Should we record the path from which an XML is loaded?

[mattwthompson]

See @uppittu11's comment: #298 (review)

Another thing we could add is an attribute to the Forcefield class which gives the path from which the forcefield was loaded. This wouldn't be saved in the forcefield xml, but is only an attribute. This might be a dumb idea, since we might have issues comparing two of the same forcefield loaded from different paths.

I think this could be useful but don't feel strongly about it one way or the other. What does everybody else think?

[ahy3nz]

In the grand scheme of conda-installable FF repos, file paths for those XMLs could be unnecessary if they're hidden away in 5 conda folders somewhere? But in the current status of git cloning and pip installing, it could be helpful for debugging purposes for a user to know exactly where the XML is? Otherwise, foyer could have a large list of entry points and force field names, but a user might not remember which force field entry point corresponds to which file path?

[ahy3nz]

related: openforcefield/openff-toolkit#477

[InnocentBug]

Just an idea, while browsing these issues.
Maybe instead of logging the file path to log a hash of the forcefield files used.
This enables at least to verify that one specific forcefield has been used.
It does not allow full reconstruction, but verification.