feat: some cleanup

[n0w0f]

@RezaAliakbari I have done some cleanup. maybe you can review the code?

[kjappelbaum]

@n0w0f and @RezaAliakbari at least for me the tests failed. Probably due to a different Python version.
For this reason, I now used PDM to also create lock files.

I also linted the package and removed other unneeded parts (I think our cookiecutter should be updated).

[kjappelbaum]

It would be good to understand the test failures and if we need some additional version pinning or processing of the data to ensure that we don't fool ourselves when we generated millions of text representations (this would especially hurt us, if we collaborate with different python environments)

[n0w0f]

It would be good to understand the test failures and if we need some additional version pinning or processing of the data to ensure that we don't fool ourselves when we generated millions of text representations (this would especially hurt us, if we collaborate with different python environments)

Thank you for reviewing the code, there were some bugs which i have fixed now. Will add more test cases as well.

[kjappelbaum]

yeah, I wonder why this wasn't added by pdm. I say that this failure is caused by this and it is also in the lockfiles. It should automatically be obtained as a tensorflow dependency and not manually by us

[n0w0f]

yeah this came with slices. Was not sure why it was not installed by pdm

[kjappelbaum]

this thus doesn't feel like the right fix ...

[n0w0f]

did pdm install absl-py for you in your local ? I had conda env with slices & m3gnet, which might have added absl-py. before i started using pdm

[kjappelbaum]

you are right, there might be a confounding. Let me clean that

[n0w0f]

pdm is not adding absl-py ?

[kjappelbaum]

no, because there should be no need. It is an indirect dependency.

[kjappelbaum]

I do not have time now to look deeper into it, I'll add it back, but it is somehow strange and should not be needed and there must be a bug somewhere

[kjappelbaum]

ok, has been some venv issue on github actions, wtf

[kjappelbaum]

this PR now really must be squashed :D

[kjappelbaum]

@n0w0f the robocrys thing is still failing. Do we need to have a harder pin on the version? Or can we make the test a bit less strict?

[n0w0f]

@n0w0f the robocrys thing is still failing. Do we need to have a harder pin on the version? Or can we make the test a bit less strict?

Not sure about the origin of this issue. Test passed for N2, locally (will cross check dependency local and remote ), and how representation changes with versions. pymatgen version in CI and local is same.
for now testing on inorganic system.

[kjappelbaum]

lgtm!

[n0w0f]

incorporate #14 #7