Can not use parallel target in CMake in pure C++ project

[cmacmackin]

Describe the bug When trying to link against the parallel HDF5 target in CMake in a C++ project, CMake complains that the MPI_C target is not found:

CMake Error at /home/cmacmack/.spack/opt/spack/linux-linuxmint21-icelake/gcc-11.3.0/hdf5-1.14.1-2-nabfxjrrsno6gmfw5fsnot4b5geztyd3/cmake/hdf5-targets.cmake:68 (set_target_properties):
  The link interface of target "hdf5-shared" contains:

    MPI::MPI_C

  but the target was not found.  Possible reasons include:

    * There is a typo in the target name.
    * A find_package call is missing for an IMPORTED target.
    * An ALIAS target is missing.

This is related to but distinct from the bug addressed by #2400. The reason for the error is that the project only enables the C++ language and CMake therefore does not import the MPI_C target. (Presumably there would be a similar issue in Fortran projects.) Adding enable_language(C) before importing the HDF5 package fixes the problem.

Expected behavior Parallel HDF5 imports the MPI::MPI_C package to satisfy its public dependency.

Platform (please complete the following information)

• HDF5 1.14.1-2
• Linux Mint 21.1 (which is based on Ubuntu 22.04)
• GCC 11.3.0
• CMake 3.24.3
• set(HDF5_PREFER_PARALLEL TRUE)
• MPICH 4.1.1

HDF5 and its dependencies were compiled using Spack v0.20.3.

[mirenradia]

I have the same problem in a pure Fortran project where Fortran is the only enabled language. I'm doing

set(HDF5_PREFER_PARALLEL TRUE)
find_package(HDF5 REQUIRED COMPONENTS Fortran)
if (NOT HDF5_IS_PARALLEL)
    message(FATAL_ERROR "Parallel HDF5 Required.")
endif ()

and my software versions are as follows

• HDF5 1.14.3
• RHEL 8.8
• Nvidia Fortran Compiler (nvfortran) 24.3
• CMake 3.27.9
• OpenMPI 4.1.6

HDF5, CMake, OpenMPI and their dependencies were built with Spack at d5c1e16e43.

Is it possible to get around this without enabling C (which slows down CMake configuration and is otherwise unneeded)? This only seems to be a problem with HDF5 1.14 and I've not had this problem with various build of 1.12 and 1.10.