Thanks for this info. Makes me feel better (for whatever reason). We actually had some guesses that the issue was caused by migrating from COBOL V4 to V6, even though that was done months ago. (The "bad code" was added after the migration, and thus had never run under COBOL V4).
The unfortunate thing is (this is my understanding, anyway), the bug in the application showed up in testing but was ignored/bypassed as an "environment issue", which of course was a bad assumption. We're now discussing using the NUMCHECK option in development, and I think that will turn up a lot of these cases. Which is a good thing. Frank ________________________________ From: IBM Mainframe Discussion List <[email protected]> on behalf of Dylan Perry <[email protected]> Sent: Thursday, March 11, 2021 10:32 PM To: [email protected] <[email protected]> Subject: Re: COBOL V6.2 possible change in behavior with recent patch level ZONEDATA(MIG) is meant to provide compatibility with the NUMPROC(MIG) option in pre-V5 versions. COBOL 4.2 with NUMPROC(MIG) abends with a S0C7 with the following instructions:- 001154 IF 0004CC 4DE0 9152 BAS 14,338(0,9) TGT TEST INFORMATION AREA +10 0004D0 GN=19 EQU * 0004D0 F211 D110 8010 PACK 272(2,13),16(2,8) TS2=0 AS-NUMS 0004D6 F911 D110 A0B2 CP 272(2,13),178(2,10) TS2=0 PGMLIT AT +162 This APAR explains that in versions 5 to just before the recent 6.2 fix it was "incorrectly" fixing the S0C7:- https://www.ibm.com/support/pages/apar/PH28379 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
