Sphinx is double-counting potentials

[liamhuber]

Summary

Sphinx is double-listing potentials.

pyiron Version and Platform

Everything bleeding-edge git.

Expected Behavior

from pyiron_atomistics import Project
pr = Project('scratch')
pr.remove_jobs_silently(recursive=True)
job = pr.create.job.Sphinx('foo')
job.structure = pr.create.structure.bulk('Al', cubic=True)
job.list_potentials()
> ['Mg_GGA', 'Al_GGA']

Actual Behavior

from pyiron_atomistics import Project
pr = Project('scratch')
pr.remove_jobs_silently(recursive=True)
job = pr.create.job.Sphinx('foo')
job.structure = pr.create.structure.bulk('Al', cubic=True)
job.list_potentials()
> ['Mg_GGA', 'Mg_GGA', 'Al_GGA', 'Al_GGA']

Further Information, Files, and Links

It's causing this test to fail. I ran into it during #425, but IMO this is actually just an issue with the source code and outside the scope of what I'm working on.

I also am having a really hard time following what these tests are doing -- there is a massive amount of stuff happening in setUpClass. Is it possible these can be re-written and massively simplified using pyiron_base._tests.TestWithProject or TestWithCleanProject?

[samwaseda]

Oops someone finally noticed it. ok I'm gonna take a look at it this afternoon 😎

[samwaseda]

I also am having a really hard time following what these tests are doing -- there is a massive amount of stuff happening in setUpClass. Is it possible these can be re-written and massively simplified using pyiron_base._tests.TestWithProject or TestWithCleanProject?

It's slightly unrelated, but one of the reasons why it's massive is because it tries to support several SPHInX versions, although I think now it's time to abandon 2.3, because it's been now 3 years since this version is not used anymore by default. Is there a guideline for not supporting certain versions?

[liamhuber]

Is there a guideline for not supporting certain versions?

Not that I'm aware of. We try to keep backwards compatibility of the pyiron code itself until there's a major version number bump (i.e. first digit), but I don't know exactly how we do/should handle called executable version support.

[samwaseda]

ok I presume for now it's safe to remove the support for v.2.3 because there was (and still is) hardly anyone using SPHInX. It might be good to talk about it at the next pyiron meeting.