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

Re: removal requests of unattended RC buggy packages?

On Tue, Jun 21, 2005 at 01:53:43PM +0200, Jeroen van Wolffelaar wrote:
> On Tue, Jun 21, 2005 at 02:26:30AM -0700, Steve Langasek wrote:
> > I'm trawling the list of build failures on various architectures, and I'm
> > finding packages that have never, ever been successfully built on an
> > autobuilder from day one, in spite of bug reports being filed early and
> > often, e.g.: nemesi, bug #303075.  What should be the policy for requesting
> > removal of such packages?  I think two months is more than enough time for a
> > maintainer to get their act together before having the package bounced back
> > out again.  Anyone disagree?

> On the one had I agree -- the package was in that case only not
> rejected from NEW because the ftp-master didn't try to rebuild it.

> But typically, when some package is too buggy and the maintainer
> doesn't fix it, one wants to do something about that maintainer in
> general (including in this particular case). I'd like to take this a
> bit more general, that if some package is left buggy for a certain
> period of time, with no maintainer activity, it can get orphaned (and
> on the opinion of QA, then removed if that sounds sane).

I find this to be a horribly demotivating amount of proceduralism just to
get a package removed to the archive which, we're all agreed, should never
have been allowed in given the state of its packaging.  I think this really
explains quite a lot about the state of unstable today (basically close to
800 source packages that are perpetually uneligible for testing, with or
without bugs filed), and I sincerely hope that now that the NEW queue is
being processed regularly, we can work towards a process for removing buggy
packages that isn't quite so heavy on the paperwork and the opportunities
for veto.  It's absolutely appropriate to give the maintainer, or would-be
adopters, an opportunity to object before the removal; but when an RC bug
goes unanswered for two months after the package is accepted into the
archive, and the average wait time for reintroducing the package to the
archive is down to a week, I don't think we should be waiting very long at
all for objections.

> In the case of nemesi by the way, it has no real-life maintainer, so
> can and will be orphaned immediately, thanks for noting. Someone else
> should then reassign to ftp.d.o to get it removed because it's too
> buggy, because I don't want to do the whole process of asking for
> removal and then doing it all by myself to prevent misjudgments.

Right, so, I've reassigned this to ftp.d.o now. :)

> So, basicly, my opinion is: 'it depends'. If you have a list, feel
> invited to send it to -qa :).

Unfortunately, these don't seem to show up in nice, consistent batches.  The
other package I've found so far that's in the "RC buggy since day one" state
is linuxsampler, but it has only been in the archive since the beginning of
June, and the maintainer has made some effort to fix it (if only in a
hit-and-miss fashion...)

Steve Langasek
postmodern programmer

Attachment: signature.asc
Description: Digital signature

Reply to: