RE: H ActGrp('QILE') vs H ActGrp(*caller)

[Joel Fritz <JFritz@xxxxxxxxxxxxxxxx>]

For a service program that you need to release all resources when it ends,
*CALLER is a good choice, providing the calling program runs in a named
activation group that ends when the program does.  If you have a lot of jobs
running in QILE, you don't have the same cleanup options you would with a
more restricted activation group.  Depends on your scoping requirements,
though.  The important thing is that QILE is only a name for an activation
group, it's not a system object.


> -----Original Message-----
> From: rob@dekko.com [mailto:rob@dekko.com]
> Sent: Friday, September 21, 2001 6:20 AM
> To: rpg400-l@midrange.com
> Subject: H ActGrp('QILE') vs H ActGrp(*caller)
>
>
>
> When I first started using a service program I set up some
> default H spec's
> to be copied in using /copy for the programs using the
> service program, and
> for RPGLE programs in general.  In this copy member I have
> the following:
> H ActGrp('QILE')
>
> Some developers have requested that we be able to change this to
> H ActGrp(*CALLER)
> This will help make
> ErrorFlag=system(cmd);
> work better.  In cases where cmd is an OVRDBF.  I know I can
> workaround
> with OVRSCOPE.
>
> Any comments on the ramifications of this change?
>
> The concepts of service programs, subprocedures and the like
> are catching
> on here and I want to do everything I can to not stand in the way.
>
> Who would have thought to document their H specs?  :-)
>
>
> Rob Berendt
>
>