On Sep 16, 2013, at 12:22 AM, Timothy Sipples wrote:
Ed Gould asks:
What is the cost benefit to do the conversion?
I guess you missed what I wrote about improved code efficiency.
Another big
benefit for many/most shops is significant constraint relief for
each and
every data division (~16-fold increase) and data item (~8-fold
increase).
IBM's announcement letter for Enterprise COBOL 5.1 has more
information on
benefits:
http://www.ibm.com/common/ssi/rep_ca/4/897/ENUS213-144/ENUS213-144.PDF
By the way, isn't IBM-MAIN the forum where, for years, I've/we've
slogged
through complaints (hundreds?) about the COBOL compiler not taking
advantage of newer, more efficient zEnterprise instructions?
Enterprise
COBOL 5.1 does, all the way up to and including zEC12/zBC12
instruction
sets as you wish.
I was wishing for support for large tables and as for instruction
complaints, I do not recall as me being the complainer(its a can of
worms that one has to be careful asking for as the issue of downward
compatibility is always a huge bottomless pit.. AFAIK the binder
needing PDSE has nothing to to do with 'enhanced" instruction usage.
What's XY% improvement in code efficiency worth? "More than zero,
probably
much more than zero" is the rational answer. Are XY% efficiency
improvement
(more to come), immediate constraint relief, a clear technical
option to
64-bit, and other benefits worth a PDSE prerequisite, a
prerequisite which
many customers have already satisfied? That was the technical
choice IBM
faced together with customers through a lot of requirements
gathering and a
long development and testing process (including customer programs)
for the
new compiler. Did IBM make the right decision, or should IBM not have
improved and advanced COBOL? That really was the choice.
I really, really think IBM made the right call here for customers.
So let's
figure out how to get to PDSEs as quickly, easily, and cost-
effectively as
possible with minimum risk. Some posters in this thread have some good
ideas. IBM also took the opportunity to provide a new 90-day trial
program
so that you can technically validate your particular business cases
with
greater precision.
The 90 day trial program (to me) sounds like a shady car dealer from
the wrong side of town.
Another question: Why is it that z/OS customers running Java have
had no
particular problems putting their Java code in PDSEs for so many
years,
running lots of mission-critical batch and online applications?
What makes
them different? "Not much" is the honest answer, isn't it? They
write code,
they test it, they deploy it, and they run it, successfully. Or, if
"not
much" is not the answer, we really don't like the other answer (John
Gilmore's?), do we?
Simply put, we *DON'T* have such animals. Nor will we in the
foreseeable future.
*EVEN* if we did have need for them currently the issue of cross
system use would be a major stumbling block (as others have noted).
Then there is the irony of having a product that needs a PDSE for
output that currently resides in a simple PDS (I could use another
term but I am being polite).
I agree with Tom Marchant by the way. All operating systems
bootstrap, and
all of them must. Mac OS X 10.7, for example, introduced so-called
"full
disk encryption" (FileVault 2). That's nice, except it's not
"full": Macs
still need a readable (i.e. unencrypted) skinny boot partition to get
everything going. Maybe someday Apple will shove that unencrypted boot
partition completely into firmware, but no matter where it comes
from it's
still the same boot sequence. If there's a useful reason PDSEs (or
DB2 as
another example) ought to be available earlier in the z/OS IPL
sequence,
let IBM know through the appropriate channels. Enterprise COBOL 5.1
isn't
one of those reasons as far as I can see.
My views are my own here. My views sometimes change upon
consideration of
new evidence. They are not necessarily those of my employer or my
dentist.
If I happen not to repeat this reminder, it still applies.
Tim,
I used to be a big believer in IBM, but between COBOL (maybe the base
ZOS is also an issue) that it cannot run in true 64 bit mode cannot
be done for a reasonable reason and now requiring a PDSE as output is
stretching anyones credibility.
Now if you were say to us that it is a stepping stone, I might buy
off on it except no such statement has been made nor even hinted at.
AND until IBM can within the GUTS of Z/OS can distribute a PDSE so
that the base OS can use it (I will hold off on the IPL library issue
as it is a major issue and IBM can't seem to even get a grip on it we
are at a stale mate).
AT the minimum IBM has to address the cross plex issues with PDSE
before demanding it for COBOL (and other products).
Ed
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
