Re: Activation group quandaries

On 24 Jul 2013 13:08, David Gibbs wrote:

<<SNIP>>

I've got a service program that makes a call to an external ILE bound
RPG program (prototyped, with EXTPGM). The RPG program is compiled
with activation group *NEW. The service program is compiled with
activation group *CALLER.

The service program is invoked from java via a remote program call
(JT400).

This RPG program makes a procedure call into the same service program
that calls it.

My questions are:

When the service program calls the external RPG program, should the
activation group change from QILE to a new activation group?

Where does QILE come in? Nothing in the above suggested QILE is defined for any code.

When the external program calls back into the service program, will
a new copy of the service program be loaded ... or will it call back
into the original service program that invoked the external program?

AFaIK the service program will be activated the second time with _the_ /new/ activation group; i.e. activated within the new\temporary ActGrp definition that was created for the ILE RPG program defined with ACTGRP(*NEW).

Given:

executable: actgrp
----------: --------
pgmRPGILE : *new
srvpgm : *caller
jt400call : unknown

Thus at run-time I would expect [and the WRKJOB OPTION(*PGMSTK) issued against the job when stopped, e.g. with debug breakpoint, at a statement in the srvpgm on the second invocation; I am too lazy to code and compile something to verify]:

stack; first is bottom:
pgm\srvpgm actgrp associated [per]
----------- -----------------------
jt400call unknown
srvpgm unknown [*caller]
pgmRPGILE tmpNbrX [*new]
srvpgm tmpNbrX [*caller]