Bug#625449: Permanent BSP patch
On 06/05/11 at 20:59 +0200, Stefano Zacchiroli wrote:
> On Wed, May 04, 2011 at 05:44:47PM +0200, Lucas Nussbaum wrote:
> > > > I think that if you want to change the NMU procedures described in
> > > > dev-ref, you should at least discuss the proposals in a similar forum
> > > > than the one where the current recommendations were discussed, i.e
> > > > debian-devel@ or debian-project@.
> > > >
> > > For example, in
> > > http://lists.debian.org/debian-project/2006/07/msg00231.html ?
> >
> > The NMU policy implemented in dev-ref was discussed on -devel@ in 2008
>
> I guess Neil actually meant
> <http://lists.debian.org/debian-devel-announce/2011/03/msg00016.html> ,
> cited at the beginning of this bug report. Quoting from there:
>
> > 0-day NMU policy
> > ----------------
> > For some time now, we have had a perpetual 0-day NMU policy, and some
> > discussion [LDO:0day] was had a while ago. We feel that this has worked
> > well for the past five years, and so will be submitting a bug against
> > dev-ref to make this official.
>
> which is a very clear quote and has been circulated via d-d-a.
>
> I've noted that you question the authority of the Release Team authority
> to decide on this, which is one thing (see below about that).
>
> Nonetheless nobody can argue that the RT didn't bring this topic
> up---with more than 1 month of advance---to the most appropriate place,
> enabling everyone to comment. (Reviewing the -devel thread you can find
> one comment about it. It is by Gregor, has been posted shortly after a
> further "ping" by Mehdi on this topic, and it's a positive one. [1])
>
> So, looking from the outside, I would be more than ready to consider
> this change consensual among d-d-a readers. That of course does not
> mean that it is consensual among devref editors, although I would expect
> them to follow consensus.
I must admit that I overlooked that part of the release team's email.
I just sent a mail to -devel@ to gather more feedback about the change.
Maybe it's just me getting old and grumpy ;)
> Anyhow, to avoid doubts and the unpleasant sensation of "sneaky-ness" in
> ratifying such an important policy, I suggest to proceed as follows:
>
> - Improve the wording, given that it has been perceived as ambiguous by
> various readers.
>
> I personally agree that it would be good to clarify that maintainers
> are not forced to ping bug logs every 7 weeks (although I believe that
> at least one message saying "I'm on it" it's at the very least a wise
> thing to do to avoid duplication of efforts). The wording might for
> instance mention that 0-day NMU-ers should do a best effort to check
> out by other means if the maintainer is working on it even if the
> buglog is silent, e.g.: by looking at the VCS. Such a precaution
> sounds reasonable for 0-day NMUs and at the same time won't add much
> of a burden upon the NMU-ers; IME I often looked at the VCS anyhow, to
> check if a patch was already available.
>
> Maybe those who have found the wording ambiguous can help out with a
> first draft of that?
Regarding my own objections to this policy change, improving the wording
and adding precautions won't make them go away.
> /me puts on his formalities hat
>
> Regarding the authoritativeness of the Release Team for deciding on
> this, I'm sorry but I have no solid formal ground to rule on that. (I've
> been planning to discuss with RT on how to clarify their delegation, but
> it's still in the TODO pipe.) So, if anyone feels strongly about that,
> this should probably go to the tech-ctte.
>
> At the same time, it's undeniable that for the past 5 years or more,
> people have accepted NMU guidelines coming from the RT. So at least by
> folklore people don't seem to have a problem with RT authority on
> this. I could look more into this if you, or anyone else want me to
> (e.g. by digging archives trying to rebuild RT delegation history), but
> I do hope we can solve this in ways where the time of everybody is
> better spent than that.
I'm really wondering whether people have really been accepting the NMU
guidelines coming from the RT, or just ignoring them and using the
dev-ref recommendations instead. In the 2010 -bugs-rc archives, there
are 390 mentions of "DELAYED/2", vs 711 mentions of "DELAYED", so
uploading to DELAYED/2 looks quite popular. (I raised that point in my
-devel@ email, you might want to reply there).
I also note that you have been following the dev-ref recommendations
yourself during RCBW. ;)
Regarding the authority of the release team, I very much respect the
release team, and highly value the opinion of its members, especially on
everything QA-related. But I don't see how defining the policy of
uploads to unstable could be formally part of their role.
- Lucas
Reply to: