Changes to Fortran interface

[iwheel]

I've made three changes each relates to one of my commits:

1: Fixes of bugs in API Fortran interface.
2: Addition of a new function PMMG_Get_NodeCommunicatorsf. Differs from the original function as it only returns the shared node information for a given neighbouring partition. This is because Fortran does not have the functionality to create a pointer pointer with different memory allocations on each index. Therefore the original function was not usable and just segfaulted.
3: Addition of a check when recovering global node/triangle numbers to check they have been calculated. Previously they are only calculated if the mesh was saved to file which always done when using the API.

I have created the pull request to the feature/tetFromTria-API as it had some added API functions required for Fortran interfacing.

With these changes ParMMG works with ElmerFEM (and Elmer/Ice) for both centralised and distributed meshes.

Let me know if you need any more information.

Best,
Iain

[Algiane]

Hi @iwheel ,

Thanks for your very clean work and contribution and for its comprehensive dexcription.

I would like to propose to rename the new PMMG_Get_NodeCommunicator_nodesf in PMMG_Get_ithNodeCommunicator_nodes as it seems more consistent with the other API names: does this look ok to you ?

Best Regards,
Algiane

[iwheel]

That makes more sense and looks good to me. I assumed the new routine would be renamed but thought I'd leave that up to you.

Best,
Iain

[Algiane]

Is it possible to authorize me to push inside your fork so I can bring modifications to the PR?
(I think that the doc to do this is here)

Thank you by advance,

[iwheel]

I don't seem to have the option described in the link. I can just give you temporary write permission to the fork if that works?

[Algiane]

Thanks again for your contribution: I will merge the PR and submit a new one to integrate the feature/tetFromTria-API branch inside the develop one.

Have a nice day.