On Apr 23,  2:14pm, Jason Quigley wrote:
...
>  2 JOB TESTREDO,MANAGER.SYS,PUB.
...
> 12 :COMMENT   STREAM FILE TESTREDO.PUB.SYS
> 13 :COMMENT     STREAMED BY JQUIGLEY,MANAGER.SYS,PUB ON LDEV 366
> 14 :COMMENT     RUNNING PROGRAM MAIN.PUB.VESOFT
> 15 :COMMENT     ON WED, APR 23, 1997,  2:05 PM
> 16 :SETVAR hpautocont true
> 17 :SHOWTIME
> 18 WED, APR 23, 1997,  2:05 PM
> 19 :SHOWTIME
> 20 WED, APR 23, 1997,  2:05 PM
> 21 :LISTREDO
> 22      1) COMMENT   STREAM FILE TESTREDO.PUB.SYS
> 23      2) COMMENT     STREAMED BY JQUIGLEY,MANAGER.SYS,PUB ON LDEV 366
> 24      3) COMMENT     RUNNING PROGRAM MAIN.PUB.VESOFT
> 25      4) COMMENT     ON WED, APR 23, 1997,  2:05 PM
> 26      5) SETVAR hpautocont true
> 27      6) SHOWTIME
> 28      7) SHOWTIME
> 29      8) LISTREDO

The CI posts commands to the redo stack when the HPREDOSIZE variable is > 0.
It doesn't matter if your are in a job or session.  Note: commands executed
from UDCs or scripts are never posted to the redo stack.

> 30 :MPEX
...
> 44 %showtime
> 45 WED, APR 23, 1997,  2:05 PM
> 46 %listredo
> 47 Error: INITREDO has not been done, REDOs are not enabled.
> 49
> 50 run mpex.pub.vesoft
> 51 Program terminated in an error state. (CIERR 976)
> 54 comment !8
> 55 Missing colon before command name. (CIERR 981)
> 56 exit
...
> 87 :EOJ
> 88 CPU sec. = 2.  elapsed min. = 1.  WED, APR 23, 1997,  2:05 PM.
>
> I also have INITREDO in my MPEXMGR file, but it seems that MPEX doesn't
> set up a redo stack for jobs!

Nonetheless, why do (or did) you get a divide by zero error (which I don't
see in this posting)?

It also seems that the CI is not correctly flushing your STDIN input to
MPEX!  Typically, when a program in a job aborts the program's STDIN
input is flushed by the CI.  This is done by scanning STDIN to find the first
line that begins with a ":".  This behavior is affected by whether or not
MPEX reopens STDIN(X) and some other parameters.  Mike P and others have
the details (hint: the CI checks some eof bits in the JIT).

regards,
Jeff Vance, CSY

--