On Thu, 21 Jan 2021 20:22:22 +0000, Jesse 1 Robinson wrote: > >Not so fast. We discovered from their install jobs that they had coded >something like this: > > BANKAPP > Lots of code and calls including the IBM module > END > >As long as the program was compiled and linked in the same run, the END >statement picked up BANKAPP as entry point and everything was cool. But when >run separately, the entry point was indeterminate, so link failed. Source of >course was not available so we could not add > > END BANKAPP > I perpetrated one of those once. I was maintaining/upgrading a FOSS Pascal compiler. Direct to SYSLIN; no Assembler. I added the effect of "END APP"; no ESD; nary a label. Obsessive; saving 4 bytes of branch around eyecatcher. But worked fine until a co-worker relinked -- he didn't like my BLKSIZE. I got better; added the branch.
Nowadays, INCLUDE -ATTR should handle that -- probably what it was invented for. Same co-worker once bypassed some test code with EXEC COND=ONLY. He was considerably upset when my program ABENDed in an earlier job step. I always use COND=(0,LE) for that purpose. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
