Re: Why not gnat Ada in gcc?

[Corey Minyard <minyard at acm dot org>]

kenner@vlsi1.ultra.nyu.edu (Richard Kenner) writes:

>     The model proposed by Robert Dewar is the right one to start with, and
>     it will put the merging burden on ACT if some contributions are indeed
>     made by people having write access to the GNAT sources. If this
>     becomes significant, I assume the natural thing will happen (move of
>     the master source to the FSF CVS repository).
> 
> I think that is indeed the issue: to what extent are there likely to
> *be* significant contributions from the FSF community?

I would contribute.  I have contributed to GNAT in the past.  The
number of people I know that would contribute are likely not a
numerous as for C or C++, but they probably number several dozen
active contributers.

> Ada is a very complex language and one which is understood to that
> extent by very few people and those few people nearly all work for Ada
> compiler vendors.  So I don't accept that putting GNAT sources in the
> repository will magically make such expertese appear.

I don't agree with this.  After programming for 10 years in C, and
learning both Ada and C++, I found Ada easy to learn and not very
complex.  Certainly much less complex than C++.  Don't re-enforce the
FUD.

I think you are saying that people will break things (violate the
reference manual) because they don't understand it well enough.
That's true.  But you can say the same for C, Java, and C++.
Especially Java.  People break it all the time.  Java itself is not
that complex, but the libaries are fairly vast.  But at least Ada has
a well-defined validation suite and I know ACT has a very extensive
test suite.  So if people break it, finding the breaks should be
easier.  Plus, GNAT is very well written and easy to follow.

And if the ease of entry gets lowered to using Ada, maybe people will
start using it.  Maybe they won't.  But if it is part of GCC, it's
much more likely that Redhat, SuSE, etc. will build it into the
standard distribution.  And if they do, the effort for entry is
greatly lowered.  I think the main reason C became so popular was the
low effort to use it (it came on everything) and the portability (it
worked pretty much the same on everything).

> In any event, it will *always* be the case that merges are necessary
> since ACT can't use the GCC repository copy as its main development
> compiler for the same reasons that Cygnus couldn't do it for GCC.

You are right, merges are always necessary.  But there is a vast
difference between the ACT development model and the Cygnus model.
Cygnus puts changes out quite often, allows experiments to go on in
the public release.  ACT holds all changes until one massive release.
I don't like the ACT model very much, I would like to see a much more
free and open interworking with the user community.  And if GNAT goes
into the GCC distibution, with their current model ACT will have to
work very hard to coordinate their releases to make sure they work in
the standard releases of GCC from the FSF.  Otherwise, GNAT will
always be broken.

I agree with Mr. Stallman about the neutral repository.  I think it
would be better to host the FSF sources in a neutral location.
Although I think Redhat/Cygnus is doing a fairly good job of remaining
neutral, it still makes other businesses a little nervous, I think.
And the point it to avoid squelching cooperation and innovation.  If I
had an OC-3 and a bunch machines, I'd host it.  But maybe I'm not
neutral :-).

Corey