Re: [gentoo-dev] Re: [gentoo-core] Marking packages stable on x86

[Chris Gianelloni]

On Thu, 2005-09-22 at 19:04 -0400, Mark Loeser wrote:
> Ciaran McCreesh wrote:
> > On Thu, 22 Sep 2005 18:36:18 -0400 Mark Loeser <halcy0n@g.o>
> > wrote:
> > | I'm sending this email because I have seen some packages marked stable
> > | on x86 without the permission of the x86 team, and would like the
> > | people who can mark stable for x86 to contact us.
> >
> > Would the x86 team prefer it if lots of people made arrangements of
> > this kind, or would they rather handle the majority of keywording
> > themselves? I'd imagine there are quite a few of developers who have
> > the hardware to do x86 testing, but who would not want to step on the
> > x86 team's toes...
> >
>
> For the time being, I believe we would like to make arrangements with a
> lot of people, instead of being in the dark about who is doing what.
> This way we could atleast easily know what is probably being neglected
> by us.

Neither of Ciaran's solutions leave us "in the dark", at all.  What he
is saying is would we rather half the developers tell us "I want to
stable my own stuff" or would we rather be like "We would prefer to
stable everything, but will make allowances for a few developers"?

Personally, I think we should try to take care of the bulk of
stabilization requests, otherwise we lose the QA benefits of having an
arch team that is aware of what is stabilized on the architecture.  It
also makes it easier to keep the architectures in sync, where possible,
as when someone files a stabilization request for x86, they can do the
same for $n other arches, too.

> I only conferred with a few other people in #gentoo-x86.  If anyone else
> on the x86 team has opinions about this, I'd love to hear them, but as
> the GLEP is outlined, I believe this way is the best way to go about it.

I think if we start blanket allowing people to stabilize their own
packages, we won't be any better off than we were before.  In many
cases, the maintainer *does* know best and should be listened to, but
there are good reasons for making stabilization requests go through an
arch team.  After all, how many of these maintainers asking to mark
their own packages stable *run* a stable system?  Maybe a stable chroot?
I'm willing to bet nearly *none* of them do, which is a problem when it
comes to the QA of the x86 arch, which is the x86 arch team's
responsibility.

-- 
Chris Gianelloni
Release Engineering - Strategic Lead
Games - Developer
Gentoo Linux