Re: Compiling SQLRPG programs in BPCS - one sentence corrected

Correction to one sentence to avoid confusion....this sentence should have
read as follows (corrected word shown with << >>):

"Additionally, I would think that CLOSQLCUR should be *ENDPGM unless you
have yourselves
completely verified that cursor handling is done properly for <<*ENDJOB>>
processing (believe me, it was NOT done properly in early BPCS releases!)."

Sorry about that...

Genyphyr Novak
SSA GT
----- Original Message -----
From: "Genyphyr Novak" <novakg@ssax.com>
To: <bpcs-l@midrange.com>
Sent: Wednesday, October 16, 2002 5:38 PM
Subject: Re: Compiling SQLRPG programs in BPCS


> Hello,
>
> What kind of errors did you receive? I would do a DSPPGM of the new object
> and compare all the compile settings to a prior object version (if you
have
> one on the system or can restore one, or even view a similar one that does
> work) and then recompile accordingly.
>
> I do not know what the original default compile settings were for BPCS
5.1,
> however as already suggested, commitment control should be *NONE (you'll
get
> file open errors in the joblog if this is set differently). Additionally,
I
> would think that CLOSQLCUR should be *ENDPGM unless you have yourselves
> completely verified that cursor handling is done properly for *ENDPGM
> processing (believe me, it was NOT done properly in early BPCS releases!).
> The value of *ENDJOB on this parm often is tried to improve performance,
but
> if cursors are not properly opened/closed as the job runs, you will find
> this can result in various and sundry file positioning errors on cursors
> left in a given position in the file from a prior read when you re-enter
the
> program. This will bring the infamous 'unpredictable results' when running
> said program.
>
> Later BPCS versions were fixed so that cursors are always handled properly
> and were delivered with the CLOSQLCUR *ENDJOB setting for better
performance
> (starting at V6100 and up). In V6004 a selected set of programs (selected
on
> basis of bad performance and with cursors that were good candidates to
leave
> 'soft' closed for the run of the job) were reviewed, fixed, tested and are
> now also delivered with this compile setting as default from SSA GT -- but
> other programs in that release should not be compiled with that setting
> (unless you test/mod/etc. yourself to ensure proper cursor handling).
>
> Hope this helps.
>
> Thanks,
>
> Genyphyr Novak
> SSA GT
>