Re: MVT COBOL under MVS3.8j.

"Gary [email protected] [H390-MVS]" <[email protected]>
Newsgroups gmane.comp.emulators.hercules390.mvs
Message-ID <[email protected]>
OK, I did some "light" reading of the 
LY28-6486-2_OS_VS_COBOL_Compiler_Rel_2_PLM manual.  Fun stuff. The XREF 
listing is produced by Phase 6A (pgs 190 & 191).  Of interest is that if 
XREF was specified, then Phase 6A uses the GETALL routine in the TAMER 
to get all available storage...

Going back to GC28-6399-2.  There are 3 parameters of interest. The JCL 
REGION, COBOL SIZE and COBOL BUF.  Appendix D is of informative reading 
if you what all of the facts.  Here is the short version, COBOL SIZE is 
the maximum amount of storage to be used by the compiler.  COBOL BUF is 
the amount of storage to use from SIZE for use by I/O file buffers and 
the like.  The JCL REGION value must be 6K greater than the COBOL SIZE 
value. Appendix D has a nice long formula for calculating the necessary 
memory required by the compiler.  The key information is that the 
resident phase 00 takes 80K and the minimum buffer size is 2768 BYTES.

Peppe's TK4- values where 9M for REGION, 8192K for SIZE and 2048K for 
BUF.  There is a large difference between 2768 bytes and 2097152 bytes (2M).

I would suggest that Peppe overide the COBOL BUF to be 4096K or don't 
specify it at all and let the compiler calculate the required size for 
the buffer pool.  Using a quarter of the available REGION for buffers 
seems to be a waste.  The SIZE parm could also be bumped up to 9M - 6K 
to get a little more room.

If it still fails, then a more detailed reading of the Logic Manual and 
of memory management modules for MVS and MVT to look for differences.

Hope this helps.

Gary


On 9/12/19 6:41 PM, Gregg Levine [email protected] [H390-MVS] wrote:
>
> Hello!
> I don't know either. But there's still a rumor circulating that all of
> the tools for program building for MVS (or Z/OS or OS/390) were built
> on the matching time period of VM. And as for these problems? Two
> things to say, one is I arranged it. And the second is that of if this
> issue isn't resolved by the end of the month, then we are going to
> leave the issue as is, and from there it will be revisited by the
> person who started this, when he has the time, and report it as such.
> -----
> Gregg C Levine [email protected]
> "This signature fought the Time Wars, time and again."
>
> On Thu, Sep 12, 2019 at 6:13 PM Gary [email protected] [H390-MVS]
> <[email protected]> wrote:
> >
> > <snip>
> >
> > By the way, the manual, GC28-6399-2, page 40, specify how
> > to code the REGION statement for the compiler.
> >
> > I've never seen used the REGION statement in this way:
> >
> > REGION=(nnnnnxK[,nnnnnyK])
> >
> > It looks like the second parameter "nnnnnyK" is rather
> > important for this COBOL compiler, but I can't understand
> > why.
> >
> > May I ask for an explanation?
> >
> > > The answer you are seeking is found on the same page in the manual 
> under the heading " Specifying Main Storage Requirements for a Job 
> (Region)". The short answer is unless you have an IBM 2361 Core 
> Storage "attached" or your MVS has been "genned" for storage 
> hierarchies, then you need not worry about the second memory 
> allocation. There is a Max storage size of 16383k, 1k less than 16M.
> >
> > I don't remember who stated that the compiler was built from 
> "source" under VM. Is that a viable option for MVS?
> >
> > Regards,
> >
> > Gary
> >
> >
> > don't remember
> >
> > ________________________________
> > Posted by: Gary <[email protected]>
> > ________________________________
>
> Sign on the door of the datacenter, "Remember to trust the fish who
> runs the site!"
>
>
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.