Re: What ada 83 compiler is *best*

[Chris Morgan <mihalis@ix.netcom.com>]

to.reply@read.my.sig (Rick Thorne) writes:

> Ada and C++/Java operate in the same domain.  They're all used for large
> system development, real-time processing, applications development, etc. 
> Which you choose depends on the development environment you want.  If you
> want to get close to the machine and control the memory usage, us C++
> because it provides incredible memory allocation capabilities. 

Yes, and incredible opportunities to shoot yourself in every reference
to a pointer to a virtual base class template of your foot. I speak as
someone with plenty of foot injuries in this area.

 If you
> want simplicity in memory allocation and want enourmous flexibility in
> portability and UI development, use Java.

Nice language, but lacks templates and still has severe performance
problems. Ada is portable via recompilation instead of interpretation,
a deliberate design decision.

>  If you really don't care about
> performance or vendor support and want to comply with obscure and obsolete
> government standards, use Ada.  Your call.

Complete twaddle.

If you compare the performance of C and Ada both compiled with gcc
based compilers you will generally see equivalent performance with
equivalent code, there is no inherent speed cost with Ada. If you
chose to turn on run-time checks with the Ada compiler you can, and
therefore pay a price in speed in return for greater reliability. The
fact that there is no equivalent facility in C, or C++ is simply a
point against those languages.

There are at least two or three Ada vendors with excellent support in
my own personal experience. There are also C++ vendors with rather
poor support unless you cough up big bucks.

As for obscure and obsolete standards, an ISO standard for a language
is a large win and not in the least obscure. It's also not a
government standard any more. When Java or C++ standards are ratified
they too become ISO standards not Sun or AT&T property.


> 
> > Of course, the computing world loves convergence to standards (C++, Windows).
> > It's a good thing.
> 
> ...and you're saying that Ada isn't compliant to a standard, and that Ada
> hasn't tried to impose its standards on the rest of the known universe? 
> If so, you haven't read much about the intent of the language you seem to
> love so much!  I think one of the reasons Ada has failed so miserably in
> commercial US software development is precisely BECAUSE it is a standard
> the government has tried to bully on us. 

The government didn't try to bully anyone into using Ada. Requiring
it's use on government contracts is hardly bullying. Additionally Ada
has standardised interfaces to other languages (at least the ones that
are themselves standardised) to encourage reuse of existing
libraries).

 C++/Java and others have
> considerable strengths of their own that make Ada unnecessary.  YES -
> unnecessary.  C++ and Java are perfect forms of protest.  They were
> developed by a handful of people (not a government bureaucracy like Ada
> was) AND they're incredible languages, whether or not YOU agree.

Ada was not developed by a bureaucracy, again you reveal your
ignorance and prejudice here as you lecture others. In both cases
(Ada83 and Ada95) the language definition was developed by small
highly-focused design teams. For example in Ada95 although there was a
large amount of consensus building and requirement gathering, there
was a strong lead from a chief architect who, in the end, had final
say.

In fact the original Ada lent some inspiration to C++, and it has
often been said that semantically Java is closer to Ada than C or C++
for obvious reasons.

> Can you possibly be implying that C++/C compilers don't find bugs, and the
> Ada somehow produces code without runtime errors by virtue of superior
> compiler technology?
> 
> To make the statement that Ada compilers - by definition and/or
> technological superiority - make Ada a virtual bug-free language is simply
> ludicrous.

True - you've built up the other poster's statement into an easy to
rebut straw man, impressing nobody. There are some factual aspects of
the difference between Ada and, say, C++ that make the range of
mistakes catchable by an Ada compiler far greater than a C++
compiler. Aliasing analysis, array bounds checking, dangling pointer
prevention are major wins, but even the simple fact of a clean
compilation model with no macro-processing helps a lot.

> Again, puh-lease.  Some of the worst code I've ever seen is Ada code.  And
> again - to make the statement that Ada by virtue of its own merits always
> produced easy to read code isn't even worth discussing.  I've been there. 
> I know too well to be patient with this argument.

Bad Ada code is possible, after all we are discussing fairly complete
languages, but good Ada code _is_ very very readable, even to a
non-Ada programmer. Good C++ is less readable to a non-C++ programmer
in my opinion. I think this has to do with it being less encoded,
there is closer to a 1-1 between word and concept. Additionally C++
has many more unpleasant warts such as the over use of the word
static, the polluted messy global namespace inherited from C etc.

> 
> There as another point in all this, sir: easy-to-read code isn't that
> answer to one-tenth of the problems that plague software development. 
> Most of the serious problems are requirements analysis and transmittal and
> software architecture & design. 

True.

> As a programming language, Ada doesn't begin to address these issues
> except in the most obtuse way.  If we've learned anything from Ada,
> we've learned that languages AREN'T at the heart of the software
> engineering crisis.

False. Ada as part of a well-controlled software engineering process
is a big win. If you don't start with a good process you will have
problems no matter what language you use.

> 
> My advice: before you make the public statement that Ada solves software
> engineering's nightmares because the compilers are great and it produces
> easy to read source code, I suggest you read "No Silver Bullet" by
> Fredrick Brooks.  Ada is just a programming language.  Languages are NOT
> at the heart of the software engineering crisis.  They are peripheral
> co-conspirators at best.

The designers of Ada work with the assumption that the basics such as
source code control/configuration management, requirements
traceability, testing etc are also addressed. Without all that stuff
too you are correct, Ada is as unhelpful as a high-performance car
where there are no roads.

> AND AGAIN - are you actually implying that there's no Ada code out there
> that wasn't years late, $$millions over budget, and virtually
> unmaintainable?  If you believe this, I suggest you read some of the GAO
> reports written in the last 10 years on this topic.  Ever wonder why the
> Ada Initiative was dropped by the DoD?  The reason is somple: Ada code
> isn't any less expensive, buggy, slow, or difficult to read than anyone
> ELSE's code.

Well the word is that the C++ zealots aren't doing any better at all
in all those projects who leaped the fence.

> Cute closure, and I'm certain the Ada worshippers are laughing with you. 
> The rest of us are only too happy with our own stuff to simply smile at
> your pointless jingoism and return to technologies with a future. 
> Interesting how few of us regard Ada as one of them!

From the posting you've made on this thread, not much you have to say
about Java, C++ or Ada will be at all interesting to anyone. I haven't
worked with Ada for a living for four years (now doing C++, Java, Perl
etc) so it may be you think "we" share your views. Well I don't.

Chris
-- 
Chris Morgan <mihalis at ix.netcom.com>		http://www.mihalis.net
      NP:- Angus Dei - AC/DC do choral music