On Tue, 19 Sep 2006 13:40:58 -0300, Shmuel Metz (Seymour J.) <[EMAIL PROTECTED]> wrote: > > In <[EMAIL PROTECTED]>, on 09/19/2006 > at 10:34 AM, Binyamin Dissen <[EMAIL PROTECTED]> said: > > >The issue is that AC=1 programs expect to be called as job-step > >programs and may not completely clean up after themselves (expecting > >the initiator to do it). > > Not the ones that I've seen. The standard used to be that all IBM > utilities were written to run as subroutines and they typically > allowed additional parameters when called as subroutines, e.g., ddname > lists. If anything, there are more eyeballs on the behavior of the > authorized programs. > Binyamin utters considerable sooth. I prevailed on our sysprog to add GIMSMP to AUTHPGM. This helped lots of things when I run SMP/E from an EXEC:
o I can now specify WAIT on DDDEFs because S99WTDSN no longer ABENDs. o RETRY with IEBCOPY compress works. Sort of. But in the case of a successful RETRY, SMP/E fails to close at least one of its output data sets. I'm very much aware of this because I'm piping them through /bin/awk filters so I can treat certain SMP/E warnings as errors. (I also strip trailing blanks for compactness.) since the awk filter's input is never closed, I "SYSCALL wait" indefinitely for the awk process to terminate. I suppose I could try to "SYSCALL spawn" SMP/E into a separate address space -- the DCB's should be closed at end-of-memory, shouldn't they? But this would require significant reorganization of the EXEC. But I'll try to reproduce the behavior in a compact test case to see whether IBM thinks it's worth an APAR. -- gil -- StorageTek INFORMATION made POWERFUL ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
