The cpp and hl_cpp packages are not build when 'with_mpi == true', but the Requires in the main package unconditionally specifies %{libname -l _cpp} for the hpc flavors.
This makes e.g. science:netcdf unresolvable for the hpc flavors.
Is there actually a reason not to build _cpp and _hl_cpp for mpi? I know MPI itself deprecated C++ bindings with MPI-3, but this should not exclude the hdf5 mpi bindings from being build, correct?
The cpp and hl_cpp packages are not build when 'with_mpi == true', but the Requires in the main package unconditionally specifies %{libname -l _cpp} for the hpc flavors. This makes e.g. science:netcdf unresolvable for the hpc flavors.
Is there actually a reason not to build _cpp and _hl_cpp for mpi? I know MPI itself deprecated C++ bindings with MPI-3, but this should not exclude the hdf5 mpi bindings from being build, correct?