MPA-Ada. Again

MPA-Ada. Again

[Richard Riehle]

Caution: This is a rather long non-technical essay devoted to my personal.
         opinion on the captioned topic. You may want to skip it for some
         other time, or delete it all together.

It is my occasional good fortune to enjoy the brief company of people
who make decsions about the future of their coporate enterprise. This
is purely an accident deriving from my visibility as a published
scribbler, and it sometimes affords me an opportunity to buttonhole the
startled decision-maker and sometimes to be buttonholed myself.

In a recent such encounter with a industry official whose company has
relied heavily on defense appropriations, the subject of leveraging
existing corporate software resources came up.  Since I am rarely bereft
of an opinion on such things, I shared my views.

To my astonishment, the response was something like, "But who can we find
to fund the development?"  My reply was,  "You take a corporate risk and
fund it yourselves. Then you sell the product on the open market."  This
is an odd concept for people who have lived their entire lives in the
build-it-only-if-funded world.

Where is this going, Richard?

On this forum, last month, the subject of Fortran and massively-parallel
computers (MPA) enagaged the attention of some readers.  The question was
whether Ada had a chance to make a difference in the world of massively-
parallel architectures.  Many correspondents to this forum believe Ada
would be perfect for MPA, even though MPA-Fortran is already in place.

Now, HP has further legitimized MPA with a brand new system. It was
announced today.  Looks like a good architecture.

This is a perfect opportunity for a Lockheed, TRW, Loral, or whoever is
left in the MDE (Military-Defense Establishment), to invest in a project
to build an MPA-Ada compiler for the new HP architecture. As of right now,
there is no programming language better suited to MPA, and such a product
would have the multiple benefits of,

            1) providing a unique commercial software opportunity
               for a large investor such as one of the MDE coporations,

            2) put a fair number of software developers to work on some
               really interesting projects,

            3) create a superior environment for MPA software development
               because nothing comes close to Ada for this domain,

            4) open another avenue of commercialization for Ada

I suspect that MPA is unique enough that this is not just a simple port
of GNAT.  Oh, GNAT might be part of the underlying technology, but it is
going to require a whole set of supporting tools.

Lest anyone think this is a small market, think again.  The potential is
huge.  The kinds of problems that can only be solved using something akin
to MPA is growing.  The constraint on using it for more problems is the
lack of programming environments and tools.

Here is one of those opportunities which can be most effectively tackled
by someone with the economic resources to make it a reality.  TRW,
Lockheed, Westinghouse, United Defense, Loral, Grumman, and brother
MDE'ers ...   any one of them could make the investment and reap the
future rewards.  Of course, there is always the question, "But who is
going to provide the funding?"

Richard Riehle

Re: MPA-Ada. Again

[Ken Garlington]

Richard Riehle wrote:
> 
> In a recent such encounter with a industry official whose company has
> relied heavily on defense appropriations, the subject of leveraging
> existing corporate software resources came up.  Since I am rarely bereft
> of an opinion on such things, I shared my views.
> 
> To my astonishment, the response was something like, "But who can we find
> to fund the development?"  My reply was,  "You take a corporate risk and
> fund it yourselves. Then you sell the product on the open market."  This
> is an odd concept for people who have lived their entire lives in the
> build-it-only-if-funded world.

A couple of comments:

1. There's a lot of examples of companies over-diversifying and killing themselves.
   Why would the average defense contractor want to diversify into the business
   of selling Ada compilers? How is that part of their core competencies?

   (Granted, TI did it, but only to support their existing core competency of
   building DSPs.)

2. The claim is that there is a large potential market for this product. If there's
   a business case that support this, then there should be no problem getting
   venture capital. Why should a defense contractor get into the venture capital
   business?

3. Doesn't appealing to the defense sector to fund new Ada markets undercut the
   claim of Ada's viability outside that sector?

My company paid for the development of new Ada compilers in the early 80's. I like
the COTS model much better.

-- 
LMTAS - "Our Brand Means Quality"