Re: Ada The Best Language?

["John R. Strohm" <strohm@airmail.net>]

Marin David Condic answered you, but you apparently didn't completely
understand his answer.

Taking "developer productivity" as your metric of interest basically
destroys the ability to conduct a meaningful language comparison.  If you go
back far enough, to studies that may well have been conducted before you
were born, you will find the interesting result that programmers, on
average, produce, over the life of a development project, ten lines of code
per day, and (here is the interesting part) that number is INDEPENDENT of
the language used.  This discovery went a long way to push the move from
assembly language to higher-order languages in the 1960s.

When you take "developer productivity" as your metric, you basically ignore
lifecycle costs.

Ada was designed to minimize lifecycle costs, at the POTENTIAL risk of
slightly higher initial development costs.  That risk did not really
materialize: development costs using Ada are typically not significantly
higher than development costs using C++.  Where Ada shines, as Condic
observes, is that the total lifecycle costs, the ongoing support costs, the
bugfinding and fixing costs in the out years, are MUCH lower than with
C/C++.  This is an issue that some commercial software companies
fundamentally choose to ignore: once the product is sold to the end loser,
it isn't their problem any more, as long as the buglist is not so egregious
as to turn the loser off forever.  (Even then, if the company enjoys an
effective monopoly on that product, it doesn't matter HOW bad the buglist
is: the loser is stuck.)

With military embedded systems, on the other hand, this isn't true.  The
company doesn't just do the initial development.  They also do the out-year
support and upgrade.  That system may well have a life exceeding 30 years,
and those software systems will still be needed.  The B-52 first flew in the
1950s, as I recall, and is still the mainstay of the fleet.  The F-16, the
first truly software-intensive bird, first flew in the late 1970s, and is
still on the front lines 20+ years later.  For systems like that, initial
development costs, which you are measuring under "developer productivity",
vanish when compared to total lifecycle costs, which you are by definition
not measuring.

C++ (and C, for that matter) take the viewpoint of "trust the programmer"
and allow the programmer to do whatever he wants, EVERY TIME, on the
assumption that once in a while the programmer needs to do something weird.
Ada aims at reducing bug generation and propagation, by attempting to catch
the bugs that are easy to commit and sometimes hard to track down.  Ada
allows the programmer to do whatever he wants, if he wants it bad enough,
but MAKES him go through some contortions if he is apparently doing
something that historically is known to cause problems.  These contortions
both force the programmer to think about it, and, much more importantly in
Ada's philosophy, warn the maintenance programmer TEN YEARS LATER that there
is heavy-duty black magic lurking in those contorted, tortured lines of
code.  (I've done low-level device drivers in Ada, including machine code
insertions as defined in the LRM.  I know what kind of contortions I had to
pull to do them.  They forced me to think CAREFULLY about the problem and
how to solve it.)

codesavvy <codesavvy@aol.com> wrote in message
news:5be89e2f.0107191355.534211d0@posting.google.com...
> I can't believe some of the posters in this thread.  Again I said that
> Ada 95 has nothing to offer that is substantially better than C++.  If
> there was then logically higher productivity would result.  I can't
> believe that you have a difficult time grasping this concept.  Some
> posters have been kind enough to provide me with anecdotal data and in
> one case a link to some data that Capers Jones has which I appreciate.
>
> Kilgallen@eisner.decus.org.nospam (Larry Kilgallen) wrote in message
news:<X+JoNS25futh@eisner.encompasserve.org>...
> > In article <5be89e2f.0107181237.4ab3594@posting.google.com>,
codesavvy@aol.com (codesavvy) writes:
> > > "Marin David Condic" <marin.condic.auntie.spam@pacemicro.com> wrote in
message news:<9j46bt$3qj$1@nh.pace.co.uk>...
> > >> Well, if Ada has nothing to offer and this is obvious then why are
you
> > >> bothering to a) read this newsgroup and b) post to it at all?
> > >>
> > >
> > > Here is what I wrote:
> > >
> > > I think the answer is rather obvious, Ada has nothing to offer that is
> > > substantially better than what C++ offers.
> > >
> > > I didn't say that Ada had nothing to offer, just nothing that is
> > > substantially better than C++.
> > >
> > >> If you have a serious question about Ada and its potential benefits
and are
> > >> willing to entertain the possibility that maybe Ada *is* a better
choice
> > >> than C++, then we will be more than happy to point to resources for
you or
> > >> help you learn the language or answer questions about the language.
But a
> > >> blanket statement that seems to be saying "Ada is s**t! Why are you
guys
> > >> bothering???" seems more calculated to start a flame war than to get
a
> > >> serious question answered.
> > >>
> > >
> > > You must have missed my other post where I stated that Ada 95 is an
> > > excellent language and may actually be better than C++.  However, I
> > > don't feel that developer productivity is significantly enhanced with
> > > Ada 95 as opposed to C++.
> >
> > You started with "nothing to offer", but seem to have devolved into
> > "no productivity advantage to offer".  The reason I use Ada is not
> > productivity, but correctness.