allow formatting the diff of ndarray attributes #3728
Conversation
|
I think a unit test for the use case is in order? |
keewis
left a comment
keewis
left a comment
There was a problem hiding this comment.
I added the test which helped me find the warning raised when comparing arrays with different shapes
| def is_array_like(value): | ||
| return ( | ||
| hasattr(value, "ndim") | ||
| and hasattr(value, "shape") | ||
| and hasattr(value, "dtype") | ||
| ) |
There was a problem hiding this comment.
I'm not really sure if this is how array_like is defined
There was a problem hiding this comment.
Looks reasonable at the moment; unless anyone knows better?
This could move into utils if we get more confident (almost surprised we don't do this test elsewhere...)
There was a problem hiding this comment.
we have mostly been using __array_function__ instead, but that is not really sufficient: pint defines __array_function__ even if it wraps python scalars. In that case pint.Quantity is not array-like.
|
should be ready to merge, too |
|
Looks great! Thanks! |
|
Thanks @keewis |
…under * upstream/master: (71 commits) Optimize isel for lazy array equality checking (pydata#3588) pin msgpack (pydata#3793) concat now handles non-dim coordinates only present in one dataset (pydata#3769) Add new h5netcdf backend phony_dims kwarg (pydata#3753) always use dask_array_type for isinstance calls (pydata#3787) allow formatting the diff of ndarray attributes (pydata#3728) Pint support for variables (pydata#3706) Format issue template comment as md comment (pydata#3790) Avoid running test_open_mfdataset_list_attr without dask (pydata#3780) remove seaborn.apionly compatibility (pydata#3749) Python 3.8 CI (pydata#3727) PKG: Explicitly add setuptools dependency (pydata#3628) update whats-new Typo in Universal Functions section (pydata#3663) Release v0.15.0 fix setup.cfg Documentation fixes (pydata#3732) Remove extra && in PR template (pydata#3730) Remove garbage text inserted in DASK_LICENSE (pydata#3729) Avoid unsafe use of pip (pydata#3726) ...
4 participants
This allows to have
ndarrayinstances in attributes, which is the case for the dataset fromxarray/tests/data/example.uamiv. It won't fix the cause for the error of #3711, though, because that seems to non-deterministic.isort -rc . && black . && mypy . && && flake8whats-new.rstfor all changes andapi.rstfor new API