Re: Proving Correctness (was Java Portability)

[Leif Roar Moldskred <lrm@huldreheim.provida.no>]

nicolas <n.brunot@cadwin.com> wrote:

> Ada doesn't  have a lot of standard, cross-compiler, cross-platform
> libraries, this is in my opinion, one of the main reason why so few people
> use Ada.
> That's the subject of the dicussion.

Yes, but then neither does C or C++. I really think Java is the odd
man out here, rather than Ada. (And I wouldn't be at all surprised if
some of the 'standard libraries' for Java have problem with the ":"
file separator on Macintosh, or similar glitches.)

C and C++ certainly have a much wider selection of libraries, but how
many of those can really be said to be cross-compiler, cross-platform?
Sure, you get some that can be used across a couple of different
compilers, and a couple of different platforms - but in the context of
_standard_ libraries, very few would really be wide-reaching enough to
be valid.

> I've seen those aspects extremely highly hyped for about 12 years ...
> And if you forget them, I think Ada loose the most part of its reason to
> exist.

Well, personally I consider robustness, safety and ease of maintenance
to be more central - so I wouldn't agree that it would lose "most part
of its reason to exist."

I've no first-hand experience, so this is an honest question and not
rethoric, but how many Ada projects today have chosen Ada mainly
because of software-reuse or portability reasons, as opposed to those
who've chosen it mainly for other reasons?

> There is few demand for 2 basic reasons
> - There are very few Ada users
> - Ada users are often very interested in making their own version of already
> existing things, forgetting that goes against all their claims to justify
> Ada use.

Hopefully, the open-source movements might alleviate the second reason
some. If Ada software packages becomes easily available on the
Internet, people might get into the habit of looking before
leaping. 

Remember, languages like Java and Perl grew together with the wide
spread of the Internet, and the large, easily available code-base that
makes these languages so handy is probably a direct result of that.

> From my experience, well written Ada code portability is almost
> straightforward as long as you care about not using specific compiler
> packages or pragmas.
> The additional work is almost nothing, The benefits are enormous.
> After all, that's what Ada says about maintenance to justify its use.
> When you've tested your code against Windows and Linux, with 2 compilers or
> let's say just with Gnat public version, if you care about not using
> anything compiler specific, any problem you find is very likely a compiler
> bug.

I think that was my point, really. ;-)


-- 
Leif Roar Moldskred