Re: It seems I have messed up mpi-default in team upload
Hi Mattia,
On Thu, Apr 19, 2018 at 03:03:26PM +0200, Mattia Rizzolo wrote:
> On Thu, Apr 19, 2018 at 02:50:39PM +0200, Andreas Tille wrote:
> > index 16e8518..6a59dc9 100644
> > --- a/debian/changelog
> > +++ b/debian/changelog
> > @@ -1,3 +1,13 @@
> > +mpi-defaults (1.11) unstable; urgency=medium
> > +
> > + * Team upload.
> > + * Remove "Adam C. Powell, IV" <hazelsct@debian.org> from uploaders since
> > + he is not active in Debian Science team since 5 years.
> > + Closes: #547460
> > + * Standards-Version: 4.1.4
> > +
> > + -- Andreas Tille <tille@debian.org> Thu, 19 Apr 2018 13:31:37 +0200
>
> You forgot a bit here, I pushed it for the next time.
> I don't remember a policy in our team regarding "generate changelog
> before upload" vs "edit changelog at each change", but I believe the
> latter is more common here, so I guess it's my fault…
I see more and more people doing the latter and I'm in the process of
convincing myself to do so as well. I have no idea whether we should
fix this somewhere in our team policy since its not really visible in
the resulting package.
> > The error is:
> >
> > if [ "" = "openmpi" ]; then \
> > echo "mpi=openmpi-bin" > debian/mpi-default-bin.substvars; \
> > echo "mpi-dev=libopenmpi-dev" > debian/mpi-default-dev.substvars; \
> > elif [ "" = "mpich" ]; then \
> > echo "mpi=mpich" > debian/mpi-default-bin.substvars; \
> > echo "mpi-dev=libmpich-dev" > debian/mpi-default-dev.substvars; \
> > else \
> > echo "Unknown MPI implementation, stopping"; \
> > exit 1; \
> > fi
> > Unknown MPI implementation, stopping
> > debian/rules:109: recipe for target 'override_dh_gencontrol' failed
> >
> > Any idea what might went wrong here?
>
>
> Apparently neither mpich nor openmpi are providing
> /etc/alternatives/mpicc anymore, which broke mpi-defaults' assumptions.
> Guess a quick fix would be to rely on some other link, like `mpi` or I
> notice there is a `mpiCC` now (for openmpi, didn't look at mpich).
OK, fix pushed. I've seen you fixed the changelog so I'll upload
after rebuilding again.
> > Sorry for the mess - I assumed it would be a real trivial change
>
> The change by itself is trivial and fine, but didn't you test build it
> before uploading?... :)
Its worse: I test build here but I assumed some MPI related failure on
my test machine since I was remembering some trouble with this before.
I checked the build logs, was unable to believe that my changes could
have created the bug I've seen here and did a source only build. So
realising a build problem and uploading anyway is something you can
blame me about and I regret that I did so. Hope that my new upload
which builds here now is better ...
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: