Closed michaelleerilee closed 1 year ago
@NiklasPhabian We should bump version and create a release if this update is acceptable.
Note I found and corrected a bunch of errors when running pytest from code not updated since we started adding nom_res to variable names.
I'm getting a clean pytest locally, but pytest is only finding 95 items, not 96.
by clean pytest, do you mean passing all tests? That would be curious because the GH actions and my local fails two tests. Both possibly caused by pandas>=2.0.0
uuuf. Indeed, this was caused by pandas. Now unfortunately, pickled dataframes are not compatible between versions. I.e. we would have to re-pod data if we switched to pandas>=2.0.
Is there a path for reading pickles at one version and writing to a new version?
Apparently the pickle lib can be instructed which protocol to use.
Ah, but we're talking about serializing dataframes...
https://pandas.pydata.org/docs/reference/api/pandas.read_pickle.html says that read_pickle is compatible back to pandas 0.20.3...
I'm getting a clean pytest locally, but pytest is only finding 95 items, not 96.
we had been skipping one of the examples
@michaelleerilee @mbauer288 can you verify that my edits do not break your usecases?
…_res=None. Propagated changes through tests and some granule readers. Added set_nom_res() method to base Granule class.
Ref. #144