Hi Emily,
I've got two RPG IV programs. RPGPGM1 executes an ovrprtf on file
PRT1 and then opens it and writes data to it. This executes fine and
no errors are generated. RPGPGM1 then calls RPGPGM2. This program
closes PRT1 and then executes a DLTOVR on PRT1.
I don't really understand this. You say RPGPGM2 closes PRT1... does it
also open it? Please keep in mind that a CLOSE in one RPG program won't
affect a file that's opened in a different program. They open separate
"paths" to that file...
As far as the DLTOVR... it sounds like you want it to delete the
override from RPGPGM1? Is that what you truly want, if so, why? In
order to accomplish THAT, you'll have to either use real activation
groups, or you'll have to use OVRSCOPE(*JOB) (which I really don't
recommend)
When you're using a *CALLLVL override scope, the overrides are scoped to
the call stack level. OPM programs can only choose between *CALLLVL and
*JOB scoping, and *CALLLVL is the default. ILE programs can also use
scoping by the activation group -- but it sounds like you're
deliberately using ACTGRP(*CALLER) to try to prevent your ILE programs
from using activation groups.
With a *CALLLVL override, RPGPGM2 can never affect the override that's
in RPGPGM1 because it's at a higher call stack level. In fact, unless
you specify SECURE(*YES) on the override, RPGPGM2's overrides will be
ignored, because RPGPGM1's override is at a lower call-stack level and
therefore takes precedence.
If you change it to OVRSCOPE(*JOB), then the DLTOVR in RPGPGM2 will
delete the override issued by RPGPGM1, and RPGPGM2's override will take
affect (and possibly affect RPGPGM1, if it doesn't issue it's own
override before re-opening the file).
Does this seem overly complex and confusing? Wouldn't it be more
convenient if overrides only affected the one program they're issued in?
Wouldn't that make things much less complex? Then USE ACTIVATION GROUPS!
When an override is scoped to the activation group, it only affects
other programs that are in the same activation group. Therefore, if
RPGPGM1 and RPGPGM2 run in separate activation groups, their overrides
won't affect each other at all. Think of how that simplifies things!
Alternately, just specify SECURE(*YES) in RPGPGM2. Then both overrides
will be in affect for RPGPGM2, but the SECURE(*YES) one will take
precedence over the one program PGM1, and therefore will probably
accomplish what you want. To me, that's a great deal more complicated,
but it should also work.
Personally, I believe in encapsulation. Programs should stand alone,
they shouldn't have to know or care how other programs work. It just
makes maintenance so much simpler when you can change the logic in one
program without having to re-test everything else in every conceivable
job-stream to make sure you didn't break the other programs!
midrange.com
