Re: The GNU Ada compiler

["Steve Whalen" <SteveWhalen001@hotmail.com>]

Martin Krischik wrote:
>
> >> There are 700kb worth of special changes to the standart GCC in order to
> >> make GNAT work pleasantly on OpenVMS. So just compiling a standart GCC on
> >> OpenVMS might not be enough to have an well working Ada on OpenVMS.
> >
> > So the GNAT which is on the publically accessible source repository
> > does not represent something that _ever_ worked on VMS ?
>
> I don't know! You have to ask AdaCore about that. Only they know if there is
> any release in the FSF repository which is also an officially released GNAT
> -  or at least works well with VMS.
>
> But this is a problem for any OS. If you download the GNAT/GPL sources:
> There is a 700kb patch file for all OSes. Now the README.BUILD of GNAT/GPL
> says:
>
> |  as well as a set of patches that you may need to apply to the
> | GCC sources, depending on your target platform. One of these patches is
> | "gcc-34.dif" and should be applied whatever the target.
>
> Only there are no target specific patch files in GNAT/GPL. There is a
> district 2 class society here and it is more the just the difference
> between GPL and GMGPL.
>
> The way it works: AdaCore commits there sources to HEAD of the FSF tree -
> but they don't actually use HEAD for GNAT/Pro and GNAT/GPL. Instead they
> backport the changes to some older - and presumably more stable GCC.

... and later ...

> I think AdaCore could do with some competition in the binary distribution
> area. Especial for targets for which they don't offer a GNAT/GPL edition.

First, thank you for getting this project started again.

I agree that it is badly needed and appreciate your efforts. I think
we need these compilers for all targets, whether or not GNAT/GPL
editions are available. Much of the real world lives outside the GPL
universe and a GPL only compiler is not going to help get Ada into
such a shop, particularly a smaller shop.

I already suggested that one small shop take a look at the GPL version
of Ada to see what Ada could do, but was told "It won't do any good if
it's GPL only; we don't write GPL software here".

In the past, I would have given the person in the shop who was willing
to take the time to look at Ada, a CD or a link to a GNAT download for
(in this instance) the Windows platform and they could try it out.
The fact that they knew that if they wrote an interesting prototype in
Ada they could sell it without restriction, made it much easier to get
a "foot in the door" for Ada.

It doesn't do any good to try to convince a small shop to start using
Ada by having them try the GNAT GPL version, when they know that it's
going to cost them several thousand dollars (or euros) per seat for a
version of Ada that allows them to write non-GPL proprietary
applications.  Especially in non-defense oriented smaller shops, they
just don't have that kind of money to throw at a technology they
aren't' sure they can make use of.  We need free GMGPL compilers to
bridge that gap.

To proselytize Ada requires being able to put a fully usable Ada
compiler on whatever hardware and software environment is used in a
shop you want to impress without cost or restriction, and without
worrying about whether the Ada compiler is as brain dead as some of
them were / are when the gcc maintainers and the distribution release
people decided to pull the gcc tree for a distribution's release.

We DO very much need a separate effort to be able to package Ada at
those times when it is stable and then make THAT compiler and those
libraries available with the GMGPL exceptions in all the right places.

The gcc that is chosen for release by distribution maintainers is
based on many factors, the very least of which is the state of the Ada
compiler in that particular flavor of gcc.

If we could point the distribution maintainers to a sourceforge site
and have them know that at any point in time the most stable version
of Ada was available there, we might have a very good chance that
quality Ada compilers would be distributed with more than just debian
(for which I thank Ludovic Brenta).

In particular for proselytizing Ada, stable GMGPL Ada compilers need
to be available for Windows (Ming and cygwin), Redhat, SUSE, Mandrake,
and Solaris.

Steve