Mike sent me a test case.  As I suspected, the first program check is
in the prologue of the routine that traps console output in the
COMMAND interface.  This leads to program checks in DMSMGM undoubtedly
when I try to report the program check in the pipeline code.

Running multiple CMS commands concurrently is asking for trouble; and
running multiple COMMAND stages concurrently even more so, as the
magic and mirrors that make it happen is in parts of CMS that is not
re-entrant.

Like, would you contemplate running two HLASMs concurrently?  Or
COPYFILE?  Or just about any CMS command?  Most of them are not
re-entrant.

I shall update the preface accordingly.

   j.

On 27 August 2010 06:31, John P. Hartmann <[email protected]> wrote:
> I should think that running multiple COMMAND stages concurrently would
> lead to trouble, so don't do that.
>
> Please send me the test case with instructions.
>
>   j.
>

Reply via email to