This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: 4.2 Backport fix for PR libfortran/31196
On Sun, Apr 15, 2007 at 11:23:41PM +0200, Gerald Pfeifer wrote:
> On Sun, 15 Apr 2007, Steve Kargl wrote:
> > IMHO, there should a current stable release branch (aka 4.1.x) and
> > mainline. When a new release (aka 4.2) is set to occur, mainline
> > is branched and a moratorium on commits to mainline is then put
> > into place. No one is allowed to commit mainline (ie., SC, global
> > write maintainers, zip, nada, no one).
>
> How does this differ from only allowing release-relevant changes on
> mainline for those two weeks and branching after, rather than before
> this period?
>
The difference is mainline is still open for anyone to slip in
a non-release critical commit. A rigid rule of no commits to
mainline means no commits. Period. The idea is to reduce the
burden on markm on making judgments on patches:
Dev: markm, I've worked up a patch for PR 31415926. It's not
a regression, but fixes a really important bug. I've
tested this patch on i386. Ok to commit?
markm is now in a position to say yes or no (or get more testing),
which is a position that I think we should not be imposing on
markm.
There are two possible outcomes: (1) everyone concentrates on
getting 4.2 release. (2) everyone sits on their hands and mainline
stays closed. As of today, there are 125 regressions in 4.2. This
is after markm has already requested a special effort to get the
bug count below 100. I'll also note that 4.2 was created on 20 Oct
06. That's 6 months and there appears to be no release forthcoming.
We've also had the discussion about skipping a 4.2 release, which
is detrimental to gfortran's exposure/image.
--
Steve