Re: Activation groups for beginners

["John Taylor" <jtaylor@xxxxxxxxxxxx>]

Nelson,

It's specifically the behaviour of the trigger within a commitment control
environment that made me question your decision. With the trigger running in
a separate NAG, you're forced to use a *JOB level commitment scope -
something that may be undesirable from the perspective of the overall
application. At the very least, it forces the programmer to remember to
change the commitment definition scope on every STRCMTCTL statement (the
default is *ACTGRP).

Without a compelling reason to use a NAG, I would recommend that triggers be
compiled to run in *CALLER.


Regards,

John Taylor

----- Original Message -----
From: "Smith, Nelson" <NSmith@lincare.com>
To: <rpg400-l@midrange.com>
Sent: Wednesday, March 06, 2002 8:19 AM
Subject: RE: Activation groups for beginners


> Just ease of recognition.  Could just as easily have been QILE.
>
> The important point is that it is a separate named activation group.  You
> definitely would not want to use *NEW here due to the overhead of building
> and destroying the activation group on EACH record that is triggered.
> *CALLER may get flaky if the triggering program is OPM, although I'm not
> real clear on the implications when Data Management gets in between the
the
> triggering program and the trigger.  The stack listings I've looked at
would
> seem to indicate that the actual trigger calling routines from Data
> Management run in *dftactgrp (although that may be just coming forward
from
> the triggering program, I'm not sure).