It should, but even if it starts in CLLE, and everything is running in the same named activation group, if it then runs back to an OPM program you could have issues if you don't set CMTSCOPE(*JOB).
Mark Murphy
STAR BASE Consulting, Inc.
mmurphy@xxxxxxxxxxxxxxx
-----rpg400-l-bounces@xxxxxxxxxxxx wrote: -----
To: "RPG programming on the IBM i / System i" <rpg400-l@xxxxxxxxxxxx>
From: "Wintermute, Sharon" <Sharon.Wintermute@xxxxxxx>
Sent by: rpg400-l-bounces@xxxxxxxxxxxx
Date: 10/06/2010 02:02PM
Subject: RE: avoid running ILE programs in the DAG...
I wonder if the OP is using CLP instead of CLLE routines to start
commitment control. The CLLE should use the same named AG as the rpg
routine.
Sharon Wintermute
-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx
[mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Mark Murphy/STAR
BASE Consulting Inc.
Sent: Wednesday, October 06, 2010 12:55 PM
To: RPG programming on the IBM i / System i
Subject: RE: avoid running ILE programs in the DAG...
Yes it works on named activation groups. The commit scope can be the
activation group it was started in (this is the default), or the job.
If Alan was using mixed activation groups (some programs OPM, some
programs ILE with named activation group), he might think that
commitment control was not working, but in actuality it was doing just
what he told it to do. To cover the entire job in such a mixed
environment you need to use STRCMTCTL CMTSCOPE(*JOB).
Mark Murphy
STAR BASE Consulting, Inc.
mmurphy@xxxxxxxxxxxxxxx
-----rpg400-l-bounces@xxxxxxxxxxxx wrote: -----
To: "RPG programming on the IBM i / System i" <rpg400-l@xxxxxxxxxxxx>
From: "Wintermute, Sharon" <Sharon.Wintermute@xxxxxxx>
Sent by: rpg400-l-bounces@xxxxxxxxxxxx
Date: 10/06/2010 01:39PM
Subject: RE: avoid running ILE programs in the DAG...
I use commitment control within named AGs all the time. It does work.
Sharon Wintermute
-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx
[mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Bryce Martin
Sent: Wednesday, October 06, 2010 12:31 PM
To: RPG programming on the IBM i / System i
Subject: Re: avoid running ILE programs in the DAG...
Why not have a cleanup program for each service program so that if it
ends
in error you won't have any junk laying around. What exactly were you
pointing to that you got a pointer crash? I write all my new code in
named AG's, and trust me, when I'm developing they crash and burn. But
I
don't ever have trouble running the program again. Are you trying to
trap
the error and continue? Or are you describing just rerunning the
program?
I'm guessing you are missing a critical piece. You should no way no
how
need to run your service programs in the DAG.
Can anyone else verify if commitment control will or won't work in named
AG's? I can't hardly believe that. I know most i shops don't use CC,
but
this would seem like a huge oversight on IBM's part.
Thanks
Bryce Martin
Programmer/Analyst I
570-546-4777
Alan Campin <alan0307d@xxxxxxxxx>
Sent by: rpg400-l-bounces@xxxxxxxxxxxx
10/06/2010 12:35 PM
Please respond to
RPG programming on the IBM i / System i <rpg400-l@xxxxxxxxxxxx>
To
"RPG programming on the IBM i / System i" <rpg400-l@xxxxxxxxxxxx>
cc
Subject
Re: avoid running ILE programs in the DAG...
We had to put our service programs back into DAG because commitment
control
would not work unless the program was running DAG.
We also ran into big problems with errors in named activation groups.
Every
time that an error is thrown in a named activation group the system
destroys
the activation group and when you try to go back to it the program
crashes
because the pointer is now pointing at a object that doesn't exist
anymore.
The only way we could seem to get around this was to run the service
program
and programs in the DAG so that it could not destroy the activation
group.
Totally bizarre behavior on IBM's part. Errors occur all the time. Why
destroy the activation group but that what they do.
I don't like running everything in the DAG but pretty much have no
choice.
On Wed, Oct 6, 2010 at 8:12 AM, Stefano Arrighini - Smea s.r.l. <
s.arrighini@xxxxxxx> wrote:
Hi to all
Some years ago all ours programs ran in the DAG.
In the past two years we "introduced" Activation Group in some
programs
(some in named AG, other in temporary ones).
We made this for various reason (avoid recursion, improve
performance,
etc.).
Unfortunately (or better... by ignorance...) we made this operation
without having well understood Activation Groups.
Now I'm trying to do spring cleaning...
In all (or almost) technical articles and forum I read, I found the
advice
to avoid running ILE programs in the DAG.
So we modified our programs to run all of them in Named Activation
Group
(or Temporary) and not in the DAG.
None of our programs are compilated with DFTACTGRP(*YES)!!
So I simply replaced *CALLER with Named-one in the ACTGRP option of
ours
"starting program" (the first program in the call stack of ours
job).
I found lots of advantages... but this operation introduced some
problem
with OVRDBF scoping.
To avoid this problem I changed the OVRDBF (and OVRPRTF and DLTOVR)
with
scoping problem.
A make an OVRPRTF and calls B
B calls C
C is in a different AG and isn't affected by OVRPRTF. But I want it.
OVRPRTF OVRSCOPE(*CALLLVL) was my solution..
Now my MAIN question is...
Compiling all my existing program in a proper Activation Group (not
in
the
DAG), how many other problem can produce?
For example I found some difference in "free resources".
The RCLRSC LVL(*CALLER) used in the DAG can't be "replaced" with a
similar
command in a Named AG...
I have to close the Activation Group... Right?
Any help or observation would be greatly appreciated.
Thanks in advance
Stefano
--
This is the RPG programming on the IBM i / System i (RPG400-L) mailing
list
To post a message email: RPG400-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/rpg400-l
or email: RPG400-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/rpg400-l.
midrange.com
