Re: The GNU Ada compiler

[Ludovic Brenta <ludovic@ludovic-brenta.org>]

Martin Krischik <krischik@users.sourceforge.net> writes:

[about patching the default GCC until it supports Ada well, instead of
adding a new GNAT]

> Before SuSE 10 this was no option at all. Up until then SuSE used
> gcc 3.3.x.  And we know that Ada in gcc only became usable *again*
> with 3.4.x.
>
> With SuSE 10 they jumped right to 4.0.2 - and a prerelease snapshot
> on top.  I am unsure if 4.0.2 is a good release either since AdaCore
> used 3.4.x as basis for GNAT/Pro and GNAT/GPL.

It is true that each distribution's policy WRT compiler and toolchain
affects Ada support.  When SuSE's policy was to ship only GCC 3.3, it
would have been a good idea to try and stop them from shipping GCC
3.3's Ada support at all, and instead ship gnat 3.15p.  But it is too
late now.

>> I would like to add that, in 2001, I switched from Red Hat to
>> Debian for precisely two reasons: no Ada support in Red Hat 8, and
>> no possibility that I could influence them.  And now, I do much
>> more than influence Ada in Debian :)
>
> But we would not want that Ada is only available to debian. That
> would leat to lots of FUD.

I never suggested that Debian should be the only distribution with Ada
support.  On the contrary, I am quite pleased to see that you take it
upon yourself to improve the situation.

Sadly, there is already a lot of FUD created by poor support for Ada
in most distributions.  Do you not agree that SuSE, for example,
created a lot of FUD by shipping gnat-3.3?

I still think that the best way to reduce this confusion is not to
create a new binary distribution of GNAT (which would more or less
compete with AdaCore's), but instead to join your distribution's GCC
maintenance team and influence them so they provide good Ada support
*by default*.  Of course, if you also provide a central repository for
patches that can be shared between distributions, this is all for the
better.

As to how GCC 4.0.2 fares Ada-wise, I am not sure either but I have
decided to skip it entirely in Debian, because the next stable release
of Debian (due in december 2006) will probably not include it.  I am
going for GCC 4.1 instead, and I am even prepared to backport Ada
fixes from HEAD to GCC 4.1 if necessary.

-- 
Ludovic Brenta.