EBERARD@USC-ISIF.ARPA (Edward V. Berard) (03/31/86)
I am writing in response to your message, "Not really unsigned and not
trash." While I do not expect that I can change your opinion of Ada
technology, some clarifications are in order.
Regarding your experience in Ada applications:
While there have been a good number of Ada success stories (contact
Maj. Al Kopp at the Ada Joint Program Office), there have also been a
number of failures. For example, Owen McOmber reported in February of
1985 that the SUBACS program was a failure (the Navy would be lucky,
said McOmber, if they could salvage 50,000 lines of Ada code out of an
estimated 2,000,000 lines written). McOmber went on to say that
"continuing to push Ada for Ada's sake will kill the language." However,
McOmber went on to acknowledge that, " the SUBACS failure was due to its
ambitious schedule and that it would have failed regardless of the
implementation language used."
We technologists often have a hard time differentiating between a bad
idea and a bad implementation of a good idea. Any concept, including Ada
technology, can be implemented so poorly that it appears that the
original idea was ill-conceived. Further, since technologists and their
managers often have a hard time accepting blame for any failures, it is
often convenient to blame inanimate objects (e.g., the Ada programming
language) for the failure of a project. (Remember the cliche about a
poor workman blaming his tools.)
Regarding your breaking of Ada compilers:
As recently as this year, I have had experiences with bug-ridden
FORTRAN and C compilers. If you meant to imply that only Ada compilers
have bugs, or that they are any more bug-ridden than compilers for other
languages, you have an almost impossible task. A visit to any computer
science department will turn up dozens of students who can break the
compilers on the university's compilers (for all programming
languages).
Quite a lot is demanded of Ada compilers. It has barely been three
years since the ANSI/MIL-STD 1815A was established (February 17, 1983).
It has been less than three years since the first Ada translator was
validated (April 11, 1983). Considering that short time span, I am
impressed with the state of Ada compiler technology. [As as exercise for
your students, you might have them attempt to establish and compare the
maturity of compilers for other languages, e.g., C, LISP, Pascal, and
FORTRAN, when they were as old as Ada is today.]
If the fact that certain technology originates from the federal
government, or from the Department of Defense, offends you, you may find
that your offense diminishes with time. It is no secret that the
technology associated with items such as airplanes, computers, and computer
programming languages (remember COBOL?) was either originated or
accelerated when it was deemed to be in the interest of national
security. Fortunately, in time, most of the DoD's pet technologies lose
their direct association with the DoD and are then evaluated by the
masses on their own merits. [You might find it interesting to note that
in Europe over 80% of the Ada applications are non-military, or that the
Japanese are making extensive use of Ada technology (to my knowledge,
none of it for military purposes).]
Regarding your comments about the time to compile and link an Ada
program:
This is a classic case of being unable to distinguish between a
language and its implementation. For example, the Rational Ada
environment makes such changes surprisingly fast, especially for large
programs. Further, even with other Ada systems, the compile and linking
time can be significantly reduced via such things as subunits.
Regarding the observation that Ada is " a verbose hodgepodge formed by a
committee who wanted it all but didn't know what all was available.":
Any time I hear this kind of remark, I can be fairly certain that the
originator is viewing Ada from a syntax-only perspective. Yes, if one
has little or no formal software engineering education, Ada appears to
be one of the most complex languages ever created. However, if one views
the language as a vehicle for implementing modern software engineering,
the language is very simple and straightforward. [For more of a software
engineering perspective on Ada, you might wish to contact Dr. Charles
McKay at the University of Houston , Clear Lake.]
Finally, while many people contributed to the design of the language, it
is primarily the work of one man: Jean Ichbiah (now at Alsys, S.A.).
-- Ed Berard
(301) 695 - 6960
-------