🚀 go-pugleaf

In my opinion a properly designed software stack can have drastic
_functional_ advantages over the straight Getmain/Freemain approach. For
example, it may enable "persistent" working storage, i.e. such working
storage that survives accross the calls. Isn't it handy? And oh, by the
way, the CPU time goes down too, especially for the subroutines called in a
loop.

By replacing DFHEISTG [the standard working storage obtained/released by
the CICS prolog/epilog] with the above "persistent" working storage chained
off the TWA, we had a sizable CPU reduction in our CICS transactions that,
by the business necessity call the same module up to 1000 times per task.
We are now exploring the same method to enable "persistent" LINKAGE SECTION
items for the COBOL subroutines.

-Victor-

On Mon, 17 Oct 2005 12:12:23 EDT, Bill Fairchild <DASDBill2@aol.com> wrote:

>In a message dated 10/1/2005 9:30:17 A.M. Central Daylight Time,
>bob.shannon@rocketsoftware.com writes:
>
>... establishing a software stack for saveareas/workareas that will
eliminate the need >for Getmains...

I think this is a solution in search of a problem.  22 years  ago I wrote an
ASM program that used GETMAIN to build a very large number of  fairly small
control blocks.  Thinking that there was too much overhead  using the
ancient SVC form of GETMAIN (this was before the PC form was  invented), I
changed the code to do a single huge GETMAIN at the beginning and  then do
my own suballocation.  I ran some benchmarks and found a  negligible
difference in CPU time.
<snipped>
Bill Fairchild