Re: clear policy discussions

[Graham Percival]

On Fri, Jul 13, 2012 at 06:23:57PM +0100, Phil Holmes wrote:
> ----- Original Message ----- From: "Trevor Daniels"
> <address@hidden>
> 
> >a) a release-meister with responsibility to take all decisions.
> >That's what Han-Wen used to do, I believe.  It worked
> >quite well, although Graham will have more of an inside view
> >than I.

It worked well in general, but not without a few quirks.  The
2.12 release caught many people off guard:
http://lists.gnu.org/archive/html/lilypond-devel/2008-12/msg00553.html
http://lists.gnu.org/archive/html/lilypond-devel/2008-12/msg00602.html
I also recall that the translators wanted additional notice so
that they could clear up some loose ends before a major stable
release.

The current setup was in large part a reaction to that confusion
-- make it absolutely clear to absolutely everybody when a release
would happen, and therefore avoid any "panic list" of major
problems discovered in the actual release.

> In principle, I'm not against giving people authority to take
> decisions where there aren't hard-and-fast rules.  However, there is
> a very substantial downside with this in an environment such as
> LilyPond.  Given our poor recent history of being able to achieve
> consensus on fairly simple things (<> vs s1*0 anyone) there's a
> significant possibility of large clashes where the Release Meister
> makes a value judgement with which others disagree.  I could
> envisage a stable release being pushed out because the RM thought it
> optimum, but another team member being significantly annoyed
> because, say, it had a bug that stopped his music setting.  All hell
> then starts.
> 
> That's the benefit of rules - we argue over the rules slightly
> dispassionately, not over the value judgements.

Agreed entirely!

That said, I am leaning towards the "benevolent dictator of 2.16"
approach.  Limit it to, say, 6 months (at which point we'd discuss
it and probably renew the "appointment"), and make it specific to
a certain branch.  I don't think this is a perfect solution for
all the reasons you gave, but we've tried the experiment of very
rigid rules for the past 3.5 years.  Let's let the pendulum swing
the other way and try it for a few months.

I'll write an official proposal along these lines and introduce it
on Tuesday, July 17.  We'll have a week of discussion, then I'll
make a final proposal on Wednesday, July 25.  Assuming there's no
huge objections, it will be accepted on Thursday, August 2.

- Graham