Re: Question regarding Trigger programs and recursive calls

[Scott Klement <rpg400-l@xxxxxxxxxxxxxxxx>]

Charles,

All ILE programs remain activated until the activation group ends 
(unless you specify DFTACTGRP(*YES) on the CRTBNDxxx command, that is)
That includes all of CL, RPG, and C.

(Yes, an ILE RPG program remains active regardless of whether *INLR is 
on or off. However, if INLR is on, the files are closed, and the 
variables are reset to their initial values on the subsequent calls -- 
but either way the program remains active until the activation group is 
reclaimed.)

What I think you're driving at is that RPG's main procedure stores it's 
variables in static storage.  So they can keep their values from call to 
call.  CL uses automatic storage, so the storage of the variables is 
freed when the procedure ends, and has no chance of keeping it's value 
from call to call.

That, by the way, is one of the reasons why CL can be called recursively 
and RPG can not!  :)   So that's a GOOD feature in this case, not a bad 
one.  Because CL's variables are in automatic storage, there's a 
separate copy created for each call stack level -- which is a 
requirement for recursion.

C will also, by default, use automatic storage (like CL does) though C 
has more options to control how the storage is allocated than CL has.

However, if it's a simple wrapper that does nothing by call an RPG 
procedure, I really think this bit about automatic vs. static storage is 
a moot point.  There probably won't be any variables scoped to the main 
procedure, since all it's going to do is pass it's parameters on to the 
RPG subprocedure.

Also, just to straighten out the terminology a little bit...  we are 
talking about using C or CL for the UEP, not the PEP.  You can't create 
your own PEP, the PEP is generated by the compiler.  We are working on 
the UEP (user entry procedure) that's called when the program is invoked.



Charles Wilt wrote:
> Good point Scott, just to expand upon your reply...
>
> The OP would want to use ILE CL in order to use the CALLPRC command to
> call the RPG IV procedure.
>
> However, since we're talking about triggers, I'm wondering how big the
> performance hit would be using ILE CL.  Correct me if I'm wrong Scott,
> but ILE CL doesn't support a program staying active like RPG does by
> returning with LR = *OFF.  C programs on the other hand always remain
> active till the activation group ends.
>
> Thus, it would seem to me that using C for the PEP (mainline) would be
> preferable to using ILE CL.  Even if a shop doesn't have C experience,
> the C routine is so simple and easily copied for use in other
> triggers, picking up what little knowledge is required to use it
> shouldn't be a big deal in a modern RPG IV shop.
>
> Charles