Re: DM application for Adam Cécile (Le_Vert)
On Sun, 2008-02-03 at 20:34 -0500, David Nusinow wrote:
> Further, I would argue that you are absolutely flat out wrong when you say
> that being a DM means you just need to maintain your packages well.
Maybe it does right now. But if so I think that is wrong,
for a number of reasons.
Firstly, requiring someone to "a good maintainer" is a
fairly broad brush. It means your packages follow Debian
policy, it means you respond quickly and effectively to
bug reports, it usually requires technical skills, and
maintaining something generally means you have already
social skills to persuade whoever let you maintain the
package. In other words, having a nice set of well
maintained packages implies the many of the things that
you seem to say aren't related to "just maintaining your
packages well", so here I disagree with you.
Secondly, and more importantly, it isn't hard to check
if someone's packages are maintained well, provided the
criteria is kept simple. Debian runs lots of
_automated_ checks along these lines now. This makes
it really easy to verify if someone is fit to become a
DM. This has already been demonstrated when an
applicant didn't handle his gpg keys correctly. He
was rejected with what I thought was relatively little
discussion because in the end it is a fairly straight
forward technical point, as opposed to a ideological
one.
Thirdly, becoming a DM is a light weight process compared
to the marathon that is necessary to become a DD. It is
obvious one of the reasons for this is _because_ the
checks are currently light weight.
Fourthly, Debian isn't giving much away here. You are
given the right to upload an existing packages, provided
your sponsor thinks its a good thing, and that right can
be withdrawn easily. So why the need for checks other
than "do the job well"?
These things all tie in together to give a whole that is
bigger than the sum of the parts. Consider the marathon
that is the process to become a DD. I think one of the
reasons it is a marathon is this is related to the fact
as once you are granted DD status you are effectively
granted it for life. Yes, there are ways to get rid of
a DD, but in reality it is almost never done as the
criteria to becoming a DD is based on all sorts of touchy
feelly things like "following the spirit of the social
contract" and a persons "contributions". How do you
unambiguously measure whether a person is still acting
in the spirit of Debian's social contract, and whose
definition of spirit do you use anyway? And what is a
relevant contribution and how do you measure the many
ways a person can make one? In practise its very hard
to objectively measure these things so it is equally
hard to eject someone for not conforming to them. The
end result is the up-front checks must be fierce to
compensate. You can't fix mistakes, so you try and
make as few of them as possible. Hence the DD marathon.
In contrast, a Debian Maintainer status is renewed each
year. If someone's work starts causing problems for the
archive - things that cause a noticeable drop in quality
like not fixing rc bugs, or straining the infrastructure
then the privilege can be easily withdrawn. Provided you
leave airy fairy things such "as being strong promoter of
Debian's principles" out of the criteria for being a
Debian Maintainer, measuring these things is relatively
uncontentious, and so acting on them also becomes less
contentious.
So, I am not saying upholding and believing in Debian's
principles isn't important. It is very important. And
people who have demonstrated they do understand and
follow the Debian way should be acknowledged and rewarded.
It is just that demonstrating those things necessarily
takes a long time. In comparison it does not take a
good programmer that long to lean how to competently
package something for Debian. Well maintained packages
are Debian's lifeblood - its reason for existence. We
need all the good packagers we can get - providing we
can be sure they are doing a good job. The DM process
is a way of getting these packagers into the project
quickly, and weeding out the ones don't work out. In
order for this to work you need simple criteria, and
I think "do your packaging job well" is about is
simple as you can make it. We will have enough long
winded arguments about how to measure that. Don't
complicate it.
Reply to: