Activation Group

I usually compile programs to use either QILE or *Caller for the activation group. Never really thought about it until today so I did some further reading. I would appreciate confirmation or correction to what I think I read.


An activation group (AG) is scoped to a job. Meaning that, if two interactive jobs (A and B) both use the same ILE program(say PGMA) running in AG named, say, AG01, there are two versions of the AG and two copies of the program running (one in each AG). If job A runs a RCLACTGRP on AG01, only its AG is deleted and its version of PGMA is ended. Also, if job A simply signs off, AG01 is deleted and all of its programs are closed and any resources, such as tables, are released. But job B's AG01 and its programs are active, as well as its other resources.


The same thing would apply to a batch job. When it ends, the AG is deleted, any open programs are closed, and locks on resources are released.


From both my reading and experiences, this basic description appears to apply. The reason that I got to wondering about it is because I wrote a trigger program (which is achieving its function). But I wrote it so that (a) it is in its own activation group, and (b) never sets on Last Record. The idea being that, after the first invocation, it would still be available to the job without having to reload it. That is, even if two different processes within the job caused the trigger to fire, the second process would just be able to use the (still running?) program. But that, when the job ended (user signs off) the trigger program is shut down and tables are released.

Was it reasonable to write this trigger program this way to save invocations (time)?