On Tue, Apr 29, 2025 at 02:07:56PM +0200, Santiago Vila wrote: > Thanks for the report, I was actually thinking about reporting this myself > to get some help. > > El 29/4/25 a las 3:09, Chris Hofstaedtler escribió: > > esys-particle previously built on all archs, but now cannot build on armel, > > armhf, i386, as its Build-Depends are not satisfiable: > > > > ∙ ∙ esys-particle unsatisfiable Build-Depends(-Arch) on armel: openmpi-bin > > ∙ ∙ esys-particle unsatisfiable Build-Depends(-Arch) on armel: > > libopenmpi-dev > > ∙ ∙ esys-particle unsatisfiable Build-Depends(-Arch) on armhf: openmpi-bin > > ∙ ∙ esys-particle unsatisfiable Build-Depends(-Arch) on armhf: > > libopenmpi-dev > > ∙ ∙ esys-particle unsatisfiable Build-Depends(-Arch) on i386: openmpi-bin > > ∙ ∙ esys-particle unsatisfiable Build-Depends(-Arch) on i386: libopenmpi-dev > > > > Either this needs to be fixed or the old binaries need to be removed, to > > allow testing migration. > > > > If r-deps exist, please coordinate with them. > > I made a team upload (2.3.5+dfsg2-9) from a NMU, then fixed a problem in the > NMU > in the next upload (2.3.5+dfsg2-10). I believe the root of the problem is > this one: > > * Build depend on libopenmpi-dev until someone figures out what > goes wrong with MPICH. (Closes: #1101991) > > Am I right to think that openmpi does only exist for 64-bit archs, and > therefore > replacing mpich by openmpi is actually equivalent to making the package > 64-bit only?
This matches my understanding. > Can this be fixed by asking ftpmasters to drop the now unbuildable binaries, > or does it require more uploads? (there are no reverse build-depends). Without knowing anything about the package in question, this seems correct. The only r-deps shown by apt-cache use Recommends:, which should be acceptable. > Cc: to Anton Gladky, the usual uploader. I'll try to fix this myself, > since it was triggered by an upload which I made, but I still > welcome help and guidance about what to do. Chris
