Importance of Ada in DARPA, other federal R&D

Importance of Ada in DARPA, other federal R&D

[Bruce Weide]

There has been no shortage of discussion here about the role of Ada in
the defense and commercial software development sectors.  Does anyone
have any opinion -- or even better, knowledge -- about the importance
of Ada in the R&D plans of DARPA (apparently again to be called ARPA)
and/or other federal agencies?  Is it important to them?  As a CS
researcher, this question is almost as important to me as Ada's
position in the software development community.

Hoping to see some (but not too much!) discussion...

Cheers,
    -Bruce

Re: Importance of Ada in DARPA, other federal R&D

[Gregory Aharonian]

>There has been no shortage of discussion here about the role of Ada in
>the defense and commercial software development sectors.  Does anyone
>have any opinion -- or even better, knowledge -- about the importance
>of Ada in the R&D plans of DARPA (apparently again to be called ARPA)
>and/or other federal agencies?  Is it important to them?  As a CS
>researcher, this question is almost as important to me as Ada's
>position in the software development community.

   As much as I have a low regard for DARPA's technology transfer capabilities,
(after all when you spend over a billion dollars funding technical software
development, but don't have a directory of where all this taxpayer funded
stuff is (unlike Japan's Fifth Generation project), you're incompetent), I
do have a high regard for the software they have had developed at their
direction. 
    And as DARPA tends to be interested in getting something that works,
as opposed to being politically correct, Ada does not tend to be a big
consideration.   Considering that much of the advanced software that
DARPA develops: parallel processing, artificial intelligence, sensor
fusion, distributed computing, CAD/CAM/CAE/, etc, is all done by software
communities wholly disinterested in Ada (most of whom associate, maybe
prophetically, Ada with the American Disability Act), most of the software
that DARPA funds is not written in Ada, and if forced to do so, would
cripple these efforts to meet DoD high tech needs.
    So just by a surface examination of the results of DARPA funding, I
doubt pushing Ada is big on their agenda, especially when a few of them
went on record (as reported in Government Computer News last fall) as
distancing themselves from Ada.
    Most state of the art in the United States for software and software
related stuff (such as imaging through the atmosphere) is done by people
using Fortran or C/C++ in communities that have for the most part rejected
Ada.

    In fact, when Ada first hit the scene, I predicted that Ada would never
gain much acceptance in these communities, simply because these peoples
rely on hugh libraries of existing Fortran and C codes to do their studies,
and at the time, the DoD (including DARPA) had no plans to convert these
libraries over to Ada to encourage people to switch languages, since no
matter how fantastically, incredibly, postively great Ada is or could be,
these people won't switch because they need there existing large libraries
of reusable code (like I said, the DoD is years behind some of these groups
in large scale software reuse).

    At the time, some argued that converting these libraries (many of which
I collect, totalling over ten million lines of code a billion dollar cost
to convert at DoD contractor rates) should be the role of the private
sector that I so champion here.  And in fact, I was involved with three
ventures to start such a business going, all three of which folded because
of lack of business.  Ada software reuse as a business has always been an
impossible idea because of the losuy economics - trust me - I have lost
enough of MY money trying.

    Too much of what DARPA needs depends on such processes for which Ada
has had no impact.  Thus there probably is little interest inside DARPA
for pushing or caring about Ada because it interferes with its basic
mission.

   Besides which, DARPA tends to go its own way from the rest of the DoD.
For example, while it's idiotic for the DoD to have two major software
reuse centers, DSRO and ASSET, it's even more idiotic for DARPA to start
up a DoD software reuse center at VCOE (Virginia Center of Excellence).
I mean, do you guys hate each other, or what?  Let' see the Pentagon has
a reuse center (DSRO), the Air Force has a reuse center (ASSET), the
Army has a reuse effort (SIMTEL) and DARPA has a reuse center (VCOE).
That leaves the Navy - hey, I am available for adopting, and I come much
cheaper.

   As a CS researcher, don't worry about Ada.  By the time you ramp upp
you own efforts to incorporate Ada into your activities over the next
few years, the DoD will have dropped the Ada mandate, and the language
will disappear within a few years (to the Forth level).

Greg Aharonian
Source Translation & Optimization

"If you are reading this, you can't deny knowledge of the Internet!"
-- 
**************************************************************************
Greg Aharonian
Source Translation & Optimiztion
P.O. Box 404, Belmont, MA 02178

Re: Importance of Ada in DARPA, other federal R&D

[David Emery]

At the time Ada first came out, existing libraries of C code weren't
much more common than Ada libraries.  One of the big differences was
the availability of cheap/free C compilers, as compared to expensive
Ada compilers.  C grew in popularity due in large part to the spread
of free Unix to graduate schools.  Its popularity in the PC world
comes from all of the grad (and undergrad) students who learned C in
college, and then went into the PC world.

Hopefully Ada has learned its lesson with the GNAT Ada 9X project.
It's too bad that DOD didn't fund a free Ada compiler and distribute
it in the early-mid '80s.  Part of this was due to the fear of selling
good software to the Commies...

Incidentally, DARPA funded much of the UCB work on Berkeley Unix, so
we can "blame" DARPA for the widespread adoption of C in universities.

				dave