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

Re: RFC: Changing the NM system



Hi,

On 00-12-16 Adrian Bunk wrote:
> My impression is that currently maintainers are accepted too early. For
> some AMs it's enough that they build one package (and thanks to debhelper
> it's relatively easy to build a package) and even if they make a buggy

Especially since people can build very easy a debian package from an
original source package with the help of deb-make. :(

> package this is sometimes enough to pass the "Tasks & Skills" test (e.g.

Well, do yo have some other examples too? I think one example is nice,
but some others would be good to have. 

> in [2]). We have currently over 600 developers and at about 6000 packages

s/600/700 according to ldap.

> that have over 600 RC bugs. If we don't have severe look at the quality of
> the work of new maintainers it will become very hard to retain our
> reputation of being a high quality distribution.

Absolutely agreed. Especially since a lot of bugs are often very easy to
fix, even if I look at the RCB list, I see some of this bugs.

> Another important thing: We trust every single developer:
> - We trust him when he signs the GPG key of a prospective developer.

Well, I think we have to decide at this point, if the developer knows
the prospective developer for a longer time now or not. 

> - A developer can make an upload for every single package in Debian.
>   We trust him that he doesn't do any harm.

Do you want a system where you can only upload packages where you are
the Maintainer? This would make it hard to do bugfixes and NMUs, if the
maintainer went MIA.

> - Every Debian developer represents Debian (e.g. at exhibitions).

Agreed.

> And another small point: When we let people become developers very fast
> it's more likely that someone who has the idea "I want to get a
> @debian.org address" but doesn't want to do do long term work for Debian
> becomes a member of Debian - and we get more packages whose maintainer is
> AWOL.

Point taken.

> I suggest the following instead (important is the general idea, not the
> exact details):
> Someone starts working for Debian (e.g. by maintaining packages or helping
> at a port). After some months his sponsor, one of the port maintainers,...
> (or wherever the applicant participates in Debian) says that he does good
> work (e.g. high quality packages, quick reaction at bug reports,...) and

Well, I think this should be also based on the decision of the person
itself and not only be a decision of the sponsor. They both should make
this decision. But then you need to make sure that the sponsors have a
good knowledge about the debian build system and are qualified to make a
judgement about a package.

> suggests to the NM team that he should become a Debian account. The NM
> team (perhaps the current NM-Committee plus other interested Debian
> developers) then looks critical at the work of the applicant, makes a
> "Philosophy and Procedures" check,... and discusses and decides whether
> the applicant will be a valuable gain for the project.

Interesting idea.

> Additionally, we'll need a better organized sponsorship program to help
> applicants to work for Debian.

Absolutely agreed.

Ciao
     Christian
-- 
          Debian Developer and Quality Assurance Team Member
    1024/26CC7853 31E6 A8CA 68FC 284F 7D16  63EC A9E6 67FF 26CC 7853

Attachment: pgpd5Q4oUdHLl.pgp
Description: PGP signature


Reply to: