Re: .3 releases

["Luis Villa" <luis villa gmail com>]

On 1/24/07, Kjartan Maraas <kmaraas broadpark no> wrote:
> ons, 24.01.2007 kl. 12.51 -0500, skrev Luis Villa:
> > On 1/24/07, Elijah Newren <newren gmail com> wrote:
> > > On 1/24/07, Vincent Untz <vuntz gnome org> wrote:
> > > > Discussion on IRC right now...
> > > > Some maintainers have a hard time finding time to do stable .3 releases.
> > > > And they already don't have enough time for 2.17.
> > > >
> > > > Some pointed out that no distro is using .3 releases, so it looks like
> > > > much effort for nothing.
> > >
> > > Hmmm...seems to reflect my habits a bit as well; I've often treated .3
> > > releases as optional, only making them if there were important
> > > bugfixes I wanted to make available.  I know that some module
> > > maintainers may feel like fixing bugs on the stable branch is useless
> > > if we don't make a release (I would), and even if distributors don't
> > > use a full GNOME 2.even.3 release, they might include individual
> > > package updates.
> > >
> > > Maybe we should just explicitly state somewhere that additional
> > > package releases are optional for .3 releases, and that we're just
> > > trying to publicize any important fixes in modules that do make
> > > releases?  Does that sound sane, or am I talking crazy?
> >
> > FWIW, as a distributor, I wouldn't ship anything except .0, since all
> > the freezes go up in smoke after .0, making them likely to be less
> > stable than .0, not more so. But I'm quality-paranoid.
> >
> If this is a real concern among distros

I don't think it is, but then again I don't think most of them have a
real clue about interacting with GNOME QA or really measuring how
stable/unstable GNOME is. I'm just saying that is what I would do were
*I* a distributor. Note that I couldn't even convince Novell of this
;)

> I think we should rethink how we
> spend our resources. I don't really agree though since all patches
> should be reviewed by the module maintainers anyway so QA surely isn't
> supposed to be worse than what it is up until the .0 release?

Once we reach .0, we're making a certain guarantee that the quality
will not regress after that point. I believe that demands more than
just one set of eyes on reviewing patches, since we know that having
only one set of eyes (the maintainer) is insufficient.

But realistically, none of that matters to distros, AFAICT- are any of
them doing anything other than taking whatever the latest release is
when they freeze and shipping that? Any testing to say '.0 better than
.1' or vice-versa, or any policies other than 'whatever is latest when
we freeze'? And after their ships, are any of them shipping further
stable releases of ours? I'm guessing not, but I'm not really paying
attention in detail like I used to.

Luis