Re: (unverified) Ada mandate cancelled (Greg A would be proud)

[dewar@merv.cs.nyu.edu (Robert Dewar)]

Emmet Paige says

<<I THINK ADA WILL COMPETE BETTER WITHOUT THE MANDATE>>

Robert Dewar comments

I hope that everyone realizes that my post making the point about
"preferred technology" agrees with this position. The trouble with
an absolute mandate, even in the area where it makes best sense, i.e.
war-fighting systems in the narrow sense, is that it generates antagonism,
and worse, it sends a message that Ad is not suitable in other areas.

Consider for example the isue of reengineering of large scale COBOL
applications. Now some simple applications can be reasonably approached
using high level tools, but large complex applications need to be
reprogrammed in an appropriate technology. To me Ada is clearly a preferred
technology for such reengineering compared to C++ or COBOL or Smalltalk.
But I would not for a moment suggest mandating its use, because that would
have a negative effect. On the other hand, I would not want to send a signal
that it was inappropriate because the DoD has determined that Ada is only
useful for warfighting software.

I actually think the NRC report is somewhat flawed in this area. It swallows
hook, line, and sinker, the salesman's bogus pitch that 4GL's can be used
for all major informatoin systems development, a fallacy that has been
understood in IS circles for some time. Yes, the 4GL tools are very useful
for simple applications, but are by no means panaceas. The report makes
me wonder whether anyone on the panel was really familiar first hand with
the use of 4GL's. If so, they sure have an unusual view.

So reacting to the NRC report with a narrowed mandate might, as Mr. Paige
suggests, be counter productive to the goal of making the best possible
use of Ada technology in the DoD, not to mention outside the DoD.

What is important to me in the formulation of any new replacement policy
is that it be one that is meaningful. If there is a statement that Ada is
a preferred technology, then this should have some teeth behind it that
are relected in procurement policies, both for hardware and for softwar.e
The GSA model, formulated many years ago, that the government could only
acquire computers with validated COBOL technology is an example of such
a preference in action -- this rule did not mandate the use of COBOL, but
it did make sure that no one was in a position of not being able to choose
COBOL when it *was* the best tool, because of finding that they were forced
to work on hardware that did not support COBOL.

The hardware vendors, with some notable exceptions, are not supporting
Ada 95 to the extent that they supported Ada 83. This reflects their
belief that the DoD is not really serious in preferring Ada. Even a
few instances of large hardware orders being even partially decided
by the level of Ada support would have a salutory effect!

Similarly, even a few cases in which competitive contracts for software
systems were awarded even partly on the basis of preferring Ada to other
less preferred non-validated technologies would have a significant effect.

I got a call about a year ago from a project director of a large DoD
project, who was looking for an Ada to C++ translator. He told me that
they were an Ada shop, and were extremely upset to have to convert to
C++, but that "the general in charge of the project" was insisting that
the code be written in C++, and would not allow Ada to be used. This kind
of case (I don't think it is unique), not only makes a mockery of the
idea of a mandate, but also is the antithesis of a policy of preferring
the use of Ada.

When I related this story to several people, they were angry that the
general in question was "breaking the law". But that is not my main concern
in such a case. My concern is that here is a case where the technical
people involved have made a judgment that Ada is the best tool, and that
judgment was being overridden for non-technical reasons with no substance.
This means that in such a case, DoD is ending up with inferior systems and
not taking advantage of its investment in new technology. It is that
situation that is most urgent to avoid.


Robert Dewar
Ada Core Technologies

Note: I am speaking on behalf of both myself and Ada Core Technologies, but
I believe that the interests here go far beyond those of any one company,
and indeed far beyond the interests of the entire Ada vendor and user
community. What is important here is that all DoD systems (not just
war-fighting systems) are in a position to take maximum advantage of
the relatively modest, but highly productive, investment that DoD has
made in the development of the Ada 95 technnology.

This statement may be quoted anywhere in its entirety.