Re: Developers vs Uploaders
On Sun, 18 Mar 2007, Erinn Clark wrote:
> > The problem is to allow more small-scale contributors. We have volunteers
> > who would like to maintain only few specific packages and who don't want
> > to go the burden to go through NM to be able to do that job. The skills
> > required to maintain one or two small packages are different from the
> > skills that the NM process requires.
> How are they different? You have to demonstrate technical ability and
> some level of comprehension about Debian policy. That seems to be true
> for both NMs and potential DMs. NMs just have to write more essays.
Technical ability to maintain any Debian package != technical ability to
maintain a (documentation|perl module|latex class) package.
Of course all maintainers have to know the Debian policy and have to apply
it in their packaging. However many people have advanced skills in a
particular domain and can be trusted to handle sanely that domain but they
can't be trusted to do a reasonable work on all Debian packages.
The NM process check that:
- the candidate has the skills to maintain his/her packages
- the candidate is able to learn new skills when faced to new problems
- the candidate is able to sponsor someone else
- the candidate is able to do NMUs
- the candidate is familiar with autobuilding and porting issues like
endianness (which is not needed when you maintain an arch: all package)
Only the first is important to check for a DM which maintains an arch: all
package. Only the first and the last is important for a DM which maintains
an arch: any package, and even then most packages are no-brainers
concerning auto-building and porting (and porters are here to help when
there's an arch-specific problem).
So clearly, there are differences in what need to be checked for DD and
> > The checks for the NM process are more strict and rightly so, because the
> > rights of a DD are important.
> Which rights do you mean? (I assume voting rights here -- are there others
> you're referring to?)
I mean mainly right to NMU and the right to package a new software. A DM
can't package a new software without going through a round of sponsorship.
This is rightly required because we haven't check their abilities to work
on any domain...
> > Additionnaly, once this possibility exists, it just make sense to use it
> > for NM who have already proved their skills. I fail to see why it would
> > augment the length of their NM process. On the contrary, the time won for
> > his sponsors can be reused to train other people and/or process more
> > people in the NM queue.
> Well, most people in NM are just waiting for someone else -- this goes
> for AMs as well. And the more people you add, the more likely you are to
> have to wait for them to get to you.
Adding an NM to the DM keyring is parallel to the NM process, nobody needs
to wait until it has been done, life goes on as usual...
> > The NM team awaits your contributions. :-)
> I'm still in therapy for the time I spent in it. Maybe later. ;)
Sure. But then, the fact that you're unhappy with the current NM process
shouldn't be used as an excuse to be against the DM idea. "The perfect is
the ennemy of the good" do we say in French.
I'd rather have an improved situation now with DM, than have to
wait until NM is adapted to all cases. We should all be interested in
concrete results for Debian, more than in the perfection of the internal
> > > - Overly bureaucratic
> > I would prefer drafting a jetring entry for a good sponsoree of mine
> > instead of continuing the sponsorship... sponsorship is also bureaucracy
> > past a certain point.
> But above all of that, would you prefer they got through NM?
I don't care. It's up to them to see how much they care about the project,
and how much they want to be involved.
My personal opinion varies on the technical ability that I see. Some
people are very good and I'd like to have them as DD, others are not ready
yet but still manage to handle correctly some packages, I'd like to have
them as DM.
> > > - More power structures
> > I see that as good thing when we have more people able to empower other
> > people to do Debian work.
> In theory it's good -- in practice? ... Well, have you ever been a part
> of a management-heavy system? At some level it becomes a process fetish
> and I think that's something worth fighting.
I will reconsider the question when we have that kind of problems. For
now, the problem is the contrary... when elmo doesn't add keys to the
keyring, we're blocked.
> > But you know agreeing to the SC and DFSG (and checking the GPG key) is the
> > shortest part of the NM procedure.
> I do. So what I'm seeing is:
> - They agree to the DFSG
> - They agree to the SC
> - They have their key signed
> - They demonstrate enough packaging ability and community integration to get into
> this DM keyring and earn the right to autonomously upload packages
> Why not just make them DDs?
They demonstrated packaging ability for a limited set of packages only.
See my long explanation at the beginning of this answer.
Premier livre français sur Debian GNU/Linux :