LOOP;
CLOSE c1;
END;
/

-- Examine the values.

SELECT region, P.Sample.TotGridDemand, P.Sample.MaxCellDemand,
P.Sample.MinCellDemand,
to_char(P.sample.sampletime, 'MM-DD-YYYY HH')
FROM PowerDemand_Tab P;

15.6.2 Querying Without the Index

The queries used here are executed by applying the underlying function
PowerEqualsSpecific_Func() for every row in the table, because the index has not
yet been defined.
The example file includes queries that check, both for a specific cell number and for
any cell number, for values equal to, greater than, and less than a specified value. For
example, the equality queries are demonstrated in Example 15-37.
The execution plans show that a full table scan is performed in each case:
OPERATIONS OPTIONS OBJECT_NAME
--------------- --------------- ---------------
SELECT STATEMENT
TABLE ACCESS FULL POWERDEMAND_TAB

Example 15-37 Making Equality Queries for Power Demand Cartridge

SET SERVEROUTPUT ON
-------------------------------------------------------------------
-- Query, referencing the operators (without index)
-------------------------------------------------------------------
explain plan for
SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
P.Sample.MinCellDemand
FROM PowerDemand_Tab P
WHERE Power_Equals(P.Sample,2,10) = 1;
@tkoqxpll

SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,

P.Sample.MinCellDemand
FROM PowerDemand_Tab P
WHERE Power_Equals(P.Sample,2,10) = 1;

explain plan for

SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
P.Sample.MinCellDemand
FROM PowerDemand_Tab P
WHERE Power_Equals(P.Sample,1,25) = 1;
@tkoqxpll

15-52
 Chapter 15
Testing the Domain Index

SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,

P.Sample.MinCellDemand
FROM PowerDemand_Tab P
WHERE Power_Equals(P.Sample,1,25) = 1;

explain plan for

SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
P.Sample.MinCellDemand
FROM PowerDemand_Tab P
WHERE Power_Equals(P.Sample,2,8) = 1;
@tkoqxpll

SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,

P.Sample.MinCellDemand
FROM PowerDemand_Tab P
WHERE Power_Equals(P.Sample,2,8) = 1;

explain plan for

SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
P.Sample.MinCellDemand
FROM PowerDemand_Tab P
WHERE Power_EqualsAny(P.Sample,9) = 1;
@tkoqxpll

SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,

P.Sample.MinCellDemand
FROM PowerDemand_Tab P
WHERE Power_EqualsAny(P.Sample,9) = 1;

15.6.3 Creating the Index

The index is created on the Sample column in the power demand table, as
demonstrated in Example 15-38.
Example 15-38 Creating an Index in PowerDemand_Tab Table for Power
Demand Cartridge
CREATE INDEX PowerIndex ON PowerDemand_Tab(Sample)
INDEXTYPE IS power_idxtype;

15.6.4 Querying with the Index

The queries used here are identical to those in "Querying Without the Index", but this
time the index is used.
The execution plans show that Oracle is using the domain index and not performing
full table scans, thus resulting in more efficient execution, as demonstrated in
Example 15-39.
Example 15-39 Making Equality Queries with Index for Power Demand
Cartridge
SQLPLUS> -------------------------------------------------------------------
SQLPLUS> -- Query, referencing the operators (with index)
SQLPLUS> -------------------------------------------------------------------
SQLPLUS> explain plan for
2> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
3> P.Sample.MinCellDemand
4> FROM PowerDemand_Tab P
5> WHERE Power_Equals(P.Sample,2,10) = 1;

15-53
 Chapter 15
Testing the Domain Index

Statement processed.
SQLPLUS> @tkoqxpll
SQLPLUS> set echo off
Echo OFF
Charwidth 15
OPERATIONS OPTIONS OBJECT_NAME
--------------- --------------- ---------------
SELECT STATEMEN
TABLE ACCESS BY ROWID POWERDEMAND_TAB
DOMAIN INDEX POWERINDEX
3 rows selected.
Statement processed.
Echo ON
SQLPLUS>
SQLPLUS> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
2> P.Sample.MinCellDemand
3> FROM PowerDemand_Tab P
4> WHERE Power_Equals(P.Sample,2,10) = 1;
REGION SAMPLE.TOT SAMPLE.MAX SAMPLE.MIN
---------- ---------- ---------- ----------
0 rows selected.
ODCIIndexStart>>>>>
ODCIIndexInfo
Index owner : POWERCARTUSER
Index name : POWERINDEX
Table owner : POWERCARTUSER
Table name : POWERDEMAND_TAB
Indexed column : "SAMPLE"
Indexed column type :POWERDEMAND_TYP
Indexed column type schema:POWERCARTUSER
ODCIPredInfo
Object owner : POWERCARTUSER
Object name : POWER_EQUALS
Method name :
Predicate bounds flag :
Exact Match
Include Start Key
Include Stop Key
start key : 1
stop key : 1
compare position : 2
compare value : 10
ODCIIndexStart>>>>>select r from POWERCARTUSER.POWERINDEX_pidx where cpos ='2' and
cval ='10'
ODCIIndexFetch>>>>>
Nrows : 2000
ODCIIndexClose>>>>>
SQLPLUS>
SQLPLUS> explain plan for
2> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
3> P.Sample.MinCellDemand
4> FROM PowerDemand_Tab P
5> WHERE Power_Equals(P.Sample,2,8) = 1;
Statement processed.
SQLPLUS> @tkoqxpll
SQLPLUS> set echo off
Echo OFF
Charwidth 15
OPERATIONS OPTIONS OBJECT_NAME
--------------- --------------- ---------------
SELECT STATEMEN

15-54
 Chapter 15
Testing the Domain Index

TABLE ACCESS BY ROWID POWERDEMAND_TAB

DOMAIN INDEX POWERINDEX
3 rows selected.
Statement processed.
Echo ON
SQLPLUS>
SQLPLUS> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
2> P.Sample.MinCellDemand
3> FROM PowerDemand_Tab P
4> WHERE Power_Equals(P.Sample,2,8) = 1;
REGION SAMPLE.TOT SAMPLE.MAX SAMPLE.MIN
---------- ---------- ---------- ----------
1 90 55 5
1 89 56 3
1 88 55 3
1 87 54 3
1 86 54 3
2 49 16 5
2 53 20 5
7 rows selected.
ODCIIndexStart>>>>>
ODCIIndexInfo
Index owner : POWERCARTUSER
Index name : POWERINDEX
Table owner : POWERCARTUSER
Table name : POWERDEMAND_TAB
Indexed column : "SAMPLE"
Indexed column type :POWERDEMAND_TYP
Indexed column type schema:POWERCARTUSER
ODCIPredInfo
Object owner : POWERCARTUSER
Object name : POWER_EQUALS
Method name :
Predicate bounds flag :
Exact Match
Include Start Key
Include Stop Key
start key : 1
stop key : 1
compare position : 2
compare value : 8
ODCIIndexStart>>>>>select r from POWERCARTUSER.POWERINDEX_pidx where cpos ='2' and
cval ='8'
ODCIIndexFetch>>>>>
Nrows : 2000
ODCIIndexClose>>>>>
SQLPLUS>
SQLPLUS> explain plan for
2> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
3> P.Sample.MinCellDemand
4> FROM PowerDemand_Tab P
5> WHERE Power_EqualsAny(P.Sample,9) = 1;
Statement processed.
SQLPLUS> @tkoqxpll
SQLPLUS> set echo off
Echo OFF
Charwidth 15
OPERATIONS OPTIONS OBJECT_NAME
--------------- --------------- ---------------
SELECT STATEMEN
TABLE ACCESS BY ROWID POWERDEMAND_TAB

15-55
 Chapter 15
Testing the Domain Index

DOMAIN INDEX POWERINDEX

3 rows selected.
Statement processed.
Echo ON
SQLPLUS>
SQLPLUS> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
2> P.Sample.MinCellDemand
3> FROM PowerDemand_Tab P
4> WHERE Power_EqualsAny(P.Sample,9) = 1;
REGION SAMPLE.TOT SAMPLE.MAX SAMPLE.MIN
---------- ---------- ---------- ----------
1 90 55 5
1 89 56 3
1 88 55 3
1 87 54 3
1 86 54 3
2 49 16 5
2 53 20 5
7 rows selected.
ODCIIndexStart>>>>>
ODCIIndexInfo
Index owner : POWERCARTUSER
Index name : POWERINDEX
Table owner : POWERCARTUSER
Table name : POWERDEMAND_TAB
Indexed column : "SAMPLE"
Indexed column type :POWERDEMAND_TYP
Indexed column type schema:POWERCARTUSER
ODCIPredInfo
Object owner : POWERCARTUSER
Object name : POWER_EQUALSANY
Method name :
Predicate bounds flag :
Exact Match
Include Start Key
Include Stop Key
start key : 1
stop key : 1
compare value : 9
ODCIIndexStart>>>>>select distinct r from POWERCARTUSER.POWERINDEX_pidx where cval
='9'
ODCIIndexFetch>>>>>
Nrows : 2000
ODCIIndexClose>>>>>
SQLPLUS>
SQLPLUS> explain plan for
2> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
3> P.Sample.MinCellDemand
4> FROM PowerDemand_Tab P
5> WHERE Power_GreaterThanAny(P.Sample,50) = 1;
Statement processed.
SQLPLUS> @tkoqxpll
SQLPLUS> set echo off
Echo OFF
Charwidth 15
OPERATIONS OPTIONS OBJECT_NAME
--------------- --------------- ---------------
SELECT STATEMEN
TABLE ACCESS BY ROWID POWERDEMAND_TAB
DOMAIN INDEX POWERINDEX
3 rows selected.

15-56
 Chapter 15
Testing the Domain Index

Statement processed.
Echo ON
SQLPLUS>
SQLPLUS> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
2> P.Sample.MinCellDemand
3> FROM PowerDemand_Tab P
4> WHERE Power_GreaterThanAny(P.Sample,50) = 1;
REGION SAMPLE.TOT SAMPLE.MAX SAMPLE.MIN
---------- ---------- ---------- ----------
1 90 55 5
1 89 56 3
1 88 55 3
1 87 54 3
1 86 54 3
5 rows selected.
ODCIIndexStart>>>>>
ODCIIndexInfo
Index owner : POWERCARTUSER
Index name : POWERINDEX
Table owner : POWERCARTUSER
Table name : POWERDEMAND_TAB
Indexed column : "SAMPLE"
Indexed column type :POWERDEMAND_TYP
Indexed column type schema:POWERCARTUSER
ODCIPredInfo
Object owner : POWERCARTUSER
Object name : POWER_GREATERTHANANY
Method name :
Predicate bounds flag :
Exact Match
Include Start Key
Include Stop Key
start key : 1
stop key : 1
compare value : 50
ODCIIndexStart>>>>>select distinct r from POWERCARTUSER.POWERINDEX_pidx where cv
al >'50'
ODCIIndexFetch>>>>>
Nrows : 2000
ODCIIndexClose>>>>>
SQLPLUS>
SQLPLUS> explain plan for
2> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,
3> P.Sample.MinCellDemand
4> FROM PowerDemand_Tab P
5> WHERE Power_LessThanAny(P.Sample,50) = 0;
Statement processed.
SQLPLUS> @tkoqxpll
SQLPLUS> set echo off
Echo OFF
Charwidth 15
OPERATIONS OPTIONS OBJECT_NAME
--------------- --------------- ---------------
SELECT STATEMEN
TABLE ACCESS BY ROWID POWERDEMAND_TAB
DOMAIN INDEX POWERINDEX
3 rows selected.
Statement processed.
Echo ON
SQLPLUS>
SQLPLUS> SELECT P.Region, P.Sample.TotGridDemand ,P.Sample.MaxCellDemand,

15-57
 Chapter 15
Testing the Domain Index

2> P.Sample.MinCellDemand
3> FROM PowerDemand_Tab P
4> WHERE Power_LessThanAny(P.Sample,50) = 0;
REGION SAMPLE.TOT SAMPLE.MAX SAMPLE.MIN
---------- ---------- ---------- ----------
0 rows selected.
ODCIIndexStart>>>>>
ODCIIndexInfo
Index owner : POWERCARTUSER
Index name : POWERINDEX
Table owner : POWERCARTUSER
Table name : POWERDEMAND_TAB
Indexed column : "SAMPLE"
Indexed column type :POWERDEMAND_TYP
Indexed column type schema:POWERCARTUSER
ODCIPredInfo
Object owner : POWERCARTUSER
Object name : POWER_LESSTHANANY
Method name :
Predicate bounds flag :
Exact Match
Include Start Key
Include Stop Key
start key : 0
stop key : 0
compare value : 50
ODCIIndexStart>>>>>select distinct r from POWERCARTUSER.POWERINDEX_pidx minus se
lect distinct r from POWERCARTUSER.POWERINDEX_pidx where cval <'50'
ODCIIndexFetch>>>>>
Nrows : 2000
ODCIIndexClose>>>>>

15-58
16
PSBTREE: Extensible Indexing Example
This an example of extensible indexing, with C-language implementation of ODCIIndex
interface routines.

16.1 About the PSBTREE Example

Consider how to implement the extensible indexing interface routines in C. The
example's focus is on topics that are common to all implementations; it does not
expose domain-specific details.
The code for the example is in the demo directory, in the file extdemo6.sql. It extends
an earlier example (extdemo2.sql, also in demo directory) by adding to the indextype
support for local domain indexes on range partitioned tables.

16.2 Design of the Indextype

The indextype implemented here, called PSBtree, operates like a b-tree index. It
supports three user-defined operators: eq (equals), lt (less than), and gt (greater
than). These operators operate on operands of VARCHAR2 data type.

The index data consists of records of the form <key, rid> where key is the value of
the indexed column and rid is the row identifier of the corresponding row. To simplify
the implementation of the indextype, the index data is stored in an system-partitioned
table.
When an index is a system-managed local domain index, one partition in a system-
partitioned table is created for each partition to store the index data for that partition.
Thus, the index manipulation routines merely translate operations on the PSBtree into
operations on the table partition that stores the index data.
When a user creates a PSBtree index (a local index), n table partitions are created
consisting of the indexed column and a rowid column, where n is the number of
partitions in the base table. Inserts into the base table cause appropriate insertions
into the affected index table partition. Deletes and updates are handled similarly.
When the PSBtree is queried based on a user-defined operator (one of gt, lt and eq),
an appropriate query is issued against the index table partitions to retrieve all the
satisfying rows. Appropriate partition pruning occurs, and only the index table
partitions that correspond to the relevant, or "interesting", partitions are accessed.

16.3 Implementing Operators

The PSBtree indextype supports three operators: eq, gt and lt. Each operator has a
corresponding functional implementation.

16-1
 Chapter 16
Implementing Operators

16.3.1 Functional Implementations

Consider the functional implementation of comparison operators. The Implementing
the EQUALS Operator section explains how to implement eq (equals), the
Implementing the LESS THAN Operator section explains how to implement lt (less
than), and the Implementing the GREATER THAN Operator section explains how to
implement gt (greater than) operators.

16.3.1.1 Implementing the EQUALS Operator

The functional implementation for eq is provided by a function (bt_eq) that takes in two
VARCHAR2 parameters and returns 1 if they are equal and 0 otherwise.
CREATE FUNCTION bt_eq(a VARCHAR2, b VARCHAR2) RETURN NUMBER AS
BEGIN
IF a = b then
RETURN 1;
ELSE
RETURN 0;
END IF;
END;

16.3.1.2 Implementing the LESS THAN Operator

The functional implementation for lt is provided by a function (bt_lt) that takes in two
VARCHAR2 parameters and returns 1 if the first parameter is less than the second, 0
otherwise.
CREATE FUNCTION bt_lt(a VARCHAR2, b VARCHAR2) RETURN NUMBER AS
BEGIN
IF a < b then
RETURN 1;
ELSE
RETURN 0;
END IF;
END;

16.3.1.3 Implementing the GREATER THAN Operator

The functional implementation for gt is provided by a function (bt_gt) that takes in two
VARCHAR2 parameters and returns 1 if the first parameter is greater than the second, 0
otherwise.
CREATE FUNCTION bt_gt(a VARCHAR2, b VARCHAR2) RETURN NUMBER AS
BEGIN
IF a > b then
RETURN 1;
ELSE
RETURN 0;
END IF;
END;

16-2
 Chapter 16
Implementing the ODCIIndex Interfaces

16.3.2 Operators
To create the operator, you must specify the signature of the operator along with its
return type and its functional implementation. Example 16-1 shows how to create eq
(equals), Example 16-2 shows how to create lt (less than), and Example 16-3 shows
how to create gt (greater than) operators.

Example 16-1 Creating the EQUALS Operator

CREATE OPERATOR eq
BINDING (VARCHAR2, VARCHAR2) RETURN NUMBER
USING bt_eq;

Example 16-2 Creating the LESS THAN Operator

CREATE OPERATOR lt
BINDING (VARCHAR2, VARCHAR2) RETURN NUMBER
USING bt_lt;

Example 16-3 Creating the GREATER THAN Operator

CREATE OPERATOR gt
BINDING (VARCHAR2, VARCHAR2) RETURN NUMBER
USING bt_gt;

16.4 Implementing the ODCIIndex Interfaces

To implement the PSBTREE, you must implement the ODCIIndexXXX() routines. You can
implement the index routines in any language supported by Oracle. This discussion
implements the ODCIGetInterfaces() routine in the C programming language. Note
that these require advance setup, such as creating a library object, extdemo6l, for your
compiled C code.

16.4.1 Defining an Implementation Type for PSBTREE

Define an implementation type that implements the ODCIIndex interface routines, as
demonstrated in Example 16-4.
Example 16-4 Creating a PSBTREE Index Type
CREATE TYPE psbtree_im AUTHID CURRENT_USER AS OBJECT
(
scanctx RAW(4),
STATIC FUNCTION ODCIGetInterfaces(ifclist OUT SYS.ODCIObjectList)
RETURN NUMBER,
STATIC FUNCTION ODCIIndexCreate (ia SYS.ODCIIndexInfo, parms VARCHAR2,
env SYS.ODCIEnv) RETURN NUMBER,
STATIC FUNCTION ODCIIndexAlter (ia sys.ODCIIndexInfo,
parms IN OUT VARCHAR2, altopt number, env sys.ODCIEnv) RETURN NUMBER,
STATIC FUNCTION ODCIIndexDrop(ia SYS.ODCIIndexInfo, env SYS.ODCIEnv)
RETURN NUMBER,
STATIC FUNCTION ODCIIndexExchangePartition(ia SYS.ODCIIndexInfo,
ia1 SYS.ODCIIndexInfo, env SYS.ODCIEnv) RETURN NUMBER,
STATIC FUNCTION ODCIIndexUpdPartMetadata(ia sys.ODCIIndexInfo,
palist sys.ODCIPartInfoList, env sys.ODCIEnv) RETURN NUMBER,
STATIC FUNCTION ODCIIndexInsert(ia SYS.ODCIIndexInfo, rid VARCHAR2,
newval VARCHAR2, env SYS.ODCIEnv) RETURN NUMBER,

16-3
 Chapter 16
Implementing the ODCIIndex Interfaces

STATIC FUNCTION ODCIIndexDelete(ia SYS.ODCIIndexInfo, rid VARCHAR2,

oldval VARCHAR2, env SYS.ODCIEnv) RETURN NUMBER,
STATIC FUNCTION ODCIIndexUpdate(ia SYS.ODCIIndexInfo, rid VARCHAR2,
oldval VARCHAR2, newval VARCHAR2, env SYS.ODCIEnv) RETURN NUMBER,
STATIC FUNCTION ODCIIndexStart(sctx IN OUT psbtree_im, ia SYS.ODCIIndexInfo,
op SYS.ODCIPredInfo, qi sys.ODCIQueryInfo, strt number, stop number,
cmpval VARCHAR2, env SYS.ODCIEnv) RETURN NUMBER,
MEMBER FUNCTION ODCIIndexFetch(nrows NUMBER, rids OUT SYS.ODCIridlist,
env SYS.ODCIEnv) RETURN NUMBER,
MEMBER FUNCTION ODCIIndexClose(env SYS.ODCIEnv) RETURN NUMBER
);
/
SHOW ERRORS

16.4.2 Creating the Implementation Type Body

Define the implementation type body, as demonstrated in Example 16-5.
Example 16-5 Creating the Implementation Body for PBSTREE
CREATE OR REPLACE TYPE BODY psbtree_im IS

16.4.3 Defining PL/SQL Routines in the Implementation Body

Consider how to implement the index definition routines in PL/SQL.

16.4.3.1 Implementing ODCIGetInterfaces() for PBSTREE in PL/SQL

The ODCIGetInterfaces() routine returns the expected interface name through its OUT
parameter.
STATIC FUNCTION ODCIGetInterfaces(
ifclist OUT sys.ODCIObjectList)
RETURN NUMBER IS
BEGIN
ifclist := sys.ODCIObjectList(sys.ODCIObject('SYS','ODCIINDEX2'));
RETURN ODCIConst.Success;
END ODCIGetInterfaces;

16.4.3.2 Implementing ODCIIndexCreate() for PBSTREE in PL/SQL

The ODCIIndexCreate() routine creates a system-partitioned index storage table with
two columns. The first column stores the VARCHAR2 indexed column value. The routine
makes use of the information passed in to determine the context in which it is invoked.
Dynamic SQL is used to execute the dynamically constructed SQL statement.
STATIC FUNCTION ODCIIndexCreate (
ias sys.ODCIIndexInfo,
parms VARCHAR2,
env sys.ODCIEnv)
RETURN NUMBER IS
i INTEGER;
stmt VARCHAR2(2000);
cursor cur1(ianame VARCHAR2) IS
SELECT partition_name, parameters
FROM user_ind_partitions
WHERE index_name = ianame
ORDER BY partition_position;

16-4
 Chapter 16
Implementing the ODCIIndex Interfaces

cursor cur2(ianame VARCHAR2) IS

SELECT subpartition_name, parameters
FROM user_ind_subpartitions
WHERE index_name = ianame
ORDER BY partition_position, subpartition_position;
BEGIN
stmt := '';

IF (env.CallProperty is null) THEN

stmt := 'create table ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbtree(f1 VARCHAR2(1000), f2 rowid)';

ELSIF (env.CallProperty = sys.ODCIConst.FirstCall) THEN

stmt := '';
i := 1;

IF (bitand(ia.IndexInfoFlags, ODCIConst.CompPartn) = 0) THEN

FOR c1 in cur1(ia.IndexName) LOOP
IF (i > 1) THEN
stmt := stmt || ',';
END IF;
stmt := stmt || 'partition ' || c2.partition_name;
i := i + 1;
END LOOP;
ELSE
FOR c1 in cur1(ia.IndexName) LOOP
IF (i > 1) THEN
stmt := stmt || ',';
END IF;
stmt := stmt || 'partition ' || c2.subpartition_name;
i := i + 1;
END LOOP;
END IF;
stmt := 'create table ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbtree (f1 VARCHAR2(1000), f2 rowid) partition by system ' ||
'(' || stmt || ')';

ELSIF (env.CallProperty = sys.ODCIConst.FinalCall) THEN

stmt := 'create index ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbti on ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbtree (f1) local';
END IF;

dbms_output.put_line('Create');
dbms_output.put_line(stmt);

-- execute the statement

IF ((env.CallProperty is null) OR
(env.CallProperty = sys.ODCIConst.FirstCall) OR
(env.CallProperty = sys.ODCIConst.FinalCall) ) THEN
execute immediate stmt;

IF (env.CallProperty is null) THEN

execute immediate 'insert into ' || ia.IndexSchema || '.' || ia.IndexName
|| '_sbtree select ' || ia.IndexCols(1).ColName || ', ROWID from ' ||
ia.IndexCols(1).TableSchema || '.' || ia.IndexCols(1).TableName;
execute immediate 'create index ' || ia.IndexSchema || '.' ||
ia.IndexName || '_sbti on ' || ia.IndexSchema || '.' ||
ia.IndexName || '_sbtree (f1)';
END IF;
END IF;

16-5
 Chapter 16
Implementing the ODCIIndex Interfaces

RETURN ODCIConst.Success;
END ODCIIndexCreate;

16.4.3.3 Implementing ODCIIndexDrop() for PBSTREE in PL/SQL

The ODCIIndexDrop() routine drops the index storage tables.
STATIC FUNCTION ODCIIndexDrop(
ia sys.ODCIIndexInfo,
env sys.ODCIEnv)
RETURN NUMBER IS
stmt VARCHAR2(1000);
cnum INTEGER;
junk INTEGER;
BEGIN
-- construct the sql statement
stmt := '';

IF (env.CallProperty is null) THEN

stmt := 'drop table ' || ia.IndexSchema || '.' || ia.IndexName || '_sbtree';
dbms_output.put_line('Drop');
dbms_output.put_line(stmt);
execute immediate stmt;
END IF;
RETURN ODCIConst.Success;
END ODCIIndexDrop;

16.4.3.4 Implementing ODCIIndexAlter() for PSBTREE in PL/SQL

The ODCIIndexAlter() routine can perform many index alteration tasks, such as
rebuilding and renaming an index.
STATIC FUNCTION ODCIIndexAlter (
ia sys.ODCIIndexInfo,
parms IN OUT VARCHAR2,
altopt NUMBER,
env sys.ODCIEnv)
RETURN NUMBER IS
stmt VARCHAR2(2000);
BEGIN
stmt := '';
IF (altopt = ODCIConst.AlterIndexRebuild) THEN
IF (ia.IndexPartition is null) THEN
stmt := 'insert into ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbtree select ' || ia.IndexCols(1).ColName || ', ROWID from ' ||
ia.IndexCols(1).ColName || ', ROWID from ' ||
ia.IndexCols(1).TableSchema || '.' || ia.IndexCols(1).TableName;
ELSIF (bitand(ia.IndexInfoFlags, ODCIConst.CompPartn) = 0) THEN
stmt := 'insert into ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbtree select partition (' || ia.IndexPartition || ')' ||
ia.IndexCols(1).ColName || ', ROWID from ' ||
ia.IndexCols(1).TableSchema || '.' || ia.IndexCols(1).TableName ||
' partition (' || ia.IndexCols(1).TablePartition || ')';
ELSE
stmt := 'insert into ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbtree select partition (' || ia.IndexPartition || ')' ||
ia.IndexCols(1).ColName || ', ROWID from ' ||
ia.IndexCols(1).TableSChema || '.' || ia.IndexCols(1).TableName ||
' subpartition (' || ia.IndexCols(1).TablePartition || ')';

16-6
 Chapter 16
Implementing the ODCIIndex Interfaces

END IF;

ELSIF (altopt = ODCIConst.AlterIndexRename) THEN

IF (ia.IndexPartition is not null) THEN
stmt := 'alter table ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbtree rename partition ' || ia.IndexPartition || ' to ' || parms;
ELSE
stmt := 'alter table ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbtree rename to ' || parms || '_sbtree';
END IF;
END IF;

dbms_output.put_line('Alter');
IF ((altopt = ODCIConst.AlterIndexRebuild) OR (altopt =
ODCIConst.AlterIndexRename))
THEN
dbms_output.put_line(stmt);
execute immediate stmt;
END IF;
RETURN ODCIConst.Success;
END ODCIIndexAlter;

16.4.3.5 Implementing ODCIIndexUpdPartMetadata() for PSBTREE in PL/SQL

To handle partition maintenance operations, the kernel performs the maintenance
tasks on behalf of the user. The indextype, to maintain its metadata, should have the
ODCIIndexUpdPartMetadata() routine.
STATIC FUNCTION ODCIIndexUpdPartMetadata(
ia sys.ODCIIndexInfo,
palist sys.ODCIPartInfoList,
env sys.ODCIEnv)
RETURN NUMBER IS
col number;
BEGIN
dbms_output.put_line('ODCIUpdPartMetadata');
sys.ODCIIndexInfoDump(ia);
sys.ODCIPartInfoListDump(palist);
sys.ODCIEnvDump(env);
RETURN ODCIConst.Success;
END ODCIIndexUpdPartMetadata;

16.4.3.6 Implementing ODCIIndexExchangePartition() for PSBTREE in PL/SQL

The ODCIIndexExchangePartition() exchanges the index storage tables for the index
partition being exchanged, with the index storage table for the global domain index.
STATIC FUNCTION ODCIIndexExchangePartition(
ia sys.ODCIIndexInfo,
ia1 sys.ODCIIndexInfo,
env sys.ODCIEnv)
RETURN NUMBER IS
stmt VARCHAR2(2000);
cnum INTEGER;
junk INTEGER;
BEGIN
stmt := '';
dbms_output.put_line('Exchange Partition');

-- construct the sql statements

16-7
 Chapter 16
Implementing the ODCIIndex Interfaces

IF bitand(ia.IndexInfoFlags, ODCIConst.CompPartn) = 0 OR
bitand(ia.IndexInfoFlags, ODCIConst.SubPartn) = ODCIConst.SubPartn THEN
-- non-composite partitioned or exchanging subpartition
stmt := 'alter table ' || ia.IndexSchema || '.' || ia.IndexName ||
'_sbtree exchange partition ' || ia.IndexPartition || ' with table ' ||
ia1.IndexSchema || '.' || ia1.IndexName || '_sbtree';

dbms_output.put_line(stmt);
execute immediate stmt;
ELSE
-- composite partition exchange
stmt := 'create table temp_exch (f1 VARCHAR2(1000), f2 rowid)';
dbms_output.put_line(stmt);
execute immediate stmt;

stmt := 'alter table ' || ia.IndexSchema || '.' || ia.IndexName ||

'_sbtree exchange partition ' || ia.IndexPartition || ' with table ' ||
'temp_exch';
dbms_output.put_line(stmt);
execute immediate stmt;

stmt := 'alter table ' || ia1.IndexSchema || '.' || ia1.IndexName ||

'_sbtree exchange partition ' || ia1.IndexPartition || ' with table ' ||
'temp_exch';
dbms_output.put_line(stmt);
execute immediate stmt;

stmt := 'alter table ' || ia.IndexSchema || '.' || ia.IndexName ||

'_sbtree exchange partition ' || ia.IndexPartition || ' with table ' ||
'temp_exch';
dbms_output.put_line(stmt);
execute immediate stmt;

-- exchange done, drop temporal table

stmt := 'drop table temp_exch';
dbms_output.put_line(stmt);
execute immediate stmt;
END IF;

RETURN ODCIConst.Success;
END ODCIIndexExchangePartition;

16.4.4 Registering the C Implementation of the ODCIIndexXXX()

Methods
After creating the extdemo6l library object for the compiled C methods, you must
register the implementations of each of the routines. The Registering the
Implementation of ODCIIndexInsert() section demonstrates how to register the
ODCIIndexInsert() implementation, the Registering the Implementation of
ODCIIndexDelete() section registers the ODCIIndexDelete() implementation, the
Registering the Implementation of ODCIIndexUpdate() section registers the
ODCIIndexUpdate() implementation, the Registering the Implementation of
ODCIIndexStart() section registers the ODCIIndexStart() implementation, the
Registering the Implementation of ODCIIndexFetch() section registers the
ODCIIndexFetch() implementation, and the Registering the Implementation of
ODCIIndexClose() section registers the ODCIIndexClose() implementation.

16-8
 Chapter 16
Implementing the ODCIIndex Interfaces

16.4.4.1 Registering the Implementation of ODCIIndexInsert()

Register the implementation of the ODCIIndexInsert() routine.
STATIC FUNCTION ODCIIndexInsert(
ia SYS.ODCIIndexInfo,
rid VARCHAR2,
newval VARCHAR2,
env SYS.ODCIEnv)
RETURN NUMBER AS EXTERNAL
name "qxiqtbspi"
library extdemo6l
with context
parameters (
context,
ia,
ia indicator struct,
rid,
rid indicator,
newval,
newval indicator,
env,
env indicator struct,
return OCINumber
);

16.4.4.2 Registering the Implementation of ODCIIndexDelete()

Register the implementation of the ODCIIndexDelete() routine.
STATIC FUNCTION ODCIIndexDelete(
ia SYS.ODCIIndexInfo,
rid VARCHAR2,
oldval VARCHAR2,
env SYS.ODCIEnv)
RETURN NUMBER AS EXTERNAL
name "qxiqtbspd"
library extdemo6l
with context
parameters (
context,
ia,
ia indicator struct,
rid,
rid indicator,
oldval,
oldval indicator,
env,
env indicator struct,
return OCINumber
);

16.4.4.3 Registering the Implementation of ODCIIndexUpdate()

Register the implementation of the ODCIIndexUpdate() routine.
STATIC FUNCTION ODCIIndexUpdate(
ia SYS.ODCIIndexInfo,
rid VARCHAR2,

16-9
 Chapter 16
Implementing the ODCIIndex Interfaces

oldval VARCHAR2,
newval VARCHAR2,
env SYS.ODCIEnv)
RETURN NUMBER AS EXTERNAL
name "qxiqtbspu"
library extdemo6l
with context
parameters (
context,
ia,
ia indicator struct,
rid,
rid indicator,
oldval,
oldval indicator,
newval,
newval indicator,
env,
env indicator struct,
return OCINumber
);

16.4.4.4 Registering the Implementation of ODCIIndexStart()

Register the implementation of the ODCIIndexStart() routine.
STATIC FUNCTION ODCIIndexStart(
sctx IN OUT psbtree_im,
ia SYS.ODCIIndexInfo,
op SYS.ODCIPredInfo,
qi SYS.ODCIQueryInfo,
strt NUMBER,
stop NUMBER,
cmpval VARCHAR2,
env SYS.ODCIEnv)
RETURN NUMBER AS EXTERNAL
name "qxiqtbsps"
library extdemo6l
with context
parameters (
context,
sctx,
sctx indicator struct,
ia,
ia indicator struct,
op,
op indicator struct,
qi,
qi indicator struct,
strt,
strt indicator,
stop,
stop indicator,
cmpval,
cmpval indicator,
env,
env indicator struct,
return OCINumber
);

16-10
 Chapter 16
Implementing the ODCIIndex Interfaces

16.4.4.5 Registering the Implementation of ODCIIndexFetch()

Register the implementation of the ODCIIndexFetch() routine.
MEMBER FUNCTION ODCIIndexFetch(
nrows NUMBER,
rids OUT SYS.ODCIRidList,
env SYS.ODCIEnv)
RETURN NUMBER AS EXTERNAL
name "qxiqtbspf"
library extdemo6l
with context
parameters (
context,
self,
self indicator struct,
nrows,
nrows indicator,
rids,
rids indicator,
env,
env indicator struct,
return OCINumber
);

16.4.4.6 Registering the Implementation of ODCIIndexClose()

Register the implementation of the ODCIIndexClose() routine.
MEMBER FUNCTION ODCIIndexClose (
env SYS.ODCIEnv)
RETURN NUMBER AS EXTERNAL
name "qxiqtbspc"
library extdemo6l
with context
parameters (
context,
self,
self indicator struct,
env,
env indicator struct,
return OCINumber
);

16.4.5 Defining Additional Structures in C Implementation

The stuct qxiqtim, struct qciqtin, and struct qxiqtcx are used for mapping the
object type and its null value (demonstrated in Example 16-6), and for keeping state
during fetching calls (demonstrated in Example 16-7). These structures are used by
the methods described in Defining C Methods in the Implementation Body.
The C structs for mapping the ODCI types are defined in the file odci.h. For
example, the C struct ODCIIndexInfo is the mapping for the corresponding ODCI
object type. The C struct ODCIIndexInfo_ind is the mapping for the null object.

16-11
 Chapter 16
Implementing the ODCIIndex Interfaces

Example 16-6 Defining Mappings for the Object Type and Its Null Value
We have defined a C struct, qxiqtim, as a mapping for the object type. There is an
additional C struct, qxiqtin, for the corresponding null object. The C structs for the
object type and its null object can be generated from the Object Type Translator
(OTT).
/* The index implementation type is an object type with a single RAW attribute
* used to store the context key value.
* C mapping of the implementation type : */

struct qxiqtim{
OCIRaw *sctx_qxiqtim;
};
typedef struct qxiqtim qxiqtim;

struct qxiqtin{
short atomic_qxiqtin;
short scind_qxiqtin;
};
typedef struct qxiqtin qxiqtin;

Example 16-7 Keeping the Scan State During Fetching Calls

There are a set of OCI handles that must be cached away and retrieved during fetch
calls. A C struct, qxiqtcx, is defined to hold all the necessary scan state. This
structure is allocated out of OCI_DURATION_STATEMENT memory to ensure that it persists
till the end of fetch. After populating the structure with the required info, a pointer to
the structure is saved in OCI context. The context is identified by a 4-byte key that is
generated by calling an OCI routine. The 4-byte key is stashed away in the scan
context - exiting. This object is returned back to the Oracle server and is passed in
as a parameter to the next fetch call.
/* The index scan context - should be stored in "statement" duration memory
* and used by start, fetch and close routines.
*/
struct qxiqtcx
{
OCIStmt *stmthp;
OCIDefine *defnp;
OCIBind *bndp;
char ridp[19];
};
typedef struct qxiqtcx qxiqtcx;

16.4.6 Defining C Methods in the Implementation Body

Consider how to implemented PSBEETree methods in the C language.

16.4.6.1 Implementing a Common Error Processing Routine in C

This function is used to check and process the return code from all OCI routines. It
checks the status code and raises an exception in case of errors.
static int qxiqtce(
OCIExtProcContext *ctx,
OCIError *errhp,
sword status)
{

16-12
 Chapter 16
Implementing the ODCIIndex Interfaces

text errbuf[512];
sb4 errcode = 0;
int errnum = 29400; /* choose some oracle error number */
int rc = 0;

switch (status)
{
case OCI_SUCCESS:
rc = 0;
break;
case OCI_ERROR:
(void) OCIErrorGet((dvoid *)errhp, (ub4)1, (text *)NULL, &errcode,
errbuf, (ub4)sizeof(errbuf), OCI_HTYPE_ERROR);
/* Raise exception */
OCIExtProcRaiseExcpWithMsg(ctx, errnum, errbuf, strlen((char *)errbuf));
rc = 1;
break;
default:
(void) sprintf((char *)errbuf, "Warning - some error\n");
/* Raise exception */
OCIExtProcRaiseExcpWithMsg(ctx, errnum, errbuf, strlen((char *)errbuf));
rc = 1;
break;
}
return (rc);
}

16.4.6.2 Implementing ODCIIndexInsert() for PSBTREE in C

The insert routine, ODCIIndexInsert(), parses and executes a statement that inserts a
new row into the index table. The new row consists of the new value of the indexed
column and the rowid that have been passed in as parameters.
OCINumber *qxiqtbspi(
OCIExtProcContext *ctx,
ODCIIndexInfo *ix,
ODCIIndexInfo_ind *ix_ind,
char *rid,
short rid_ind,
char *newval,
short newval_ind,
ODCIEnv *env,
ODCIEnv_ind *env_ind)
{
OCIEnv *envhp = (OCIEnv *) 0; /* env. handle */
OCISvcCtx *svchp = (OCISvcCtx *) 0; /* service handle */
OCIError *errhp = (OCIError *) 0; /* error handle */
OCIStmt *stmthp = (OCIStmt *) 0; /* statement handle */
OCIBind *bndp = (OCIBind *) 0; /* bind handle */

int retval = (int)ODCI_SUCCESS; /* return from this function */

OCINumber *rval = (OCINumber *)0;

char insstmt[2000]; /* sql insert statement */

ODCIColInfo *colinfo; /* column info */
ODCIColInfo_ind *colinfo_ind;
boolean exists = TRUE;
unsigned int partiden; /* table partition iden */
unsigned int idxflag; /* index info flag

16-13
 Chapter 16
Implementing the ODCIIndex Interfaces

/* allocate memory for OCINumber first */

rval = (OCINumber *)OCIExtProcAllocCallMemory(ctx, sizeof(OCINumber));

/* Get oci handles */

if (qxiqtce(ctx, errhp, OCIExtProcGetEnv(ctx, &envhp, &svchp, &errhp)))
return(rval);

/* set up return code */

if (qxiqtce(ctx, errhp, OCINumberFromInt(errhp, (dvoid *)&retval,
sizeof(retval), OCI_NUMBER_SIGNED, rval)))
return(rval);

/* Convert idxflag to integer from OCINumber */

if (qxiqtce(ctx, errhp, OCINumberToInt(errhp, &(ix->IndexInfoFlags),
sizeof(idxflag), OCI_NUMBER_UNSIGNED, ( void *)&idxflag)))
return(rval);

/*****************************
* Construct insert Statement *
******************************/
if ((idxflag & ODCI_INDEX_RANGE_PARTN) != ODCI_INDEX_RANGE_PARTN)
(void)sprintf(insstmt, "INSERT into %s.%s_sbtree values (:newval, :mrid)",
OCIStringPtr(envhp, ix->IndexSchema), OCIStringPtr(envhp, ix->IndexName));
else
{
if (qxiqtce(ctx, errhp, OCICollGetElem(envhp, errhp, (OCIColl *)ix->IndexCols,
(sb4)0, &exists, (void **) &colinfo, (void **) &colinfo_ind)))
return(rval);

(void)sprintf(insstmt,
"INSERT into %s.%s_sbtree partition (DATAOBJ_TO_PARTITION(%s, :partiden))
VALUES (:newval, :mrid)",
OCIStringPtr(envhp, ix->IndexSchema), OCIStringPtr(envhp, ix->IndexName),
OCIStringPtr(envhp, colinfo->TableName));
}

/***************************************
* Parse and Execute Create Statement *
****************************************/

/* allocate stmt handle */

if (qxiqtce(ctx, errhp, OCIHandleAlloc((dvoid *)envhp, (dvoid **)&stmthp,
(ub4)OCI_HTYPE_STMT, (size_t)0, (dvoid **)0)))
return(rval);

/* prepare the statement */

if (qxiqtce(ctx, errhp, OCIStmtPrepare(stmthp, errhp, (text *)insstmt,
(ub4)strlen(insstmt), OCI_NTV_SYNTAX, OCI_DEFAULT)))
return(rval);

if ((idxflag & ODCI_INDEX_RANGE_PARTN) == ODCI_INDEX_RANGE_PARTN)

{
/* Convert partiden to integer from OCINumber */
if (qxiqtce(ctx, errhp, OCINumberToInt(errhp,
&(colinfo->TablePartitionIden), sizeof(partiden), OCI_NUMBER_UNSIGNED,
( void *)&partiden)))
return(rval);

/* Set up bind for partiden */

if (qxiqtce(ctx, errhp, OCIBindByName(stmthp, &bndp, errhp,
text *)":partiden", sizeof(":partiden")-1, (dvoid *)&partiden,

16-14
 Chapter 16
Implementing the ODCIIndex Interfaces

(sb4)(sizeof(partiden)), (ub2)SQLT_INT, (dvoid *)0, (ub2 *)0,

(ub2 *)0, (ub4)0, (ub4 *)0, (ub4)OCI_DEFAULT)))
return(rval);
}

/* Set up bind for newval */

if (qxiqtce(ctx, errhp, OCIBindByName(stmthp, &bndp, errhp, (text *)":newval",
sizeof(":newval")-1, (dvoid *)newval, (sb4)(strlen(newval)+1),
(ub2)SQLT_STR, (dvoid *)0, (ub2 *)0, (ub2 *)0, (ub4)0, (ub4 *)0,
(ub4)OCI_DEFAULT)))
return(rval);

/* Set up bind for rid */

if (qxiqtce(ctx, errhp, OCIBindByName(stmthp, &bndp, errhp, (text *)":mrid",
sizeof(":mrid")-1, (dvoid *)rid, (sb4)(strlen(rid)+1), (ub2)SQLT_STR,
(dvoid *)0, (ub2 *)0, (ub2 *)0, (ub4)0, (ub4 *)0, (ub4)OCI_DEFAULT)))
return(rval);

/* Execute statement */
if (qxiqtce(ctx, errhp, OCIStmtExecute(svchp, stmthp, errhp, (ub4)1,
(ub4)0, (OCISnapshot *)NULL, (OCISnapshot *)NULL, (ub4)OCI_DEFAULT)))
return(rval);

/* free stmt handle */

if (qxiqtce(ctx, errhp, OCIHandleFree((dvoid *)stmthp, (ub4)OCI_HTYPE_STMT)))
return(rval);

return(rval);
}

16.4.6.3 Implementing ODCIIndexDelete() for PSBTREE in C

The delete routine constructs a SQL statement to delete a row from the index table
corresponding to the row being deleted from the base table. The row in the index table
is identified by the value of rowid that is passed in as a parameter to this routine.
OCINumber *qxiqtbspd(
OCIExtProcContext *ctx,
ODCIIndexInfo *ix,
ODCIIndexInfo_ind *ix_ind,
char *rid,
short rid_ind,
char *oldval,
short oldval_ind,
ODCIEnv *env,
ODCIEnv_ind *env_ind)
{
OCIEnv *envhp = (OCIEnv *) 0; /* env. handle */
OCISvcCtx *svchp = (OCISvcCtx *) 0; /* service handle */
OCIError *errhp = (OCIError *) 0; /* error handle */
OCIStmt *stmthp = (OCIStmt *) 0; /* statement handle */
OCIBind *bndp = (OCIBind *) 0; /* bind handle */

int retval = (int)ODCI_SUCCESS; /* return from this function */

OCINumber *rval = (OCINumber *)0;

char delstmt[2000]; /* sql delete statement */

ODCIColInfo *colinfo; /* column info */
ODCIColInfo_ind *colinfo_ind;
boolean exists = TRUE;

16-15
 Chapter 16
Implementing the ODCIIndex Interfaces

unsigned int partiden; /* table partition iden */

unsigned int idxflag; /* index info flag

/* Get oci handles */

if (qxiqtce(ctx, errhp, OCIExtProcGetEnv(ctx, &envhp, &svchp, &errhp)))
return(rval);

/* set up return code */

rval = (OCINumber *)OCIExtProcAllocCallMemory(ctx, sizeof(OCINumber));
if (qxiqtce(ctx, errhp, OCINumberFromInt(errhp, (dvoid *)&retval,
sizeof(retval), OCI_NUMBER_SIGNED, rval)))
return(rval);

/* Convert idxflag to integer from OCINumber */

if (qxiqtce(ctx, errhp, OCINumberToInt(errhp, &(ix->IndexInfoFlags),
sizeof(idxflag), OCI_NUMBER_UNSIGNED, ( void *)&idxflag)))
return(rval);

/*****************************
* Construct delete Statement *
******************************/
if ((idxflag & ODCI_INDEX_RANGE_PARTN) != ODCI_INDEX_RANGE_PARTN)
(void)sprintf(delstmt, "DELETE FROM %s.%s_sbtree WHERE f2 = :rr",
OCIStringPtr(envhp, ix->IndexSchema), OCIStringPtr(envhp, ix->IndexName));
else
{
if (qxiqtce(ctx, errhp, OCICollGetElem(envhp, errhp, (OCIColl *)ix->IndexCols,
(sb4)0, &exists, (void **) &colinfo, (void **) &colinfo_ind)))
return(rval);

(void)sprintf(delstmt,
"DELETE FROM %s.%s_sbtree partition (DATAOBJ_TO_PARTITION(%s, :partiden))
WHERE f2 = :rr",
OCIStringPtr(envhp, ix->IndexSchema), OCIStringPtr(envhp, ix->IndexName),
OCIStringPtr(envhp, colinfo->TableName));
}

/***************************************
* Parse and Execute delete Statement *
****************************************/

/* allocate stmt handle */

if (qxiqtce(ctx, errhp, OCIHandleAlloc((dvoid *)envhp, (dvoid **)&stmthp,
(ub4)OCI_HTYPE_STMT, (size_t)0, (dvoid **)0)))
return(rval);

/* prepare the statement */

if (qxiqtce(ctx, errhp, OCIStmtPrepare(stmthp, errhp, (text *)delstmt,
(ub4)strlen(delstmt), OCI_NTV_SYNTAX, OCI_DEFAULT)))
return(rval);

if ( (idxflag & ODCI_INDEX_RANGE_PARTN) == ODCI_INDEX_RANGE_PARTN)

{
/* Convert partiden to integer from OCINumber */
if (qxiqtce(ctx, errhp, OCINumberToInt(errhp, &(colinfo->TablePartitionIden),
sizeof(partiden), OCI_NUMBER_UNSIGNED, ( void *)&partiden)))
return(rval);

/* Set up bind for partiden */

if (qxiqtce(ctx, errhp, OCIBindByName(stmthp, &bndp, errhp,
(text *)":partiden", sizeof(":partiden")-1, (dvoid *)&partiden,

16-16
 Chapter 16
Implementing the ODCIIndex Interfaces

sb4)(sizeof(partiden)), (ub2)SQLT_INT, (dvoid *)0, (ub2 *)0,

(ub2 *)0, (ub4)0, (ub4 *)0, (ub4)OCI_DEFAULT)))
return(rval);
}

/* Set up bind for rid */

if (qxiqtce(ctx, errhp, OCIBindByName(stmthp, &bndp, errhp, (text *)":rr",
sizeof(":rr")-1, (dvoid *)rid, (sb4)(strlen(rid)+1), (ub2)SQLT_STR,
(dvoid *)0, (ub2 *)0, (ub2 *)0, (ub4)0, (ub4 *)0, (ub4)OCI_DEFAULT)))
return(rval);

/* Execute statement */
if (qxiqtce(ctx, errhp, OCIStmtExecute(svchp, stmthp, errhp, (ub4)1, (ub4)0,
(OCISnapshot *)NULL, (OCISnapshot *)NULL, (ub4)OCI_DEFAULT)))
return(rval);

/* free stmt handle */

if (qxiqtce(ctx, errhp, OCIHandleFree((dvoid *)stmthp, (ub4)OCI_HTYPE_STMT)))
return(rval);

return(rval);
}

16.4.6.4 Implementing ODCIIndexUpdate() for PSBTree in C

The update routine constructs a SQL statement to update a row in the index table
corresponding to the row being updated in the base table. The row in the index table is
identified by the value of rowid that is passed in as a parameter to this routine. The old
column value (oldval) is replaced by the new value (newval).
OCINumber *qxiqtbspu(
OCIExtProcContext *ctx,
ODCIIndexInfo *ix,
ODCIIndexInfo_ind *ix_ind,
char *rid,
short rid_ind,
char *oldval,
short oldval_ind,
char *newval,
short newval_ind,
ODCIEnv *env,
ODCIEnv_ind *env_ind)
{
OCIEnv *envhp = (OCIEnv *) 0; /* env. handle */
OCISvcCtx *svchp = (OCISvcCtx *) 0; /* service handle */
OCIError *errhp = (OCIError *) 0; /* error handle */
OCIStmt *stmthp = (OCIStmt *) 0; /* statement handle */
OCIBind *bndp = (OCIBind *) 0; /* bind handle */

int retval = (int)ODCI_SUCCESS; /* return from this function */

OCINumber *rval = (OCINumber *)0;

char updstmt[2000]; /* sql upate statement */

ODCIColInfo *colinfo; /* column info */
ODCIColInfo_ind *colinfo_ind;
boolean exists = TRUE;
unsigned int partiden; /* table partition iden */
unsigned int idxflag; /* index info flag

/* Get oci handles */

16-17
 Chapter 16
Implementing the ODCIIndex Interfaces

if (qxiqtce(ctx, errhp, OCIExtProcGetEnv(ctx, &envhp, &svchp, &errhp)))

return(rval);

/* set up return code */

rval = (OCINumber *)OCIExtProcAllocCallMemory(ctx, sizeof(OCINumber));
if (qxiqtce(ctx, errhp, OCINumberFromInt(errhp, (dvoid *)&retval,
sizeof(retval), OCI_NUMBER_SIGNED, rval)))
return(rval);

/* Convert idxflag to integer from OCINumber */

if (qxiqtce(ctx, errhp, OCINumberToInt(errhp, &(ix->IndexInfoFlags),
sizeof(idxflag), OCI_NUMBER_UNSIGNED, ( void *)&idxflag)))
return(rval);

/*****************************
* Construct update Statement *
******************************/
if ( (idxflag & ODCI_INDEX_RANGE_PARTN) != ODCI_INDEX_RANGE_PARTN)
(void)sprintf(updstmt, "UPDATE %s.%s_sbtree SET f1 = :newval WHERE f2 = :rr",
OCIStringPtr(envhp, ix->IndexSchema), OCIStringPtr(envhp, ix->IndexName));
else
{
if (qxiqtce(ctx, errhp, OCICollGetElem(envhp, errhp, OCIColl *)ix->IndexCols,
(sb4)0, &exists, (void **) &colinfo, (void **) &colinfo_ind)))
return(rval);

(void)sprintf(updstmt, "UPDATE %s.%s_sbtree partition

(DATAOBJ_TO_PARTITION(%s, :partiden)) SET f1 = :newval WHERE f2 = :rr",
OCIStringPtr(envhp, ix->IndexSchema), OCIStringPtr(envhp, ix->IndexName),
OCIStringPtr(envhp, colinfo->TableName));
}

/****************************************
* Parse and Execute Create Statement *
****************************************/

/* allocate stmt handle */

if (qxiqtce(ctx, errhp, OCIHandleAlloc((dvoid *)envhp, (dvoid **)&stmthp,
(ub4)OCI_HTYPE_STMT, (size_t)0, (dvoid **)0)))
return(rval);

/* prepare the statement */

if (qxiqtce(ctx, errhp, OCIStmtPrepare(stmthp, errhp, (text *)updstmt,
(ub4)strlen(updstmt), OCI_NTV_SYNTAX, OCI_DEFAULT)))
return(rval);

if ( (idxflag & ODCI_INDEX_RANGE_PARTN) == ODCI_INDEX_RANGE_PARTN)

{
/* Convert partiden to integer from OCINumber */
if (qxiqtce(ctx, errhp, OCINumberToInt(errhp,
&(colinfo->TablePartitionIden), sizeof(partiden), OCI_NUMBER_UNSIGNED,
( void *)&partiden)))
return(rval);

/* Set up bind for partiden */

if (qxiqtce(ctx, errhp, OCIBindByName(stmthp, &bndp, errhp,
(text *)":partiden", sizeof(":partiden")-1, (dvoid *)&partiden,
(sb4)(sizeof(partiden)), (ub2)SQLT_INT, (dvoid *)0, (ub2 *)0,
(ub2 *)0, (ub4)0, (ub4 *)0, (ub4)OCI_DEFAULT)))
return(rval);
}

16-18
 Chapter 16
Implementing the ODCIIndex Interfaces

/* Set up bind for newval */

if (qxiqtce(ctx, errhp, OCIBindByName(stmthp, &bndp, errhp, (text *)":newval",
sizeof(":newval")-1, (dvoid *)newval, (sb4)(strlen(newval)+1),
(ub2)SQLT_STR, (dvoid *)0, (ub2 *)0, (ub2 *)0, (ub4)0, (ub4 *)0, (
ub4)OCI_DEFAULT)))
return(rval);

/* Set up bind for rid */

if (qxiqtce(ctx, errhp, OCIBindByName(stmthp, &bndp, errhp, (text *)":rr",
sizeof(":rr")-1, (dvoid *)rid, (sb4)(strlen(rid)+1), (ub2)SQLT_STR,
(dvoid *)0, (ub2 *)0, (ub2 *)0, (ub4)0, (ub4 *)0, (ub4)OCI_DEFAULT)))
return(rval);

/* Execute statement */
if (qxiqtce(ctx, errhp, OCIStmtExecute(svchp, stmthp, errhp, (ub4)1,
ub4)0, (OCISnapshot *)NULL, (OCISnapshot *)NULL, (ub4)OCI_DEFAULT)))
return(rval);

/* free stmt handle */

if (qxiqtce(ctx, errhp, OCIHandleFree((dvoid *)stmthp, (ub4)OCI_HTYPE_STMT)))
return(rval);

return(rval);
}

16.4.6.5 Implementing ODCIIndexStart() for PSBTREE in C

The start routine performs the setup for an psbtree index scan. The query information
in terms of the operator predicate, its arguments, and the bounds on return values are
passed in as parameters to this function. The scan context that is shared among the
index scan routines is an instance of the type psbtree_im.

This function sets up a cursor that scans the index table. The scan retrieves the stored
rowids for the rows in the index table that satisfy the specified predicate. The predicate
for the index table is generated based on the operator predicate information that is
passed in as parameters. For example, if the operator predicate is of the form eq(col,
'joe') = 1, then the predicate on the index table is set up to be f1 = 'joe'.

This function uses the structs qxiqtim, qxiqtin, and qxiqtcx, which were
demonstrated in Example 16-6 and Example 16-7.
OCINumber *qxiqtbsps(
OCIExtProcContext *ctx,
qxiqtim *sctx,
qxiqtin *sctx_ind,
ODCIIndexInfo *ix,
ODCIIndexInfo_ind *ix_ind,
ODCIPredInfo *pr,
ODCIPredInfo_ind *pr_ind,
ODCIQueryInfo *qy,
ODCIQueryInfo_ind *qy_ind,
OCINumber *strt,
short strt_ind,
OCINumber *stop,
short stop_ind,
char *cmpval,
short cmpval_ind,
ODCIEnv *env,
ODCIEnv_ind *env_ind)

16-19
 Chapter 16
Implementing the ODCIIndex Interfaces

{
sword status;
OCIEnv *envhp = (OCIEnv *) 0; /* env. handle */
OCISvcCtx *svchp = (OCISvcCtx *) 0; /* service handle */
OCIError *errhp = (OCIError *) 0; /* error handle */
OCISession *usrhp = (OCISession *) 0; /* user handle */
qxiqtcx *icx = (qxiqtcx *) 0; /* state to be saved for later calls */

int strtval; /* start bound */

int stopval; /* stop bound */

int errnum = 29400; /* choose some oracle error number */

char errmsg[512]; /* error message buffer */
size_t errmsglen; /* Length of error message */

char relop[3]; /* relational operator used in sql stmt */

char selstmt[2000]; /* sql select statement */

int retval = (int)ODCI_SUCCESS; /* return from this function */

OCINumber *rval = (OCINumber *)0;
ub4 key; /* key value set in "sctx" */

ub1 *rkey; /* key to retrieve context */

ub4 rkeylen; /* length of key */
ODCIColInfo *colinfo; /* column info */
ODCIColInfo_ind *colinfo_ind;
boolean exists = TRUE;
unsigned int partiden; /* table partition iden */
unsigned int idxflag; /* index info flag

/* Get oci handles */

if (qxiqtce(ctx, errhp, OCIExtProcGetEnv(ctx, &envhp, &svchp, &errhp)))
return(rval);

/* set up return code */

rval = (OCINumber *)OCIExtProcAllocCallMemory(ctx, sizeof(OCINumber));
if (qxiqtce(ctx, errhp, OCINumberFromInt(errhp, (dvoid *)&retval,
sizeof(retval), OCI_NUMBER_SIGNED, rval)))
return(rval);

/* get the user handle */

if (qxiqtce(ctx, errhp, OCIAttrGet((dvoid *)svchp, (ub4)OCI_HTYPE_SVCCTX,
(dvoid *)&usrhp, (ub4 *)0, (ub4)OCI_ATTR_SESSION, errhp)))
return(rval);

/**********************************************/
/* Allocate memory to hold index scan context */
/**********************************************/
if (sctx_ind ->atomic_qxiqtin == OCI_IND_NULL ||
sctx_ind ->scind_qxiqtin == OCI_IND_NULL)
{
if (qxiqtce(ctx, errhp, OCIMemoryAlloc((dvoid *)usrhp, errhp, (dvoid **)&icx,
OCI_DURATION_STATEMENT, (ub4)(sizeof(qxiqtcx)), OCI_MEMORY_CLEARED)))
return(rval);

icx->stmthp = (OCIStmt *)0;

icx->defnp = (OCIDefine *)0;
icx->bndp = (OCIBind *)0;
}

else

16-20
 Chapter 16
Implementing the ODCIIndex Interfaces

{
/*************************/
/* Retrieve scan context */
/*************************/
rkey = OCIRawPtr(envhp, sctx->sctx_qxiqtim);
rkeylen = OCIRawSize(envhp, sctx->sctx_qxiqtim);

if (qxiqtce(ctx, errhp, OCIContextGetValue((dvoid *)usrhp, errhp,

rkey, (ub1)rkeylen, (dvoid **)&(icx))))
return(rval);
}

/***********************************/
/* Check that the bounds are valid */
/***********************************/
/* convert from oci numbers to native numbers */
if (qxiqtce(ctx, errhp, OCINumberToInt(errhp, strt, sizeof(strtval),
OCI_NUMBER_SIGNED, (dvoid *)&strtval)))
return(rval);

if (qxiqtce(ctx, errhp, OCINumberToInt(errhp, stop, sizeof(stopval),

OCI_NUMBER_SIGNED, (dvoid *)&stopval)))
return(rval);

/* verify that strtval/stopval are both either 0 or 1 */

if (!(((strtval == 0) && (stopval == 0)) || ((strtval == 1) && (stopval == 1))))
{
strcpy(errmsg, (char *)"Incorrect predicate for sbtree operator");
errmsglen = (size_t)strlen(errmsg);
if (OCIExtProcRaiseExcpWithMsg(ctx, errnum, (text *)errmsg, errmsglen)
!= OCIEXTPROC_SUCCESS)
/* Use cartridge error services here */;
return(rval);
}

/*********************************************/
/* Generate the SQL statement to be executed */
/*********************************************/
if (memcmp((dvoid *)OCIStringPtr(envhp, pr->ObjectName), (dvoid *)"EQ", 2) == 0)
if (strtval == 1)
strcpy(relop, (char *)"=");
else
strcpy(relop, (char *)"!=");
else if
(memcmp((dvoid *)OCIStringPtr(envhp, pr->ObjectName), (dvoid *)"LT",2) == 0)
if (strtval == 1)
strcpy(relop, (char *)"<");
else
strcpy(relop, (char *)">=");
else
if (strtval == 1)
strcpy(relop, (char *)">");
else
strcpy(relop, (char *)"<=");

/* Convert idxflag to integer from OCINumber */

if (qxiqtce(ctx, errhp, OCINumberToInt(errhp, &(ix->IndexInfoFlags),
sizeof(idxflag), OCI_NUMBER_UNSIGNED, ( void *)&idxflag)))
return(rval);

if ( (idxflag & ODCI_INDEX_RANGE_PARTN) != ODCI_INDEX_RANGE_PARTN)

16-21
 Chapter 16
Implementing the ODCIIndex Interfaces

(void)sprintf(selstmt, "select f2 from %s.%s_sbtree where f1 %s :val",

OCIStringPtr(envhp, ix->IndexSchema), OCIStringPtr(envhp, ix->IndexName),
relop);
else
{
if (qxiqtce(ctx, errhp, OCICollGetElem(envhp, errhp, OCIColl *)ix->IndexCols,
(sb4)0, &exists, (void **) &colinfo, (void **) &colinfo_ind)))
return(rval);

/* Convert partiden to integer from OCINumber */

if (qxiqtce(ctx, errhp, OCINumberToInt(errhp, &(colinfo->TablePartitionIden),
sizeof(partiden), OCI_NUMBER_UNSIGNED, ( void *)&partiden)))
return(rval);

(void)sprintf(selstmt, "select f2 from %s.%s_sbtree partition

(DATAOBJ_TO_PARTITION(%s, %d)) where f1 %s :val",
OCIStringPtr(envhp, ix->IndexSchema), OCIStringPtr(envhp, ix->IndexName),
OCIStringPtr(envhp, colinfo->TableName), partiden, relop);
}

/***********************************/
/* Parse, bind, define and execute */
/***********************************/
if (sctx_ind ->atomic_qxiqtin == OCI_IND_NULL ||
sctx_ind ->scind_qxiqtin == OCI_IND_NULL)
{
/* allocate stmt handle */
if (qxiqtce(ctx, errhp, OCIHandleAlloc((dvoid *)envhp,
(dvoid **)&(icx->stmthp), (ub4)OCI_HTYPE_STMT, (size_t)0, (dvoid **)0)))
return(rval);
}

/* prepare the statement */

if (qxiqtce(ctx, errhp, OCIStmtPrepare(icx->stmthp, errhp, (text *)selstmt,
(ub4)strlen(selstmt), OCI_NTV_SYNTAX, OCI_DEFAULT)))
return(rval);

/* Set up bind for compare value */

if (qxiqtce(ctx, errhp, OCIBindByName(icx->stmthp, &(icx->bndp), errhp,
(text *)":val", sizeof(":val")-1, (dvoid *)cmpval, (sb4)(strlen(cmpval)+1),
(ub2)SQLT_STR, (dvoid *)0, (ub2 *)0, (ub2 *)0, (ub4)0, (ub4 *)0,
(ub4)OCI_DEFAULT)))
return(rval);

/* Set up define */
if (qxiqtce(ctx, errhp, OCIDefineByPos(icx->stmthp, &(icx->defnp), errhp,
(ub4)1, (dvoid *)(icx->ridp), (sb4) sizeof(icx->ridp), (ub2)SQLT_STR,
(dvoid *)0, (ub2 *)0, (ub2 *)0, (ub4)OCI_DEFAULT)))
return(rval);

/* execute */
if (qxiqtce(ctx, errhp, OCIStmtExecute(svchp, icx->stmthp, errhp, (ub4)0,
(ub4)0, (OCISnapshot *)NULL, (OCISnapshot *)NULL, (ub4)OCI_DEFAULT)))
return(rval);

/************************************/
/* Set index context to be returned */
/************************************/
if (sctx_ind ->atomic_qxiqtin == OCI_IND_NULL ||
sctx_ind ->scind_qxiqtin == OCI_IND_NULL)
{

16-22
 Chapter 16
Implementing the ODCIIndex Interfaces

/* generate a key */
if (qxiqtce(ctx, errhp, OCIContextGenerateKey((dvoid *)usrhp, errhp, &key)))
return(rval);

/* set the memory address of the struct to be saved in the context */

if (qxiqtce(ctx, errhp, OCIContextSetValue((dvoid *)usrhp, errhp,
OCI_DURATION_STATEMENT, (ub1 *)&key, (ub1)sizeof(key), (dvoid *)icx)))
return(rval);

/* statement duration memory alloc for key */

if (qxiqtce(ctx, errhp, OCIMemoryAlloc(( void *)usrhp, errhp,
( void **)&(sctx->sctx_qxiqtim), OCI_DURATION_STATEMENT,
(sb4)(sizeof(key)+sizeof(ub4)), OCI_MEMORY_CLEARED)))
return(rval);

/* set the key as the member of "sctx" */

if (qxiqtce(ctx, errhp, OCIRawAssignBytes(envhp, errhp, (ub1 *)&key,
ub4)sizeof(key), &(sctx->sctx_qxiqtim))))
return(rval);

sctx_ind->atomic_qxiqtin = OCI_IND_NOTNULL;
sctx_ind->scind_qxiqtin = OCI_IND_NOTNULL;

return(rval);
}

return(rval);
}

16.4.6.6 Implementing ODCIIndexFetch() for PSBTREE in C

The scan context set up by the start routine is passed in as a parameter to the fetch
routine. This function first retrieves the 4-byte key from the scan context. The C
mapping for the scan context is qxiqtim (see Example 16-6). Next, key is used to look
up the OCI context. This gives the memory address of the qxiqtcx structure (see
Example 16-7) that holds the OCI handles.
This function returns the next batch of rowids that satisfy the operator predicate. It
uses the value of the nrows parameter as the size of the batch. It repeatedly fetches
rowids from the open cursor and populates the rowid list. When the batch is full or
when there are no more rowids left, the function returns them back to the Oracle
server.
OCINumber *qxiqtbspf(
OCIExtProcContext *ctx,
qxiqtim *self,
qxiqtin *self_ind,
OCINumber *nrows,
short nrows_ind,
OCIArray **rids,
short *rids_ind,
ODCIEnv *env,
ODCIEnv_ind *env_ind)
{
sword status;
OCIEnv *envhp = (OCIEnv *) 0; /* env. handle */
OCISvcCtx *svchp = (OCISvcCtx *) 0; /* service handle */
OCIError *errhp = (OCIError *) 0; /* error handle */
OCISession *usrhp = (OCISession *) 0; /* user handle */
qxiqtcx *icx = (qxiqtcx *) 0; /* state to be saved for later calls */

16-23
 Chapter 16
Implementing the ODCIIndex Interfaces

int idx = 1;
int nrowsval;

OCIArray *ridarrp = *rids; /* rowid collection */

OCIString *ridstr = (OCIString *)0;

int done = 0;
int retval = (int)ODCI_SUCCESS;
OCINumber *rval = (OCINumber *)0;

ub1 *key; /* key to retrieve context */

ub4 keylen; /* length of key */

/*******************/
/* Get OCI handles */
/*******************/
if (qxiqtce(ctx, errhp, OCIExtProcGetEnv(ctx, &envhp, &svchp, &errhp)))
return(rval);

/* set up return code */

rval = (OCINumber *)OCIExtProcAllocCallMemory(ctx, sizeof(OCINumber));
if (qxiqtce(ctx, errhp, OCINumberFromInt(errhp, (dvoid *)&retval,
sizeof(retval), OCI_NUMBER_SIGNED, rval)))
return(rval);

/* get the user handle */

if (qxiqtce(ctx, errhp, OCIAttrGet((dvoid *)svchp, (ub4)OCI_HTYPE_SVCCTX,
(dvoid *)&usrhp, (ub4 *)0, (ub4)OCI_ATTR_SESSION, errhp)))
return(rval);

/********************************/
/* Retrieve context from key */
/********************************/
key = OCIRawPtr(envhp, self->sctx_qxiqtim);
keylen = OCIRawSize(envhp, self->sctx_qxiqtim);

if (qxiqtce(ctx, errhp, OCIContextGetValue((dvoid *)usrhp, errhp, key,

(ub1)keylen, (dvoid **)&(icx))))
return(rval);

/* get value of nrows */

if (qxiqtce(ctx, errhp, OCINumberToInt(errhp, nrows, sizeof(nrowsval),
OCI_NUMBER_SIGNED, (dvoid *)&nrowsval)))
return(rval);

/****************/
/* Fetch rowids */
/****************/
while (!done)
{
if (idx > nrowsval)
done = 1;
else
{
status =OCIStmtFetch(icx->stmthp, errhp, (ub4)1, (ub2) 0, (ub4)OCI_DEFAULT);
if (status == OCI_NO_DATA)
{
short col_ind = OCI_IND_NULL;
/* have to create dummy oci string */
OCIStringAssignText(envhp, errhp, (text *)"dummy", (ub2)5, &ridstr);

16-24
 Chapter 16
Implementing the ODCIIndex Interfaces

/* append null element to collection */

if (qxiqtce(ctx, errhp, OCICollAppend(envhp, errhp, (dvoid *)ridstr,
(dvoid *)&col_ind, (OCIColl *)ridarrp)))
return(rval);
done = 1;
}
else if (status == OCI_SUCCESS)
{
OCIStringAssignText(envhp, errhp, (text *)icx->ridp, (ub2)18,
OCIString **)&ridstr);
/* append rowid to collection */
if (qxiqtce(ctx, errhp, OCICollAppend(envhp, errhp, (dvoid *)ridstr,
(dvoid *)0, (OCIColl *)ridarrp)))
return(rval);
idx++;
}
else if (qxiqtce(ctx, errhp, status))
return(rval);
}
}

/* free ridstr finally */

if (ridstr &&
(qxiqtce(ctx, errhp, OCIStringResize(envhp, errhp, (ub4)0, &ridstr))))
return(rval);

*rids_ind = OCI_IND_NOTNULL;

return(rval);
}

16.4.6.7 Implementing ODCIIndexClose() for PSBTREE in C

The scan context set up by the start routine is passed in as a parameter to the close
routine. This function first retrieves the 4-byte key from the scan context. The C
mapping for the scan context is qxiqtim (see Example 16-6). Next, the OCI context is
looked up based on the key. This gives the memory address of the structure that holds
the OCI handles, the qxiqtcx structure (see Example 16-7).

This function closes and frees all the OCI handles. It also frees the memory that was
allocated in the start routine.
OCINumber *qxiqtbspc(
OCIExtProcContext *ctx,
qxiqtim *self,
qxiqtin *self_ind,
ODCIEnv *env,
ODCIEnv_ind *env_ind)
{
sword status;
OCIEnv *envhp = (OCIEnv *) 0; /* env. handle */
OCISvcCtx *svchp = (OCISvcCtx *) 0; /* service handle */
OCIError *errhp = (OCIError *) 0; /* error handle */
OCISession *usrhp = (OCISession *) 0; /* user handle */
qxiqtcx *icx = (qxiqtcx *) 0; /* state to be saved for later calls */

int retval = (int) ODCI_SUCCESS;

OCINumber *rval = (OCINumber *)0;

ub1 *key; /* key to retrieve context */

16-25
 Chapter 16
Implementing the ODCIIndex Interfaces

ub4 keylen; /* length of key */

if (qxiqtce(ctx, errhp, OCIExtProcGetEnv(ctx, &envhp, &svchp, &errhp)))

return(rval);

/* set up return code */

rval = (OCINumber *)OCIExtProcAllocCallMemory(ctx, sizeof(OCINumber));
if (qxiqtce(ctx, errhp, OCINumberFromInt(errhp, (dvoid *)&retval,
sizeof(retval), OCI_NUMBER_SIGNED, rval)))
return(rval);

/* get the user handle */

if (qxiqtce(ctx, errhp, OCIAttrGet((dvoid *)svchp, (ub4)OCI_HTYPE_SVCCTX,
(dvoid *)&usrhp, (ub4 *)0,
(ub4)OCI_ATTR_SESSION, errhp)))
return(rval);
/********************************/
/* Retrieve context using key */
/********************************/
key = OCIRawPtr(envhp, self->sctx_qxiqtim);
keylen = OCIRawSize(envhp, self->sctx_qxiqtim);

if (qxiqtce(ctx, errhp, OCIContextGetValue((dvoid *)usrhp, errhp, key,

(ub1)keylen, (dvoid **)&(icx))))
return(rval);

/* Free handles and memory */

if (qxiqtce(ctx, errhp, OCIHandleFree((dvoid *)icx->stmthp,
(ub4)OCI_HTYPE_STMT)))
return(rval);

if (qxiqtce(ctx, errhp, OCIMemoryFree((dvoid *)usrhp, errhp, (dvoid *)icx)))

return(rval);

/* free the memory allocated for the index context. */

if (qxiqtce(ctx, errhp, OCIContextClearValue((dvoid *)usrhp, errhp, key,
(ub1)keylen)))
return(rval);

return(rval);
}

16.4.7 Implementing the Indextype for PSBTREE

You should create the indextype object and specify the list of operators that it
supports. In addition, specify the name of the implementation type that implements the
ODCIIndexXXX() interface routines.
CREATE INDEXTYPE psbtree
FOR
eq(VARCHAR2, VARCHAR2),
lt(VARCHAR2, VARCHAR2),
gt(VARCHAR2, VARCHAR2)
USING psbtree_im
WITH LOCAL RANGE PARTITION
WITH SYSTEM MANAGED STORAGE TABLES

16-26
 Chapter 16
Using PSBTREE

16.5 Using PSBTREE

You should next create the indextype object and specify the list of operators that it
supports. In addition, specify the name of the implementation type that implements the
ODCIIndexXXX() interface routines. This step is demonstrated in the Implementing the
Indextype for PSBTREE section.
One typical usage scenario is to create a range partitioned table and populate it, as
demonstrated in the Creating and Populating a Partitioned Table for PSBTREE
section.
You can then create a psbtree index on column f2. The CREATE INDEX statement
specifies the indextype that should be used, as demonstrated in the Creating a
PSBTREE Index on a Column section.
To execute a query that uses one of the psbtree operators, use the code in the Using
PSBTREE Operators in a Query section.

16.5.1 Creating and Populating a Partitioned Table for PSBTREE

CREATE TABLE t1 (f1 NUMBER, f2 VARCHAR2(200))
PARTITION BY RANGE(f1)
(
PARTITION p1 VALUES LESS THAN (101),
PARTITION p2 VALUES LESS THAN (201),
PARTITION p3 VALUES LESS THAN (301),
PARTITION p4 VALUES LESS THAN (401)
);
INSERT INTO t1 VALUES (10, 'aaaa');
INSERT INTO t1 VALUES (200, 'bbbb');
INSERT INTO t1 VALUES (100, 'cccc');
INSERT INTO t1 VALUES (300, 'dddd');
INSERT INTO t1 VALUES (400, 'eeee');
COMMIT;

16.5.2 Creating a PSBTREE Index on a Column

CREATE INDEX it1 ON t1(f2) iINDEXTYPE IS psbtree LOCAL
(PARTITION pe1 PARAMETERS('test1'), PARTITION pe2,
PARTITION pe3, PARTITION pe4 PARAMETERS('test4'))
PARAMETERS('test');

16.5.3 Using PSBTREE Operators in a Query

SELECT * FROMM t1 WHERE eq(f2, 'dddd') = 1 AND f1>101 ;

The explain plan output for this query should look like this:
OPERATION OPTIONS PARTITION_START PARTITION_STOP
--------------------------------------------------------------------------------
SELECT STATEMENT
PARTITION RANGE ITERATOR 2 4
TABLE ACCESS BY LOCAL INDEX ROWID 2 4
DOMAIN INDEX

16-27
17
Pipelined Table Functions: Interface
Approach Example
Two complete implementations of the StockPivot pipelined table function using the
interface approach are described. One implementation is in C, and the other is in Java.
The function StockPivot converts a row of the type (Ticker, OpenPrice,
ClosePrice) into two rows of the form (Ticker, PriceType, Price). For example,
from an input row ("ORCL", 41, 42), the table function returns the two rows ("ORCL",
"O", 41) and ("ORCL", "C", 42).

Tip:
Consider the table functions described in Using Pipelined and Parallel Table
Functions .

17.1 Pipelined Table Functions Example: C Implementation

In this example, the three ODCITable interface methods of the implementation type are
implemented as external functions in C. These methods must first be declared in SQL.

17.1.1 Making SQL Declarations for C Implementation

Example 17-1 shows how to make SQL declarations for the methods implemented in
C language in Implementation ODCITable Methods in C.
Example 17-1 Making SQL Declarations for Implementing ODCITableXXX() in C
-- Create the input stock table
CREATE TABLE StockTable (
ticker VARCHAR(4),
openprice NUMBER,
closeprice NUMBER
);

-- Create the types for the table function's output collection

-- and collection elements

CREATE TYPE TickerType AS OBJECT

(
ticker VARCHAR2(4),
PriceType VARCHAR2(1),
price NUMBER
);
/

CREATE TYPE TickerTypeSet AS TABLE OF TickerType;

/

17-1
 Chapter 17
Pipelined Table Functions Example: C Implementation

-- Create the external library object

CREATE LIBRARY StockPivotLib IS '/home/bill/libstock.so';
/

-- Create the implementation type

CREATE TYPE StockPivotImpl AS OBJECT
(
key RAW(4),

STATIC FUNCTION ODCITableStart(

sctx OUT StockPivotImpl,
cur SYS_REFCURSOR)
RETURN PLS_INTEGER
AS LANGUAGE C
LIBRARY StockPivotLib
NAME "ODCITableStart"
WITH CONTEXT
PARAMETERS (context, sctx, sctx INDICATOR STRUCT, cur, RETURN INT),

MEMBER FUNCTION ODCITableFetch(

self IN OUT StockPivotImpl,
nrows IN NUMBER,
outSet OUT TickerTypeSet)
RETURN PLS_INTEGER
AS LANGUAGE C
LIBRARY StockPivotLib
NAME "ODCITableFetch"
WITH CONTEXT
PARAMETERS (context, self, self INDICATOR STRUCT, nrows, outSet,
outSet INDICATOR, RETURN INT),

MEMBER FUNCTION ODCITableClose(

self IN StockPivotImpl)
RETURN PLS_INTEGER
AS LANGUAGE C
LIBRARY StockPivotLib
NAME "ODCITableClose"
WITH CONTEXT
PARAMETERS (context, self, self INDICATOR STRUCT, RETURN INT)
);
/

-- Define the ref cursor type

CREATE PACKAGE refcur_pkg IS
TYPE refcur_t IS REF CURSOR RETURN StockTable%ROWTYPE;
END refcur_pkg;
/

-- Create table function

CREATE FUNCTION StockPivot(p refcur_pkg.refcur_t) RETURN TickerTypeSet
PIPELINED USING StockPivotImpl;
/

17.1.2 Implementation ODCITable Methods in C

Example 17-2 implements the three ODCITable methods as external functions in C.

17-2
 Chapter 17
Pipelined Table Functions Example: C Implementation

Example 17-2 Implementing ODCTableXXX() Methods in C

#ifndef OCI_ORACLE
# include <oci.h>
#endif
#ifndef ODCI_ORACLE
# include <odci.h>
#endif

/*---------------------------------------------------------------------------
PRIVATE TYPES AND CONSTANTS
---------------------------------------------------------------------------*/

/* The struct holding the user's stored context */

struct StoredCtx
{
OCIStmt* stmthp;
};
typedef struct StoredCtx StoredCtx;

/* OCI Handles */

struct Handles_t
{
OCIExtProcContext* extProcCtx;
OCIEnv* envhp;
OCISvcCtx* svchp;
OCIError* errhp;
OCISession* usrhp;
};
typedef struct Handles_t Handles_t;

/********************** SQL Types C representation **********************/

/* Table function's implementation type */

struct StockPivotImpl
{
OCIRaw* key;
};
typedef struct StockPivotImpl StockPivotImpl;

struct StockPivotImpl_ind
{
short _atomic;
short key;
};
typedef struct StockPivotImpl_ind StockPivotImpl_ind;

/* Table function's output collection element type */

struct TickerType
{
OCIString* ticker;
OCIString* PriceType;
OCINumber price;
};
typedef struct TickerType TickerType;

struct TickerType_ind

17-3
 Chapter 17
Pipelined Table Functions Example: C Implementation

{
short _atomic;
short ticker;
short PriceType;
short price;
};
typedef struct TickerType_ind TickerType_ind;

/* Table function's output collection type */

typedef OCITable TickerTypeSet;

/*--------------------------------------------------------------------------*/
/* Static Functions */
/*--------------------------------------------------------------------------*/

static int GetHandles(OCIExtProcContext* extProcCtx, Handles_t* handles);

static StoredCtx* GetStoredCtx(Handles_t* handles, StockPivotImpl* self,

StockPivotImpl_ind* self_ind);

static int checkerr(Handles_t* handles, sword status);

/*--------------------------------------------------------------------------*/
/* Functions definitions */
/*--------------------------------------------------------------------------*/

/* Callout for ODCITableStart */

int ODCITableStart(OCIExtProcContext* extProcCtx, StockPivotImpl* self,

StockPivotImpl_ind* self_ind, OCIStmt** cur)
{
Handles_t handles; /* OCI hanldes */
StoredCtx* storedCtx; /* Stored context pointer */

ub4 key; /* key to retrieve stored context */

/* Get OCI handles */

if (GetHandles(extProcCtx, &handles))
return ODCI_ERROR;

/* Allocate memory to hold the stored context */

if (checkerr(&handles, OCIMemoryAlloc((dvoid*) handles.usrhp, handles.errhp,
(dvoid**) &storedCtx,
OCI_DURATION_STATEMENT,
(ub4) sizeof(StoredCtx),
OCI_MEMORY_CLEARED)))
return ODCI_ERROR;

/* store the input ref cursor in the stored context */

storedCtx->stmthp=*cur;

/* generate a key */
if (checkerr(&handles, OCIContextGenerateKey((dvoid*) handles.usrhp,
handles.errhp, &key)))
return ODCI_ERROR;

/* associate the key value with the stored context address */

if (checkerr(&handles, OCIContextSetValue((dvoid*)handles.usrhp,
handles.errhp,
OCI_DURATION_STATEMENT,

17-4
 Chapter 17
Pipelined Table Functions Example: C Implementation

(ub1*) &key, (ub1) sizeof(key),

(dvoid*) storedCtx)))
return ODCI_ERROR;

/* stored the key in the scan context */

if (checkerr(&handles, OCIRawAssignBytes(handles.envhp, handles.errhp,
(ub1*) &key, (ub4) sizeof(key),
&(self->key))))
return ODCI_ERROR;

/* set indicators of the scan context */

self_ind->_atomic = OCI_IND_NOTNULL;
self_ind->key = OCI_IND_NOTNULL;

*cur=(OCIStmt *)0;

return ODCI_SUCCESS;
}

/***********************************************************************/

/* Callout for ODCITableFetch */

int ODCITableFetch(OCIExtProcContext* extProcCtx, StockPivotImpl* self,

StockPivotImpl_ind* self_ind, OCINumber* nrows,
TickerTypeSet** outSet, short* outSet_ind)
{
Handles_t handles; /* OCI hanldes */
StoredCtx* storedCtx; /* Stored context pointer */
int nrowsval; /* number of rows to return */

/* Get OCI handles */

if (GetHandles(extProcCtx, &handles))
return ODCI_ERROR;

/* Get the stored context */

storedCtx=GetStoredCtx(&handles,self,self_ind);
if (!storedCtx) return ODCI_ERROR;

/* get value of nrows */

if (checkerr(&handles, OCINumberToInt(handles.errhp, nrows, sizeof(nrowsval),
OCI_NUMBER_SIGNED, (dvoid *)&nrowsval)))
return ODCI_ERROR;

/* return up to 10 rows at a time */

if (nrowsval>10) nrowsval=10;

/* Initially set the output to null */

*outSet_ind=OCI_IND_NULL;

while (nrowsval>0)
{

TickerType elem; /* current collection element */

TickerType_ind elem_ind; /* current element indicator */

OCIDefine* defnp1=(OCIDefine*)0; /* define handle */

OCIDefine* defnp2=(OCIDefine*)0; /* define handle */
OCIDefine* defnp3=(OCIDefine*)0; /* define handle */

sword status;

17-5
 Chapter 17
Pipelined Table Functions Example: C Implementation

char ticker[5];
float openprice;
float closeprice;
char PriceType[2];

/* Define the fetch buffer for ticker symbol */

if (checkerr(&handles, OCIDefineByPos(storedCtx->stmthp, &defnp1,
handles.errhp, (ub4) 1,
(dvoid*) &ticker,
(sb4) sizeof(ticker),
SQLT_STR, (dvoid*) 0, (ub2*) 0,
(ub2*) 0, (ub4) OCI_DEFAULT)))
return ODCI_ERROR;

/* Define the fetch buffer for open price */

if (checkerr(&handles, OCIDefineByPos(storedCtx->stmthp, &defnp2,
handles.errhp, (ub4) 2,
(dvoid*) &openprice,
(sb4) sizeof(openprice),
SQLT_FLT, (dvoid*) 0, (ub2*) 0,
(ub2*) 0, (ub4) OCI_DEFAULT)))
return ODCI_ERROR;

/* Define the fetch buffer for closing price */

if (checkerr(&handles, OCIDefineByPos(storedCtx->stmthp, &defnp3,
handles.errhp, (ub4) 3,
(dvoid*) &closeprice,
(sb4) sizeof(closeprice),
SQLT_FLT, (dvoid*) 0, (ub2*) 0,
(ub2*) 0, (ub4) OCI_DEFAULT)))
return ODCI_ERROR;

/* fetch a row from the input ref cursor */

status = OCIStmtFetch(storedCtx->stmthp, handles.errhp, (ub4) 1,
(ub4) OCI_FETCH_NEXT, (ub4) OCI_DEFAULT);

/* finished if no more data */

if (status!=OCI_SUCCESS && status!=OCI_SUCCESS_WITH_INFO) break;

/* Initialize the element indicator struct */

elem_ind._atomic=OCI_IND_NOTNULL;
elem_ind.ticker=OCI_IND_NOTNULL;
elem_ind.PriceType=OCI_IND_NOTNULL;
elem_ind.price=OCI_IND_NOTNULL;

/* assign the ticker name */

elem.ticker=NULL;
if (checkerr(&handles, OCIStringAssignText(handles.envhp, handles.errhp,
(text*) ticker,
(ub2) strlen(ticker),
&elem.ticker)))
return ODCI_ERROR;

/* assign the price type */

elem.PriceType=NULL;
sprintf(PriceType,"O");
if (checkerr(&handles, OCIStringAssignText(handles.envhp, handles.errhp,
(text*) PriceType,
(ub2) strlen(PriceType),

17-6
 Chapter 17
Pipelined Table Functions Example: C Implementation

&elem.PriceType)))
return ODCI_ERROR;

/* assign the price */

if (checkerr(&handles, OCINumberFromReal(handles.errhp, &openprice,
sizeof(openprice), &elem.price)))
return ODCI_ERROR;

/* append element to output collection */

if (checkerr(&handles, OCICollAppend(handles.envhp, handles.errhp,
&elem, &elem_ind, *outSet)))
return ODCI_ERROR;

/* assign the price type */

elem.PriceType=NULL;
sprintf(PriceType,"C");
if (checkerr(&handles, OCIStringAssignText(handles.envhp, handles.errhp,
(text*) PriceType,
(ub2) strlen(PriceType),
&elem.PriceType)))
return ODCI_ERROR;

/* assign the price */

if (checkerr(&handles, OCINumberFromReal(handles.errhp, &closeprice,
sizeof(closeprice), &elem.price)))
return ODCI_ERROR;

/* append row to output collection */

if (checkerr(&handles, OCICollAppend(handles.envhp, handles.errhp,
&elem, &elem_ind, *outSet)))
return ODCI_ERROR;

/* set collection indicator to not null */

*outSet_ind=OCI_IND_NOTNULL;

nrowsval-=2;
}

return ODCI_SUCCESS;
}

/***********************************************************************/

/* Callout for ODCITableClose */

int ODCITableClose(OCIExtProcContext* extProcCtx, StockPivotImpl* self,

StockPivotImpl_ind* self_ind)
{
Handles_t handles; /* OCI hanldes */
StoredCtx* storedCtx; /* Stored context pointer */

/* Get OCI handles */

if (GetHandles(extProcCtx, &handles))
return ODCI_ERROR;

/* Get the stored context */

storedCtx=GetStoredCtx(&handles,self,self_ind);
if (!storedCtx) return ODCI_ERROR;

/* Free the memory for the stored context */

if (checkerr(&handles, OCIMemoryFree((dvoid*) handles.usrhp, handles.errhp,

17-7
 Chapter 17
Pipelined Table Functions Example: C Implementation

(dvoid*) storedCtx)))
return ODCI_ERROR;

return ODCI_SUCCESS;
}

/***********************************************************************/

/* Get the stored context using the key in the scan context */

static StoredCtx* GetStoredCtx(Handles_t* handles, StockPivotImpl* self,

StockPivotImpl_ind* self_ind)
{
StoredCtx *storedCtx; /* Stored context pointer */
ub1 *key; /* key to retrieve context */
ub4 keylen; /* length of key */

/* return NULL if the PL/SQL context is NULL */

if (self_ind->_atomic == OCI_IND_NULL) return NULL;

/* Get the key */

key = OCIRawPtr(handles->envhp, self->key);
keylen = OCIRawSize(handles->envhp, self->key);

/* Retrieve stored context using the key */

if (checkerr(handles, OCIContextGetValue((dvoid*) handles->usrhp,
handles->errhp,
key, (ub1) keylen,
(dvoid**) &storedCtx)))
return NULL;

return storedCtx;
}

/***********************************************************************/

/* Get OCI handles using the ext-proc context */

static int GetHandles(OCIExtProcContext* extProcCtx, Handles_t* handles)

{
/* store the ext-proc context in the handles struct */
handles->extProcCtx=extProcCtx;

/* Get OCI handles */

if (checkerr(handles, OCIExtProcGetEnv(extProcCtx, &handles->envhp,
&handles->svchp, &handles->errhp)))
return -1;

/* get the user handle */

if (checkerr(handles, OCIAttrGet((dvoid*)handles->svchp,
(ub4)OCI_HTYPE_SVCCTX,
(dvoid*)&handles->usrhp,
(ub4*) 0, (ub4)OCI_ATTR_SESSION,
handles->errhp)))
return -1;

return 0;
}

/***********************************************************************/

17-8
 Chapter 17
Pipelined Table Functions Example: Java Implementation

/* Check the error status and throw exception if necessary */

static int checkerr(Handles_t* handles, sword status)

{
text errbuf[512]; /* error message buffer */
sb4 errcode; /* OCI error code */

switch (status)
{
case OCI_SUCCESS:
case OCI_SUCCESS_WITH_INFO:
return 0;
case OCI_ERROR:
OCIErrorGet ((dvoid*) handles->errhp, (ub4) 1, (text *) NULL, &errcode,
errbuf, (ub4) sizeof(errbuf), (ub4) OCI_HTYPE_ERROR);
sprintf((char*)errbuf, "OCI ERROR code %d",errcode);
break;
default:
sprintf((char*)errbuf, "Warning - error status %d",status);
break;
}

OCIExtProcRaiseExcpWithMsg(handles->extProcCtx, 29400, errbuf,

strlen((char*)errbuf));

return -1;
}

17.2 Pipelined Table Functions Example: Java

Implementation
In this example, the declaration of the implementation type references Java methods
instead of C functions. This is the only change from the preceding, C example: all the
other objects (TickerType, TickerTypeSet, refcur_pkg, StockTable, and StockPivot)
are the same. These methods must first be declared in SQL.

17.2.1 Making SQL Declarations for Java Implementation

Example 17-3 shows how to make SQL declarations for the methods implemented in
C language in Implementing the ODCITable Methods in Java.
Example 17-3 Making SQL Declarations for Implementing OCITableXXX() in
Java
// create the directory object

CREATE OR REPLACE DIRECTORY JavaDir AS '/home/bill/Java';

// compile the java source

CREATE AND COMPILE JAVA SOURCE NAMED source01

USING BFILE (JavaDir,'StockPivotImpl.java');
/
show errors

-- Create the implementation type

17-9
 Chapter 17
Pipelined Table Functions Example: Java Implementation

CREATE TYPE StockPivotImpl AS OBJECT

(
key INTEGER,

STATIC FUNCTION ODCITableStart(sctx OUT StockPivotImpl, cur SYS_REFCURSOR)

RETURN NUMBER
AS LANGUAGE JAVA
NAME 'StockPivotImpl.ODCITableStart(oracle.sql.STRUCT[], java.sql.ResultSet)
return java.math.BigDecimal',

MEMBER FUNCTION ODCITableFetch(self IN OUT StockPivotImpl, nrows IN NUMBER,

outSet OUT TickerTypeSet) RETURN NUMBER
AS LANGUAGE JAVA
NAME 'StockPivotImpl.ODCITableFetch(java.math.BigDecimal, oracle.sql.ARRAY[])
return java.math.BigDecimal',

MEMBER FUNCTION ODCITableClose(self IN StockPivotImpl) RETURN NUMBER

AS LANGUAGE JAVA
NAME 'StockPivotImpl.ODCITableClose() return java.math.BigDecimal'

);
/
show errors

17.2.2 Implementing the ODCITable Methods in Java

Example 17-4 implements the three ODCITable methods as external functions in Java.

Example 17-4 Implementing ODCITableXXX() Methods in Java

import java.io.*;
import java.util.*;
import oracle.sql.*;
import java.sql.*;
import java.math.BigDecimal;
import oracle.CartridgeServices.*;

// stored context type

public class StoredCtx

{
ResultSet rset;
public StoredCtx(ResultSet rs) { rset=rs; }
}

// implementation type

public class StockPivotImpl implements SQLData

{
private BigDecimal key;

final static BigDecimal SUCCESS = new BigDecimal(0);

final static BigDecimal ERROR = new BigDecimal(1);

// Implement SQLData interface.

String sql_type;
public String getSQLTypeName() throws SQLException
{
return sql_type;
}

17-10
 Chapter 17
Pipelined Table Functions Example: Java Implementation

public void readSQL(SQLInput stream, String typeName) throws SQLException

{
sql_type = typeName;
key = stream.readBigDecimal();
}

public void writeSQL(SQLOutput stream) throws SQLException

{
stream.writeBigDecimal(key);
}

// type methods implementing ODCITable interface

static public BigDecimal ODCITableStart(STRUCT[] sctx,ResultSet rset)

throws SQLException
{
Connection conn = DriverManager.getConnection("jdbc:default:connection:");

// create a stored context and store the result set in it

StoredCtx ctx=new StoredCtx(rset);

// register stored context with cartridge services

int key;
try {
key = ContextManager.setContext(ctx);
} catch (CountException ce) {
return ERROR;
}

// create a StockPivotImpl instance and store the key in it

Object[] impAttr = new Object[1];
impAttr[0] = new BigDecimal(key);
StructDescriptor sd = new StructDescriptor("STOCKPIVOTIMPL",conn);
sctx[0] = new STRUCT(sd,conn,impAttr);

return SUCCESS;
}

public BigDecimal ODCITableFetch(BigDecimal nrows, ARRAY[] outSet)

throws SQLException
{
Connection conn = DriverManager.getConnection("jdbc:default:connection:");

// retrieve stored context using the key

StoredCtx ctx;
try {
ctx=(StoredCtx)ContextManager.getContext(key.intValue());
} catch (InvalidKeyException ik ) {
return ERROR;
}

// get the nrows parameter, but return up to 10 rows

int nrowsval = nrows.intValue();
if (nrowsval>10) nrowsval=10;

// create a vector for the fetched rows

Vector v = new Vector(nrowsval);
int i=0;

StructDescriptor outDesc =

17-11
 Chapter 17
Pipelined Table Functions Example: Java Implementation

StructDescriptor.createDescriptor("TICKERTYPE", conn);
Object[] out_attr = new Object[3];

while(nrowsval>0 && ctx.rset.next()){

out_attr[0] = (Object)ctx.rset.getString(1);
out_attr[1] = (Object)new String("O");
out_attr[2] = (Object)new BigDecimal(ctx.rset.getFloat(2));
v.add((Object)new STRUCT(outDesc, conn, out_attr));

out_attr[1] = (Object)new String("C");

out_attr[2] = (Object)new BigDecimal(ctx.rset.getFloat(3));
v.add((Object)new STRUCT(outDesc, conn, out_attr));

i+=2;
nrowsval-=2;
}

// return if no rows found

if(i==0) return SUCCESS;

// create the output ARRAY using the vector

Object out_arr[] = v.toArray();
ArrayDescriptor ad = new ArrayDescriptor("TICKERTYPESET",conn);
outSet[0] = new ARRAY(ad,conn,out_arr);

return SUCCESS;
}

public BigDecimal ODCITableClose() throws SQLException {

// retrieve stored context using the key, and remove from ContextManager
StoredCtx ctx;
try {
ctx=(StoredCtx)ContextManager.clearContext(key.intValue());
} catch (InvalidKeyException ik ) {
return ERROR;
}

// close the result set

Statement stmt = ctx.rset.getStatement();
ctx.rset.close();
if(stmt!=null) stmt.close();

return SUCCESS;
}

17-12
Part IV
Reference
This part contains reference information on cartridge-related APIs.
• Cartridge Services Using C_ C++ and Java
• Extensibility Constants_ Types_ and Mappings
• Extensible Indexing Interface
• Extensible Optimizer Interface
• User-Defined Aggregate Functions Interface
• Pipelined and Parallel Table Functions
18
Cartridge Services Using C, C++ and Java
Cartridge services are available to programmers using C/C++ and Java.

See Also:
Oracle Call Interface Programmer's Guide for more details on cartridge
services using C

18.1 OCI Access Functions for External Procedures

When called from an external procedure, a service routine can raise exceptions,
allocate memory, and get OCI handles for callbacks to the server. To use the
functions, you must specify the WITH CONTEXT clause, which lets you pass a context
structure to the external procedure. The context structure is declared in header file
ociextp.h as follows:
typedef struct OCIExtProcContext OCIExtProcContext;

Consider how service routines use context information. For more information and
examples of usage, see Oracle Database Advanced Application Developer's Guide.

18.1.1 OCIExtProcAllocCallMemory
This service routine allocates n bytes of memory for the duration of the external
procedure call. Any memory allocated by the function is freed as soon as control
returns to PL/SQL.
Do not use any other function to allocate or free memory.
The C prototype for this function follows:
void *OCIExtProcAllocCallMemory(
OCIExtProcContext *with_context,
size_t amount);

The parameters with_context and amount are the context pointer and number of bytes
to allocate, respectively. The function returns an untyped pointer to the allocated
memory. A return value of zero indicates failure.

18.1.2 OCIExtProcRaiseExcp
This service routine raises a predefined exception, which must have a valid Oracle
error number in the range 1 to 32767. After doing any necessary cleanup, the external
procedure must return immediately. (No values are assigned to OUT or IN OUT
parameters.) The C prototype for this function follows:

18-1
 Chapter 18
OCI Access Functions for External Procedures

int OCIExtProcRaiseExcp(
OCIExtProcContext *with_context,
size_t error_number);

The parameters with_context and error_number are the context pointer and Oracle
error number. The return values OCIEXTPROC_SUCCESS and OCIEXTPROC_ERROR indicate
success or failure.

18.1.3 OCIExtProcRaiseExcpWithMsg
This service routine raises a user-defined exception and returns a user-defined error
message. The C prototype for this function follows:
int OCIExtProcRaiseExcpWithMsg(
OCIExtProcContext *with_context,
size_t error_number,
text *error_message,
size_t len);

The parameters with_context, error_number, and error_message are the context

pointer, Oracle error number, and error message text. The parameter len stores the
length of the error message. If the message is a null-terminated string, len is zero. The
return values OCIEXTPROC_SUCCESS and OCIEXTPROC_ERROR indicate success or failure.

18.1.4 OCIExtProcGetEnv
This service routine enables OCI callbacks to the database during an external
procedure call. Use the OCI handles obtained by this function only for callbacks. If you
use them for standard OCI calls, the handles establish a new connection to the
database and cannot be used for callbacks in the same transaction. In other words,
during an external procedure call, you can use OCI handles for callbacks or a new
connection but not for both.
The C prototype for this function follows:
sword OCIExtProcGetEnv(
OCIExtProcContext *with_context,
OCIEnv **envh,
OCISvcCtx **svch,
OCIError **errh);

The parameter with_context is the context pointer, and the parameters envh, svch,
and errh are the OCI environment, service, and error handles, respectively. The return
values OCIEXTPROC_SUCCESS and OCIEXTPROC_ERROR indicate success or failure.

"Using Callbacks" shows how OCIExtProcGetEnv might be used in callbacks. For a

working example, see the script extproc.sql in the PL/SQL demo directory. (For the
location of this directory, see your Oracle installation or user's guide.) This script
demonstrates the calling of an external procedure. The companion file extproc.c
contains the C source code for the external procedure. To run the demo, follow the
instructions in extproc.sql. You must use an account that has CREATE LIBRARY
privileges.

18-2
 Chapter 18
Installing Java Cartridge Services Files

18.2 Installing Java Cartridge Services Files

The ODCI.jar and CartridgeServices.jar files must be installed into the SYS schema
to use the Java classes described here.
If you installed the Java option, then you must install the ODCI.jar and
CartridgeServices.jar files. You do not have to perform this task if you did not install
the Java option.

To install ODCI.jar and CartridgeServices.jar files

1. Activate the SQL*Plus prompt.
C:\sqlplus
2. When prompted, login using the system account.
Enter user-name: system
Enter password: password
3. Use the server-side loadjava command to install the classes and create the
synonyms in the SYSTEM schema.
SQL> call dbms_java.loadjava('-resolve -synonym -grant public
-verbose vobs/jilip/Cartridge Services.jar');
SQL> call dbms_java.loadjava('-resolve -synonym -grant public
-verbose vobs/jlib/ODCI.jar');
See Oracle Database Upgrade Guide for further details on installing the jar files.

18.3 Cartridge Services-Maintaining Context

The Java cartridge service is used for maintaining context. It is similar to the OCI
context management service. This class is necessary when switching context between
the server and the cartridge code.

18.3.1 ContextManager
ContextManager is a Constructor in class Oracle that extends Object.

Class Interface
public static Hashtable ctx extends Object

Variable
ctx public static Hashtable ctx

Constructors
ContextManager public ContextManager()

Methods
The following methods are available:

18-3
 Chapter 18
Cartridge Services-Maintaining Context

setContext (static method in class oracle)

getContext (static method in class oracle)
clearContext (static method in class oracle)

18.3.2 CountException()
Constructor that extends Exception.
Class oracle.CartridgeServices.CountException

18.3.3 CountException(String)
Constructor that extends Exception.
public CountException(String s)

18.3.4 InvalidKeyException()
Constructor that extends Exception.
public InvalidKeyException(String s)

18.3.5 InvalidKeyException(String)
Constructor that extends Exception.
public InvalidKeyException(String s)

18-4
19
Extensibility Constants, Types, and
Mappings
System Defined Constants and System Defined Types apply generically to all
supported languages, as well as mappings that are specific to the PL/SQL, C, and
Java languages.

19.1 System Defined Constants

Consider the constants defined in the ODCIConst package installed as part of the
catodci.sql script. There are equivalent definitions for use within C routines in
odci.h. You should use these constants instead of hard coding their underlying values
in your routines.
The system defined constants fall into two broad categories:
• Bit-field values that may be combined using the OR operator: Table 19-3,
Table 19-4, Table 19-5, Table 19-6, Table 19-7, Table 19-8, Table 19-9,
Table 19-10
• Distinct values, where only one option may be specified: Table 19-1, Table 19-2,
Table 19-12, Table 19-11

19.1.1 ODCIArgDesc.ArgType System Defined Constants

Table 19-1 ODCIArgDesc.ArgType Values

Name Description
Argument is other expression
ArgOther

Argument is a column name

ArgCol

Argument is a literal value

ArgLit

ArgAttr Argument is an ADT attr column

ArgCursor Argument is a CURSOR expression

ArgNull Argument is NULL

19-1
 Chapter 19
System Defined Constants

19.1.2 ODCIEnv.CallProperty System Defined Constants

Table 19-2 ODCIEnv.CallProperty Values

Name Description
Default option
None

First partition call

FirstCall

Intermediate partition call

Intermediate Call

Final call after last partition

FinalCall

Used to specify global statistics gathering

StatsGlobal

Used to specify global and partition-level statistics gathering

StatsGlobalAndPartition

Used to specify partition-level statistics gathering

StatsPartition

19.1.3 ODCIIndexAlter System Defined Constants

Table 19-3 ODCIIndexAlter Options

Name Description
Default option
AlterIndexNone

Rename Partition option

AlterIndexRename

Rebuild Index option

AlterIndexRebuild

IOT update block references

AlterIndexUpdBlockRefs

Migrate user-managed domain index to a system-managed

AlterIndexMigrate
domain index.
Rename the column on which the domain index is based
AlterIndexRenameCol

Rename the table on which the domain index is based

AlterIndexRenameTab

19-2
 Chapter 19
System Defined Constants

19.1.4 ODCIIndexInfo.Flags System Defined Constants

Table 19-4 ODCIIndexInfo.Flags Bits

Name Bits Value Description

CONSTANT INTEGER := 1; Indicates a local domain index
Local

Parallel CONSTANT INTEGER := 16; Indicates that a parallel degree was

specified for the index creation or
alter operation

Unusable CONSTANT INTEGER := 32; Indicates that UNUSABLE was

specified during index creation, and
that the index is marked unusable

IndexOnIOT CONSTANT INTEGER := 64; Indicates that the domain index is

defined on an index-organized table

CONSTANT INTEGER := 2; For a local domain index, indicates

RangePartn
that the base table is range-
partitioned. Is set only in conjunction
with the Local bit
CONSTANT INTEGER := For a local domain index, indicates
ListPartn
512; that the base table is list-partitioned.
Is set only in conjunction with the
Local bit.
CONSTANT INTEGER := 4; For a local domain index, indicates
HashPartn
that the base table is hash-
partitioned. Is set only in conjunction
with the Local bit
CONSTANT INTEGER := Indicates that the domain index is
TransTblspc
128; created in a transportable tablespace
session.
CONSTANT INTEGER := Indicates that the index is a function-
FunctionIdx
256; based domain index

CONSTANT INTEGER := 8; Indicates an online domain index

Online

CONSTANT INTEGER := Indicates that the index is an updated

UpdateGlobalIndexes
1024; global domain index

CONSTANT INTEGER := For a local domain index, indicates

RefPartn
2048; that the base table is reference-
partitioned. Is set only in conjunction
with the Local bit.
CONSTANT INTEGER := For a local domain index, indicates
CompPartn
8192; that the base table is composite
partitioned. Is set in conjunction with
the Local bit and the RangePartn,
ListPartn, or HashPartn bits.

19-3
 Chapter 19
System Defined Constants

Table 19-4 (Cont.) ODCIIndexInfo.Flags Bits

Name Bits Value Description

CONSTANT INTEGER := For a local domain index, indicates
SubPartn
16384 that the partition information for the
current call refers to a subpartition. Is
set in conjunction with the
CompPartn bit.

19.1.5 ODCIIPartInfo.PartOp System Defined Constants

Table 19-5 Description of the ODCIIPartInfo.PartOp System Defined Constant

Name Description
The partition to be added
AddPartition

The partition to be dropped

DropPartition

19.1.6 ODCIIPredInfo.Flags System Defined Constants

Table 19-6 ODCIIPredInfo.Flags Bits

Name Description
Equality predicate
PredExactMatch

PredPrefixMatch LIKE predicate

Include start value in index range scan

PredIncludeStart

Include stop value in index range scan

PredIncludeStop

Left hand side of predicate is a standalone function

PredObjectFunc

Left hand side of predicate is a package function

PredObjectPkg

Left hand site of predicate is a type method

PredObjectType

Predicate contains columns from several tables

PredMultiTable

Inequality predicate
PredNotEqual

19-4
 Chapter 19
System Defined Constants

19.1.7 ODCIFuncInfo.Flags System Defined Constants

Table 19-7 ODCIFuncInfo.Flags Bits

Name Description
Standalone function
ObjectFunc

Package function
ObjectPkg

Type method
ObjectType

19.1.8 ODCIQueryInfo.Flags System Defined Constants

Table 19-8 ODCIQueryInfo.Flags Bits

Name Description

QueryFirstRows Optimizer mode is FIRST_ROWS

QueryAllRows Optimizer mode is ALL_ROWS

19.1.9 ODCIStatsOptions.Flags System Defined Constants

Table 19-9 ODCIStatsOptions.Flags Bits

Name Description
Estimate statistics option
EstimateStats

Compute exact statistics option

ComputeStats

Validate index option

Validate

19.1.10 ODCIStatsOptions.Options System Defined Constants

Table 19-10 ODCIStatsOptions.Options Bits

Name Description
Compute statistics by sampling
PercentOption

Compute statistics based on all rows

RowOption

19-5
 Chapter 19
System-Defined Types

19.1.11 Return Status System Defined Constants

Table 19-11 Return Status Values

Name Description
Indicates a successful operation.
Success

Indicates an error.
Error

Indicates a warning.
Warning

Indicates that there is an error in an index partition, but continues

ErrContinue
to work on the next partition.

Indicates that all dictionary entries of the index are cleaned up,
Fatal
and that the CREATE INDEX operation is rolled back

19.1.12 ScnFlg System Defined Constants

Table 19-12 ScnFlg Values; Function with Index Context

Name Description
User defined operator regular call
RegularCall

User defined operator cleanup call

CleanupCall

19.2 System-Defined Types

Several system-defined types are defined by Oracle and must be created by running
the catodci.sql catalog script. The C mappings for these object types are defined in
odci.h. The ODCIIndex and ODCIStats routines described in Extensible Indexing
Interface and Extensible Optimizer Interface use these types as parameters.
Unless otherwise mentioned, the names parsed as type attributes are unquoted
identifiers.

19.2.1 ODCIArgDesc
Object type. Stores function or operator arguments.

19-6
 Chapter 19
System-Defined Types

Table 19-13 ODCIArgDesc Function and Operator Argument Description -

Attributes

Name Data Type Description

Argument type
ArgType NUMBER

Name of table
TableName VARCHAR2(30)

Schema containing the table

TableSchema VARCHAR2(30)

Name of column. This could be top level

ColName VARCHAR2(4000)
column name such as "A", or a nested
column "A"."B" Note that the column name
are quoted identifiers.
Contains the name of the lowest table
TablePartitionLower VARCHAR2(30)
partition that is accessed in the query

Contains the name of the highest table

TablePartitionUpper VARCHAR2(30)
partition that is accessed in the query

Cardinality NUMBER Cardinality value for CURSOR expressions

19.2.2 ODCIArgDescList
Contains a list of argument descriptors

Data Type
VARRAY(32767) of ODCIArgDesc

19.2.3 ODCIRidList
Stores list of rowids. The rowids are stored in their character format.

Data Type
VARRAY(32767) OF VARCHAR2("M_URID_SZ")

19.2.4 ODCIColInfo
Stores column related information.

Data Type
Object type.

19-7
 Chapter 19
System-Defined Types

Table 19-14 ODCIColInfo Column Related Information - Attributes

Name Data Type Purpose

Schema containing table
TableSchema VARCHAR2(30)

Name of table
TableName VARCHAR2(30)

Name of column. This could be top level

ColName VARCHAR2(4000)
column name such as "A", or a nested column
"A"."B" Note that the column name are quoted
identifiers.
Data Type of column
ColTypeName VARCHAR2(30)

Schema containing data type if user-defined

ColTypeSchema VARCHAR2(30)
data type

For a local domain index, contains the name

TablePartition VARCHAR2(30)
of the specific base table partition
Base table partition physical identifier
TablePartitionIden NUMBER

Total number of partitions in a table

TablePartitionTotal NUMBER

19.2.5 ODCIColInfoList
Stores information related to a list of columns.

Data Type
VARRAY(32) OF ODCIColInfo

19.2.6 ODCICost
Object type. Stores cost information.

Table 19-15 ODCICost Cost Information - Attributes

Name Data Type Purpose

CPU cost
CPUCost NUMBER

I/O cost
IOCost NUMBER

Communication cost
NetworkCost NUMBER

Optional user-supplied information about the

IndexCostInfo VARCHAR2(255)
domain index for display in the PLAN table (255
characters maximum)

19-8
 Chapter 19
System-Defined Types

19.2.7 ODCIEnv
Object type. Contains general information about the environment in which the
extensibility routines are executing.

Table 19-16 ODCIEnv Environment Variable Descriptor Information - Attributes

Name Data Type Purpose

EnvFlags NUMBER • 1 = Debugging On
• 2 = NoData; used in ODCIIndexAlter() method with
alter_option = AlterIndexRebuild to indicate that
there is no data in the base partition. It is set only
when ODCIIndexAlter() is used as part of
TRUNCATE TABLE and partition management
operations.
• 4 = UserParamString; used in ODCIIndexAlter()
method with alter_option = AlterIndexRebuild to
indicate that the parameter string was specified by
the user. It is set only when ODCIIndexAlter is
invoked when rebuilding an index partition.
• 8 = RowMigration; used in ODCIIndexInsert() and
ODCIIndexDelete() to indicate that the respective
call to these interfaces is done because an
UPDATE operation caused the row to be migrated
from one partition to another.
CallProperty NUMBER • 0 = None
• 1 = First Call
• 2 = Intermediate Call
• 3 = Final Call
• 6 = Global Statistics
• 7 = Global and Partition Statistics
• 8 = Partition Statistics
DebugLevel NUMBER Indicates the level of debugging

Usage Notes
CallProperty is used only for CREATE INDEX, DROP INDEX, TRUNCATE TABLE, and for
some extensible optimizer-related calls. In all other cases, including DML and query
routines for local domain indexes, it is set to 0.

19.2.8 ODCIFuncInfo
Object type. Stores functional information.

Table 19-17 ODCIFuncInfo Function Information - Attributes

Name Data Type Purpose

Object schema name
ObjectSchema VARCHAR2(30)

Function/package/type name
ObjectName VARCHAR2(30)

19-9
 Chapter 19
System-Defined Types

Table 19-17 (Cont.) ODCIFuncInfo Function Information - Attributes

Name Data Type Purpose

Method name for package/type
MethodName VARCHAR2(30)

Function flags - see ODCIConst

Flags NUMBER

19.2.9 ODCIIndexInfo
Object type. Stores the metadata information related to a domain index. It is passed as
a parameter to all ODCIIndex routines.

Table 19-18 ODCIIndexInfo Index Related Information - Attributes

Name Data Type Purpose

Schema containing domain index
IndexSchema VARCHAR2(30)

Name of domain index

IndexName VARCHAR2(30)

List of indexed columns

IndexCols ODCIColInfoList

For a local domain index, contains the

IndexPartition VARCHAR2(30)
name of the specific index partition

Possible flags are:

IndexInfoFlags NUMBER
• Local
• RangePartn
• Parallel
• Unusable
• IndexOnIOT
• ListPartn
• TransTblspc
• FunctionIdx
• HashPartn
The degree of parallelism, if one is specified
IndexParaDegree NUMBER
when creating or rebuilding a domain index
or local domain index partition in parallel
The index partition object identifier, for local
IndexPartitionIden NUMBER
domain indexes
The total number of partitions in an index
IndexPartitionTotal NUMBER

19-10
 Chapter 19
System-Defined Types

19.2.10 ODCIIndexCtx
Object type. Stores the index context, including the domain index metadata and the
rowid. It is passed as parameter to the functional implementation of an operator that
expects index context.

Table 19-19 ODCIIndexCtx Index Context Related Information - Attributes

Name Data Type Purpose

Stores the metadata information about the
IndexInfo ODCIIndexInfo
domain index

Row identifier of the current row

rid VARCHAR2("M_URID_SZ")

19.2.11 ODCIObject
Object type. Stores information about a schema object.

Table 19-20 ODCIObject Index Context Related Information - Attributes

Name Data Type Purpose

Name of schema in which object is located
ObjectSchema VARCHAR2(30)

Name of object
ObjectName VARCHAR2(30)

19.2.12 ODCIObjectList
Stores information about a list of schema objects.

Data Type
VARRAY(32) OF ODCIObject

19.2.13 ODCIPartInfo
Object type. Contains the names of both the table partition and the index partition.

Table 19-21 ODCIPartInfo Index-Related Information - Attributes

Name Data Type Purpose

VARCHAR2(30) Table partition name
TablePartition

VARCHAR2(30) Index partition name

IndexPartition

19-11
 Chapter 19
System-Defined Types

Table 19-21 (Cont.) ODCIPartInfo Index-Related Information - Attributes

Name Data Type Purpose

NUMBER Index partition object identifier
IndexPartitionIden

NUMBRER Partition operation that is being performed

PartOp

19.2.14 ODCIPartInfoList
Stores information related to a list of partitions.

Data Type
VARRAY(64000) OF ODCIPartInfo

19.2.15 ODCIPredInfo
Object type. Stores the metadata information related to a predicate containing a user-
defined operator or function. It is also passed as a parameter to the ODCIIndexStart()
query routine.

Table 19-22 ODCIPredInfo Operator Related Information - Attributes

Name Data Type Purpose

Schema of operator/function
ObjectSchema VARCHAR2(30)

Name of operator/function
ObjectName VARCHAR2(30)

Name of method, applies only to package methods

MethodName VARCHAR2(30)
type

Possible flags are:

Flags NUMBER
• PredExactMatch - Exact Match
• PredPrefixMatch - Prefix Match
• PredIncludeStart - Bounds include the start
key value
• PredIncludeStop - Bounds include the stop key
value
• PredMultiTable - Predicate involves multiple
tables
• PredNotEqual - Predicate inequality
• PredObjectFunc - Object is a function
• PredObjectPkg - Object is a package
• PredObjectType - Object is a type

19-12
 Chapter 19
System-Defined Types

19.2.16 ODCIQueryInfo
Object type. Stores information about the context of a query. It is passed as a
parameter to the ODCIIndexStart() routine.

Table 19-23 ODCIQueryInfo Index Context Related Information - Attributes

Name Data Type Purpose

The following flags can be set:
Flags NUMBER
• QueryFirstRows - Set when the optimizer hint
FIRST_ROWS is specified in the query
• QueryAllRows - Set when the optimizer hint
ALL_ROWS is specified in the query
Ancillary operators referenced in the query
AncOps ODCIObjectList

19.2.17 ODCIStatsOptions
Object type. Stores options information for DBMS_STATS.

Table 19-24 ODCIStatsOptions Cost Information - Attributes

Name Data Type Purpose

Sample size
Sample NUMBER

DBMS_STATS options - see "ODCICost"

Options NUMBER

DBMS_STATS flags - see "ODCICost"

Flags NUMBER

19.2.18 ODCITabFuncStats
Object type. Stores cardinality information for a table function.

Table 19-25 ODCITabFuncStats Parameter

Parameter Data Type Purpose

num_rows NUMBER Contains the number of rows expected to be returned by the
table function

19.2.19 ODCITabStats
Stores table statistics for a table function.

Data Type
NUMBER

19-13
 Chapter 19
System-Defined Types

Table 19-26 ODCITabStats - Attributes

Name Data Type Purpose

Num_rows NUMBER Number of rows in table

19.2.20 ODCIBFileList
Stores varrays of BFILEs.

Data Type
VARRAY(32767) OF BFILE

19.2.21 ODCITabFuncInfo
Object type. Stores information on which attributes of user-defined types in a collection
must be set by a table function.

Table 19-27 ODCITabFuncInfo Parameters

Name Data Type Purpose

Attrs ODCINumberList Indicates the attributes that must be set
RetType AnyType For AnyDataSet table functions, indicates the actual
return type to be expected in the AnyDataSet collection

19.2.22 ODCIDateList
Stores varrays of DATEs.

Data Type
VARRAY(32767) OF DATE

19.2.23 ODCINumberList
Stores varrays of NUMBERs.

Data Type
VARRAY(32767) OF NUMBER

19.2.24 ODCIRawList
Stores varrays of Raws.

Data Type
VARRAY(32767) OF Raw(2000)

19-14
 Chapter 19
Mappings of Constants and Types

19.2.25 ODCIVarchar2List
Stores varrays of VARCHAR2s

Data Type
VARRAY(32767) OF VARCHAR2(4000)

19.2.26 ODCIFuncCallInfo
Object type. Stores information about the functional implementation of an operator.

Table 19-28 ODCIFuncCallInfo - Attributes

Name Data Type Purpose

ColInfo ODCIColInfo Information about the column on which the operator is invoked

Usage Notes
A functional implementation can be defined with this parameter only if the operator
binding is declared WITH COLUMN CONTEXT. This is useful if the functional
implementation requires information about the column it was invoked on, and there is
no domain index defined on the column. This argument is only populated in the
function invocation if the first argument of the operator invocation is a column and
there is no domain index defined on that column.

19.3 Mappings of Constants and Types

Consider language-specific constant and type mappings.

19.3.1 Mappings in PL/SQL

A variety of PL/SQL mappings are common to both Extensible Indexing and the
Extensible Optimizer.
• Constants are defined in the ODCIConst package found in catodci.sql
• Types are defined as object types found in catodci.sql

19.3.2 Mappings in C
Mappings of constants and types are defined for C in the public header file odci.h.
Each C structure to which a type is mapped has a corresponding indicator structure
called structname_ind and a reference definition called structname_ref.

19-15
20
Extensible Indexing Interface
Various Oracle Data Cartridge Interface extensible indexing parameters are described.

20.1 Extensible Indexing - System-Defined Interface

Routines
Table 20-1 summarizes the extensible indexing routines.

Caution:
These routines are invoked by Oracle at the appropriate times based on SQL
statements executed by the end user. Do not invoke these routines directly
as this may result in corruption of index data.

Table 20-1 Summary of System-Defined Extensible Indexing Interface Routines

Routine Description
ODCIGetInterfaces() Invoked when an INDEXTYPE is created by a CREATE
INDEXTYPE... statement or is altered.
ODCIIndexAlter() Invoked when a domain index or a domain index partition is
altered using an ALTER INDEX, an ALTER INDEX
PARTITION, a TRUNCATE TABLE, a RENAME TABLE, an
ALTER TABLE RENAME COLUMN, or an ALTER TABLE
[ADD|TRUNCATE|SPLIT|MERGE] PARTITION statement.
ODCIIndexClose() Invoked to end the processing of an operator.
ODCIIndexCreate() Invoked when a domain index is created by a CREATE
INDEX...INDEXTYPE IS...PARAMETERS... statement
issued by the user.
ODCIIndexDelete() Invoked when a row is deleted from a table that has a
domain index defined on one or more of its columns.
ODCIIndexDrop() Invoked when a domain index is dropped explicitly using a
DROP INDEX statement, or implicitly through a DROP TABLE
or DROP USER statement.
ODCIIndexExchangePartition() Invoked when an ALTER TABLE EXCHANGE
PARTITION...INCLUDING INDEXES is issued on a
partitioned table on which a local domain index is defined.
ODCIIndexFetch() Invoked repeatedly to retrieve the rows satisfying the
operator predicate.
ODCIIndexGetMetadata() Returns a series of strings of PL/SQL code that comprise
the non-dictionary metadata associated with the index.

20-1
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

Table 20-1 (Cont.) Summary of System-Defined Extensible Indexing Interface

Routines

Routine Description
ODCIIndexInsert() Invoked when a row or a set of rows is inserted into a table
that has a domain index defined on one or more of its
columns.
ODCIIndexStart() Invoked to start the evaluation of an operator on an indexed
column.
ODCIIndexUpdate() Invoked when a row is updated in a table and the updated
column has a domain index defined on.
ODCIIndexUpdPartMetadata() Invoked during partition maintenance operations. Patches
the indextype metadata tables to correctly reflect the
partition maintenance operation.
ODCIIndexUtilCleanup() Cleans up temporary states created by
ODCIIndexUtilGetTableNames().
ODCIIndexUtilGetTableNames() Determines if the secondary tables storing the index data
should be transported.

20.1.1 ODCIGetInterfaces()
Invoked when an INDEXTYPE is created by a CREATE INDEXTYPE... statement or is
altered.

Syntax
FUNCTION ODCIGetInterfaces(
ifclist OUT ODCIObjectList)
RETURN NUMBER

Parameter Description
Contains information about the interfaces it supports
ifclist

Returns
ODCIConst.Success on success or ODCIConst.Error on error

Usage Notes
This function should be implemented as a static type method.
This function must return SYS.ODCIINDEX2 in the ODCIObjectList if the indextype uses
the second version of the ODCIIndex interface.

20.1.2 ODCIIndexAlter()
Invoked when a domain index or a domain index partition is altered using one of the
following methods:
• ALTER INDEX

20-2
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

• ALTER INDEX PARTITION

• TRUNCATE TABLE table_name
• RENAME TABLE
• ALTER TABLE...[ADD|TRUNCATE|SPLIT|MERGE]...PARTITION
• ALTER TABLE RENAME
• ALTER TABLE RENAME COLUMN
To populate the index partitions when creating local domain indexes, this method is
invoked for each partition of the base table.

Syntax
STATIC FUNCTION ODCIIndexAlter(
ia ODCIIndexInfo,
parms IN OUT VARCHAR2,
alter_option NUMBER,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Contains information about the index and the indexed column
ia

Parameter string
parms (IN)
• With ALTER INDEX PARAMETERS or ALTER INDEX REBUILD,
contains the user specified parameter string
• With ALTER INDEX RENAME, contains the new name of the domain
index
• With ALTER TABLE RENAME COLUMN, contains the new domain-
indexed column name
• With ALTER TABLE RENAME or RENAME TABLE, contains the
new table name
Parameter string
parms (OUT)
Valid only with ALTER INDEX PARAMETERS or ALTER INDEX REBUILD;
contains the resultant string to be stored in system catalogs
Specifies one of the following options:
alter_option
• AlterIndexNone if ALTER INDEX [PARTITION] PARAMETERS
• AlterIndexRename if ALTER INDEX RENAME [PARTITION]
• AlterIndexRebuild if ALTER INDEX REBUILD [PARTITION]
[PARAMETERS]
• AlterIndexRenameCol if ALTER TABLE RENAME COLUMN
• AlterIndexRenameTab if ALTER TABLE RENAME or RENAME
TABLE
• AlterIndexUpdBlockRefs if ALTER TABLE UPDATE BLOCK
REFERENCES
• AlterIndexMigrate if ALTER INDEX COMPILE when the domain
index is user-managed, but its indextype is system-managed
The environment handle passed to the routine
env

20-3
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

Returns
ODCIConst.Success on success, ODCIConst.Error on error, or ODCIConst.Warning
otherwise. When invoked to rebuild local index partitions, may also return
ODCIConst.ErrContinue.

Usage Notes
• This function should be implemented as a static type method.
• An ALTER INDEX statement can be invoked for domain indexes in multiple ways.
ALTER INDEX index_name
PARAMETERS (parms);

or
ALTER INDEX index_name
REBUILD PARAMETERS (parms);

The precise behavior in these two cases is defined by the implementation. One
possibility is that the first statement would merely reorganize the index based on
the parameters while the second would rebuild it from scratch.
• The maximum length of the input parameters string is 1000 characters. The OUT
value of the parms argument can be set to resultant parameters string to be stored
in the system catalogs.
• The ALTER INDEX statement can also be used to rename a domain index in the
following way:
ALTER INDEX index_name
RENAME TO new_index_name
• When the name of the table on which a domain index is created changes,
ODCIIndexAlter() is invoked with alter_option=AlterIndexRenameTab, and
new_table_name is passed to the parms argument:
ALTER TABLE table_name
RENAME new_table_name

or
RENAME table_name
TO new_table_name
• When the name of the column on which a domain index is created changes,
ODCIIndexAlter() is invoked with alter_option=AlterIndexRenameCol, and
new_column_name is passed to the parms argument:
ALTER TABLE table_name
RENAME COLUMN column_name
TO new_column_name
• If the PARALLEL clause is omitted, then the domain index or local domain index
partition is rebuilt sequentially.
• If the PARALLEL clause is specified, the parallel degree is passed to the
ODCIIndexAlter() invocation in the IndexParaDegree attribute of ODCIIndexInfo,
and the Parallel bit of the IndexInfoFlags attribute is set. The parallel degree is
determined as follows:

20-4
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

– If PARALLEL DEGREE deg is specified, deg is passed.

– If only PARALLEL is specified, then a constant is passed to indicate that the
default degree of parallelism was specified.
• If the ODCIIndexAlter routine returns with the ODCIConst.Success, the index is
valid and usable. If the ODCIIndexAlter() routine returns with ODCIConst.Warning,
the index is valid and usable but a warning message is returned to the user. If
ODCIIndexAlter() returns with an error (or exception), the domain index is
marked FAILED.
• When the ODCIIndexAlter() routine is being executed, the domain index is
marked LOADING.
• Every SQL statement executed by ODCIIndexAlter() is treated as an
independent operation. The changes made by ODCIIndexCreate() are not
guaranteed to be atomic.
• The AlterIndexUpdBlockRefs alter option applies only to domain indexes on
index-organized tables. When the end user executes an ALTER INDEX
domain_index UPDATE BLOCK REFERENCES, ODCIIndexAlter() is called with the
AlterIndexUpdBlockRefs bit set to give the cartridge developer the opportunity to
update guesses as to the block locations of rows, stored in logical rowids.
• The AlterIndexMigrate alter options applies only to migration of user-managed
domain indexes to system-managed domain indexes. When the user-managed
domain index is marked INVALID, but its indextype is system-managed, you must
make an ALTER INDEX domain_index COMPILE call to re-validate the domain index.
This calls the ODCIIndexAlter() method with alter_option=AlterIndexMigrate,
to allow an opportunity to migrate the domain index to the system-managed
approach.

See Also:

• ODCIIndexAlter()
• ODCIIndexCreate()

20.1.3 ODCIIndexClose()
Invoked to end the processing of an operator.

Syntax
FUNCTION ODCIIndexClose(
self IN impltype,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Is the value of the context returned by the previous invocation of
self(IN)
ODCIIndexFetch()

The environment handle passed to the routine

env

20-5
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

Returns
• ODCIConst.Success on success
• ODCIConst.Error on error

Usage Notes
The index implementor can perform any appropriate actions to finish up the processing
of an domain index scan, such as freeing memory and other resources.

20.1.4 ODCIIndexCreate()
Invoked when a domain index is created by a CREATE INDEX...INDEXTYPE
IS...PARAMETERS... statement issued by the user. The domain index can be either a
non-partitioned index or a local partitioned domain index. The local partitioned domain
index can be created in either a system- or a user-managed scheme.

Syntax
FUNCTION ODCIIndexCreate(
ia ODCIIndexInfo,
parms VARCHAR2,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Contains information about the index and the indexed column
ia

parms The PARAMETERS string passed in not interpreted by Oracle. The

maximum size of the parameter string is 1,000 characters.
The environment handle passed to the routine
env

Returns
ODCIConst.Success , ODCIConst.Error, ODCIConst.Warning, ODCIConst.ErrContinue
if the method is invoked at the partition level for creation of a local partitioned index, to
continue to the next partition even in case of an error, or ODCIConst.Fatal to signify
that all dictionary entries for the index are cleaned up and that the CREATE INDEX
operation is rolled back. Returning this status code assumes that the cartridge code
has not created any objects (or cleaned up any objects created).

Usage Notes
• This function should be implemented as a STATIC type method.
• Creates objects (such as tables) to store the index data, generate the index data,
and store the data in the index data tables.
• This procedure should handle creation of indexes on both empty and non-empty
tables. If the base table is not empty, the procedure can scan the entire table and
generate index data.

20-6
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

• When the ODCIIndexCreate() routine is running, the domain index is marked

LOADING.
• Every SQL statement executed by ODCIIndexCreate() is treated as an
independent operation. The changes made by ODCIIndexCreate() are not
guaranteed to be atomic.
• To create a non-partitioned domain index, the ODCIIndexCreate() method is
invoked, and the only valid return codes are ODCIConst.Success,
ODCIConst.Warning, ODCIConst.Error, or ODCIConst.Fatal. If the operation
returns ODCIConst.Fatal, the CREATE INDEX statement is rolled back by the
server.
• In a non-partitioned domain index, the IndexPartition, TablePartition name,
and the callProperty should be NULL.
• For a non-partitioned domain index, the parallel degree is passed to the
ODCIIndexCreate() invocation in the IndexParaDegree attribute of ODCIIndexInfo,
and the Parallel bit of the IndexInfoFlags is set. The parallel degree is
determined as follows:
– If PARALLEL DEGREE deg is specified, deg is passed.
– If only PARALLEL is specified, then a constant indicating that the default degree
of parallelism was specified, is passed.
– If the PARALLEL clause is omitted entirely, the operation is performed
sequentially.
• If the ODCIIndexCreate() routine returns with the ODCIConst.Success, the index is
valid and usable. If the ODCIIndexCreate() routine returns with
ODCIConst.Warning, the index is valid and usable but a warning message is
returned to the user. If the ODCIIndexCreate() routine returns with an
ODCIConst.Error (or exception), the domain index is marked FAILED.
• The only operations permitted on FAILED domain indexes is DROP INDEX, TRUNCATE
TABLE or ALTER INDEX REBUILD.
• If a domain index is created on an column of object type which contains a REF
attribute, do not dereference the REFs while building your index. Dereferencing a
REF fetches data from a different table instance. If the data in the other table is
modified, the domain index becomes incorrect. Note that the user is not notified.
• The ODCIIndexCreate() method is invoked twice for the creation of system
managed local domain indexes and the only valid return codes are
ODCIConst.Success, ODCIConst.Warning or ODCIConst.Error. ODCIConst.Fatal
can be returned by the first call and results in the CREATE INDEX statement being
rolled back by the server. The number of partitions is passed in as an argument
ODCIIndexInfo.IndexPartitionTotal. The first call should create all the index
storage tables. All the index storage tables should preferably be system partitioned
to get the benefits of local domain indexes. Also:
– These tables must have the same number of partitions as the base table. If the
base table is composite partitioned, then the domain index storage table
should have the same number of total subpartitions as the base table.
– The users should generate the create table statement with both object and
partition level attributes.
• Note that the object level create routine only passes in the object level parameter
string. However, to construct the storage attributes for all the partitions, it needs

20-7
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

the partition level parameter strings. The cartridge indexing code must obtain them
by querying the *_ind_partitions views on the dictionary tables. The system
partitioned tables should not be populated in this phase. The user should wait for
the subsequent calls ODCIIndexAlter() to populate the partitions. Also, it is
recommended that the users should derive the names of the storage tables and its
partitions from the index name and the index partition names. In this case, the
user should fetch the index partition names from the *_ind_partitions view and
construct the partition names for the storage table.
• In the second ODCIIndexCreate() call, the user can create domain index storage
table dependent objects, such as indexes, constraints, and triggers. These can be
created as before by directly using the SQL callbacks. However, for system
partitioned storage tables, the following types of indexes are disallowed:
– non-partitioned index
– globally partitioned index
• Sequence numbers and synonyms can be created using callbacks and they are
assumed to be partition-independent. The set of objects created for non-
partitioned domain index is identical to that of a local partitioned index and these
objects are not impacted when a table or partition maintenance operation is done.
It is the users responsibility to drop these objects when the index is dropped.
• Other (transient) objects needed for temporary use can be created using callbacks
as before. It is the responsibility of user-supplied code to drop them by the end of
the create call.
• Temporary tables can be created for holding intermediate data. The server does
not perform maintenance operations on these tables.
• External Objects, such as files, can be created for temporary use.
• All the tables left after the invocation of ODCIIndexCreate() or ODCIIndexAlter()
are supposed to be system-managed, and the server takes appropriate actions on
them during drop, truncate, or the partition maintenance operations.
• Since this routine handles multiple things, such as creation of a non-partitioned
index or creation of a local index, you must take special care to code it
appropriately.

See Also:

• ODCIIndexAlter()
• ODCIIndexCreate()

20.1.5 ODCIIndexDelete()
Invoked when a row is deleted from a table that has a domain index defined on one or
more of its columns.

Syntax
FUNCTION ODCIIndexDelete(
ia ODCIIndexInfo,
rid VARCHAR2,

20-8
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

oldval icoltype,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Contains information about the index and the indexed column
ia

The row identifier of the deleted row

rid

The value of the indexed column in the deleted row. The data type is
oldval
identical to that of the indexed column.

The environment handle passed to the routine

env

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

Usage Notes
• This function should be implemented as a STATIC type method.
• This method should delete index data corresponding to the deleted row from the
appropriate tables or files storing index data.
• Note that the index partition object identifier ODCIIndexInfo.IndexPartitionIden
and the base table partition physical identifier
ODCIIndexInfo.IndexCols(1).TablePartitionIden is passed in for local domain
index. The indextype must use the new DML syntax using the partition number
and the provided DATAOBJ_TO_MAT_PARTITION function to delete data from the
storage system partitioned table:
DELETE FROM SP PARTITION (
DATAOBJ_TO_MAT_PARTITION(
base_table_name,
:tab_physical_partid))
VALUES(…)
WHERE rowid = :rowid;

20.1.6 ODCIIndexDrop()
The ODCIIndexDrop() procedure is invoked when a domain index is dropped explicitly
using a DROP INDEX statement, or implicitly through a DROP TABLE or DROP USER
statement.

Syntax
FUNCTION ODCIIndexDrop(
ia ODCIIndexInfo,
env ODCIEnv)
RETURN NUMBER

20-9
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

Parameter Description
Contains information about the index and the indexed column
ia

The environment handle passed to the routine

env

Returns
ODCIConst.Success on success, or ODCIConst.Error on error, or ODCIConst.Warning.

Usage Notes
• This method should be implemented as a static type method.
• This method should drop the tables storing the domain index data.
• For both a non-partitioned domain index and system managed local domain index,
the ODCIIndexDrop() method is invoked only one time. The user need not drop the
index storage tables if the system-managed approach is used. This is done
automatically by the kernel after the call is completed.
• Since it is possible that the domain index is marked FAILED (due to abnormal
termination of some DDL routine), the ODCIIndexDrop() routine should be capable
of cleaning up partially created domain indexes. When the ODCIIndexDrop()
routine is being executed, the domain index is marked LOADING.
• Note that if the ODCIIndexDrop() routine returns with an ODCIConst.Error or
exception, the DROP INDEX statement fails and the index is marked FAILED. In that
case, there is no mechanism to get rid of the domain index except by using the
FORCE option. If the ODCIIndexDrop() routine returns with ODCIConst.Warning in
the case of an explicit DROP INDEX statement, the operation succeeds but a
warning message is returned to the user.
• Every SQL statement executed by ODCIIndexDrop() is treated as an independent
operation. The changes made by ODCIIndexDrop() are not guaranteed to be
atomic.
• For both a non-partitioned domain index and system managed local domain index,
the ODCIIndexDrop() method is invoked only one time. With the system-managed
approach, the index storage tables don't have to be dropped. This is done
automatically by the kernel after the call is completed.

See Also:
ODCIIndexDrop()

20.1.7 ODCIIndexExchangePartition()
This method is invoked when an ALTER TABLE EXCHANGE PARTITION...INCLUDING
INDEXES command is issued on a partitioned table that has a defined local domain
index.

20-10
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

Syntax
FUNCTION ODCIIndexExchangePartition(
ia ODCIIndexInfo,
ia1 ODCIIndexInfo,
env ODCIEnv)
RETURN NUMBER

Parameter Description
ia Contains information about the domain index partition to exchange.
ia1 Contains information about the non-partitioned domain index.
env The environment handle passed to the routine

Returns
ODCIConst.Success on success, or ODCIConst.Error on error, or ODCIConst.Warning.

Usage Notes
• The function should be implemented as a STATIC type method.
• This method should handle both converting a partition of a domain index into a
non-partitioned domain index and converting a non-partitioned index to a partition
of a partitioned domain index.
• When exchanging composite partitions of a composite partitioned table, a call is
made to ODCIIndexExchangePartition for each subpartition of the partition to
exchange.

20.1.8 ODCIIndexFetch()
This procedure is invoked repeatedly to retrieve the rows satisfying the operator
predicate.

Syntax
FUNCTION ODCIIndexFetch(
self IN [OUT] impltype,
nrows IN NUMBER,
rids OUT ODCIRidList,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Is the value of the context returned by the previous call (to
self(IN)
ODCIIndexFetch or to ODCIIndexStart() if this is the first time fetch
is being called for this operator instance
The context that is passed to the next query-time call. Note that this
self(OUT)
parameter does not have to be defined as OUT if the value is not
modified in this routine.
Is the maximum number of result rows that can be returned to Oracle in
nrows
this call

20-11
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

Parameter Description
Is the array of row identifiers for the result rows being returned by this
rids
call

The environment handle passed to the routine

env

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

Usage Notes
• ODCIIndexFetch() returns rows satisfying the operator predicate. That is, it returns
the row identifiers of all the rows for which the operator return value falls within the
specified bounds.
• Each call to ODCIIndexFetch() can return a maximum of nrows number of rows.
The value of nrows passed in is decided by Oracle based on some internal factors.
However, the ODCIIndexFetch() routine can return lesser than nrows number of
rows. The row identifiers are returned through the output rids array. A NULL ROWID
(as an element of the rids array) indicates that all satisfying rows have been
returned.
Assume that there are 3000 rows which satisfy the operator predicate, and that the
value of nrows = 2000. The first invocation of ODCIIndexFetch() can return the
first 2000 rows. The second invocation can return a rid list consisting of the
remaining 1000 rows followed by a NULL element. The NULL value in rid list
indicates that all satisfying rows have now been returned.
• If the context value is changed within this call, the new value is passed in to
subsequent query-time calls.

See Also:

• ODCIIndexFetch()
• ODCIIndexStart()

20.1.9 ODCIIndexGetMetadata()
Returns a series of strings of PL/SQL code that comprise the non-dictionary metadata
associated with the index in ia. The routine can pass whatever information is required
at import time. For example, policy, version, preferences, and so on. This method is
optional unless implementation-specific metadata is required.

Syntax
FUNCTION ODCIIndexGetMetadata(
ia IN ODCIIndexInfo,
version IN VARCHAR2,
new_block OUT PLS_INTEGER,

20-12
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

env ODCIEnv)
RETURN VARCHAR2;

Parameter Description
Specifies the index on which export is currently working
ia

version Version of export making the call in the form 11.2.0.1.00

Non-zero (TRUE): Returned string starts a new PL/SQL block. Export

new_block
terminates the current block (if any) with END; and open a new block
with BEGIN before writing strings to the dump file. The routine is called
again.
0 (FALSE): Returned string continues current block. Export writes only
the returned string to the dump file then calls the routine again.
The environment handle passed to the routine
env

Returns
• A null-terminated string containing a piece of an opaque block of PL/SQL code
• A zero-length string indicates no more data; export stops calling the routine

Usage Notes
• This function should be implemented as a static type method.
• The routine is called repeatedly until the return string length is 0. If an index has no
metadata to be exported using PL/SQL, it should return an empty string upon first
call.
• This routine can be used to build one or more blocks of anonymous PL/SQL code
for execution by import. Each returned block is invoked independently by import.
That is, if a block fails for any reason at import time, subsequent blocks are still
invoked. Therefore any dependent code should be incorporated within a single
block. The size of an individual block of PL/SQL code is limited only by the size of
import's read buffer controlled by its BUFFER parameter.
• The execution of these PL/SQL blocks at import time is considered part of the
associated domain index's creation. Therefore, their execution is dependent upon
the successful import of the index's underlying base table and user's setting of
import's INDEXES=Y/N parameter, as is the creation of the index.
• The routine should not pass back the BEGIN/END strings that open and close the
individual blocks of PL/SQL code; export adds these to mark the individual units of
execution.
• The parameter version is the version number of the currently executing export
client. Since export and import can be used to downgrade a database to the
previous functional point release, it also represents the minimum server version
you can expect to find at import time; it may be higher, but never lower.
• The cartridge developer can use this information to determine what version of
information should be written to the dump file. For example, assume the current
server version is 11.2.0.1.0, but the export version handed in is 11.1.0.1.0. If a
cartridge's metadata changed formats these version, it would know to write the
data to the dump file in 11.1 format, anticipating an import into an 11.2 system.

20-13
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

• The data contained within the strings handed back to export must be completely
platform-independent. That is, they should contain no binary information that may
reflect the endian nature of the export platform, which may be different from the
import platform. Binary information may be passed as hex strings and converted
through RAWTOHEX and HEXTORAW.
• The strings are translated from the export server to export client character set and
are written to the dump file as such. At import time, they are translated from export
client character set to import client character set, then from import client char set
to import server character set when handed over the UPI interface.
• Specifying a target schema in the execution of any of the PL/SQL blocks must be
avoided because it frequently causes an error if you use import's FROMUSER ->
TOUSER schema replication feature. For example, a procedure prototype such as:
PROCEDURE AQ_CREATE ( schema IN VARCHAR2, que_name IN VARCHAR2) ...

should be avoided because it fails if you have remapped schema A to schema B

on import. You can assume at import time that you are connected to the target
schema.
• Export dump files from a particular version must be importable into all future
versions. This means that all PL/SQL routines invoked within the anonymous
PL/SQL blocks written to the dump file must be supported for all time. You may
wish to encode some version information to assist with detecting when conversion
may be required.
• Export operates in a read-only transaction if its parameter CONSISTENT=Y. In this
case, no writes are allowed from the export session. Therefore, this method must
not write any database state.
• You can attempt to import the same dump file multiple times, especially when
using import's IGNORE=Y parameter. Therefore, this method must produce PL/SQL
code that is idempotent, or at least deterministic when executed multiple times.
• Case on database object names must be preserved; that is, objects named 'Foo'
and 'FOO' are distinct objects. Database object names should be enclosed within
double quotes ("") to preserve case.

Error Handling
Any unrecoverable error should raise an exception allowing it to propagate back to
get_domain_index_metadata and thence back to export. This causes export to
terminate the creation of the current index's DDL in the dump file and to move on to
the next index.
At import time, failure of the execution of any metadata PL/SQL block causes the
associated index not to be created under the assumption that the metadata creation is
an integral part of the index creation.

20.1.10 ODCIIndexInsert()
Invoked when a row or a set of rows is inserted into a table that has a domain index
defined on one or more of its columns.

20-14
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

Syntax Description
Inserts a single row
FUNCTION ODCIIndexInsert(
ia ODCIIndexInfo,
rid VARCHAR2,
newval icoltype,
env ODCIEnv)
RETURN NUMBER

Inserts a set of rows

FUNCTION ODCIIndexInsert(
ia ODCIIndexInfo,
ridlist ODCIRidList,
newvallist varray_of_column_type,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Contains information about the index and the indexed column
ia

The row identifier of the new row in the table

rid

The value of the indexed column in the inserted row

newval

ridlist A varray (maximum size 32767) containing the list of rowids for the
rows being inserted into the base table

newvallist A varray (maximum size 32767) containing the list of values being
inserted into the indexed column in the base table; these entries have a
one-to-one correspondence with the entries in ridlist
The environment handle passed to the routine
env

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

Usage Notes
• This function should be implemented as a STATIC type method.
• This method should insert index data corresponding to the row or set of rows
passed in into the appropriate tables or files storing index data. A NULL value in
ridlist indicates the end of the varray.
• If the indextype is defined WITH ARRAY DML, a batch of rows can be inserted into
the table. In this case, ODCIIndexInsert() is invoked using the second of the two
syntax synopses. Otherwise, the single-row syntax is used.
• Note that the index partition object identifier ODCIIndexInfo.IndexPartitionIden
and the base table partition physical identifier
ODCIIndexInfo.IndexCols(1).TablePartitionIden is passed in for local domain
index. The indextype must use the new DML syntax using the partition number

20-15
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

and the provided DATAOBJ_TO_MAT_PARTITION function to insert into the storage

system partitioned table:
INSERT INTO SP PARTITION (
DATAOBJ_TO_MAT_PARTITION(
base_table_name,
:tab_physical_partid))
VALUES(…);

20.1.11 ODCIIndexStart()
Invoked to start the evaluation of an operator on an indexed column.

Syntax
FUNCTION ODCIIndexStart(
sctx IN OUT <impltype>,
ia ODCIIndexInfo,
pi ODCIPredInfo,
qi ODCIQueryInfo,
strt <opbndtype>,
stop <opbndtype>,
<valargs>,
env ODCIEnv)
RETURN NUMBER

Parameter Description
The value of the scan context returned by some previous related query-
sctx(IN)
time call (such as the corresponding ancillary operator, if invoked
before the primary operator); NULL otherwise
The context that is passed to the next query-time call; the next query-
sctx(OUT)
time call is to ODCIIndexFetch()
Contains information about the index and the indexed column
ia

Contains information about the operator predicate

pi

Contains query information (hints plus list of ancillary operators

qi
referenced)

The start value of the bounds on the operator return value. The data
strt
type is identical to that of the operator's return value

The stop value of the bounds on the operator return value. The data
stop
type is identical to that of the operator's return value.
The value arguments of the operator invocation. The number and data
valargs
types of these arguments are identical to those of the value arguments
to the operator.
The environment handle passed to the routine
env

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

20-16
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

Usage Notes
• The function should be implemented as a static method.
• ODCIIndexStart() is invoked to begin the evaluation of an operator on an indexed
column. In particular, the following conditions hold:
– The first argument to the operator is a column which has a domain index
defined on it.
– The indextype of the domain index (specified in ODCIIndexInfo parameter)
supports the current operator.
– All other arguments to the operator are value arguments (literals) which are
passed in through the <valargs> parameters.
• The ODCIIndexStart() method should initialize the index scan as needed (using
the operator-related information in the pi argument) and prepare for the
subsequent invocations of ODCIIndexFetch().
• The strt and stop parameters, with the bndflg value in ODCIPredInfo parameter,
specify the range of values within which the operator return value should lie.
• Bounds for operator return values are specified as follows:
– If the predicate to be evaluated is of the form op LIKE val, the
ODCIIndexPrefixMatch flag is set. In this case, the start key contains the value
<val> and the stop key value is irrelevant.
– If the predicate to be evaluated is of the form op = val, the
ODCIIndexExactMatch flag is set. In this case, the start key contains the value
<val> and the stop key value is irrelevant.
– If the predicate to be evaluated is of the form op > val, startkey contains the
value <val> and stop key value is set to NULL. If the predicate is of the form op
>= <val>, the flag ODCIIndexIncludeStart is also set.
– If the predicate to be evaluated is of the form op < val, stop key contains the
value <val> and the start key value is set to NULL. If the predicate is of the form
op <= val, the flag ODCIIndexIncludeStop is also set.
• A context value can be returned to Oracle (through the SELF argument) which is
then passed back to the next query-time call. The next call is to ODCIIndexFetch()
if the evaluation continues, or to ODCIIndexStart() if the evaluation is restarted.
The context value can be used to store the entire evaluation state or just a handle
to the memory containing the state.
• Note that if the same indextype supports multiple operators with different
signatures, multiple ODCIIndexStart() methods must be implemented, one for
each distinct combination of value argument data types. For example, if an
indextype supports three operators:
1. op1(number, number)
2. op1(varchar2, varchar2)
3. op2(number, number)
two ODCIIndexStart routines must be implemented:
– ODCIIndexStart(...., NUMBER)— handles cases (1) and (3) which has a
NUMBER value argument

20-17
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

– ODCIIndexStart(...., VARCHAR2)— handles case (2) which has a VARCHAR2

value argument
• The query information in qi parameter can be used to optimize the domain index
scan, if possible. The query information includes hints that have been specified for
the query and the list of relevant ancillary operators referenced in the query block.
• The index partition object identifier ODCIIndexInfo.IndexPartitionIden and the
base table partition physical identifier
ODCIIndexInfo.IndexCols(1).TablePartitionIden is passed in for local domain
index. The indextype must use the new SQL syntax using the partition number and
the provided DATAOBJ_TO_MAT_PARTITION function to query the corresponding
partition of the storage system partitioned table:
SELECT FROM SP PARTITION(
DATAOBJ_TO_MAT_PARTITION(
base_table_name,
:tab_physical_partid))
WHERE ...;

See Also:

• ODCIIndexFetch()
• ODCIIndexStart()

20.1.12 ODCIIndexUpdate()
Invoked when a row is updated in a table that has a defined domain index on one or
more of its columns.

Syntax
FUNCTION ODCIIndexUpdate(
ia ODCIIndexInfo,
rid VARCHAR2,
oldval icoltype,
newval icoltype,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Contains information about the index and the indexed column
ia

The row identifier of the updated row

rid

The value of the indexed column before the update. The data type is
oldval
identical to that of the indexed column.

The value of the indexed column after the update. The data type is
newval
identical to that of the indexed column.
The environment handle passed to the routine
env

20-18
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

Usage Notes
• The function should be implemented as a static type method.
• This method should update the tables or files storing the index data for the
updated row.
• In addition to a SQL UPDATE statement, a LOB value can be updated through a
variety of WRITE interfaces (see Oracle Database SecureFiles and Large Objects
Developer's Guide). If a domain index is defined on a LOB column or an object type
containing a LOB attribute, the ODCIIndexUpdate routine is called when a LOB
locator is implicitly or explicitly closed after one or more write operations.
• The index partition object identifier, ODCIIndexInfo.IndexPartitionIden, and the
base table partition physical identifier,
ODCIIndexInfo.IndexCols(1).TablePartitionIden, is passed in for local domain
indexes. The indextype must use the new DML syntax with the partition number,
and the provided DATAOBJ_TO_MAT_PARTITION() function to update data in the
storage system partitioned table:
UPDATE SP PARTITION
(DATAOBJ_TO_MAT_PARTITION(
base_table_name, :tab_physical_partid))
VALUES(…) SET val = :newval WHERE rowid + :rowid;

20.1.13 ODCIIndexUpdPartMetadata()
Invoked during partition maintenance operations. Patches the indextype metadata
tables to correctly reflect the partition maintenance operation.

Syntax
FUNCTION ODCIIndexUpdPartMetadata(
ia ODCIIndexInfo,
palist ODCIPartInfoList,
env ODCIEnv)

Parameter Description
ia The information about the domain index; does not contain partition-
specific information
palist The information about the dropped or added partitions
env The environment handle

Usage Notes
• This method should be implemented as a STATIC type method.
• When an indextype is specified with the SYSTEM MANAGED approach, this method is
invoked on the local domain index of this indextype during partition management
operations.
• SQL DDLs are not allowed in this method.

20-19
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

• The indextype should update its metadata mapping specific to the partitions, if
any.
• The palist argument contains a list of partitions that should be dropped or added.
For example, if the base table operation is ALTER TABLE SPLIT PARTITTION P1
INTO P11 AND P12, then the palist would have information about 3 partitions: P1
(drop), P11(add) and P12(add), along with their index partition names and index
partition object identifiers.
• If the ODCIIndexUpdPartMetadata() call raises or returns an error, then the
partition management operation on the base table is rolled back.

20.1.14 ODCIIndexUtilCleanup()
Cleans up temporary states created by ODCIIndexUtilGetTableNames().

Syntax
FUNCTION ODCIIndexUtilCleanup (
context PLS_INTEGER)

Parameter Description

context The number created by ODCIIndexUtilGetTableNames() that

uniquely identifies state information for a particular index.

Usage Notes
• The procedure should be implemented as a static type method.
• ODCIIndexUtilCleanup() deletes any temporary state associated with the
parameter context.
• Exceptions raised by ODCIIndexUtilCleanup() are ignored by its caller.

See Also:

• ODCIIndexUtilGetTableNames()
• ODCIIndexUtilCleanup()

20.1.15 ODCIIndexUtilGetTableNames()
Determines if the secondary tables of the domain index should be exported/imported.
By default, secondary objects of the domain are not imported or exported. However, if
this interface and ODCIIndexUtilCleanup() are present, the system invokes them.

If this interface is implemented, your application can also invoke it for transportable
tablespace operations.

Syntax
FUNCTION ODCIIndexUtilGetTableNames(
ia sys.odciindexinfo,
read_only PLS_INTEGER,

20-20
 Chapter 20
Extensible Indexing - System-Defined Interface Routines

version varchar2,
context OUT PLS_INTEGER)
RETURN BOOLEAN

Parameter Description
Contains information about the index and the indexed column
ia

Specify 1 if the encompassing transaction is read-only, meaning no

read_only
writes allowed. Otherwise 0.
Version of export making the call.
version

context A unique number that is used by ODCIIndexUtilCleanup() to

facilitate the clean up of any state held open between
ODCIIndexUtilGetTableNames() and ODCIIndexUtilCleanup()

Returns
TRUE if the domain indexes' secondary tables should be exported/imported. Otherwise,
the function returns FALSE.

Usage Notes
• This function should be implemented as a static type method.
• This function should return TRUE or FALSE based on whether the secondary tables
should be exported/imported.
• This function should return TRUE or FALSE based on whether the secondary tables
should be transported. Secondary objects other than tables do not participate in
transportable tablespaces. They must be recreated on the import side when the
ODCIIndexCreate() method is invoked with the ODCI_INDEX_TRANS_TBLSPC bit set
in the ODCIIndexInfo.IndexInfoFlags.

See Also:

• ODCIIndexUtilCleanup()
• ODCIIndexCreate()
• ODCIIndexUtilGetTableNames()

20-21
21
Extensible Optimizer Interface
Consider the functions and procedures that comprise the interface to the extensible
optimizer.

21.1 Extensible Optimizer Interface

The extensible optimizer interfaces support working with partitioned tables and domain
indexes. This is accomplished in two ways:
• Additional attributes have been added to some system-defined object types that
are parameters to the ODCIStats interface methods. For example, the
ODCIColInfo type is enhanced to add information about the column's partition.
• Arguments or semantics of the arguments have changed for some ODCIStats
methods.
Note that you must update your code for ODCIStats2 version of the ODCIStats
interfaces to use your statistics type with an indextype that implements the ODCIIndex2
version of the extensible indexing interfaces.

21.1.1 Using Statistics Functions in an Extensible Optimizer Interface

Consider an example of how the statistics functions might be used.
1. In the schema HR, we define the following operators:
CREATE OPERATOR Contains binding (VARCHAR2(4000), VARCHAR2(30))
RETURN NUMBER USING Contains_fn;

CREATE TYPE stat1 (

...,
STATIC FUNCTION ODCIStatsSelectivity(pred ODCIPredInfo, sel OUT NUMBER,
args ODCIArgDescList, start NUMBER, stop NUMBER, doc VARCHAR2(4000),
key VARCHAR2(30)) return NUMBER,
STACTIC FUNCTION ODCIStatsFunctionCost(func ODCIFuncInfo, cost OUT
ODCICost, args ODCIArgDescList, doc VARCHAR2(4000), key VARCHAR2(30))
return NUMBER,
STATIC FUNCTION ODCIStatsIndexCost(ia ODCIIndexInfo, sel NUMBER,
cost OUT ODCICost, qi ODCIQueryInfo, pred ODCIPredInfo,
args ODCIArgDescList, start NUMBER, stop NUMBER,
key VARCHAR2(30)) return NUMBER,
...
);

CREATE TABLE T (resume VARCHAR2(4000));

CREATE INDEX T_resume on T(resume) INDEXTYPE IS indtype;

ASSOCIATE STATISTICS WITH FUNCTIONS Contains_fn USING stat1;

ASSOCIATE STATISTICS WITH INDEXTYPE indtype USING stat1

WITH SYSTEM MANAGED STORAGE TABLES;

21-1
 Chapter 21
Extensible Optimizer Interface

2. The optimizer encounters the query

SELECT * FROM T WHERE Contains(resume, 'ORACLE') = 1,
3. It computes the selectivity of the predicate by invoking the user-defined selectivity
function for the functional implementation of the Contains operator. In this case,
the selectivity function is stat1.ODCIStatsSelectivity. It is called as follows:
stat1.ODCIStatsSelectivity (
ODCIPredInfo('HR', 'Contains_fn', NULL, 29),
sel,
ODCIArgDescList(
ODCIArgDesc(ODCIConst.ArgLit,
NULL, NULL, NULL, NULL, NULL, NULL),
ODCIArgDesc(ODCIConst.ArgLit,
NULL, NULL, NULL, NULL, NULL, NULL),
ODCIArgDesc(ODCIConst.ArgCol, 'T', 'HR', '"RESUME"', NULL,
NULL, NULL),
ODCIArgDesc(ODCIConst.ArgLit,
NULL, NULL, NULL, NULL, NULL, NULL)),
1,
1,
NULL,
'ORACLE')
4. Suppose the selectivity function returns a selectivity of 3 (percent). When the
domain index is being evaluated, then the optimizer calls the user-defined index
cost function as follows:
stat1.ODCIStatsIndexCost ( ODCIIndexInfo('HR', 'T_RESUME',
ODCIColInfoList(ODCIColInfo('HR', 'T', '"RESUME"', NULL, NULL, NULL, 0, 0, 0,
0)), NULL, 0, 0, 0, 0), 3, cost, NULL, ODCIPredInfo('HR',
'Contains', NULL, 13),
ODCIArgDescList( ODCIArgDesc(ODCIConst.ArgLit,
NULL, NULL, NULL, NULL, NULL, NULL),
ODCIArgDesc(ODCIConst.ArgLit, NULL, NULL,
NULL, NULL, NULL, NULL),
ODCIArgDesc(ODCIConst.ArgLit, NULL, NULL,
NULL, NULL, NULL, NULL)), 1, 1, 'ORACLE')
5. Suppose that the optimizer decides not to use the domain index because it is too
expensive. Then it calls the user-defined cost function for the functional
implementation of the operator as follows:
stat1.ODCIStatsFunctionCost (
ODCIFuncInfo('HR', 'Contains_fn', NULL, 1),
cost,
ODCIArgDescList( ODCIArgDesc(ODCIConst.ArgCol,
'T', 'HR', '"RESUME"', NULL, NULL, NULL),
ODCIArgDesc(ODCIConst.ArgLit,
NULL, NULL, NULL, NULL, NULL, NULL)),
NULL,
'ORACLE')

21.1.2 EXPLAIN PLAN

EXPLAIN PLAN shows the user-defined CPU and I/O costs for domain indexes in the
CPU_COST and IO_COST columns of PLAN_TABLE. For example, suppose we have a table
Emp_tab and a user-defined operator Contains. Further, suppose that there is a

21-2
 Chapter 21
User-Defined ODCIStats Functions

domain index EmpResume_indx on the Resume_col column of Emp_tab, and that the
indextype of EmpResume_indx supports the operator Contains. Then, the query
SELECT * FROM Emp_tab WHERE Contains(Resume_col, 'Oracle') = 1

might have the following plan:

OPERATION OPTIONS OBJECT_NAME CPU_COST IO_COST

SELECT STATEMENT

TABLE ACCESS BY ROWID EMP_TAB

DOMAIN INDEX EMPRESUME_INDX 300 4

21.1.3 INDEX Hint

The index hint applies to domain indexes. In other words, the index hint forces the
optimizer to use the hinted index for a user-defined operator, if possible.

21.1.4 ORDERED_PREDICATES Hint

The hint ORDERED_PREDICATES forces the optimizer to preserve the order of predicate
evaluation (except predicates used for index keys) as specified in the WHERE clause of
a SQL DML statement.

21.2 User-Defined ODCIStats Functions

User-defined ODCIStats functions are used for table columns, functions, package,
type, indextype or domain indexes.

Table 21-1 Summary of User-Defined ODCIStats Functions

Function Description
ODCIGetInterfaces() Discover which version of the ODCIStats interface
the user has implemented.
ODCIStatsCollect() Called by the DBMS_STATS package to collect user-
defined statistics on a table, a partition of a table, an
index, or a partition of an index.
ODCIStatsDelete() Deletes user-defined statistics on a table, a partition of
a table, an index, or a partition of an index.
ODCIStatsFunctionCost() Computes the cost of a function.
ODCIStatsExchangePartition() Exchanges domain index statistics when an ALTER
TABLE EXCHANGE PARTITION ... INCLULDING
INDEXES command is issued.
ODCIStatsIndexCost() Calculates the cost of a domain index scan.
ODCIStatsSelectivity() Specifies the selectivity of a predicate.

21-3
 Chapter 21
User-Defined ODCIStats Functions

Table 21-1 (Cont.) Summary of User-Defined ODCIStats Functions

Function Description
ODCIStatsTableFunction() Provides cardinality statistics for table functions and
input cursor expressions.
ODCIStatsUpdPartStatistics() Updates statistics during partition maintenance
operations. Patches the domain index statistics.

21.2.1 ODCIGetInterfaces()
ODCIGetInterfaces is invoked by the server to discover which version of the
ODCIStats interface the user has implemented in the methods of the user-defined
statistics type.

Syntax
FUNCTION ODCIGetInterfaces(
ifclist OUT ODCIObjectList)
RETURN NUMBER;

Parameter IN/OUT Description

ifclist OUT The version of the ODCIStats interfaces implemented by the

statistics type. This value should be SYS.ODCISTATS2.

Returns
ODCIConst.Success on success, ODCIConst.Error otherwise.

21.2.2 ODCIStatsCollect()
Called by the DBMS_STATS package to collect user-defined statistics.

Syntax Description

FUNCTION ODCIStatsCollect( Called by the DBMS_STATS package to collect user-

col ODCIColInfo, defined statistics on a table or a partition of a table.
options ODCIStatsOptions,
statistics OUT RAW,
env ODCIEnv)
return NUMBER;

Called to collect user-defined statistics on an index or a

FUNCTION ODCIStatsCollect(
partition of an index.
ia ODCIIndexInfo,
options ODCIStatsOptions,
statistics OUT RAW,
env ODCIEnv)
return NUMBER;

21-4
 Chapter 21
User-Defined ODCIStats Functions

Parameter IN/OUT Description

col Column for which statistics are being collected

options Options passed to DBMS_STATS

statistics User-defined statistics collected

env Contains general information about the environment in which the

routine is executing

ia Domain index for which statistics are being collected

Returns
The function returns ODCIConst.Success, ODCIConst.Error, or ODCIConst.Warning.

Usage Notes
• This function should be implemented as a STATIC type method.
• If statistics are being collected for only one partition, the TablePartition field in
the ODCIColInfo type is filled in with the name of the partition. Otherwise (if
statistics must be collected for all the partitions or for the entire table), the
TablePartition field is null.
• If the DBMS_STATS package methods are executed to collect user-defined statistics
on a partitioned table, then n+1 ODCIStatsCollect calls are made, where n is the
number of partitions in the table. The first n calls are made with the
TablePartition attribute in ODCIColInfo filled in with the partition name and the
ODCIStatsOptions.CallProperty set to IntermediateCall. The last call is made
with ODCIEnv.CallPropertyflag set to FinalCall to allow you to collect
aggregate statistics for the entire table.
• If user-defined statistics are being collected for only one partition of the table, two
ODCIStatsCollect calls are made. In the first, you should collect statistics for the
partition. For this call, the TablePartition attribute of the ODCIColInfo structure is
filled in and the ODCIEnv.CallProperty is set to FirstCall.
• In the second call you can update the aggregate statistics of the table based upon
the new statistics collected for the partition. In this call, the
ODCIEnv.CallPropertyflag is set to FinalCall to indicate that it is the second
call. The ODCIColInfo.TablePartition is filled in with the partition name in both
the calls.
• The ODCIStatsCollect() method is invoked only one time for a non-partitioned
domain index, a partitioned domain index and a partition in a domain index. If the
statistics are being collected only for one partition in a domain index, the
IndexPartitionNum field in the ODCIIndexInfo type is filled in with the partition
number. Otherwise, the IndexPartitionNum field is null.
• Because the statistics OUT RAW argument of statistics is not used in the new
interface, the cartridge developer should store the user-defined statistics result in
some user-defined tables.

21-5
 Chapter 21
User-Defined ODCIStats Functions

• If a non-partitioned domain index is being ANALYZEd, the user should collect

statistics for the domain index.
• If a partitioned domain index is being ANALYZEd,
– ODCIEnv.CallProperty = StatsGlobalAndPartition means that the user
should collect statistics for all partitions in the domain index and then
aggregate statistics of the domain index based upon the statistics collected for
all the partitions
– ODCIEnv.CallProperty = StatsGlobal means that the user should aggregate
domain index statistics from the statistics of all the domain index partitions.
– ODCIEnv.CallProperty = StatsPartition means that the user should collect
statistics for all index partitions in the domain index.
• If only one partition of the domain index is being ANALYZEd,
– ODCIEnv.CallProperty = StatsGlobalAndPartition means that the user
should collect statistics for the single index partition and then aggregate
statistics of the domain index based upon the statistics of all the partitions.
– ODCIEnv.CallProperty = StatsGlobal means that the user should aggregate
domain index statistics from the statistics of all the index partitions.
– ODCIEnv.CallProperty = StatsPartition means that the user should collect
statistics for the single index partition.
• Note that when ODCIEnv.CallProperty = StatsGlobalAndPartition or
StatsGlobal, the user should aggregate statistics for the domain index, depending
on the availability of the statistics collected for the other index partitions. If the
statistics for all the index partitions are available, aggregate these statistics. If any
one statistics for an index partition is absent, do nothing.
• To properly collect the statistics for a local system-managed domain index, the
IndexInfoFlags and IndexPartition fields in the ODCIIndexInfo argument and
the CallProperty in the ODCIEnv argument must be selected.
• The values in ODCIIndexInfo.IndexPartition can be used to determine whether
statistics need to be collected for all index partitions or subpartitions. Select the
following flags to determine the partition level for statistics collection.
– If ODCIConst.SubPartn and ODCIConst.CompPartn are set, then statistics need
to be collected for a single subpartition.

Note:
The ODCIConst.SubPartn will never be set if
ODCIIndexInfo.IndexPartition is not set.

– If ODCIConst.CompPartn is set, but ODCIConst.SubPartn is not set; then

statistics need to be collected for all subpartitions of the composite partition
and the composite partition itself. If ODCIIndexInfo.IndexPartition is not set,
then statistics are collected in the same manner for all composite partitions of
the index.
– If neither ODCIConst.CompPartn nor ODCIConst.SubPartn are set, then the
index is not composite partitioned and statistics need to be collected for the

21-6
 Chapter 21
User-Defined ODCIStats Functions

specified partition. If ODCIIndexInfo.IndexPartition is not set, then statistics

are collected for all partitions.

21.2.3 ODCIStatsDelete()
ODCIStatsDelete is called to delete user-defined statistics.

Syntax Description
Deletes user-defined statistics on a table or a partition of
FUNCTION ODCIStatsDelete(
a table.
col ODCIColInfo,
statistics OUT RAW,
env ODCIEnv)
return NUMBER;

Deletes user-defined statistics on an index or a partition

FUNCTION ODCIStatsDelete(
of an index.
ia ODCIIndexInfo,
statistics OUT RAW,
env ODCIEnv)
return NUMBER;

Parameter IN/OUT Description

col Column for which statistics are being deleted

statistics Contains table-level aggregate statistics for a partitioned table or

OUT
index

env Contains general information about the environment in which the

routine is executing

ia Domain index for which statistics are deleted

Returns
ODCIConst.Success, ODCIConst.Error, or ODCIConst.Warning.

Usage Notes
• This function should be implemented as a STATIC method.
• When the function is called for a non-partitioned table, the statistics argument in
the ODCIStatsDelete interface is ignored.
• If the statistics are being deleted for a partitioned table, the ODCIStatsDelete is
called n+1 times. The first n calls are with the partition name filled in the
ODCIColInfo structure and the ODCIEnv.CallProperty set to IntermediateCall.
The last call is made with the ODCIEnv.CallProperty set to FinalCall.
• In the first call, delete the statistics for the specific partitions; and in the last call
drop or clean up any structures created for holding statistics for the deleted table.
The ODCIColInfo.TablePartition is set to null in the last call. In the first call, the
TablePartition field is filled in.

21-7
 Chapter 21
User-Defined ODCIStats Functions

• If statistics are being deleted for only one partition and the
_minimal_stats_aggregation parameter is set to FALSE, two ODCIStatsDelete
calls are made. In each call, ODCIColInfo.TablePartition is filled in with the
partition name. On the first call, delete any user-defined statistics collected for that
partition. On the second call, update the aggregate statistics for the table.
• If statistics are being deleted for one partition and _minimal_stats_aggregation is
set to TRUE, ODCIStatsDelete is only called one to delete any user-defined
statistics collected for that partition.
• The initial value of _minimal_stats_aggregation is TRUE.
• The ODCIStatsDelete() method is invoked only one time for non-partitioned
domain index, partitioned domain index, or an index partition.
• If the statistics is being deleted for a non-partitioned domain index, the user should
delete user-defined statistics for the domain index.
• If the statistics is being deleted for a partitioned domain index, the user should
delete the aggregated statistics of the domain index and optionally delete user-
defined statistics for all domain index partitions, depending on Options in
ODCIEnv.CallProperty:
– ODCIEnv.CallProperty = StatsGlobalAndPartition means that the user
should delete statistics for all the domain index partitions and aggregated
statistics of the domain index.
– ODCIEnv.CallProperty = StatsGlobal means that the user should delete the
aggregated statistics of the domain index.
– ODCIEnv.CallProperty = StatsPartition is not valid option.
• If the statistics is being deleted for only one partition of the index, the user should
delete user-defined statistics for the index partition.

21.2.4 ODCIStatsFunctionCost()
Computes the cost of a function.

Syntax
FUNCTION ODCIStatsFunctionCost(
func ODCIFuncInfo,
cost OUT ODCICost,
args ODCIArgDescList,
list,
env ODCIEnv)
return NUMBER;

Parameter IN/OUT Description

Function or type method for which the cost is being computed
func

Computed cost (must be positive whole numbers)

cost OUT

Descriptor of actual arguments with which the function or type

args
method was called. If the function has n arguments, the args
array contains n elements, each describing the actual arguments
of the function or type method

21-8
 Chapter 21
User-Defined ODCIStats Functions

Parameter IN/OUT Description

List of actual parameters to the function or type method; the
list
number, position, and type of each argument must be identical in
the function or type method.
Contains general information about the environment in which the
env
routine is executing

Returns
ODCIConst.Success, ODCIConst.Error, or ODCIConst.Warning.

Usage Notes
This function should be implemented as a static type method.

21.2.5 ODCIStatsExchangePartition()
Exchanges domain index statistics when an ALTER TABLE EXCHANGE PARTITION ...
INCLULDING INDEXES command is issued.

Syntax
FUNCTION ODCIStatsExchangePartition(
ia ODCIIndexInfo,
ia1 ODCIIndexInfo,
env ODCIEnv)
return NUMBER;

Parameter IN/OUT Description

Information about the index partition that must be exchanged
ia

Information about the index of the non-partitioned table with which

ia1
the partition is exchanged

Contains general information about the environment in which the

env
routine is executing

Returns
ODCIConst.Success, ODCIConst.Error, or ODCIConst.Warning

Usage Notes
• This method should be implemented as a STATIC type.
• This method should be capable of converting the statistics associated with a
domain index partition into statistics associated with a non-partitioned domain
index, and the reverse. If the statistics are missing for one of the indexes or index
partitions, the user should be able to delete these statistics.

21-9
 Chapter 21
User-Defined ODCIStats Functions

21.2.6 ODCIStatsIndexCost()
Calculates the cost of a domain index scan, either a scan of the entire index or a scan
of one or more index partitions if a local domain index has been built.

Syntax
FUNCTION ODCIStatsIndexCost(
ia ODCIIndexInfo,
sel NUMBER,
cost OUT ODCICost,
qi ODCIQueryInfo,
pred ODCIPredInfo,
args ODCIArgDescList,
start operator_return_type,
stop operator_return_type,
list,
env ODCIEnv)
return NUMBER;

Parameter IN/OUT Description

domain index for which statistics are being collected
ia

the user-computed selectivity of the predicate

sel

computed cost (must be positive whole numbers)

cost

Information about the query

qi

Information about the predicate

pred

args Descriptor of start, stop, and actual value arguments with which
the operator was called. If the operator has n arguments, the args
array contains n+1 elements, the first element describing the start
value, the second element describing the stop value, and the
remaining n-1 elements describing the actual value arguments of
the operator (that is, the arguments after the first)

start Lower bound of the operator (for example, 2 for a predicate

fn(...) > 2)
Upper bound of the operator (for example, 5 for a predicate
stop
fn(...) < 5)
List of actual parameters to the operator (excluding the first); the
list
number, position, and type of each argument must be identical to
the one in the operator.
Contains general information about the environment in which the
env
routine is executing

Returns
ODCIConst.Success, ODCIConst.Error, or ODCIConst.Warning

21-10
 Chapter 21
User-Defined ODCIStats Functions

Usage Notes
• For each table in the query, the optimizer uses partition pruning to determine the
range of partitions that may be accessed. These partitions are called interesting
partitions. The set of interesting partitions for a table is also the set of interesting
partitions for all domain indexes on that table. The cost of a domain index can
depend on the set of interesting partitions, so the optimizer passes a list of
interesting index partitions to ODCIStatsIndexCost in the args argument (the type
of this argument, ODCIArgDescList, is a list of ODCIArgDesc argument descriptor
types) for those arguments that are columns. For non-partitioned domain indexes
or for cases where no partition pruning is possible, no partition list is passed to
ODCIStatsIndexCost, and you should assume that the entire index is accessed.
• The domain index key can contain multiple column arguments (for example, the
indexed column and column arguments from other tables appearing earlier in a
join order). For each column appearing in the index key, the args argument
contains the list of interesting partitions for the table. For example, for an index key
op(T1.c1, T2.c2) = 1

the optimizer passes a list of interesting partitions for tables T1 and T2 if they are
partitioned and there is partition pruning for them.
• This function should be implemented as a static type method.
• Only a single call is made to the ODCIStatsIndexCost() function for queries on
partitioned or non-partitioned tables. For queries on partitioned tables, additional
information is passed in the ODCIStatsIndexCost() function. Note that some
partitions in the list passed to ODCIStatsIndexCost() may not actually be
accessed by the query. The list of interesting partitions chiefly serves to exclude
partitions that are definitely not accessed.
• When the ODCIStatsIndexCost() function is invoked, users can fill in a string in
the IndexCostInfo field of the cost attribute to supply any additional information
that might be helpful. The string (255 characters maximum) is displayed in the
OPTIONS column in the EXPLAIN PLAN output when an execution plan chooses a
domain index scan.
• Users implementing this function must return 'SYS.ODCISTATS2' in the
ODCIGetInterfaces() call.

See Also:

• ODCIGetInterfaces()
• ODCIStatsIndexCost()

21.2.7 ODCIStatsSelectivity()
Specifies the selectivity of a predicate. The selectivity of a predicate involving columns
from a single table is the fraction of rows of that table that satisfy the predicate. For
predicates involving columns from multiple tables (for example, join predicates), the
selectivity should be computed as a fraction of rows in the Cartesian product of those
tables.

21-11
 Chapter 21
User-Defined ODCIStats Functions

Syntax
FUNCTION ODCIStatsSelectivity(
pred ODCIPredInfo,
sel OUT NUMBER,
args ODCIArgDescList,
start function_return_type,
stop function_return_type,
list,
env ODCIEnv)
return NUMBER;

Parameter IN/OUT Description

pred Predicate for which the selectivity is being computed

sel The computed selectivity, expressed as a number between (and

including) 0 and 100, represents a percentage.
The optimizer ignores numbers less than 0 or greater than 100 as
invalid values. If the computed selectivity is less than 0.5%, a
value of 0 may be returned in the output parameter sel. A
selectivity of 0 does not mean that the predicate will be removed
args Descriptor of start, stop, and actual arguments with which the
function, type method, or operator was called. If the function has n
arguments, the args array contains n+2 elements, the first
element describing the start value, the second element describing
the stop value, and the remaining n elements describing the actual
arguments of the function, method, or operator
start Lower bound of the function (for example, 2 for a predicate
fn(...) > 2)
stop Upper bound of the function (for example, 5 for a predicate
fn(...) < 5)
list List of actual parameters to the function or type method; the
number, position, and type of each argument must be identical to
the one in the function, type method, or operator.
env Contains general information about the environment in which the
routine is executing

Returns
ODCIConst.Success, ODCIConst.Error, or ODCIConst.Warning

Usage Notes
• As in ODCIStatsIndexCost, the args argument contains a list of interesting
partitions for the tables whose columns are referenced in the predicate for which
the selectivity has to be computed. These interesting partitions are partitions that
cannot be eliminated by partition pruning as possible candidates to be accessed.
The set of interesting partitions is passed to the function only if partition pruning
has occurred (in other words, the interesting partitions are a strict subset of all the
partitions).
• For example, when ODCIStatsSelectivity is called to compute the selectivity of
the predicate:

21-12
 Chapter 21
User-Defined ODCIStats Functions

f(T1.c1, T2.c2) > 4

the optimizer passes the list of interesting partitions for the table T1 (in the
argument descriptor for column T1.c1) if partition pruning is possible; similarly for
the table T2.
If a predicate contains columns from several tables, this information is indicated by
the flag bit PredMultiTable, set in the Flags attribute of the pred argument.
• This function should be implemented as a static type method.
• Users implementing this interface must return 'SYS.ODCISTATS2' in the
ODCIGetInterfaces call.
• The selectivity of a predicate involving columns from a single table is the fraction
of rows of that table that satisfy the predicate. For predicates involving columns
from multiple tables (for example, join predicates), the selectivity should be
computed as a fraction of rows in the Cartesian product of those tables. For tables
with partition pruning, the selectivity should be expressed relative to the
cardinalities of the interesting partitions of the tables involved.
The selectivity of predicates involving columns on partitioned tables is computed
relative to the rows in the interesting partitions. Thus, the selectivity of the
predicate
g(T1.c1) < 5

is the percentage of rows in the set of interesting partitions (or all partitions if no
partition pruning is possible) that satisfies this predicate. For predicates with
columns from multiple tables, the selectivity must be relative to the number of rows
in the cartesian product of the tables.
• For example, consider the predicate:
f(T1.c1, T2.c2) > 4

Suppose that the number of rows in the interesting partitions is 1000 for T1 and
5000 for T2. The selectivity of this predicate must be expressed as the percentage
of the 5,000,000 rows in the Cartesian product of T1 and T2 that satisfy the
predicate.
• If a predicate contains columns from several tables, this information is indicated by
the flag bit PredMultiTable set in the Flags attribute of the pred argument.
• A selectivity expressed relative to the base cardinalities of the tables involved may
be only an approximation of the true selectivity if cardinalities (and other statistics)
of the tables have been reduced based on single-table predicates or other joins
earlier in the join order. However, this approximation to the true selectivity should
be acceptable to most applications.
• Only one call is made to the ODCIStatsSelectivity function for queries on
partitioned or non-partitioned tables. In the case of queries on partitioned tables,
additional information is passed while calling the ODCIStatsSelectivity function.

21.2.8 ODCIStatsTableFunction()
This function provides cardinality statistics for table functions and input cursor
expressions.

21-13
 Chapter 21
User-Defined ODCIStats Functions

Syntax
STATIC FUNCTION ODCIStatsTableFunction(
func IN SYS.ODCIFuncInfo,
outStats OUT SYS.ODCITabFuncStats,
argDesc IN SYS.ODCIArgDescList,
list)
RETURN NUMBER;

Parameter IN/OUT Description

func Table function name

outStats Number of rows expected to be returned

argDesc Description of the arguments to the table function

list The arguments' compile-time values. Expressions that only have

values at run time are represented by nulls.

Returns
ODCIConst.Success, ODCIConst.Error, or ODCIConst.Warning.

21.2.9 ODCIStatsUpdPartStatistics()
Updates statistics during partition maintenance operations. This lets the statistics type
patch up the domain index statistics to correctly reflect the partition maintenance
operation.

Syntax
STATIC FUNCTION ODCIStatsCollect(
ia ODCIIndexInfo,
palist ODCIPartInfoList,
env ODCIEnv)
RETURN NUMBER

Parameter IN/OUT Description

ia Contains information about the domain index. It does not contain
any partition specific information

palist Contains information about the partitions that are to be dropped or

added
Environment handle passed to the routine
env

Returns
ODCIConst.Success, ODCIConst.Error, or ODCIConst.Warning.

• When the statistics type is specified by the SYSTEM MANAGED approach, then the
ODCIStatsUpdPartStatistics() method is invoked only one time during PMO. Only
DML and query are allowed in the method implementation.

21-14
 Chapter 21
User-Defined ODCIStats Functions

• If the user maintains the domain index statistics in a global non-partitioned table,
then the user should delete the entry for the user-defined statistics for the dropped
partition (and optionally add a NULL entry for added partition). They can then check
if ODCIEnv.CallProperty is StatsGlobalAndPartition or StatsPartition. If
ODCIEnv.CallProperty is StatsGlobalAndPartition then they should aggregate
all the available index partition statistics. If ODCIEnv.CallProperty is
StatsPartition they can simply delete the aggregate statistics, or leave the
aggregate statistics as they are. ODCIEnv.CallProperty cannot be StatsGlobal
for this call.
• The user should use the information passed in by the ODCIEnv.CallProperty to
determine the type of statistics to delete and adjust.
• If the method returns ODCIConst.Error, the error is ignored and the partition
management operation continues.

21-15
22
User-Defined Aggregate Functions
Interface
These routines must be implemented to define a user-defined aggregate function. The
routines are implemented as methods in an object type. Then the CREATE FUNCTION
statement is used to actually create the aggregate function.

See Also:
Using User-Defined Aggregate Functions

22.1 User-Defined Aggregate Functions

The methods described here are implemented as methods in an object type. The
CREATE FUNCTION statement is used to actually create the aggregate function.
Table 22-1 summarizes these functions.

Table 22-1 Summary of User-Defined Aggregate Functions

Function Description
ODCIAggregateDelete() Removes an input value from the current group.
ODCIAggregateInitialize() Initializes the aggregation context and instance of the
implementation object type, and returns it as an OUT
parameter.
ODCIAggregateIterate() Iterates through input rows by processing the input
values, updating and then returning the aggregation
context.
Note : ODCIAggregateIterate is not invoked for null
values.
ODCIAggregateMerge() Merges two aggregation contexts into a single object
instance during either serial or parallel evaluation of
the user-defined aggregate.
ODCIAggregateTerminate() Calculates the result of the aggregate computation
and performs all necessary cleanup, such as freeing
memory.
ODCIAggregateWrapContext() Integrates all external pieces of the current
aggregation context to make the context self-
contained.

22-1
 Chapter 22
User-Defined Aggregate Functions

22.1.1 ODCIAggregateDelete()
Removes an input value from the current group. The routine is invoked by Oracle by
passing in the aggregation context and the value of the input to be removed during It
processes the input value, updates the aggregation context, and returns the context.
This is an optional routine and is implemented as a member method.

Syntax
MEMBER FUNCTION ODCIAggregateDelete(
self IN OUT <impltype>,
val <inputdatatype>)
RETURN NUMBER

Parameter IN/OUT Description

As input, the value of the current aggregation context; as output, the
self IN OUT
updated value.

The input value that is being removed from the current group.
val IN

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

22.1.2 ODCIAggregateInitialize()
Initializes the aggregation context and instance of the implementation object type, and
returns it as an OUT parameter. Implement this routine as a static method.

Syntax
STATIC FUNCTION ODCIAggregateInitialize(
actx IN OUT <impltype>)
RETURN NUMBER

Parameter In/Out Description

The aggregation context that is initialized by the routine. This value is
actx IN OUT
NULL for regular aggregation cases. In aggregation over windows,
actx is the context of the previous window. This object instance is
passed in as a parameter to the next aggregation routine.

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

22.1.3 ODCIAggregateIterate()
Iterates through input rows by processing the input values, updating and then returning
the aggregation context. This is a mandatory routine and is implemented as a member
method.

22-2
 Chapter 22
User-Defined Aggregate Functions

Syntax
MEMBER FUNCTION ODCIAggregateIterate(
self IN OUT <impltype>,
val <inputdatatype>)
RETURN NUMBER

Parameter IN/OUT Description

As input, the value of the current aggregation context; as output, the
self IN OUT
updated value.

The input value that is being aggregated.

val IN

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

22.1.4 ODCIAggregateMerge()
Merges two aggregation contexts into a single object instance during either serial or
parallel evaluation of the user-defined aggregate. This is a mandatory routine and is
implemented as a member method.

Syntax
MEMBER FUNCTION ODCIAggregateMerge(
self IN OUT <impltype>,
ctx2 IN <impltype>)
RETURN NUMBER

Parameter IN/OUT Description

On input, the value of the first aggregation context; on output, the
self IN OUT
resulting value of the two merged aggregation contexts.
The value of the second aggregation context.
ctx2 IN

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

If the user-defined aggregate is a window function, and it is not possible to make an

implementation of ODCIAggregateMerge(), ODCIConst.Errorshould be returned. This
error is translated as an Oracle user error.

22.1.5 ODCIAggregateTerminate()
Calculates the result of the aggregate computation and performs all necessary
cleanup, such as freeing memory. Invoked by Oracle as the last step of aggregate
computation. This is a mandatory routine and is implemented as a member method.

22-3
 Chapter 22
User-Defined Aggregate Functions

Syntax
MEMBER FUNCTION ODCIAggregateTerminate(
self IN <impltype>,
ReturnValue OUT <return_type>,
flags IN number)
RETURN NUMBER

Parameter IN/OUT Description

The value of the aggregation context.
self IN

The resultant aggregation value.

ctx2 OUT

A bit vector that indicates various options. A set bit of

flags IN
ODCI_AGGREGATE_REUSE_CTX indicates that the context is reused
and any external context should not be freed.

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

See Also:
"Reuse of Aggregation Context for Analytic Functions" for details on setting
the ODCI_AGGREGATE_REUSE_CTX flag bit.

22.1.6 ODCIAggregateWrapContext()
Integrates all external pieces of the current aggregation context to make the context
self-contained. Invoked by Oracle if the user-defined aggregate has been declared to
have external context and is transmitting partial aggregates from slave processes. This
is an optional routine and is implemented as a member method.

Syntax
MEMBER FUNCTION ODCIAggregateWrapContext(
self IN OUT <impltype>)
RETURN NUMBER

Parameter IN/OUT Description

On input, the value of the current aggregation context; on output, the
self IN
self-contained combined aggregation context.

Returns
ODCIConst.Success on success, or ODCIConst.Error on error.

22-4
 Chapter 22
User-Defined Aggregate Functions

See Also:
"Handling Large Aggregation Contexts" for more information on using this
function

22-5
23
Pipelined and Parallel Table Functions
These routines must be implemented to define pipelined and parallel table functions in
C.

See Also:
Using Pipelined and Parallel Table Functions for an overall explanation of
pipelined and parallel table functions

23.1 Routines for Pipelined and Parallel Table Functions in

C
The following C methods, summarized in support parallel and pipelined table
functions.

Table 23-1 Summary of Pipelined and Parallel Table Functions for C

Function Description
ODCITableClose() Performs cleanup operations after scanning a table function.
ODCITableDescribe() Returns describe information for a table function whose
return type is ANYDATASET.
ODCITableFetch() returns the next batch of rows from a table function.
ODCITablePrepare() Prepares the scan context and other query information at
compile time.
ODCITableStart() initializes the scan of a table function.

23.1.1 ODCITableClose()
ODCITableClose performs cleanup operations after scanning a table function.

Syntax
MEMBER FUNCTION ODCITableClose(
self IN <imptype>)
RETURN NUMBER;

Parameter In/Out Description

The scan context set up by previous scan routine invocation
self IN

23-1
 Chapter 23
Routines for Pipelined and Parallel Table Functions in C

Returns
ODCIConst.Success on success, ODCIConst.Error otherwise.

Usage Notes
• Oracle invokes ODCITableClose after the last fetch call. The scan context is
passed in as a parameter. ODCITableClose then performs any necessary cleanup
operations, such as freeing memory.
• If ODCITablePrepare is implemented, this routine is only called one time, at the
end of query execution, rather than each time the table function exits.

23.1.2 ODCITableDescribe()
ODCITableDescribe returns describe information for a table function whose return type
is ANYDATASET.

Syntax
STATIC FUNCTION ODCITableDescribe(
rtype OUT ANYTYPE,
<args>)
RETURN NUMBER;

Parameter In/Out Description

rtype OUT The AnyType value that describes the returned rows from the table
function
The set of zero or more user specified arguments for the table
args IN
function.

Returns
ODCIConst.Success on success, ODCIConst.Error otherwise.

Usage Notes
• If the optional routine ODCITableDescribe is implemented, Oracle invokes it at
query compilation time to retrieve the specific type information.
• This interface is applicable only for table functions whose return type is
ANYDATASET. The format of elements within the returned collection is conveyed to
Oracle by returning an instance of ANYTYPE. The ANYTYPE instance specifies the
actual structure of the returned rows of the specific query.
• ANYTYPE provides a data type to model the metadata of a row: the names and data
types of all the columns (fields) comprising the row. It also provides a set of
PL/SQL and C interfaces for users to construct and access the metadata
information. ANYDATASET, like ANYTYPE, contains a description of a given type, but
ANYDATASET also contains a set of data instances of that type
• The following example shows a query on a table function that uses the ANYDATASET
type:
SELECT * FROM
TABLE(CAST(AnyBooks('http://.../books.xml') AS ANYDATASET));

23-2
 Chapter 23
Routines for Pipelined and Parallel Table Functions in C

At query compilation time, Oracle invokes the ODCITableDescribe routine. The

routine typically uses the user arguments to figure out the nature of the return
rows. In this example, ODCITableDescribe consults the DTD of the XML
documents at the specified location to determine the appropriate ANYTYPE value to
return. Each ANYTYPE instance is constructed by invoking the constructor APIs with
this field name and data type information.
• Any arguments of the table function that are not constants are passed to
ODCITableDescribe as NULLs because their values are not known at compile time.

See Also:
Transient and Generic Types for a discussion of ANYTYPE, ANYDATA, and
ANYDATASET

23.1.3 ODCITableFetch()
ODCITableFetch returns the next batch of rows from a table function.

Syntax
MEMBER FUNCTION ODCITableFetch(
self IN OUT <imptype>,
nrows IN NUMBER,
rws OUT <coll-type>)
RETURN NUMBER;

Parameter In/Out Description

The in-bound is the scan context set up by previous scan routine
self IN OUT
invocation; the outbound is the scan context to be passed to later
scan routine invocations.
The number of rows the system expects in the current fetch cycle. The
nrows IN
method can ignore this value and return a different number of rows. If
fewer rows are returned, the method is called again; if more rows are
returned, they are processed in the next cycle.
The next batch of rows from the table function. This is returned as an
rws OUT
instance of the same collection type as the return type of the table
function.

Returns
ODCIConst.Success on success, ODCIConst.Error otherwise.

Usage Notes
• ODCITableFetch is invoked one or more times by Oracle to retrieve all the rows in
the collection returned by the table function. The scan context is passed in as a
parameter. Typically ODCITableFetch uses the input scan context and computes
the next set of rows to be returned to Oracle. In addition, it may update the scan
context accordingly.
• Returning more rows in each invocation of fetch() reduces the number of fetch
calls that must be made and thus improves performance.

23-3
 Chapter 23
Routines for Pipelined and Parallel Table Functions in C

• Oracle calls ODCITableFetch repeatedly until all rows in the table function's
collection have been returned. When all rows have been returned,
ODCITableFetch should return a null collection.

23.1.4 ODCITablePrepare()
Prepares the scan context and other query information at compile time.

Syntax
STATIC FUNCTION ODCITablePrepare(
sctx OUT <imptype>,
tf_info SYS.ODCITabFuncInfo,
args);

Parameter In/Out Description

The scan context returned by this routine. This value is passed in as a
sctx OUT
parameter to the later scan routines. The scan context is an instance
of the object type containing the implementation of the ODCITable
routines.
Contains the projection information and the return type's table
tf_info
descriptor object (TDO):
• Attrs (SYS.ODCINumberList): lists the positions of the
referenced attributes of the table function's output collection type
• RefType (SYS.AnyType): for AnyDataSet table functions, this is
the actual return type expected to be returned in the AnyDataSet
collection.
The arguments that are passed to the table function. This method is
args IN
invoked at compile time; thus, only literal arguments have values.
Column and expression arguments are passed as null values.

Usage Notes
• This method prepares the scan context based on the information known at compile
time. This scan context is passed to ODCITableStart when it is called at the
beginning of query execution.
• If this optional method is implemented, ODCITableClose is only called one time, at
the end of query execution. Each time the table function is restarted,
ODCITableStart is called and passed the scan context. This allows the table
function to maintain context between restarts, and to perform cleanup operations
only one time at the end of query execution.

23.1.5 ODCITableStart()
ODCITableStart initializes the scan of a table function.

Syntax
STATIC FUNCTION ODCITableStart(
sctx IN OUT <imptype>,
<args>)
RETURN NUMBER;

23-4
 Chapter 23
Routines for Pipelined and Parallel Table Functions in C

Parameter In/Out Description

The scan context returned by this routine. This value is passed in as a
self IN OUT
parameter to the later scan routines. The scan context is an instance
of the object type containing the implementation of the ODCITable
routines. If ODCITablePrepare is implemented, the scan context it
creates is passed in to ODCITableStart.
Set of zero or more arguments specified by the user for the table
args IN
function
The next batch of rows from the table function. This is returned as an
rws OUT
instance of the same collection type as the return type of the table
function.

Returns
ODCIConst.Success on success, ODCIConst.Error otherwise.

Usage Notes
• If ODCITablePrepare is not implemented, this is the first routine that is invoked to
begin retrieving rows from a table function. This routine typically performs the
setup needed for the scan. The scan context is created (as an object instance
sctx) and returned to Oracle. The arguments to the table function, specified by the
user in the SELECT statement, are passed in as parameters to this routine. If
ODCITablePrepare is implemented, it creates the scan context at compile time,
and that scan context is passed in to this routine.
• Any REF CURSOR arguments of the table function must be declared as
SYS_REFCURSOR type in the declaration of the ODCITableStart method.

23-5
A
User-Managed Local Domain Indexes
The user-managed approach for partitioning domain indexes has been the only
method available until Oracle Database 11g Release 1, when system-managed
partitioning was introduced. The user-managed approach has three significant
limitations:
• Because the extensible indexing framework does not store information about the
domain index related objects in the kernel, you must maintain tables and partitions
by invoking user-supplied routines.
• Because the kernel does not support equipartitioned tables, each partition has to
have a set of tables and dependent schema objects, which must be managed
programmatically in the user-managed indexing code.
As the number of partitions increases, the proliferation of domain index storage
objects can become an obstacle to efficient operation. To use a table that contains
images and has 1,000 partitions, an indexing schema that creates 64 bitmap
indexes on its storage table (after it is extended to support local domain indexes)
would need create and manage 1,000 domain index storage tables and 64,000
bitmap indexes.
• During DML and query processing with local domain indexes, you would need a
separate set of cursors for each partition; this is required because each partition
has its own set of tables. As a consequence, applications that use a large number
of partitions and require access to several partitions simultaneously must compile
new SQL cursors at run-time, which impacts performance.
Oracle recommends that you use the system-managed approach, as described in
Building Domain Indexes .
Oracle plans to deprecate the user-managed approach in a future release. Information
provided in this appendix documents the specific differences between the user-
managed and system managed processes and APIs.

A.1 Comparing User-Managed and System-Managed

Domain Indexes
An alternative approach would be to use system-managed domain indexes. It
addresses these limitations and has the following benefits:
• Because the kernel performs many more maintenance tasks on behalf of the user,
there is no need for programmatic support for table and partition maintenance
operations. These operations can be implemented by taking actions in the server
and by using a very minimal set of interface routines. The cartridge code can then
be relatively unaware of partition issues.
• The number of objects that must be managed to support local partitioned domain
indexes is identical to the number for non-partitioned domain indexes. For local
partitioned indexes, the domain index storage tables are equipartitioned with

A-1
 Appendix A
Truncating Domain Indexes

respect to the base tables; therefore, the number of domain index storage tables
does not increase with an increase in the number of partitions.
• A single set of query and DML statements can now access and manipulate the
system-partitioned storage tables, facilitating cursor sharing and enhancing
performance.

A.2 Truncating Domain Indexes

There is no explicit statement for truncating a domain index. However, when the
corresponding table is truncated, your indextype's truncate method is invoked. For
example:
TRUNCATE TABLE MyEmployees;

truncates ResumeTextIndex by calling your ODCIIndexTruncate() method.

A.3 Creating Indextypes

Use the following syntax to create indextypes for the user-managed domain indexes.
CREATE INDEXTYPE TextIndexType
FOR Contains (VARCHAR2, VARCHAR2)
USING TextIndexMethods;

A.4 Using Domain Indexes for the Indextype

In order for the indextype to be able to use local domain indexes, the methods have to
be declared when the indextype is created:
CREATE INDEXTYPE TextIndexType
FOR Contains (VARCHAR2, VARCHAR2)
USING TextIndexMethods
WITH LOCAL RANGE PARTITION;

A.5 Partitioning Domain Indexes

The user-managed approach uses the methods ODCIIndexMergePartition() and
ODCIIndexSplitPartition() to support local domain indexes.

A.6 APIs for User-Managed Domain Indexes

The following methods are used only in the user-managed implementation of domain
indexes.

A.6.1 ODCIIndexTruncate()
This is an index definition method. When a user issues a TRUNCATE statement against
a table that contains a column or object type attribute indexed by your indextype,
Oracle calls your ODCIIndexTruncate() method. This method should leave the domain
index empty.

A-2
 Appendix A
APIs for User-Managed Domain Indexes

Syntax
FUNCTION ODCIIndexTruncate(
ia ODCIIndexInfo,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Contains information about the indexed column
ia

The environment handle passed to the routine

env

Returns
ODCIConst.Success on success, or ODCIConst.Error on error, or ODCIConst.Warning.

While truncating a local domain index, the first N+1 calls can return
ODCIConst.ErrContinue too.

Usage Notes
• This function should be implemented as a static type method.
• After this function executes, the domain index should be empty (corresponding to
the empty base table).
• While the ODCIIndexTruncate() routine is being executed, the domain index is
marked LOADING. If the ODCIIndexTruncate() routine returns with an
ODCIConst.Error (or exception), the domain index is marked FAILED. The only
operation permitted on FAILED domain indexes is DROP INDEX, TRUNCATE TABLE or
ALTER INDEX REBUILD. If ODCIIndexTruncate() returns with ODCIConst.Warning,
the operation succeeds but a warning message is returned to the user.
• Every SQL statement executed by ODCIIndexTruncate() is treated as an
independent operation. The changes made by ODCIIndexTruncate() are not
guaranteed to be atomic.
• This method is invoked for truncating a non-partitioned index, truncating a local
domain index, and also for truncating a single index partition during ALTER TABLE
TRUNCATE PARTITION.
For truncating a non-partitioned index, the ODCIIndexTruncate() is invoked with
the IndexPartition, TablePartition and callProperty set to NULL.
For truncating a local domain index, the routine is invoked N+2 times, where N is
the number of partitions.
For truncating a single index partition during ALTER TABLE TRUNCATE PARTITION,
this routine is invoked with the IndexPartition and the TablePartition filled in
and the callProperty set to NULL.

See Also:
ODCIIndexTruncate()

A-3
 Appendix A
APIs for User-Managed Domain Indexes

A.6.2 ODCIIndexMergePartition()
Invoked when a ALTER TABLE MERGE PARTITION is issued on range partitioned table on
which a domain index is defined.

Syntax
FUNCTION ODCIIndexMergePartition(
ia ODCIIndexInfo,
part_name1 ODCIPartInfo,
part_name2 ODCIPartInfo,
parms VARCHAR2,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Contains index and table partition name for one of the partitions to be
ia
merged

Contains index and table partition name for the second partition to be
part_name1
merged
Holds index and table partition name for the new merged partition
part_name2

Contains the parameter string for the resultant merged partition,

parms
essentially the default parameter string associated with the index.

The environment handle passed to the routine

env

Returns
ODCIConst.Success on success, or ODCIConst.Error on error, or ODCIConst.Warning.

Usage Notes
• The function should be implemented as a static type method.
• You should create a new table representing the resultant merged partition and
populate it with data from the merged partitions. Then drop the tables
corresponding to the merged index partitions.
• The newly created partition should pick the default parameter string associated
with the index level. Resulting local index partitions are marked UNUSABLE; you
should not attempt to populate the data in the new partition until after an ALTER
INDEX REBUILD PARTITION call.
• The old table and the dictionary entries for the old index partitions are deleted
before the call to ODCIIndexMergePartition(), so the cartridge code for this routine
should not rely on the existence of this data in the views.

A.6.3 ODCIIndexSplitPartition()
Invoked when an ALTER TABLE SPLIT PARTITION is invoked on a partitioned table
where a domain index is defined.

A-4
 Appendix A
APIs for User-Managed Domain Indexes

Syntax
FUNCTION ODCIIndexSplitPartition(
ia ODCIIndexInfo,
part_name1 ODCIPartInfo,
part_name2 ODCIPartInfo,
parms VARCHAR2,
env ODCIEnv)
RETURN NUMBER

Parameter Description
Contains the information about the partition to be split
ia

Holds the index and table partition names for one of the new partitions
part_name1

Holds the index and table partition names for the other new partition
part_name2

Contains the parameter string for the new partitions, the string
parms
associated with the index partition that is being split.

The environment handle passed to the routine

env

Returns
ODCIConst.Success on success, or ODCIConst.Error on error, or ODCIConst.Warning.

Usage Notes
• The function should be implemented as a static type method.
• You must to drop the metadata corresponding to the partition that is split, and
create metadata for the two newly created partitions.
• The new tables should pick up the default parameter string associated with the
split partition.
• The index data corresponding to these partitions need not be computed since the
indexes are marked UNUSABLE. The indexes can be built after an ALTER INDEX
REBUILD PARTITION call makes the indexes usable again.
• The old table and the old index partition's dictionary entries are deleted before the
call to ODCIIndexSplitPartition(), so the cartridge code for this routine should not
rely on the existence of this data in the views.

A-5
Index
Symbols binary large object, see BLOB, 6-1
binding, 8-2, 9-1
.so files BLOB, 6-1
naming conventions, 2-5 EMPTY_BLOB function, 6-3

A C
aggregate function, user-defined, 12-1 C and C++
analytic functions, 12-7, 12-8 debugging DLLs, 5-11
analytic functions and external context, 12-8 differences from PL/SQL, 4-5
creating, 12-2 callback
defining, 12-1, 12-2 restrictions, 5-10
example, 12-9 Callback Restrictions, 5-10
external context, 12-8 character large object, see CLOB, 6-1
external context and parallel aggregation, character sets
12-6 support for, 2-9
implementing, 12-2 CLOB, 6-1
large aggregation contexts, 12-5 EMPTY_CLOB function, 6-3
ODCIAggregate interface, 12-1, 22-1 collection types, 1-5
ODCIAggregateDelete, 12-7, 22-2 complex data objects, 1-2
ODCIAggregateInitialize, 12-1, 22-2 configuration files
ODCIAggregateIterate, 12-1, 22-2 naming conventions, 2-5
ODCIAggregateMerge, 12-2, 22-3 configuration files for external procedures, 5-5
ODCIAggregateTerminate, 12-2, 22-3 constructor method, 3-3
ODCIAggregateWrapContext, 12-6, 22-4 content, 1-2
parallel evaluation, 12-4 content of data cartridge, 1-2
reuse for analytic functions, 12-7 context
using, 12-3 inline, 12-6
using materialized views, 12-8 WITH CONTEXT clause, 5-9
aggregate interface, 22-1 conventions
Alias library, 5-1 naming, 2-6
ALL_INDEXTYPE_COMMENTS view, 8-8 corruption of package, 4-6
analytic functions, 12-7, 12-8 cost model, 1-8
ancillary binding, 9-10 CREATE FUNCTION statement, 22-1
ANYTYPE type, 13-23 CREATE LIBRARY
Associating the Extensible Optimizer Methods CREDENTIAL option, 5-1
with Database Objects, 15-49 DIRECTORY object option, 5-1
attributes of object type, 15-1 CREATE TYPE
referencing in method, 4-4 syntax, 1-9
autonomous transaction restriction, 13-7 CREATE TYPE BODY statement, 4-1
CREATE TYPE with OID statement, 3-2
Creating Statistics Table (PowerCartUserStats),
B 15-32
B-tree indexing algorithm, 7-2
B+trees, 1-7

Index-1
 Index

D domain indexes (continued)

definition, 1-7
data cartridge exporting and importing, 8-12
complex data objects, 1-2 loading, 8-19
content, 1-2 moving, 8-12
definition, 1-1 domain of data cartridge, 1-2
development process, 2-1
domains, 1-2
external procedures (guidelines), 5-12
E
installation, 2-3 electric utility example, 15-1
interfaces, 1-9 EMPTY_BLOB function, 6-3
key characteristics, 1-1 EMPTY_CLOB function, 6-3
method, 1-4 error messages
naming conventions, 2-6 naming conventions, 2-4
Oracle Spatial, 1-3 exception
Oracle Text, 1-3 raising (OCIExtProcRaiseExcp), 18-1
scope, 1-2 raising (OCIExtProcRaiseExcpWithMsg),
suggested development approach, 2-10 18-2
data objects, 1-2 execution plan
data types defintition, 1-8
collection, 1-5 extdemo1.sql demo file (extensible indexing in
extensibility, 1-4 power demand example), 15-12
REF (relationship), 1-5 extensibility
reference, 1-5 data types, 1-4
specifying, 5-6 indexing, 1-7
user-defined type, 1-4 interfaces, 1-9
DBA_INDEXTYPE_COMMENTS view, 8-8 optimizer, 1-8
DBA_SECONDARY_OBJECTS view, 8-13 server execution environment, 1-6, 2-4
DBMS interfaces, 1-9 services, 1-4
DBMS_LOB package, 6-9 collections, 1-5
compared with OCI, 6-4 data types, 1-4
DBMS_STATS package method, 1-4
used in optimizer, 1-8 reference type, 1-5
DDL extensibility interfaces, 1-1
for LOBs, 6-1 extensibility services, 1-4
DEBUG_EXTPROC, Using, 5-11 extensible database, 1-1
debugging extensible indexing, 1-7
C code in DLLs, 5-11 necessary application processes, 1-7
common errors, 4-5 necessary database processes, 1-8
PL/SQL, 4-4 queries benefitting, 15-10, 15-11
Debugging External Procedures, 5-11 extensible indexing framework, 7-5
demo directory (PL/SQL), 18-2 extensible optimizer, 1-8
demo file (extdemo1.sql) external context, 12-8
extensible indexing in power demand external context and parallel aggregation, 12-6
example, 15-12 external LOB, 6-1
directories external procedure
installation, 2-5 configuration files for, 5-5
DIRECTORY Object Option, 5-2 guidelines, 5-12
DLL guidelines for using with data cartridge, 5-12
debugging, 5-11 how PL/SQL calls, 5-4
naming conventions, 2-5 LOBs in, 6-10
domain index, 7-5, 8-1 OCI access functions, 18-1
domain indexes, 7-6, 8-8 overview, 5-2
creating, 15-12 PARAMETERS clause, 5-8
parallelizing, with table functions, 13-22 passing parameters to, 5-6

Index-2
 Index

external procedure (continued)

registering, 5-3
specifying data types, 5-6
J
WITH CONTEXT clause, 5-9 join order, 1-8
External Procedures, Debugging, 5-11
extproc process, 5-12
K
F Knuth, 7-2

foundational data cartridges

Oracle Spatial, 1-3 L
Oracle Text, 1-3 large aggregation contexts, 12-5
large object, see LOB, 6-1
G library
alias, 5-1
generic types shared, 2-5, 5-1
See ANYTYPE type LOB
Globalization Support, 2-9 DDL for, 6-1
globals external, 6-1
naming conventions, 2-4 external procedure use, 6-10
internal, 6-1
H OCI use with, 6-4
triggers and, 6-11
hash index, 1-7 value, 6-1
LOBs
I overview, 1-5
local domain indexes, 8-15, 16-1
implementation type, 8-2
index
domain
M
creating, 15-12 Maintaining Context - Java, 18-3
metadata for, 15-30 map methods, 3-4
index scan, 9-6 materialized views
index-organized table, 7-5 user-defined aggregate function, 12-8
indexing member method, 3-1, 4-1
extensible merge, ODCIAggregate, 12-2
queries benefitting, 15-11 message files
queries not benefitting, 15-10 naming conventions, 2-5
indextype, 8-1 metadata
definition, 1-1, 1-7 index, 15-30
indextype implementation methods, 15-17 method, 1-4, 15-1
indextypes, 7-5, 16-1 constructor, 3-3
operators and, 9-4 implementing, 4-1
initialization, ODCIAggregate, 12-1 invoking, 4-3
inline, context, 12-6 map, 3-4
installation directory member, 3-1, 4-1
naming conventions, 2-5 order, 3-4
installation of data cartridge, 2-3 referencing attributes, 4-4
interfaces
data cartridge, 1-9
DBMS, 1-9 N
extensibility, 1-9
naming conventions, 2-6
service, 1-9
configuration files, 2-5
internal LOB, 6-1
error messages, 2-4
iteration, ODCIAggregate, 12-1

3
 Index

naming conventions (continued) Oracle Extensibility Architecture, 1-1

globals, 2-4 Oracle Spatial cartridge, 1-3
installation directory, 2-5 Oracle Text, 1-3
message files, 2-5 order methods, 3-4
name format, 2-6 overview, 1, 1-1
need for, 2-6
schema, 2-4
shared libraries, 2-5
P
national language support (NLS). See packages
Globalization Support, 2-9 corruption, 4-6
NCLOB, 6-1 parallel aggregation and external context, 12-6
NLS (national language support). See Parallel evaluation of user-defined aggregates,
Globalization Support, 2-9 12-4
PARAMETERS clause with external procedure,
O 5-8
partitioning method
object identifier (OID) HASH, 13-18
with CREATE TYPE, 3-2 RANGE, 13-18
object type PL/SQL
attributes, 15-1 DBMS_LOB package compared with OCI,
comparisons, 3-4 6-4
methods, 15-1 debugging, 4-4
OCI demo directory, 18-2
LOB manipulation functions, 6-4 differences from C and C++, 4-5
OCIExtProcAllocMemory routine, 18-1 power demand cartridge example, 15-1
OCIExtProcRaiseExcp routine, 18-1 demo file (extdemo1.sql), 15-12
OCIExtProcRaiseExcpWithMsg routine, 18-2 primary binding, 9-10
OCILob...() functions, 6-4
ODCIAggregate interface, 12-1, 22-1
ODCIAggregateDelete, 22-2
R
ODCIAggregateInitialize, 22-2 R-trees, 1-7
ODCIAggregateIterate, 22-2 REF operator, 1-5
ODCIAggregateMerge, 22-3 reference type, 1-5
ODCIAggregateTerminate, 22-3 registering an external procedure, 5-3
ODCIAggregateWrapContext, 22-4 Restrictions on Callbacks, 5-10
overview, 12-1 routine
ODCIAggregateDelete, 12-7, 22-2 service, 18-1
ODCIAggregateInitialize, 12-1, 22-2 RPC time out, 4-6, 5-11
ODCIAggregateIterate, 12-1, 22-2
ODCIAggregateMerge, 12-2, 22-3
ODCIAggregateTerminate, 12-2, 22-3 S
ODCIAggregateWrapContext, 12-6, 22-4 schema
ODCIGetInterfaces method, 15-19 naming conventions, 2-4
ODCIIndexClose method, 15-27 scope, 1-2
ODCIIndexCreate method, 15-20 scope of data cartridge, 1-2
ODCIIndexDelete method, 15-28 selectivity, 1-8
ODCIIndexDrop method, 15-21 SELF parameter, 4-3, 4-4
ODCIIndexFetch method, 15-26 service routine, 18-1
ODCIIndexGetMetadata method, 15-30 examples, 18-1
ODCIIndexInsert method, 15-27 shared library, 5-1
ODCIIndexStart method, 15-22, 15-24 naming conventions, 2-5
ODCIIndexUpdate method, 15-29 signature, 9-1
OID signature mismatch, 4-5
with CREATE TYPE, 3-2
operator, 7-5

Index-4
 Index

statistics type U
definition, 1-1
suggested development approach for data USER_INDEXTYPE_COMMENTS view, 8-8
cartridge, 2-10 user-defined operator, 9-1
user-defined type, 1-4
T
V
table functions
parallel execution of, 13-5, 13-15 view
partitioning input, 13-18 ALL_INDEXTYPE_COMMENTS, 8-8
pipelined, 13-3, 13-6, 13-7 DBA_INDEXTYPE_COMMENTS, 8-8
querying, 13-13 DBA_SECONDARY_OBJECTS, 8-13
tables USER_INDEXTYPE_COMMENTS, 8-8
index-organized, 7-5
termination, ODCIAggregate, 12-2
transient types
W
See ANYTYPE type WITH CONTEXT clause and external procedure,
triggers 5-9
with LOBs, 6-11