Re: Govern.Comp.News editorial says "Drop Ada Mandate"

[munck@STARS.Reston.Paramax.COM]

>From the August 2, 1993 issue of Informationweek, page 6.

>The technical arguments can go on forever.  But economical,
>commercial alternatives to Ada exist for building secure,
>reusable, well-documented code and doing the development in
>a measurable, disciplined environment.  Ada isn't a bad
>language, but events have simply passed it by.

Any thoughts on what these alternatives might be?  Although I
do believe that it's _possible_ to write "secure, reusable,
well-documented code" in _any_ language including FORTH, CMS-2,
and APL, it's clearly harder in some languages than others; C
and C++ make it harder in each category to the point that it is
unreasonable to expect the typical DoD contractor programmer to
do so.

Therefore C and C++ are NOT possible alternatives.  What is?
Modula-2?  Pascal?  TURBO-Pascal?  Of these I can only make a
case for M2, which is currently much less widely used than Ada.
So what's the argument against Ada?


>There's a message in the fact that Ada has never developed a
>commercial following.  Why should DoD be saddled with it,
>especially when the military is trying to get rid of many
>other military specifications?

At CASE'93, the wrap-up panel was titled something like "Is
CASE a Failure?"  The use of CASE tools and environments may be
about as wide-spread as the use of Ada (and may be 50% or more
the same users).  Clearly, however, the lack of popularity of
CASE is not caused by the unpopularity of Ada; they may both
have the same cause.

It is my opinion that this cause is the short-sightedness that
is increasingly built into American commercial management.  Ada
and CASE are not used because of a (real or perceived) upfront
cost that is not balanced against the long-term benefits.  In
fact, the world of PC software is so volatile that most of the
long-term benefits may actually never be realized; a company
will either fail or grow so quickly that it can afford to
devote inordinate resources to maintenance and enhancement. 
(Of course, nothing lasts forever.)

Although they sometimes do, the US Government and the DoD in
particular cannot work this way.  The success of a DoD software
project is not based on the ratio of sales to development cost,
but rather on usability and life-cycle cost.  "Usability" often
includes not causing loss of human life.  Given this huge
difference, I don't see how a case can be made for the DoD
using C just because commercial companies writing things like
computer games do.  It's the equivalent of saying that the Air
Force shouldn't have supersonic airplanes because no commercial
American airline does.

(BTW, I personally put much of the blame for industry's lousy
long-term management on Ronald Reagan and the Harvard Business
School.  Some may argue with that, but please not in Info-Ada
or c.l.a).

Bob Munck