This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: 3.4 regressions: are 2.95 regressions still actual

On Mon, 2004-01-12 at 07:41, Giovanni Bajo wrote:
> Wolfgang Bangerth wrote:
> >> I don't think it is reasonable to have only one person handle these
> >> things: we are already seeing that this is not a viable approach.
> >
> > Quite right. The RM can delegate this work, but I am not aware that
> > he did in this case.
> Yes, I agree with Arnaud. I think it's useful to allow the maintainer of the
> code affected by the bug to postpone the milestone as well.

I am just now having a chance to review this discussion.

I believe that the RM (me) and/or  his or her delegates should be the
only people to slip the release target for a bug.

Despite Arnaud's comments, I've thus far had no trouble making a pass
over the open regressions and slipping them as appropriate; I've done
that for every release to date, and I intend to do it for 3.4.

I think it is useful to slip them later, rather than sooner, because
sometimes it turns out that a bug is not so terribly hard to fix, and
then we fix it.

At this point, I don't think we should be trying to find bugs to slip to
3.5 -- we should be trying to fix the bugs for 3.4.

I've set as a goal having fewer than 100 open regressions before making
the branch, which seems pretty reasonable to me.  That's roughly one or
two new regressions per month since the 2.95 release.  I'd like to think
we'd introduce fewer regressions than that, but clearly that turns out
to be false in practice.

Mark Mitchell <>
CodeSourcery, LLC

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]