Closed mattwthompson closed 3 years ago
Some OpenEye failures as well: https://github.com/openforcefield/cmiles/pull/42/checks?check_run_id=1083890256
Today, I'm not seeing any failures with OpenEye tests
RDKit failures, as of today: https://gist.github.com/mattwthompson/8dd44b88be0551653c33fdc028c02acb
The gist of #44 is that the test file was manually updated. This fixes the tests in the sense that they run and pass with the most recent version of RDKit, but it's kinda cheating since the tests run against a file that was generating by a script using RDKit 2020. This stems from the unresolved issue of the same version of CMILES producing slightly different "canonical" atom ordering with different underlying toolkits, which has been discussed elsewhere.
Lots of drugbank molecules are failing; I assume these are related to upstream changes like https://github.com/openforcefield/openforcefield/issues/564 since these tests presumably passed months ago.
I have not dug into this at all yet
https://gist.github.com/mattwthompson/11aa758b01df287d815c2b26be29d1c6