Re: Don't use dpkg-buildpackage -m, especially not with NMUs
On Thu, Sep 16, 2004 at 01:01:26PM +0200, Daniel Kobras wrote:
> On Thu, Sep 16, 2004 at 11:41:36AM +0200, Jeroen van Wolffelaar wrote:
> > The .changes do say:
> >
> > Maintainer: Daniel Kobras <kobras@debian.org>
> > Changed-By: Daniel Kobras <kobras@debian.org>
> >
> > however, I don't know how that happened then, as 'Maintainer:' should
> > still be copied from debian/control. Maybe you've some global
> > configuration file/environment variable with the same effect?
>
> That's the correct setting for an NMU. The use of the Maintainer field
> in .changes is confusing and does not correspond to the Maintainer
> setting in control. buildd uploads for example change it to the buildd
> admin. Basically it tell where the archive maintainance software should
> send mail to. Stuff gets more interesting with sponsored uploads, but
> NMUs just work out of the box without additional options to
> dpkg-buildpackage. What was the actual problem that triggered your
> mail?
Really? Most NMU's I see have in their .changes the actual maintainer as
Maintainer:, and not the NMU'er. This way, the maintainer will get a
copy of the katie notices, which in fact is a good thing (I'm not sure,
maybe the maintainer gets that anyway). It's true for buildd's and other
binary-only uploads, as a maintainer generally doesn't care about
binary-only uploads, and the notices etc should be going to the person
uploading the binary-only, hence buildd addresses in Maintainer:
Regarding actual problem, from the archives, I see at least this
thread[1] about incorrect use of -m. By googling I learned that not
everybody is convinced about not using -m for sponsored uploads despite
the fact that bugs get marked as fixed, Andrew Suffield being an example
that keeps returning. The argument is that the sponsor then gets katie
mails too (but that can also be achieved via PTS, and a sponsor usually
should monitor the package anyway). Another thing that occasionally goes
wrong[2] is using -e for sponsored uploads, and is why I originally
thought that the difference changed-by <-> maintainer is sourceful
.changes is the reason for being considered NMU (and I now realize that
it is 'is changed-by part of maintainer or uploaders in the .dsc?').
Okay, in short: it's all quite confusing, and documentation is not very
adequate on these issues afaics. My excuses for not having checked in
your, Kobras', case whether indeed the message to the BTS had the
desired effect or not.
--Jeroen
[1] http://lists.debian.org/debian-newmaint/2003/05/msg00004.html
[2] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=222406&msg=5
--
Jeroen van Wolffelaar
Jeroen@wolffelaar.nl (also for Jabber & MSN; ICQ: 33944357)
http://Jeroen.A-Eskwadraat.nl
Reply to: