2021-04-28 Mitchell/Wagner Check-in

From last week

• Open one or more PRs to resolve the following:

  • 4 examples aren’t being tested currently because they require additional deps

  • CI testing goes slowly because it does unit tests and examples serially

  • isort/black conflict fix

• Look into building examples package on conda-forge, but don’t try too hard

• Work on refactoring/rewording existing examples.

• Possibly incorporating parts of “toolkit showcase” to end up with an example that does a protein-ligand simulation on binder.

Update

• JM fixed black/isort conflicts

• JM got examples package building, we’re not sure about next steps

  • JW – /share/openff-toolkit/examples directory is a great place for examples

  • JW – We don’t need tests in the conda packaging step. We can have a test of the file-copier in

    • the examples package creation, doing a somewhat detailed count to ensure counting the folders/files that come out and ensuring it’s the right number.

      • Should be able to compare something like sorted(glob.glob('**/*', recursive=True)) output for examples folder before and after copy

    • conda CI (this should replace running the examples from the git cloned directory) (it seems unlikely that shutil and glob would have breaking changes AFTER the packaging test above, so let’s not worry about testing this)

  •  

  • JM will make a minimal python-based examples-helper that copies the whole examples directory to the current working directory

• JM worked a bit on toolkit showcase – https://github.com/openforcefield/openff-toolkit/pull/763/files

  • (We reviewed + worked on some wording changes)

  • JM will delete the parameter-manipulation part. We can add this to the forcefield_modification example folder/notebook later

  •  

• JW – Docker-pre-build PR looks good, but can’t be tested/merged until we have OpenFF’s docker credentials. JW will contact DD to get these.

•