Re: GNAT packages in Linux distributions

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

Simon J. Wright wrote on comp.lang.ada:
> On May 10, 10:46 am, Ludovic Brenta <ludo...@ludovic-brenta.org>
> wrote:
>
>> Actually, the fixes *are* exposed to public purview.  This happens when
>> AdaCore merge their fixes into GCC.  Each patch appears as a
>> self-contained email on the gcc-patches mailing list, complete with a
>> technical description of the bug and even the AdaCore bug number; only,
>> we don't know what triggered the bug, so it is difficult to relate the
>> technical description to a publicly known bug trigger.
>
> I haven't seen any of these (I guess I joined that list between
> merges) but presumably the test suite does get updated (as with
> changes for other languages).

Here is a good, small example:

http://gcc.gnu.org/ml/gcc-patches/2010-04/msg01528.html

The patch includes a detailed description of the bug, i.e. the result
of analyzing a bug report.  The bug report itself (and any triggering
code) is absent, as is the AdaCore bug number.  The patch does contain
a new test which triggers the bug; AdaCore constructed this from the
analysis of the bug report.  As you can see from the message, AdaCore
applied this particular patch to both the mainline (i.e. the future
GCC 4.6.0) and to the GCC 4.5 branch (i.e. the future GCC 4.5.1).

GCC 4.6 entered Stage 1 on 2010-04-06 (immediately after the release
of GCC 4.5.0) and the Ada patches started pouring in.  The same
happened for the Stage 1 of GCC 4.5.0 in March and April 2009; you can
browse the mailing list archives and look for the tag "[Ada]" in the
subject lines.

PS. Thanks, Simon, for this:

2010-01-09 Simon Wright <simon@pushface.org>

 	PR ada/42626
 	* gcc-interface/Makefile.in (gnatlib-shared-darwin): Add missing
 	end-quote.

--
Ludovic Brenta.