Package: libnetcdf19 Version: 1:4.9.2-1 Severity: normal netcdf is currently built against the serial variant of hdf5 (libhdf5_serial.so.103).
But hdf5 provides a parallel build e.g. libhdf5_openmpi.so.103.3.1 via libhdf5-mpi-dev netcdf can be configured to support parallel io https://salsa.debian.org/debian-gis-team/netcdf/-/blob/master/INSTALL.md#building-with-parallel-io-support-build_parallel In principle parallel io could allow for much faster reading and writing of large data files. Should netcdf be configured to start enabling parallel io? Or would it be useful to provide both serial and parallel variants like hdf5 does?
