One "IBM Truism" that this exchange points out is that each IBM product
group is
very independent of other groups, even down to the individual product level
in
many cases. If developing a new "xyz compiler" would only sell 200 new
licenses of that compiler, it does not matter if it also enables the sale of
100 new zVM licenses and 100 additional z196 machines. The compiler
developers get no credit for "enabling" the sale of those other products,
their business case has to live or die based on their 200 licenses, even if
the
indirect benefit is much larger.
Also, we have to remember that many of the development tools are provided
to PartnerWorld for Development members at no cost. Developers producing
software for sale are eligible for this free software and I suspect this is
a
significant part of the potential compiler market. While software product
groups are frequently measured on licenses, they are also measured on
$$ revenue, and these "free" licenses bring in no revenue. So again, a new
compiler might result in new business for IBM, but since the individual
product team does not get any of this credit, the business case evaporates.
Mike Hammock
----- Original Message -----
From: "Alan Altmark" <alan_altm...@us.ibm.com>
To: <IBMVM@LISTSERV.UARK.EDU>
Sent: Thursday, December 16, 2010 11:41 PM
Subject: Re: Mandatory ESMs?
On Thursday, 12/16/2010 at 04:32 EST, Dave Jones <d...@vsoft-software.com>
wrote:
So what you are saying is that the only interest folks might have in
using modern compilers on CMS is to write "business" applications and
nothing else? Remember that IBM first sold us on PL/I as an all-purpose
language, one that could be used for systems programming applications as
well as business ones.....so I don't see thjis as an issue of "nobody
writes business applications anymore on CMS, so we don't need to provide
the compilers....".
No. What I'm saying that compiler developers are interested in what
business application developers want and need. They also take an interest
in systems application developers to the extent it is a nit on top of the
business apps. (Thinking about the METAL option on z/OS C/C++.)
Remember, the *customer* doesn't need the compiler, the developer does. If
you have only a handful of developers....
Yes, we know what's available for development work on CMS, but surely
you're not saying we should not ask for more tools.....or that we should
simply sit down, shut up and be happy with whatever IBM thinks we need?
It is certainly true that there are a relatively few of us interested in
developing such software but I believe that we make the overall z/VM
environment more attractive to potential customers, and thus fill an
important role in the zSeries ecosystem.
You should never "sit down and shut up" if your needs aren't being met.
Using your knowledge of the systems and business application development
situation on CMS, explain to the IBM PL/I compiler owners how many
licenses they can expect. Things a product owner has to consider:
- It costs money to create a product, even if it's "just a port".
Packaging is certainly *NOT* a port!
- Do there need to be any changes to deal with CMS vs. original platform?
Usually there are considerations for file specification.
- After investing in the needed additional skills and in the crank-turning
needed to release a product, will they have a good E/R ratio?
- It is a growing business? Or am I going to eventually have to get rid
of the resources I need to acquire?
- How does it compare to other investment opportunities?
- Are there any overriding issues?
and it's not a case of "the wishes of the Few or the One are ignored in
favor of the wishes of the many".....it's more of a case of out of, say,
100 VM advocates, 4 want IBM to port PL/I to CMS and the other 96 simply
don't care, and not that the other 96 are actively against it.
This isn't about advocacy. I don't know anyone who thinks you SHOULDN'T
have the compiler of your choice. The only question is whether compiler
providers can make enough money doing it. If those other 96 aren't going
to buy the compiler, they don't have standing in the Court of Economics.
...but we all know that there is no "one size fits
all" in such software and sites will continue to tweak their
capabilities with site-specific modifications (exits, glue routines,
etc.). All I want is IBM to add one more tool to VM's kit to aid those
sites.
I don't see people doing that. They want exits to drive REXX routines or
more knobs, not linked-in HLL or assembler modules requiring skills they
haven't got. And they don't want to spend money on compilers just for
exits.
BTW, all of the talk about not being able to create a "business case"
for the port is a bit ironic (again, imho) given that the most important
advances in VM over the years come from tools developed withOUT having a
hard-nosed $$$-focused business case first: 1) CMs Pipelines, 2) rexx,
3) RSK, and 4) even the Linux port. The point being that clearly tool
development can happen in the VM environment without always needing a
business case being made first.
Ahhh, the Good Ol' Days. Don't get hung up on using IBM's 1970s/1980s-era
business decisions as a model for today. Back then, cash oozed from
every pore and business cases were a lot weaker. It's a very different
environment today.
But none of this is news to anyone. It's been this way for 20 years now.
I wish things were different in the CMS world, really I do. But I think
Rick is right: If you can't find a supplier for the compiler of your
choice, look at open source. Of course, even that isn't headache-free.
Alan Altmark
z/VM and Linux on System z Consultant
IBM System Lab Services and Training
ibm.com/systems/services/labservices
office: 607.429.3323
alan_altm...@us.ibm.com
IBM Endicott
