Closed zmoon closed 1 year ago
Upon further investigation, discovered this was more a (veiled) xesmf issue. It failed to import, for reasons that have been addressed in the recent v0.7 release, but this release isn't yet on conda-forge (and the recipe currently doesn't constrain shapely).
xesmf v0.7 release is now on cf.
The scheduled CI test was passing last week but started failing this week.
Specifically, the test of remap is failing since the datasets no longer have a
data_y
variable being added, just the originaldata
.Differences between old and new envs for py38
| name | version_old | version_new | |:-------------------|:--------------|:--------------| | cartopy | 0.21.0 | 0.21.1 | | dask | 2022.12.0 | 2022.12.1 | | dask-core | 2022.12.0 | 2022.12.1 | | distributed | 2022.12.0 | 2022.12.1 | | esmf | 8.2.0 | 8.4.0 | | **esmpy** | 8.2.0 | 8.4.0 | | importlib-metadata | 5.1.0 | 5.2.0 | | importlib_metadata | 5.1.0 | 5.2.0 | | krb5 | 1.19.3 | 1.20.1 | | parallelio | - | 2.5.10 | | pyproj | 3.4.0 | 3.4.1 | | pytz | 2022.6 | 2022.7 | | setuptools | 65.5.1 | 65.6.3 | | **shapely** | 1.8.5 | 2.0.0 | | traitlets | 5.7.0 | 5.7.1 | (numba 0.56.4 in both):point_up: The differences suggest that ESMF/ESMPy is the culprit.