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

Re: Debian Maintainers GR Proposal



On Thu, Jun 21, 2007 at 07:58:34PM +0200, Joey Schulze wrote:
> If this thingy is only meant to bypass NM since it sucks or is pink
> or whatever, just say so and the proposal will be considered as such.

The NM process is designed to create new Debian Developers -- particularly
with the ability to participate fully in the project, NMUing, hijacking
packages, voting, raising and seconding GRs, following -private, creating
new .debian.net services, accounts on dozens of machines, become a DPL
delegate, run for DPL, represent Debian, do transverse activities across
the distribution, etc.

As such, we expect people who pass through NM to have a good understanding
of a lot of things that don't matter if you're just going to maintain
one or two packages. 

Changing NM to only pass on the knowledge necessary to maintain a couple
of packages gives us DDs who don't have the knowledge to participate
in the project in the way we expect DDs to. Forcing people through NM
to participate in only that simple way makes that sort of contribution
unnecessarily difficult, which means we end up with DDs, ie the people
who *do* have the knowledge to do bigger projects in Debian, having to
spend their time working on the smaller tasks because the other people
who *could* do those simpler tasks are prevented from helping directly.

> The NM process after all is meant to help new maintainers become
> skilled maintainers of packages.

It's much more than that these days -- there're extensive licensing
questions, along with detailed questions about binary formats and
complicated library packaging.

> If we want to get them maintain
> packages without going through NM we should not create a new stage
> but drop or restructure the NM process.  IMHO

If there's some way of doing that that meets the above goals, I'd be
happy to support it. I don't think there is though -- we do have two
clases of contributors these days: people who want to be involved in
hacking on the project as a whole, and people who just want to hack on
a very small part of it -- and even if there was, I haven't seen much
happening in making the n-m process easier or more efficient. Proposing
an alternative resolution as an amendment might be a way to start, though?

> When somebody becomes a DM without going through the NM process and
> thus has no skills on packaging besides those required for the very
> package they started with and now wants to package $cool_kde_application
> which requires $not_so_cool_kde_libraries they also need to package
> how are they supposed to do so?
> Just go ahead without knowing what to do?

No, they *cannot* go ahead because they're not authorised to upload any
of those packages, or to do NEW uploads.

If they want to do that, the obvious and supported way to do it would be:

	- talk to some KDE people who *do* know what to do
	- start hacking on the package outside of Debian
	- get it to a point where it's working okay
	- get a Debian KDE person to review the packaging, and fix any
	  problems identified
	- ask the Debian KDE person to co-maintain the package
	- have the Debian KDE person upload the package to unstable (as a NEW
	  package) with the DM listed as a co-maintainer
	- as problems are found in unstable, have the co-maintainers
	  fix them and reupload

If the Debian KDE person decides, at any point, that the DM is being
incompetent at maintaining the package they can remove that DM's ability
to work on that package by uploading a new version that removes the DM
from the Maintainer/Uploaders field. So can the RM team or any other DD
via NMU. Likewise if the package is removed from unstable, it can't be
reuploaded by the DM because it needs a DD upload to get through NEW.

If the DM is already experienced with KDE packaging and wants to package a
new application, then they just need to find a DD they've already worked
with on KDE stuff, and give them a packaged version to upload. In that
case the DD might not feel the need to do as thorough review or be a
co-maintainer, because they've already seen the DM in action maintaining
similar packages. And if they turn out wrong, any DD can still do an NMU
to fix problems, or, if necessary, hijack the package or get it removed
from the archive.

> I fear that the DM thingy is just invented to get more people maintain
> packages in Debian without becoming properly involved, 

People should be able to contribute at the level they feel comfortable
with; if that increases over time, that's great; if it stays constant
or decreases, we shouldn't try to force them to do more than they want,
or refuse to accept what they're willing to do.

That doesn't mean lowering our standards of what we distribute, just
being willing to accept packages that are able to be maintained to our
standards more efficiently than we currently do.

IMHO, YMMV, etc, of course.

Cheers,
aj

Attachment: signature.asc
Description: Digital signature


Reply to: