increased use of intrinsic data types and ops for communication

• use move semantics for string parameters and reuse more code

changed milestone to %v2506

added enhancement label

assigned to @Mattijs

@Prashant - I think this is now looking pretty close. Could you drop this through the standard test loop with a different MPI? I've only used openmpi + clang int32 DP for testing. May need to throw a few weirder combinations at it as well.

@Mattijs - marked as draft, but there shouldn't be much (anything) more coming. Just the final checks I hope.

requested review from @kuti

added 1 commit

• bb6f6258 - COMP: workaround for broken in-place reduce INTEL-MPI (#3331 (closed))

Compare with previous version

marked this merge request as ready

resolved all threads

approved this merge request

Thanks Mark. All compilation tests are good.

added 16 commits

• dd09aa12 - 1 commit from branch develop
• dd09aa12...d4b52807 - 5 earlier commits
• 8c395357 - STYLE: more consistency in communicator types (int vs label)
• 7ac83f22 - ENH: refine the dataTypes handling
• 151f4df5 - ENH: add opaque data types to UPstream bridge code
• f0b844eb - ENH: generalize MPI broadcast to basic and user-defined MPI types
• c7fc9d4d - ENH: extend MPI send/recv types
• 20b6aeb4 - ENH: generalize reduce/all-reduce to include MPI types
• 3bf13990 - ENH: generalize mpiGather/mpiScatter/mpiAllGather
• 5d9f8e9a - ENH: remove gatherv/scatterv direct coding
• 20b2f463 - STYLE: prefer listReduce() to using listCombineReduce() when possible
• 28818c73 - COMP: workaround for broken in-place reduce INTEL-MPI (#3331 (closed))

Compare with previous version

@Prashant and @kuti - pushed with that stupid typo fixed, but that particular function is not yet in any active code branch (it's an optimization for broadcast list) - so no need to retest anything yet.