Pat Rogers wrote:

> > The problem is one of job security:
> >
> > Do the project in Ada, it takes 30 people 2 years, then 1 person to
> > maintain.
> >
> > Do it in C++, it takes 30 people 18 months, then 20 people to maintain.

Not my experience.  Doing it in C++ takes longer, because one is tempted to,
(and often does), ignore such items like correctness and completeness of
interfaces,
the range of the data, the when-if-data-valid which discrimiated records handle so
well.

1) On one "real-time project" (response < 1/3 second) I know about, it has now
been
implemented 4 times in 4 years.  It is over 200,000 SLOC.  Every implementation
has performance, correctness, and reliability problems.  Every implementation has
been
plagued with meory-leaks, correctness under concurrency, and exception usage.
It is neither thread-safe nor exception-safe.  After 4 "failures" one would think
they
would be ready to try something more reliable.  They were.  So, the next one was
done in Java, and had even more sever performacne problems.  Didn't solve the
concurrency or excpetion problems either.  Sigh. Some people will not start
measuring or believe those who do.

2)  A second project I worked on in Ada has great performance, runs in > 10
computers,
is highly reliable, had over 40% reuse of code from previous projects, and takes
very
little maintenace.   I'm no longer there.   It is so reliable it needs only a few
people to
maintain.  Freed up all the regular employees to build other software; consultants
like
me went elsewhere.

Cheers, ..Paul S.



>
>
> <begin rant>
> No, no, no!  This idea that it generally/usually/always takes longer to
> develop the initial product in Ada -- i.e. that productivity is lower in
> Ada -- must not continue to be promulgated!   Experienced Ada people can blow
> away the competition in productivity.  Certainly in the old days when
> compilers were poor and the developers new to the language Ada would take
> longer.  Those days are long gone.  Equally certain is that there are projects
> in which so much is already available in the form of API's that Ada won't have
> an advantage.  But I strongly object to this idea that it is inherently so!
> <end rant>
>
> Your point on the maintainability of Ada is well taken, as is the problem of
> the job market.
>
> ---
> Patrick Rogers                      Consulting and Training in:
> http://www.classwide.com      Deadline Schedulability Analysis
> [log in to unmask]        Software Fault Tolerance
> (281)648-3165                       Real-Time/OO Languages
>
> Adam ... does not deserve all the credit; much is due to Eve, the first woman,
> and Satan, the first consultant.
> Mark Twain