Thomas Weber <tweber@debian.org> writes: > On Tue, Aug 14, 2012 at 08:44:41AM +0200, Rafael Laboissiere wrote: >> My knowledge of the buildd is quite sparse, but I would guess that this >> will not happen if the solution proposed by Sébastien is adopted, because >> in the alternation of the liboctave-dev dependencies >> libhdf5-dev|libhdf5-openmpi-dev|libhdf5-mpich2-dev, the buildds will >> always pick the first one. Am I wrong? > > They will only pick the first one if none of the other two isn't > installed already. Some time ago there was the idea to build packages in > as "dirty" a chroot as possible.[1] > I don't like that idea personally, but the fact is that nothing forbids > a buildd maintainer from pre-installing packages (afaik, it's done with > texlive on some embedded architectures because generating font files > takes ages). I see, there is indeed nothing in policy or other authoritative documents that requires the buildd chroots to be clean. Do you therefore consider that we should refrain from implementing the proposed change, given the potential harm? My personal opinion is unchanged because in practice buildd currently use (almost) clean chroots, and many packages rely on that expectation (I think of blas/lapack shlibs system for example). Also, the very same solution with HDF5 alternatives did not create miscompilations in the squeeze era. My impression is that in the scientific world many people use parallel HDF5 (though I personally don't). Not providing mkoctfile to these people probably undermines the usefulness of our octave package. But of course I cannot substantiate this claim with actual numbers of users. The fact that nobody opened a bug report for this issue may be the sign that my concerns are excessive. -- .''`. Sébastien Villemot : :' : Debian Maintainer `. `' http://www.dynare.org/sebastien `- GPG Key: 4096R/381A7594
Attachment:
pgpjkVrcn1lqD.pgp
Description: PGP signature