I was asked to look at this thread.
John Goerzen wrote:
JG> * Despite having Important, Critical, or Grave bugs filed against their
JG> packages in frozen, some developers still ignore them. They could
JG> at least say "I can't fix this; can somebody else help?"
Unfortunately there are too many of them. However if a bug isn't
fixed for a long time, it is obvious that either the maintainer didn't
notice the bug, doesn't have time to fix it or lacks skills to
investigate and fix it.
In either case this is a good reason for *anybody* to investigate the
situation and provide the maintainer with a patch - and offer to do an
NMU. If he likes he could also offer to take over the package.
Debian is like a bazaar of small cathedrals, each cathedral is that
small that it can be useful but doesn't need to. In this case people
should drop the idea that only the registrated package maintainer is
able and allowed to fix a bug in his package. Everybody is.
This is some sort of quality assurance we definitively lack. There
are only very few people who go through different packages and fix
bugs in them. Espy, Topi Miet...something and Torsten Landschoff come
into my mind. We need more of them.
JG> * I have recently encountered a developer that has completely ignored
JG> some bugs for over 700 days and numerous bugs for over 200. His packages
JG> are languishing, and have required NMUs simply because they're so old
JG> that they mess things up with other software. E-mails get no reply, bug
JG> reports get no reply, etc.
Time to continue with NMU's and to ask him to orphan the package or
give it to you so it is maintained again.
JG> I view both of these as developers failing to live up to the responsibility
JG> that they accepted when they accepted maintainorship of packages. The
JG> second example is particularly bad; the developer could at least reply to
JG> the person submitting a bug.
Sure. However I've learned that only very few people are doing this,
including me (err, excluding? anyway, I'm not answering every bug
report I receive, I guess). Answering would be kind but can't be enforced.
JG> In light if this, I ask if we have some mechanism for either removing a
JG> developer from Debian if they have serious lack of responsibility issues, or
JG> at least of declaring their packages orphaned and up for adoption. If not,
JG> I would like to ask that we consider drafting a policy for such a situation.
Absence (sp?) is a reason for removing maintainers from the project.
Do get people removed please investigate the issue and try to reach
the maintainer. Get in touch with debian-admin or new-maintainer
afterwards.
Presence but lack of response is not a reason for dopping the person.
It is however a reason to bring up the issue on -devel and ask to take
over his packages.
JG> I am not proposing throwing people out for having bugs in their packages, or
JG> even for having lots of them. Rather, what I am proposing is a way to
JG> prevent people that ignore their responsibility from becoming a hindrance to
JG> the quality and freshness of Debian.
If people lack responsibility, please get in touch with them and
try to resolve the issue: a) talk to them, b) provide patches, c) do
NMUs, d) take over the package.
To give you a "scary" example, I filed a bug against a package I
maintained myself 2.5 years ago. I noticed that I would need some
time to figure out how to fix it, the problem was easy to detec. The
package was taken over by s/o else. Then he had to go to the army
(poor guy btw.) and I temporarily took over the package again. The
bug still was existant. But anyway, after two years (it might be
three, I don't remember exactly) I found time on a weekend where I
fixed the bug and rearranged a lot in the package, more than would
needed for a regular patch.
All this discussion, of course, does not touch the people who have
stated that they're going to be absent for a certain amount of time
(army time, severe illness, holliday etc.).
Hartmut Koptein wrote:
HK> > I would, too, for one thing. If a maintainer has been inactive (no email to
HK> > lists, no packages uploaded) for a year (12 months). Email should be sent to
HK> 12 months ? If i send an bug-report i expect an answer within 1-2 weeks!! Ok, for
HK> some packages (like dpkg) i wait longer.
I guess Gecko wasn't refering to bug reports when writing 12 months.
I would also expect a faster reaction after sending a bug report.
However if there is none, please nag the people, I support haggie. It
would be helpful if you could provide a patch as well.
As maintainer of a lot of packages I can say that is a darn job to be
responsible for a bug that you can't reproduce or where you think that
it's not that important and you have other jobs to do as well. Look
at my list of open bugs, there are some I really have difficulties
working on them.
Thus if there is no response it might help speed things up if you
would provide a proper patch. *Please do so*
HK> So, if a maintainer doesn't answer within _one_ month he should be mailed!!! After
HK> a second month he should be set to 'hold'.
Haggie, what do you expect from maintainers set to 'hold'?
HK> 12 months <hehe> that means two versions of debian releases!
Ideally this means four versions, I remember there was a decision made
to release Debian every three months *lol*
Since it occured several times, increasing the severity doesn't fix a
bug. You should only change the severity of the bug report if the
severity of the bug changes, i.e. cripples your system or something.
If you need this bug fixed, please go, investigate and fix it
yourself.
HK> I don't understand why maintainer doesn't upload packages after an bug report, when
HK> the bug is easy to fix or it was allready included in the bug-report. Why waiting
HK> for the next upstream upload?
I don't have a proper reason for this. Please nag the maintainer.
Even worse, I guess I still have such an open bug I tend to forget
about. If you need it, nag me. (err, please don't, nag the others *g*)
HK> I'll do make some NMU in the next time ( 2 - 3 weeks after the bug-report), if i don't hear
HK> anything from the maintainer. Some of the maintainer send me a mail "please don't do
HK> an NMU, i will do it next weekend". This is then ok for me. I know then that the
HK> maintainer is active and will work on his package.
Did you offer to make an nmu? If so your reaction is proper.
John Goerzen wrote:
JG> When do we finally remove someone? When they haven't logged in to master
JG> for 5 years? Or do we just let them keep their accounts on Debian machines
JG> even though they do nothing?
This would be silly. First of all master isn't five years old, even
if it would we changed master 2-3 years ago and it was set up
completely new at least once more. Secondly a maintainer does not
need to log in into master. People can upload their packages through
upload queues if master seems to be "too far away" for them.
A maintainer is remove if he quits Debian, was kicked off (for
whatever reason) or if he "informally" left.
Oh well, and there are, for sure, people who do not maintain any
package but work for the project in any other kind. We should
probably not remove these guys'n gals.
Regards,
Joey
--
GNU does not eliminate all the world's problems, only some of them.
-- The GNU Manifesto
Please always Cc to me when replying to me on the lists.
Attachment:
pgpMT2xHHAMck.pgp
Description: PGP signature