pull_requests: 795584845

This data as json

id node_id number state locked title user body created_at updated_at closed_at merged_at merge_commit_sha assignee milestone draft head base author_association auto_merge repo url merged_by
795584845 PR_kwDOAMm_X84va6lN 6049 closed 0 Attempt datetime coding using cftime when pandas fails 132147 A netCDF4 dataset we use has a time variable defined as: ``` double time(time) ; time:axis = "T" ; time:bounds = "time_bnds" ; time:calendar = "gregorian" ; time:long_name = "time" ; time:standard_name = "time" ; time:units = "days since 1970-01-01 00:00:00 00" ; ``` Note the `units` attribute, specifically a timezone offset of `00` without any `+-` sign. xarray can successfully open this dataset and parse the time units, making a time variable with the expeced values. However, attempting to save this dataset (e.g. after slicing some geographic bounds or selecting a subset of variables), xarray would raise an error trying to reformat the time `units`. This fix applies the same logic used in the decoding step to the encoding step - specifically, attempt to use `pandas` but if that fails then use `cftime`. The decoding step catches `ValueError` to do this, but `ValueError` was not caught in the encode workflow. - [x] Tests added - [x] Passes `pre-commit run --all-files` - [x] User visible changes (including notable bug fixes) are documented in `whats-new.rst` 2021-12-06T07:12:35Z 2022-01-04T00:28:15Z 2021-12-24T11:48:22Z 2021-12-24T11:48:22Z 60f72ef4b1b8182245ee06f1758339826052bfb7     0 312f64e4da178928d309eac87f7010c281e5618b f76ea09bebba35a175868a81845c426890ec0330 CONTRIBUTOR   13221727 https://github.com/pydata/xarray/pull/6049  

Links from other tables