This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: Two C regressions in GCC 3.4.
- From: "Joseph S. Myers" <jsm at polyomino dot org dot uk>
- To: law at redhat dot com
- Cc: Mark Mitchell <mark at codesourcery dot com>, Zack Weinberg <zack at codesourcery dot com>, gcc-patches at gcc dot gnu dot org
- Date: Thu, 18 Mar 2004 21:10:04 +0000 (UTC)
- Subject: Re: PATCH: Two C regressions in GCC 3.4.
- References: <200403181931.i2IJVSvv020262@speedy.slc.redhat.com>
On Thu, 18 Mar 2004 law@redhat.com wrote:
> >>This is PR 14366, by the way.
> >>
> >>
> >OK. So, you have a better fix for 3.5, then. Great!
> Y'all just saved me the wonderful task of trying to interpret the standard.
It doesn't help with some of the more subtle issues that the response to
DR#011 doesn't actually say what the answers to the questions in that DR
are meant to be after applying the corrections there (present in C90 TC1
and C99).
> (I ran into a couple of these building FC2 with the tree-ssa branch and had
> queued them for reinvestigation).
It's clear a regression tester doing this sort of thing (building
distributions and reporting changes in what builds or diagnostics) would
be of value to see what a patch breaks (or how useful a new diagnostic in
-Wall is on real code, or how frequent some deprecated usage is) so it can
then be judged at the time whether the breakage is correct (broken
software) or incorrect (broken GCC) rather than maybe some time later.
But I guess the resource requirements for doing such builds frequently
enough would be rather too large.
--
Joseph S. Myers
jsm@polyomino.org.uk