Re: Making Triggers Resilient

["M. Lazarus" <mlazarus@xxxxxxxx>]

Scott,

 I didn't mean that the "inheritance" is global to the job.  Rather, since 
the calling program's attributes (namely activation group) are not always 
known to the trigger, it might make sense to use *CALLER, so there's never 
the extra AG creation overhead when the trigger fires.

 -mark

At 3/1/05 12:46 AM, you wrote:
> Hi Mark,
>
> > I was under the impression that triggers run in the caller's job.  That 
> > means that it will "inherit" the *LIBL and other attributes.  Following 
> > that logic, wouldn't *CALLER run in the caller's (user job) *ACTGRP, not 
> > the DB's?
>
> It would appear (based on Barbara's post) that you're correct, it does 
> inherit the activation group of the program that made the database call.
>
> I don't follow your logic. Activation groups aren't a global attribute of 
> a job. A job can contain many, many AG's, so it's not something that can 
> be inherited as part of the job.  Instead, *CALLER means that the AG is 
> inherited from the program that calls it.
>
> But, apparently triggers are the exception to that rule -- something that 
> I didn't know until now. Instead of inheriting the activation group of the 
> database manager, the system activates it in the AG of the program that 
> made the database call.
>
> I learned something new today.