After #105 is merged, add a scheduled version of the CI job that runs on a much wider matrix (up to 256 jobs) and tests everything once a week – that is, all possible HDF5 versions, debug/release, msvc/gnu, different ubuntu versions, nompi/openmpi/mpich, etc. Hopefully this would allow to detect problems before they hit the users.
After #105 is merged, add a scheduled version of the CI job that runs on a much wider matrix (up to 256 jobs) and tests everything once a week – that is, all possible HDF5 versions, debug/release, msvc/gnu, different ubuntu versions, nompi/openmpi/mpich, etc. Hopefully this would allow to detect problems before they hit the users.