RE: Trigger Program Activation Group

Scott and Duane,

Thanks. I really did mean a separate copy of the program. Shows how much I know about activation groups. Your explanations reminded me of the old MRT programs that we had on the /34 and /36; there was only one instance of the program, but we could keep certain data elements separate by user so that people didn't step on each other's toes, but the other stuff was, as it were, global to everyone.

Again, thanks, gentlemen.

Jerry C. Adams
IBM System i Programmer/Analyst
--
B&W Wholesale
office: 615-995-7024
email: jerry@xxxxxxxxxxxxxxx

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Scott Klement
Sent: Thursday, August 13, 2009 12:13 PM
To: RPG programming on the IBM i / System i
Subject: Re: Trigger Program Activation Group

I'm not sure that you really mean "copy of the program", literally.
There's always only one copy of the program itself. What you really are
asking is whether there'd be more than one *activation* of the program,
correct?

You see, the system always has only one copy of the actual program.
that is, the series of bytes that the CPU executes -- that's the
program, and it only exists once. Even if you call the program from
completely separate jobs, there's only one copy of the program.

But the separate jobs have different variable values, different open
file paths, and stuff like that. This data that's kept separate is
referred to as an "activation" of the program. And indeed, that's where
the term "activation group" comes from -- it's a group of activations.

Anyway... your question... would the trigger have the same open file
paths, same variable values, etc saved from call to call when called
from multiple activation groups?

The answer is: No. You'd have one activation per activation group that
the trigger is called from. One in the default activation group, one in
each named activation group, one in each numbered activation group, etc,



Jerry Adams wrote:

I am writing a trigger program that uses DFTACTGRP(*No) and
ACTGRP(*Caller), which is the way I usually write these (unless I
want/need to use a named activation group). We use, as a system
value, the 36 environment so our basic program set, which is RPG II
invoked from OCL, uses the default activation group. Except for
maintenance of those RPG II programs and OCL, all new development is
in ILE RPG and CL. I had never thought about it much before (if at
all), but I wondered what activation group would be used if the
trigger was fired while using one of those old programs.

So I checked the manual, which didn't have an answerm and our
archives. Responses from Scott and Simon indicated that, despite
DFTACTGRP(*No), the program would, indeed, run from the default AG.
Answered my initial question, but then another one crept into my
head: What if the user triggered the program while running an RPG II
program and then later caused it to fire from an ILE RPG program,
which uses a named AG, or vice versa. Would another copy of the
trigger program be loaded into the named AG? Bear in mind that I
never close the trigger program to improve performance; except when I
use a named AG in a trigger so I can invoke RCLACTGRP, the trigger
programs stay active until the user signs off of the System i.

Now that I think about it, while typing this, the 36 environment is
immaterial. The trigger program could have been fired as a result of
two different programs with different named AG's. The question still
remains: Would multiple copies of the trigger program be loaded? In
a relatively small shop like ours (I checked and there are currently
18 interactive jobs running; the % CPU used (DSPSYSSTS) is around
2.7%) does it even matter?