Re: Maintainership

[David Kastrup <dak@gnu.org>]

"Eric S. Raymond" <esr@thyrsus.com> writes:

> Stefan Monnier <monnier@iro.umontreal.ca>:
>> It's nice to be a dictator, but it takes too much time, so in order to
>> try and reduce this load, I'd like to dilute my dictatorship a bit.
>> 
>> To a large extent, this has already been the case, but I think it's
>> worth stating it more formally: if Glenn, Eli, Richard, Yidong, Handa, or
>> Jan agrees with a change, then you don't need my agreement.
>> IOW you only need my opinion if none of them has an opinion or if
>> there's a disagreement.
>
> OK.  The following question is *not* an attempt to be contentious; I'm
> trying to figure out how this is supposed to work, and help everyone
> else figure out too.
>
> You have expressed "100% agreement" with a post objecting to me doing /etc
> cleanup during feature freeze.
>
> On the other hand, Richard has approved the idea and actively
> assisted.

Where is the conflict?  It's something that should be done but not
committed now.

> If I understand your intention correctly, that makes the completion of
> the /etc cleanup changes an approved project.

A project is one thing.  Committing a change to the release candidate is
another.

Now I don't want to rain on Stefan's parade, but that difference is
pretty much exactly why it makes sense that any particular release
branch is best served by a _single_ responsible person making the
calls.  After the consolidation period is considered reasonably
complete, the release branch is branched off and from that time on,
_only_ the version dictator will cherry-pick changes into that branch
until the actual release happens.

The problem with multiple parallel quality managers which are possibly
partly not fully dedicated to the release quality management means that
the quality threshold of changes to go in is worse than even the
individual minimum (since the feeling of shared responsibility lowers
the diligence even more).

> I can cheerfully live with either theory - I've certainly got enough
> on my task list to occupy me for a while.  So I'm not pushing for
> either outcome in particular, I just want to know how the decision
> procedure works.

With my usual pessimism, my guess would be "not".

-- 
David Kastrup