Re: Ada-95 Success Stories

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

Mark said

"OK, OK, so maybe I was too quick in asking for Ada-95 success stories.
After reviewing the ACVC status for the various compilers from AdaIC,
I gather no compiler is ready for heavy duty software development.  Be
that as it may, I'd really like to see early Ada-95 experiences from
the trenches posted here.
"

Oh dear! More unwarranted reliance on validation. I don't know quite how
you are drawing your conclusions from the validated compiler list, but
there is no way from this list that you could tell whether or not a
given compiler is "ready for heavy duty software development". In fact
quite a bit of such development with Ada 95 is going on using GNAT, and
I expect that other Ada 95 compilers, e.g. the Patriot 2 compiler, is
also in serious use. In the case of GNAT, there are a number of successfully
completed serious projects.

But one thing is for sure -- you cannot judge from the validation state
whether or not a compiler is usable. You can have non-validated compilers
that are very much usable, and validated compilers that are completely
unusable. 

To understand this, note the following two possibilities:

A compiler that has been tweaked to get by the validation tests, and does
indeed implement pretty much all of the language, but is still completely
unreliable for large programs.

A compiler that is highly reliable for large programs in practice, but
lacks a couple of non-critical obscure features, not needed by these lare
programs, but significant enough to shoot down a whole series of ACVC tests.

The only way you can really tell whether a compiler is reliable for your
particular application domain is to try it out!