Re: AW: AW: program won't let go of file

[Joep Beckeringh <joep.beckeringh@xxxxxxxxxx>]

 David,

Overrides are always the biggest hurdle when moving from OPM to ILE, and 
even more when combining the two. The problem is that the default scope 
for overrides changes: in the default activation group (OPM) it is the 
call level, in other activation groups it is the activation group.

In your example:

CLP1
	OVRDBF FILEA FILEB
	PGM
	CALL ILE1
		CALL ILE2
			CALL ILE3

 				etc, etc, with further calls to OPM's
	RCLRSC
	ENDPGM

the override is visible to all programs, as long as CLP1 is in the call stack. When you change CLP1 to CLLE1 with ACTGRP(*NEW), the override is visible for all programs that use ACTGRP(*CALLER), but not for the OPM programs. However, that can be change by changing the override to:
OVRDBF FILE(FILEA) TOFILE(FILEB) OVRSCOPE(*CALLLVL). That changes the behaviour of OVRDBF back to the way it worked in OPM.

So, if you change CLPs to CLLEs, add OVRSCOP(*CALLLVL) to all your overrides, and you should be fine.

Joep Beckeringh



Op 09-09-10 11:13, David FOXWELL schreef:
> > -----Message d'origine-----
> > De : rpg400-l-bounces@xxxxxxxxxxxx
> > [mailto:rpg400-l-bounces@xxxxxxxxxxxx] De la part de Joep Beckeringh
> > I think your best bet for now is to change CLP1 to ILE, with
> > ACTGRP(*NEW). Leave the RCLRSC (until you get rid of your
> > OPMs). This way, when CLP1 (or CLLE1, as would be more
> > appropriate) ends, the OPMs are ended by RCLRSC and the ILEs
> > are ended because the activation group ends.
> Joep, I already tried that. The OVRDBF issued by the CLLE1 was no longer active for the first OPM which bombed.
> We've fixed the problem by stopping the program in question from attempting to use the file.
> However, I've learned on this thread that we should have used DFTACTGRP(*YES) on the converted programs and then the program would have been disactivated after RCLRSC. The fact that these converted programs are widely used ( with RCLRSC, ie, expected to act like OPM's) means that other errors of the same kind are probable.