RE: CRTDUPOBJ an ILE program

[Buck Calabro <Buck.Calabro@xxxxxxxxxxxx>]

>My simple questions is: 
>
> Are there any potential problems with  changing 
>all our RPG programs to ACTGRP(*NEW) ?  

The number one problem will be run time performance.  Creating a new AG is
roughly equivalent to starting a new job.  I can't speak for anyone else,
but *NEW was probably not intended for this use.

>New recursive calls creep up all the time.  I read 
>that with ACTGRP(*NEW) you can avoid the recursive 
>call problem.   
>
>It sounds like the answer is:  Without an understanding 
>of how AG works and without further analysis,  it would 
>not be a good idea to change the default on CRTBNDRPG 
>to ACTGRP(*NEW).

I concur with this wholeheartedly.  I'm not plagued by having to deal with
recursive calls, so this isn't much of a problem for me, but if it were, I
would probably have to re-think my application design.  If I absolutely
needed to allow recursion NOW, I'd probably stick with the system you have:
CRTDUPOBJ just before calling the potential recursor (is that a word?)

While advanced users recommend *NEW/*CALLER (like Scott, Joe and our beloved
Barbara - I couldn't resist the alliteration) I have not been able to work
that way in my very mixed OPM/ILE environment.  In my application, I use a
single named AG.  This way, I inherit the OPM overrides, and all my
ILE-issued overrides stay intact.  While we can ask simple questions, there
are no simple answers to the ILE AG question.  AGs are designed to handle
very complex issues, and the vast majority of us don't need the full power
that ILE offers.  Because of this power, there is a good bit of thinking
that needs to happen before you finalise your AG design.

Remember that programs which share an AG are designed to work together.
Programs that DON'T share an AG are designed to work separately.  That's the
key AG philosophy, I think.
  --buck