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

DAM and NEW queues processing


For years, the DAM and NEW queues have been the major source of
frustration in the Debian community. Several attempts have been made to
improve the situation, but the problems have never been really solved so
far. And the queues are again in a bad state.

First, DAM: there are 19 prospective DDs waiting for DAM approval and 2
for account creation. Some of them have been FD-approved a very, very
long time ago: Alexander Gerasiov has been FD-approved on 2009-01-10,
and Asheesh Loria on 2008-12-22 (but apparently, because of an
unanswered RC bug, DAM postponed approving him a month ago). This make a
lot of skilled people just decide not to apply for DD, because it takes
too long, and is too frustrating. I'm sure that many of the current DD
would just give up if they were asked to show the same patience and
commitment as the current applicants. And no, the fact that some of us
went through similar delays is not a reason not to make the Debian
community slightly more attractive :-)

Then NEW. Nothing out of the ordinary here: NEW delays are often raised
on -devel@ (see [1] for example), and it's apparently considered normal
to wait 2 or more weeks before one's package gets reviewed. Since this
often blocks other works, it is a major source of slowdown in Debian. Of
course, it happens that the NEW queue is nearly empty and processed
almost daily[2], but we depend on the free time of too few people, so
good times never last.

[1] http://lists.debian.org/debian-devel/2009/03/msg01556.html
[2] http://molly.corsac.net/~corsac/debian/new/

It is clear, based on the previous attempts to solve those problems,
that simply throwing more manpower on the DAM and ftpmasters team won't
solve those problems. We have tried that for years, and it has failed
for years.

We need to compromise on the level of quality we expect from our
prospective DDs and new packages. We need to accept that new maintainers
are humans, and will sometimes make mistakes, no matter how many
questions you ask them. And that packages with licensing problems will
continue to be uploaded to Debian despite all the checking done by
ftpmasters: licence problems are already being introduced in packages
that already went through NEW (through the addition of new code in
existing packages, for example). We put a lot of energy into processes
that are not totally efficient.

We should seek processes that scale. For example:

 - the NM process could be reduced to 5 to 10 questions choosen by the
   AM amongst the 50+ questions currently in the NM templates, to verify
that the applicant has some knowledge about different aspects of Debian
packaging. Then the AM would ask for comments about the applicant from
other DDs, like it is already being done for DM. That would make the AM
report a lot shorter to read, and spread the load on all DDs, that would
have to write recommandation emails about the applicant (including links
to work the applicant has done). It would also help avoid
socially-problematic applicants, because it would be a de-facto
requirement to work with several other DDs before becoming a DD.

 - the NEW queue could also be based on peer-review: one could ask one
   or two another DDs to certify that the package is OK (licensing-wise)
to be uploaded to unstable. Then ftpmasters would just be responsible
for verifying that enough DDs have reviewed the package. And ftpmasters
could still choose some "interesting" packages and check them manually.
Of course, this will lead to buggy packages being uploaded to Debian.
But that is already the case, and Debian has never been sued so far,
AFAIK. It will also create a culture of asking for reviews from other
maintainers, which, in the long term, could help improve the quality of
the distribution as a whole.

The above proposals are just examples. I'm not expecting them to be
adopted without changes, but I hope that the possibility of wide changes
in the NM and NEW processes will be discussed. One possibility to move
forward in this discussion could be to have a poll, to know how DD
currently feel about those problems, and if we should really seek
solutions, or if the status quo satisfies everybody.
| Lucas Nussbaum
| lucas@lucas-nussbaum.net   http://www.lucas-nussbaum.net/ |
| jabber: lucas@nussbaum.fr             GPG: 1024D/023B3F4F |

Attachment: signature.asc
Description: Digital signature

Reply to: