Re: RFS: ceph
On Tue, 9 Mar 2010, Asheesh Laroia wrote:
> > Reviewing now. Initial thoughts:
> > * Your diff.gz is empty. I guess that's better than making this a native
> > package, but other maintainers will prefer that the debian directory be
> > added during in the diff.gz. I think the ftpmasters might find your way a
> > little shocking.
> > * You should remove the commented-out dh_* lines in debian/rules.
> Generally a great package. Hey other mentors -- is an empty .diff.gz something
> that will make the ftpmasters upset?
> If not, I will upload it immediately. Sage, I hope we work together to improve
> the other issues, but they're minor bugs compared to the great work you've
> done so far.
Thanks! I'm moving debian/ into the diff.gz and cleaning out
debian/rules, but will need some time to remember what my problem was with
dh_strip (IIRC I couldn't make it generate multiple -dbg versions for the
various binary .debs?).
I'm having one other problem: I'm using pbuilder to build debs on multiple
platforms. So I used dpkg-source -b to generate the original .dsc and
.diff.gz. Then copy these to another platform (i386 in this case) and run
sudo pbuilder build --basetgz ~/debian-base/sid.tgz --distribution sid \
--buildresult release/0.19.1/out --debbuildopts -j`grep -c processor \
The problem is that pbuilder is generating a _new_ .dsc and .diff.gz, with
a slightly different md5sum (maybe in a timestamp in the .gz?), and the
.changes file references the new .dsc. Which then makes reprepro complain
when I try to put it in the repository. Shouldn't there be a single
version of the .dsc and .diff.gz for all platforms? Is there any reason
pbuilder is generating a new one, and/or is there a way to make it not do
that? (I'm running pbuilder on lenny, if that's relevant... maybe I
should try on a i386 sid machine.)
Also, should I be signing the .dsc?