The following sample JCL demonstrates the capability of DSNTEP2 to issue DCL, DDL, and DML dynamically.
//DB2JOBU JOB (BATCHSQL),'DB2 SAMPLE SQL',MSGCLASS=X,
// CLASS=X,NOTIFY=USER
//*
//***********************************
//*
//* DB2 SAMPLE SQL PROGRAM
//*
//***********************************
//*
//JOBLIB DD DSN=DSN810.DSNLOAD,DISP=SHR
//BATCHSQL EXEC PGM=IKJEFT01,DYNAMNBR=20
//SYSTSPRT DD SYSOUT=*
//SYSPRINT DD SYSOUT=*
//SYSUDUMP DD SYSOUT=*
//SYSTSIN DD *
DSN SYSTEM(DSN)
RUN PROGRAM(DSNTEP2) PLAN(DSNTEP81) -
LIB('DSN810.RUNLIB.LOAD')
END
/*
//SYSIN DD *
SELECT * FROM SYSIBM.SYSTABLES;
UPDATE DSN8810.DEPT
SET DEPTNAME = 'CHANGED NAME'
WHERE DEPTNO = 'D01' ;
INSERT INTO DSN8810.ACT
VALUES (129, 'XXXXXX', 'SAMPLE ACCT');
DELETE FROM DSN8810.EMP
WHERE SALARY <>
/*
The DNSTEP2 program is written in the PL/I programming language. Prior to DB2 V6, you needed to have a PL/I compiler to use DSNTEP2. However, as of V6 IBM now provides both the source code and an object code version of DSNTEP2 with DB2. So, you no longer need a PL/I compiler to use DSNTEP2.
Because DSNTEP2 is an application program, and the PL/I source code is provided with DB2, a knowledgeable PL/I programmer can easily modify the code. After doing so, of course, it must be compiled, linked, and bound before it can be used.
DSNTEP2 can process almost every SQL statement that can be executed dynamically. DSNTEP2 accepts
· The GRANT and REVOKE DCL statements,
· The ALTER, COMMENT ON, CREATE, and DROP DDL statements,
· The DELETE, INSERT, SELECT, and UPDATE DML statements, and
· The COMMIT, ROLLBACK, EXEC SQL, EXPLAIN, and LOCK statements.
The only important statement that DSNTEP2 does not support is the LABEL ON DDL statement. Of course, DSNTEP2 can be modified to support this statement if you have PL/I knowledge and a PL/I compiler.
When Does DSNTEP2 Commit?
Well, the simple answer to that question is that the results of the SQL are committed upon completion of all the SQL. A helpful answer is a little longer.
First off, you need to know that DSNTEP2 has an internal parameter named MAXERRORS that controls the number of failing statements that can occur before it stops. A failing statement is one which returns a negative SQLCODE. The value of MAXERRORS is set to 10 inside the program, so DSNTEP2 will allow 9 failing SQL statements but when it hits the 10th failing statement, it will exit, COMMITting all other work.
This is ugly because it can wreak havoc on the integrity of your data. I mean, who wants to figure out what was run, what was impacted, and then try to rebuild a job to fix data and/or restart at the right place? To rerun DSNTEP2, remember that all SQL statements that completed with a 0 SQL code were committed. These statements should not be rerun. All SQL statements completed with a negative SQL code must be corrected and reprocessed.
Certain severe errors cause DSNTEP2 to exit immediately. One severe error is a -101 “SQL statement too long or too complex".
If any SQL errors occurred during the execution of DSNTEP2, a return code of 8 is returned by the job step.
At any rate, DSNTEP2 never issues an explicit COMMIT or ROLLBACK by itself. A COMMIT occurs at the end unless the program abends.
Specify Your SQL Statements Properly
The SQL to be run by DSNTEP2 is specified in SYSIN. Be sure to code the DSNTEP2 input properly.
DSNTEP2 reads SQL statements from an input data set with 80-byte records. The SQL statements must be coded in the first 72 bytes of each input record. SQL statements can span multiple input records and are terminated by a semicolon (;). Semicolons are not permitted in the text of the SQL statement.
Liberally Comment DSNTEP2 Input
Comments can be passed to DSNTEP2 in the SQL statements using two hyphens in columns 1 and 2 or a single asterisk in column 1. Doing so is good form and helps others to understand what your DSNTEP2 job is attempting to accomplish.
Bottom Line
DSNTEP2 is especially useful for running one-off SQL statements. Use DSNTEP2 when you have ad hoc SQL to run in a batch environment. DSNTEP2 is easier than writing your own quick and dirty programs to run ad hoc SQL in batch. It is simple to set up and saves time. But be careful if you have multiple SQL modification statements (INSERT, UPDATE, DELETE) because of the above-mentioned possibility of causing data integrity problems if some of the statements fail and others succeed.
