Re: ILE 100.5

[Joel Cochran <jrc@xxxxxxxxxx>]

On Mon, 2004-04-12 at 22:07, Chris Wolcott wrote:
> <NAMED> Starts a new ACTGRP named <x> (if required) that goes away when
> the JOB ends.  (Your session, if interactive.)  This can be a source of
> confusion, when you KNOW you've made the change yet the code still errs
> like it used to. . .

True, but I find that recreating a program with a named activation group
forces the activation group to be changed.  I do this frequently on my
CGI site without ever issuing RCLACTGRP.

> You can only use *CALLER if the called routine is a standalone bound
> program, NO *SRVPGM calls.  (I guess because of the possible link to an
> OPM environment.)

I don't think this is correct... everything, and I mean everything...
runs in an activation group.  Even OPM programs run in one of the DAGs
(there are two).  As a result, you *can* run a service program in the
DAG, if it is activated by a program running in the DAG.  

> *NEW is similair to the OPM environment, except that it is ILE.  (Doh!)
> I.E. persistence of objects, initializations, etc.

In the effect that *NEW will create the activation group (read "allocate
resources"), run the program, and then release those resources, then yes
it is similar to OPM.  However, by virtue of having an activation group
OTHER than the DAG, it does NOT behave like OPM because the activation
group itself is destroyed: you can not destroy the DAG, so an OPM
program NEVER destroys its activation group.

> If the program is used once, it is a wash over *NEW or <named>.  The
> more times it is called in a job, the more you should lean towards
> <named>.      

Not at all: as pointed out on this list, a named group will hang around
and occupy space and resources in your job until the job is ended.  If
this is truly a one-time program, why would you want it hanging around? 

Joel
http://www.rpgnext.com