[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Question for Matthew Garrett (was Re: Bits (Nybbles?) from the Vancouver release team meeting)

On Mon, Mar 14, 2005 at 11:55:50AM +0000, Matthew Garrett wrote:
> Despite the lack of representation from many people affected
> (architecture porters, for example), the conclusions reached have been
> presented as something of a done deal[1]. I don't think that's the way
> Debian should work. The proposals should have been made before the
> meeting took place, and people's opinions taken into account before
> conclusions were reached.

As a matter of fact, the result of the meeting was not known beforehand.
The proposal was formulated in Vancouver and after the meeting. None of
the participants knew what the result would be. Only by knowing the
result beforehand would it have been feasible to include the people
who now feel at a disadvantage (the porters) into the meeting. But
no one knew.

Your (and some other people's) perception of this being a
done deal might be caused by it being very detailed. But that is just a
sign that it was pondered profoundly. Reading debian-devel, I see
that lots of people don't misunderstand the proposal but see it as a
headway and useful.

> I also have concerns about the organization of the meeting. The first
> public announcement appeared two days before the meeting took place. Up
> until shortly before this point, even the DPL hadn't been told that it
> was going to happen. Meetings of this sort are obviously useful tools,
> but organizing them without notifying the rest of the project does
> nothing to improve the transparency of our decision-making process.

The reason for the perceived secrecy was confidentiality: One of the
participants hadn't cleared his attendance with his company yet and
was uncomfortable with it being announced in advance of that. This
would not have kept us from properly informing the DPL, of course. I
know that I would have liked to be told if I had been DPL. I am sorry
to say that I simply forgot to do so and I apologize for that: It was
not done intentionally. As an excuse I would like to point out the
short time span we had to make this happen and the amount of
coordination and work required.

> Frankly, to a large extent this meeting was everything Debian
> /shouldn't/ be. We've had conclusions reached without the decision
> making process being made clear[2]. We've had no opportunity for people
> to raise potential issues in advance. The way this was carried out does
> nothing to improve communication or consensus.

I disagree. The release process had been extensively and productively
discussed on wiki.debian.net. There all kinds of issues were addressed
and the participants were involved. Numerous potential issues were
raised and discussed in advance. Then the people most deeply involved
in the release process met, and expected some tough days and hard work
to arrive at some agreement. From what I hear, they were pleasantly
surprised by their similar views on the subject and managed to get
through the entire agenda in half the allocated time. In that euphoric
mood they put together the proposal and even mixed in jokes (which
were removed later on, but it helps to understand why numerous changes
were made). In the time between the meeting and the proposal's
publication it was revised several times. Matthew quoted from an
earlier draft, as he pointed out himself.

Then the proposal is announced and some people get worked up about
it. I hope this to be less of a problem in the future, when people not
only look at the technical merits of a proposal but also at the
intentions and problems people had to overcome on the way.

As Debian works today, there was little hope to make a proposal like
this without a massive flame war.

> [1] The text has actually been altered since I was sent a copy to
> approve, and it's rather less objectionable now - however, I wasn't sent
> a copy of the updated one.

It changed all the time. (c:

We (Debian) had experience with real life work meetings before and I
organized some of them myself (for debian-edu and
debian-installer). Usually we announce the meeting on the mailing lists
to give everyone interested the option to come, and be it on their own
costs. But the initial setting was more difficult this time: Besides
the above quoted initial confidentiality, some people involved here
were more touchy: Both James and Ryan are frowned upon by some
people in the project and the purpose of the meeting was therefore
both work and social acclimatisation of the group since some had not
met before. Communication issues between release and ftp/buildd people
were involved when Sarge got delayed and it was therefor one desired
goal to provide the two teams with a better base to work on in the
future. Anyone involved in release work therefor got a personal offer
from Steve (and Colin?) for travel support [1], taking care not to
hurt any feelings.

What can be learned from this? What went well, what was less good?

On the positive side: 
+ It surprised even me how well the meeting went once it started. I take
  that as a hint that Debian is just starting to understand the vast
  potential those real life meetings really have.  As a result of
  that, not only the Sarge release was dealt with (as if in
  passing) but people continued on and worked out a proposal for etch
  (as pointed out by Anthony in a parallel mail in this thread). The
  people involved noticed soon enough that their meeting's scope had
  somewhat increased and made their proposal to the project.

- The wording on this announcement could have been done more smoothly
  and clearly. Unfortunately the people who noticed beforehand did not
  give a heads-up. 

- Extra care will be taken next time when people need extra time to
  make arrangements (for example with their employer). The same thing
  goes for the invitation, which should have been more public.
  Both decisions were influenced mostly by empathy with individual's
  circumstance. This (need for individual confidentiality vs project
  openness) is a hard thing to do right and is where we need to pay
  even more attention next time.

[1] Jörg, our new DAM, who had not met James yet, wanted to join in,
too, but was willing to pay the trip himself.

Reply to: