Re: Creating an operating system
Stefano Zacchiroli wrote:
On Tue, Jun 23, 2009 at 07:27:57PM +0200, Thomas Viehmann wrote:
The problem with Debian's processes is not that they are too slow,
but that they are full of people who only degrade Debian.
I agree with the concern. But the reason is twofold. On one hand we
might have lost our ability to select talented developers; on the
other we have seriously diminished our ability to attract talented
developers.
Thanks for sharpening the analysis here.
The two parts of the reason require different solutions. What I'm
concerned about here is not to loose talented people due to their fear
of the length of NM process or their frustration after years of asking
pretty please can you sponsor my package [1].
It probably is no big surprise that in my opinion is that it is not NM
that is most unattractive about Debian.
Of all of Debian's recruitment, the NM process is easily the part that
works best. The only reason people like to pick on the NM process is
because there a) it is work, b) there are statistics, c) there are 5
people to be made responsible.
Debian is, just it has been in the post-release analysis, conveniently
fooling themselves into thinking that the few people on some team are
the cause for a misery that is too hard to properly address or not
comfortable.
The questions do not make me happy. But it is the people not being able
to answer them who are depressing. And they become the
"do-I-really-have-to-know-this" maintainers who are depressing.
If you want to attack the poor selectivity problem, on quality basis,
of the NM process go ahead, I'll stand behind your reasons, but it is
orthogonal to removing (apparently) gratuitous process inefficiencies.
That is an interesting position after this thread started with a
complaint about queue backlog (as symptom of exceeded capacity) and
Bernd told us that it is candidates for who have not shown their awesome
qualities to the extent one might like. To me this sounds as
poor selectivity ----> inefficient process
as much as it gets.
http://lists.debian.org/debian-devel/2009/06/msg00466.html
<snip>
PLAIN AND SIMPLE FAIL.
It is interesting you mention that: I witnessed it in first
person. Yes, that was a total failure. But you know what? The reason
why I was able to start working on all the missing deps to bring back
turbo{json,gears2} in a usable shape is that 2 people currently in NM
have in the last year stepped in the OCaml team doing an outstanding
amount of work, relieving a lot of my duties there.
> They are both very talented and I'm unable to explain to them why,
> after 18 months in NM, and after AM + FD reviews, they have to wait
> also for DAM approval, with an average waiting time of 2 (more)
> months.
FWIW, all non-DDs whose work I admired most last year have become
developers quite a while ago. The king of ostentative understatement
took long with 9 months, but then I thought that the quality of the
uploads he submitted for review greatly increased in between end of
January and July, too, and he probably was busy cranking out all those
QA- and non-maintainer-uploads.
Comparing Stephane's (I take it is his 18 months above, Mehdi seems to
be involved since 2008, more heavily since September and applied in
October 2008) NM record with the public record of his involvement in
Debian seems to indicate that he applied and was advocated way too
early.[1] I am sure he does a lot of excellent work now, but advocating
people with a only few weeks involvement and complaining that the NM
process takes more than a few weeks without mentioning the first is
patently absurd when some sustained commitment is a prerequisite.
> [1] Yeah, I know, people will reply that we have DM right now and that
> if they do not want to wait they do not deserve to be part of the
> project. I don't care, such arguments are no excuse for delays
> that appear as unjustified to our public, i.e., potential
> developers
Debian's recruitment process should serve Debian's goals and interests.
Words like "deserve" and "(un)justified" seem to mainly satisfy the
pathos urge. Random user user that I am, I would certainly hope that
Debian exercises great care over who meddles with my operating system
and takes time to diligently evaluate applications.
Kind regards
T.
1. There is precisely one mail (ITP bug from November) from 2007 in
the lists.d.o archive and he was advocated in January 2008.
--
Thomas Viehmann, http://thomas.viehmann.net/
Reply to: