Hi Drew > It's the other way around. slepc4py depends on petsc4py, petsc4py > does not depend on slepc4py.
Right, slepc4py cannot migrate until petsc4py migrates [1], and petsc4py cannot migrate until slepc4py's autopkgtests pass [2]. We need some way to tell the package manager that these packages need to migrate together, and I think declaring that the new petsc4py breaks the old slepc4py is a way to do that. > As far as I can see the tests are working fine. slepc4py 3.9 requires > petsc4py 3.9, and therefore fails when attempting to install with > petsc4py 3.10. The versioned dependencies are already there. > > The true villain here is numpy, I think. In Ubuntu, slepc4py's tests have been passing [3], but only when we've forced them to run with the new petsc pacakges. Also, the new numpy has not been merged into Ubuntu yet, but we still see failures where the old slepc4py autokpgtests are attempted to be run against the new petsc4py package. Regards Graham [1] https://qa.debian.org/excuses.php?package=slepc4py [2] https://qa.debian.org/excuses.php?package=petsc4py [3] http://autopkgtest.ubuntu.com/packages/s/slepc4py/disco/amd64