Public SourceMage Discussion List

[Jeremy Blosser <jblosser AT firinn.org>]

On Apr 15, Jeremy Blosser [jblosser AT firinn.org] wrote:
> On Sun, 15 Apr 2007 19:34, David Brown wrote:
> > I'd love to get stable-rc out the door this weekend but no one is
> > around to do the git thing to make it work.
> >
> > Bug 13330 is fixed with a blanket swap of -Os with -O2 and that was
> > the only bug that I saw gating.
> >
> > Other than doing all the git stuff and making tarballs I don't see why
> > stable 0.9 can't get out the door this weekend.
> >
> > For those who care.
> >
> > - David Brown
>
> I'll be home in an hour or so, I can do it then.

Well, no, because I'm not sure what's supposed to be going on with bug
13330.  I see there was a lot of IRC chatter about that today and then that
commit was made and from the comments above and in the bug I think you
meant to request it be integrated to -rc-0.9, but I'm not sure because none
of the version or flags are set on the bug to request integration.

I'm also not inclined to approve integration or roll a release with a
change to a major spell like gcc at the 11th hour when no one seems to know
what the problem is or if that's even a relevant fix; the bug comments are
all over the place and far from anything definitive.  When did this issue
show up?  In this -rc cycle, or the last stable cycle, or has it been
pre-existing?

Eric, I'd like to see some kind of hard freeze added into the -rc release
process.  I don't like major changes going in at the very end just to get
past bugs flagged gating that no one has had time to look at.  If people
are going to do that they should be required to do it <n> days before the
release happens, where <n> is at least something greater than 1 (I'd really
say at least 3 or 4).  Fixes that want to go in after the freeze (such as
fixes for gating bugs) could be allowed but they'd always push the release
date back to at least <n> days from when they are applied.  This was
discussed briefly on IRC at the end of the last release cycle but nothing
was actually put into the process for it.  Thoughts?

Attachment: pgpPbiUWj9hhx.pgp
Description: PGP signature