GCC Status Report (2004-10-26)

[Mark Mitchell <mark at codesourcery dot com>]

--
Mark Mitchell
CodeSourcery, LLC
(916) 791-8304
mark@codesourcery.com

GCC 3.4.3
=========

There are 126 regressions targeted at GCC 3.4.3, which is almost
exactly the same as the 129 on October 8th.  Of these 24 are
wrong-code, 30 are ice-on-valid code, and 9 are rejects-valid.  Most
of these are also 4.0 regressions, so fixing these is a win for both
3.4.3 and 4.0.

I wish there were fewer problems, but I'm still planning to make a
3.4.3 release next week.  As of Friday, October 29th, the branch will
be slushy; all non-documentation patches after that point will require
my explicit approval.

GCC 4.0
=======

There are 200 regressions targeted at GCC 4.0, down from 223 on
October 8th.  Of these, 36 are wrong-code, 54 are ice-on-valid-code,
and 14 are rejects-valid.  However, of these many have to do with Ada,
Java, or platforms that are not on the primary platform lists.

I'm still targeting 100 regressions before we make the branch, but I
plan to make a pass over the open regressions and retarget those that
I can tomorrow.  I expect that after that, we will be much closer.

I have been informed that our SPEC benchmark results are looking
pretty good, which matches our independent observations.  However,
there are apparently regressions in perlbmk and mgrid.  Andy MacLeod
has a potential fix for the perlbmk problem here:

  http://gcc.gnu.org/ml/gcc-patches/2004-10/msg00488.html

I intend to review this patch, if nobody else gets to it.

Zdenek has a patch to improve mgrid here:

  http://gcc.gnu.org/ml/gcc-patches/2004-10/msg02007.html

but there is some controvesy about various parts of that patch.  I
intend to review this patch as well, perhaps consulting with various
other people with more expertise in this area.