Fix ThermoML parsing#756
Conversation
Codecov Report❌ Patch coverage is 🚀 New features to boost your workflow:
|
|
Well, the test errors here actually suggest I should have been blaming OpenEye, not RDKit, oops. Edit: I take it back -- I put in the oechem skip condition for the fix and then immediately forgot about it. It's RDKit! |
62bbe3f to
84084ff
Compare
for more information, see https://pre-commit.ci
There was a problem hiding this comment.
This looks good but I found two one thing] I would like another look at; neither call for re-review if the outcomes are as expected
- A blocking comment in the code about behavior when RDKit is not installed which can be resolved by you agreeing it's an okay and intentional change. No need to re-review just for that
Triggering the failing test (#757), i.e. that check failing as expected-- now resolved: test failing/skipped as expected: https://github.com/openforcefield/openff-evaluator/actions/runs/23498429535/job/68386169870?pr=757
| except (LicenseError, InvalidIUPACNameError, Exception): | ||
| pass # Fall through to original behaviour below |
There was a problem hiding this comment.
This block isn't hit by code coverage, which from my reading of the test is because the new, correct behavior requires OpenEye. I think this is okay in the current context of all relevant stakeholders having access to a license, but we should document that this case is bad without an OpenEye license
There was a problem hiding this comment.
It now raises a warning if openeye isn't available.
| to the same string for both tautomers; the common name disambiguates. | ||
| Requires an OpenEye licence for ``Molecule.from_iupac``. | ||
| """ | ||
| pytest.importorskip("openeye.oechem") |
There was a problem hiding this comment.
(non-blocking, see comment in an except block below) it would be nice if this test encoded that the old/bad behavior persists without this license, not necessarily saying that is a desirable outcome but the test should capture the behavior we think happens
There was a problem hiding this comment.
I split out this test to test the case where openeye both is and isn't installed.
Actually relates to this dimsim issue (openforcefield/dimsim#53) but it's a shared issue as the ThermoML code was simply ported.
Edit: I messed up and rebased this off of #754 by accident, sorry