gmsh.cc: add test that passes an input .dmg model

[cwsmith]

We need to that a case where the first argument, the input model, is not none:

https://github.com/SCOREC/core/blob/d7cc6c8341e8154a3e33f4746a044d4ebfd84178/test/testing.cmake#L231-L244

This caused problems with a downstream tool: https://github.com/SCOREC/gmshToPumi/issues/2

Note, the test/gmsh.cc (from_gmsh) command line interface has changed a couple times since this repo/example was created. The first change added support for deriving a model if a .null model was passed in (https://github.com/SCOREC/core/commit/30d5d9fbf064d3a807bdf1fb5a362b73c863bafc) and the second major change added support for gmsh v4 (https://github.com/SCOREC/core/commit/b27c9b2b93d01e018bfbec92d3a215a67192d19b).

[zhangchonglin]

@cwsmith: thanks for creating this issue. From looking at the gmsh.cc code: https://github.com/SCOREC/core/blob/d7cc6c8341e8154a3e33f4746a044d4ebfd84178/test/gmsh.cc#L40-L52

• it seems that with gmsh version 4, the .dmg file is not needed, where function loadMdsDmgFromGmsh does not accept .dmg input file: https://github.com/SCOREC/core/blob/d7cc6c8341e8154a3e33f4746a044d4ebfd84178/mds/mdsGmsh.cc#L469-L475

• for gmsh version 2, .dmg input file is an optional file.

It seems that the if statement corresponding to gmshVersion == 4 needs to be restructured to be consistent with all types of input arguments.

[cwsmith]

it seems that with gmsh version 4, the .dmg file is not needed

That is correct, v4 can embed sufficient info into the .msh file for creating the output .dmg. I'm not sure if all v4 gmsh meshes will have this info though.

loadMdsDmgFromGmsh does not accept .dmg input file

Yeah, thanks for digging.

Regarding the if statement in gmsh.cc for v4, I agree, there should at least be an else that prints a helpful error message.