[Request for comments] Publish in JOSS?

[awvwgk]

The MCTC library has been around for a while and is supporting quite a few packages for reading different computational chemistry input formats. Maybe it is worth to put some time to publish the current state of the repository in a software paper with the Journal of Open Source Software (JOSS).

Why is it needed?

• Single library for geometry input formats shared between several Fortran packages (xtb, tblite, dftd4, s-dftd3, etc.)
• 11 support geometry formats for input and output (xyz, sdf/mol, pdb, cjson, qcschema, gaussian, gen, fhi-aims, qchem, turbomole, vasp)

Uniqueness in this library

• Full Fortran implementation
• Detailed error reporting

Adoption outside xTB based software

• Do we know of any library outside our own libraries which adopts this library?
• Should be proactive and integrate this library with other Fortran packages? (maybe DFTB+ or MOPAC)

Alternative libraries

• https://open-babel.readthedocs.io/ (C++, ~110 formats)
• https://github.com/chemfiles/chemfiles (C++, 25 formats)
• https://github.com/cclib/cclib (Python, 16 formats)
• https://wiki.fysik.dtu.dk/ase/ (Python, ~100 formats)

[awvwgk]

ping Marvin @marvinfriede, Thomas @thfroitzheim, Eisuke @e-kwsm, Marcel @MtoLStoN

[MtoLStoN]

Publishing scientific software in a dedicated journal is definitely beneficial for the community. It certainly provides a clear, citable reference instead of relying on a GitHub link or snapshot, which makes things easier for researchers. I'm also not aware of any other easily available fortran library for handling molecular geometries. So I think publishing this repo is a good idea and worth the effort.

Having said that, I'm not sure if I would agree that my contributions to this library are substantial enough to warrant authorship.

[marvinfriede]

Publishing scientific software in a dedicated journal is definitely beneficial for the community. It certainly provides a clear, citable reference instead of relying on a GitHub link or snapshot, which makes things easier for researchers. I'm also not aware of any other easily available fortran library for handling molecular geometries. So I think publishing this repo is a good idea and worth the effort.

I agree and would be happy to help.

[thfroitzheim]

I also think that this is a good idea and I am open to help.

It could also be an idea to mention some additional features such as the CN or lattice definitions (if we move them), which make it a "tool chain".

[e-kwsm]

How about zenodo?

---

I'm not sure if I would agree that my contributions to this library are substantial enough to warrant authorship.

neither me