On Tue, 3 Feb 2015 12:35:16 -0500, Tony Harminc wrote:
>
>> "address TSO" from a Rexx EXEC spawn()ed under Unix System Services
>> runs the TMP in a separate address space, perhaps more like a job step
>> task; certainly more roundabout.
>
>This is an elegant approach; the problem is that you don't know with
>certainty what output comes from what command. If you are ATTACHing
>commands yourself, you know where the boundaries lie.
>
>Effectively you have to screen-scrape (line-scrape?) the output
>looking for READY or whatever the subcommand prompt may be. You also
>
Not really.  No more than when you OUTTRAP to a stem you need to
scrape the trapped output looking for READY.  (But I suspect the
TSO host command interface does something of the sort.)

It's a programming interface, not a terminal interface.  You can
run a command and direct its output to a data set, not much more.

>want to avoid being prompted for input, whether unavoidably as with
>RECEIVE, or in the case of an error, or have an even more complex and
>
Unfortunately true.  I can't really make RECEIVE work in that configuration.
RECEIVE ought to provide a "no prompt" mode where everything is
supplied in command arguments.

>error prone scheme of parsing and replying to prompts.
>
And I got an ABEND in a batch job when I allocated DD SYSTSPRT
to a UNIX file.  PMR.  APAR.  PRS.  Something about crossed-up
keys.  That sort of thing shouldn't be tolerated.

(I had something similar (twice) when I allocated HLASM SYSLIB to
a mixture of PDS and UNIX files.  That one they fixed, both times.

The design is not robust.  If you break it, they don't feel they need
to fix it.

--gil

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to