RE: RPG ILE and *INLR

["Joe Pluta" <joepluta@xxxxxxxxxxxxxxxxx>]

> From: Chris Rehm
>
> Yeah, in that light it becomes predictable, and really expected
> behavior. The addition of the PAG in between caller/callee is a fun new
> twist. It is a great thing, though. *CALLER PAG gives faster performance
> but you better know what you're doing. It is just another layer of the
> RETRN or LR question.

What happens is that I don't have to worry about regstering all my
subprocedures and then manually shutting them down, provided I am willing to
deal with activation boundaries.  In my case, the reason for not setting on
LR was performance, and at shutdown I didn't do anything, so I really am
best off using a *NEW activation group.  I never thought I would use *NEW,
but it makes sense to me now.

If I break my job up into discrete work units, then each work unit is a main
program with activation group *NEW.  My job stream is a series of calls to
those work units.  Once each work unit is finished, all the subprocedures
are flushed and memory is cleaned up for the next work unit.  This is not a
bad model.  The only issue is if I need to make non-bound calls, at which
point I have to decide whether the subprogram being called is part of this
work unit (AG *CALLER) or its own unit (AG *NEW).

Old dogs can occasionally kick off the fleas and learn a new trick or two
<grin>.

Joe