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

Re: Allowing QA uploads for DMs (was: A lot of pending packages)

On Tue, 15 Jun 2010 14:59:04 -0700
Steve Langasek <vorlon@debian.org> wrote:

> On Tue, Jun 15, 2010 at 08:50:28AM +0100, Neil Williams wrote:
> > What about if Debian QA packages were all to be deemed suitable for
> > DM upload, including those which have been orphaned for over 2
> > months without a change of maintainer? Maybe when an orphaned
> > package is uploaded with the change of maintainer to Debian QA, the
> > DM upload field could also be set?
> Ugh, what a terrible idea.  DMs are by definition uploaders who have
> *not* yet demonstrated, to the satisfaction of the project, their
> capability to do unsupervised uploads of arbitrary packages.  Why are
> you so eager to gut our QA processes?

I'm not trying to undermine QA, I'm trying to get more people involved
in QA instead of directing new people at new packages which become the
QA workload of tomorrow.

I'm basing this on the idea that some packages in QA only need small
updates as the majority of the packaging work is already done. Yes,
there are some that were orphaned purely because the packaging is too
hard or the upstream code is just awkward but a lot are orphaned because
the volunteer maintainer had a change in their real life priorities,
through no fault of the package itself.

Maybe there could be a way of indicating which packages in QA fall each
side of such an evaluation. i.e. Orphaned-and-borked or
orphaned-but-ready. Debtags might be a solution for that, with suitably
renamed special QA tags or maybe comments/tags in the O: bug report.
This could be similar to the low-NMU status.

DM's have not yet demonstrated the capability to do unsupervised
uploads of NEW packages but QA uploads can be *less* work than
packaging an entirely new package. If a new package is significantly
less work than fixing a few lintian issues in an orphaned package, I'd
question whether the new package concerned is worth packaging in the
first place.

> DMs should request sponsorship of QA uploads on debian-qa just like
> anybody else.  If they consistently demonstrate their competence in
> this fashion, they should be recognized for this by making them full
> DDs - not by conferring additional rights on DMs that the DM
> admissions criteria aren't set up for!

I think we have to consider redirecting new volunteers AWAY from
assuming that their work must centre on a NEW package and make it
equally (or even more) likely that new people get to learn their craft
on existing, orphaned, packages. After all, these packages are the work
of their peers, albeit inactive peers.

If a package has been orphaned long enough that it is already under QA,
it's fairly obvious that having anyone take an interest in it is better
than just leaving it bit rot. Equally, I submit that getting orphaned
packages updated is a more worthy goal than adding another NEW package
that will become a QA package if that contributor loses interest. We
need to discourage me-too packages more firmly. We also need to
dissuade new contributors from taking only a narrow interest in a
single new package and instead gain an understanding of the wider needs
of the project.

Getting new contributors to work on QA helps QA at both ends - by
drawing some packages out of QA and back into teams (or out of the
archive completely) and by discouraging new contributors from adding
new packages merely to get something done as a contributor, thereby
reducing the flow of packages into QA in the future.

This way, the results of the MIA team flow back into the project as
the work of new contributors.


Neil Williams

Attachment: pgpHsgshbwcGH.pgp
Description: PGP signature

Reply to: