Re: Slightly off-topic: development machine for z/OS wanted

---In [email protected], <bumens@...> wrote :

> first thank you for the links and hints! I'll give it a try.

Ok, cool.

Separate question.

As I explained, I have spent a lot of effort
trying to think what authors of free software
from a Unix background would require to
start targeting MVS.

Rather than requiring them to know/learn
TSO to make software changes, I instead
decided that they could do all their work
at a Unix prompt, and have minimal
interaction with MVS. Basically I have been
trying to give Unix people no valid reason
to ignore the MVS target.

You still need to write JCL though. So
you need to learn/copy that from an
existing implementation. But after that,
you can run a script to bundle up all
your JCL and pass that up to MVS to
run, and get your result back at the
safety of the command prompt.

Does that sound appealing to you?

Here is the simplest example:

iefbr14.jcl:
//HERC01A JOB CLASS=C,REGION=0K
//*
//S1       EXEC PGM=IEFBR14
//*
//

runmvs iefbr14.jcl output.txt

output.txt:
08.00.34 JOB    1  $HASP373 HERC01A  STARTED - INIT  3 - CLASS C - SYS BSP1
08.00.34 JOB    1  IEFACTRT - Stepname  Procstep  Program   Retcode
08.00.34 JOB    1  HERC01A    S1                  IEFBR14   RC= 0
08.00.34 JOB    1  $HASP395 HERC01A  ENDED


Note that once the MVS disks have been
cached by Windows, it only takes 3 seconds
to start MVS, run a batch job (that does
nothing) and then terminates.

The runmvs.bat can be easily tailored for
other scripting languages.

Does this concept sound appealing to you?
Does it reduce the barrier to making MVS a
target for Perl etc? I can help with the
construction of the JCL if a Unix software
author is willing to fix the compile errors
when encountering the pure C90
environment offered by GCCMVS.

BFN. Paul.