Re: GCC 4.0 Status Report (2004-01-05)

[Richard Guenther <richard dot guenther at gmail dot com>]

On Sun, 9 Jan 2005 18:37:41 +0000, Paul Brook <paul@codesourcery.com> wrote:
> On Sunday 09 January 2005 12:49, Toon Moene wrote:
> > Giovanni Bajo wrote:
> > > Nonetheless, I would like to bring (again) to your attention the fact
> > > that gfortran regressions (wrt to g77) are not considered in this count.
> > > I understand that a new frontend means that we are going to live with
> > > regressions for quite a while (as you correctly said here:
> > > http://gcc.gnu.org/ml/gcc/2004-12/msg00614.html), but I reckon that we
> > > should at least consider the most important problems in the new frontend
> > > (wrong-code and the such). Tobias kindly grouped all the gfortran
> > > regressions in a meta-bug, http://gcc.gnu.org/PR19292.
> >
> > Our, i.e., the gfortran maintainers', assumption is that GNU/Linux
> > distributors (and other whole-OS distributors who use the GNU
> > compiler/linker/debugger toolchain) will offer access to the g77 binary
> > via some method (e.g., by installing it in /opt/bin and the run-time
> > library in /opt/lib).  In that way, people who need g77 because they're
> > bitten by the incompatibilities in gfortran have a way out.
> 
> Right.  I think including gfortran in the 4.0 release criteria, hence implying
> a release-worthy level of quality would be a lie.
> 
> Obviously we'll continue fixing bugs, and things like the meta-bug you mention
> help.  However I don't see what benefit including this in the official gcc4
> stats gives us.

I think that at least known wrong-code bugs for primary release
platforms should be fixed or documented as such.

Richard.