How many RPG programs have been written that perform a data scan of an indexed file just to retrieve a subset of that information based on fields not in the key? In conventional file access, the coding that is required includes not only positioning the file based on the key, but also coding multiple CHAIN/READ statements to ensure that only the required records are used. Wouldn't it be nice to be able to read only the records that meet the selection criteria? There is a way! Use embedded SQL and cursor definitions.

To define a cursor within an RPG program, the following items need to be included:

• Source Member Type Definition (either SQLRPG or SQLRPGLE)
• SQL Descriptor Area (SQLDA) Definition
• SQL_NUM Field Definition
• SQL Cursor Definition
• SQL Open Statement
• SQL Close Statement
• SQL Fetch Statement

The SQLDA lays the groundwork for all other SQL processes within the RPG program. The SQL_NUM field must be defined in the program in support of the SQLDA. If this field is not defined, the compile will fail. The code below shows how to define the SQLDA and SQL_NUM in an RPG ILE program.

* Define SQL Descriptor Area
D SQL_NUM C Const(1)
C/Exec SQL
C+ Include SQLDA
C/End-Exec

The cursor definition is the key to preparing the program to perform data retrieval. With a properly defined cursor and associated indexes on the files being processed, the program will not only require fewer database I/O processes, but also enhance readability and simplify program maintenance.

The following code demonstrates how information not defined in the keys of the associated files can be retrieved in a co-joined read. This bypasses the records that do not meet the selection criteria, thereby eliminating the need to process all of the records associated with the key of the primary file.

* Define Cursor for Based On Verification Type
C/Exec SQL
C+ Declare C1 Cursor For
C+ Select a.BTITM#, b.IMDSC1, a.BTRSCD, a.BTSOT
C+ From OPIB101 a, IMIM100P b

C+ Where a.BTCO# = :LDACO#

C+ and a.BTITM# = b.IMITM#
C+ and a.BTVTYP = :#HBVTY
C+ and a.BTITM# not in (Select c.BTITM# from OPIB101 c
C+ Where c.BTCO# = :LDACO#
C+ and c.BTITM# = a.BTITM#
C+ and c.BTVTYP = :#HVTYP)
C+ Order By a.BTCO#, a.BTITM#, a.BTVTYP
C/End-Exec

Another function performed by the cursor is record matching within a file. Notice that the primary select and the sub-select compare the same field combinations within the same file in order to process only those that do not match. The Declare statement assigns the cursor name "C1" that will be utilized in the SQL statements throughout the remainder of the RPG program. Note: The cursor only prepares the file access; it does not perform the data retrieval.

In order to retrieve the data via the defined cursor, the cursor needs to first be opened. Then, the data can be retrieved. When the data retrieval is complete, the cursor needs to be closed. The code below shows what needs to be done to process all of the rows associated with the cursor.

C/Exec SQL

C+ Open C1
C/End-Exec

C ExSr $ReadC1

C DoW SQLCOD = *Zeros
C ExSr $ReadC1
C EndDo

C/Exec SQL
C+ Close C1
C/End-Exec
*=================================================================
* $ReadC1 - Read based on records *=================================================================
C $ReadC1 BegSr

C/Exec SQL
C+ Fetch C1 Into
C+ :#JITM#, :#JDSC1, :#JRSCD, :#JSOT
C/End-Exec

C EndSr

Note that the cursor name "C1" is used in the Open, Close, and Fetch statements. Also, the field SQLCOD is defined within the SQLDA. A value of 0 in SQLCOD denotes a successful completion of the SQL statement. The loop will terminate if the end of the cursor is reached (the SQLCOD will be set to 100) or if any SQL error is encountered (the SQLCOD will be set to the appropriate error code). Another requirement when coding embedded SQL is that the number of fields following the Fetch statement is the same as the number of fields following the Select statement in the cursor definition. This field returns the SQL return code.

Any field defined in embedded SQL that is prefaced by a colon (:) is a host variable. If a colon does not preface a field, it is either a constant value or a field name. Not all of the host variables have been defined in the sample code. These fields would need to be defined either as program variables or fields of files that have been defined to the program. This includes display and printer files.

This technique can save the developer a great deal of time and effort within the program. If this process was being done using conventional file access, the data retrieval would require that the file be read twice (saving the data from the first read), or separate access paths would need to be defined within the program in order to do the field matching.

Paul Weyer is a consultant with New Resources Consulting. He has worked on the iSeries (AS/400) since 1987. His background includes both hardware configuration/installation and software development. Paul is a subject matter expert (SME) with COMMON in the areas of LPAR, IT Optimization, Performance, Availability/Business Continuity, iSeries Operations, and Managing Beyond Your Server. His language background includes both COBOL and RPG, and he has been using SQL since 1987. Paul can be reached at pweyer@nrconsults.com.