Re:Re: 2 PDE copies in memory

Discussion created by ca.portal.admin on Aug 31, 2007
Last entry. Varying memory in a PDE program name (which is part of the
index) can cause lots of problems. Program not found, duplicate entries,

Cobol programs that are no longer defined as cobol etc.
So please do not vary a program name.
As Bill Allen said, it will definitely cause index corruption.
That being said, if you REALLY had to vary the name - make sure it fits
alphabetically into the index..
Ed McKinney
IDMS Public Discussion Forum


Re: [IDMSVENDOR-L] 2 PDE copies in memory
"Kay et al.
If a PDE (program) is not found, a NULL PDE is used. The sysgen
definitions are ONLY read at startup to build the pre-defined program
PDEs. If a PDE is not found, IDMS can generate a new PDE automatically.
The only time I saw someone VARY a PDE name in memory, it caused
problems because it was out of sorted order. The reasoning was that they
had defined (in sysgen) a program as NODYNAMIC and CDMSLIB and they
really wanted the program to come out of the default dictionary load
area. They could not load the program because the definition said ONLY
get it from CDMSLIB. They renamed the program PDE in memory and used
DCMT VARY DYNAMIC PROGRAM to add a new PDE. This then did not have the
NODYNAMIC CDMSLIB specified and they were able to load from dictionary
load area. I do not suggest this though. I suggest sysgen change and
You really should not mess with indexes that we do a binary search on as
it will definitely corrupt the search with bad bad results.
Ed McKinney