Test suite: explicitly ignore irrelevant warnings

[shoyer]

Includes silencing two warnings that show up in xarray's public API:

1. Dataset.update(Dataset) Regression: Dataset.update(Dataset) #2161
2. .equals() with scalar timedelta64/datetime64 arrays containing NaT:

In [1]: import numpy as np

In [2]: import xarray as xr

In [3]: array = xr.DataArray([np.datetime64('NaT')])

In [4]: array[0].equals(array[0])
/Users/shoyer/dev/xarray/xarray/core/duck_array_ops.py:148: FutureWarning: In the future, 'NAT == x' and 'x == NAT' will always be False.
  flag_array = (arr1 == arr2)
Out[16]: True

Fixes #2161

• Tests added (for all bug fixes or enhancements)
• Tests passed (for all non-documentation changes)
• Fully documented, including whats-new.rst for all changes and api.rst for new API (remove if this change should not be visible to users, e.g., if it is an internal clean-up, or if this is part of a larger project that will be documented later)

[stickler-ci]

E501 line too long (80 > 79 characters)

[stickler-ci]

E501 line too long (80 > 79 characters)

[shoyer]

@fujiisoup any thoughts?

Do you agree that we should have different behavior for overriding coords with __setitem__ vs. update()?

[fujiisoup]

I didn't notice that iterating over dataset gives a FutureWarning.

Do you agree that we should have different behavior for overriding coords with setitem vs. update()?

I actually do not yet understand well the difference between ds.update(other) and ds.merge(other).

The docs says

If any dimensions would have inconsistent sizes in the updated dataset.update raises a ValueError,

but it does not.
Maybe it should work identically with xr.merge(other)?

[shoyer]

• make a decision on coordinate conflicts with update()

[shoyer]

OK, updated to only silence the warning in Dataset.update(Dataset) (by refactoring the logic)

[fujiisoup]

Do we update coordinates if other is Dataset?

[shoyer]

Yes, if other is a Dataset coordinates get overwritten. We probably already had test coverage for this, but I also added some explicit tests.

…

On Thu, May 24, 2018 at 9:21 PM Keisuke Fujii ***@***.***> wrote: Do we update coordinates if other is Dataset? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#2162 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABKS1v2WCnW59lZwnW_nc2oQftwaZTgVks5t14bNgaJpZM4UFEGE> .

[shoyer]

I plan to merge this shortly (with a day or two) unless anyone else has comments

[fujiisoup]

I am slightly wondering this behavior may change after we finally support #884, i.e. we will not include coordinates in .key().

[shoyer]

This line only applies to non-Dataset objects (see the isintance check on the line above), so I think we are OK here.

…

On Mon, May 28, 2018 at 9:05 PM Keisuke Fujii ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In xarray/core/merge.py <#2162 (comment)>: > """ from .dataset import Dataset from .dataarray import DataArray - other = other.copy() - for k, obj in other.items(): - if isinstance(obj, (Dataset, DataArray)): - # drop duplicated coordinates - coord_names = [c for c in obj.coords - if c not in obj.dims and c in dataset.coords] - if coord_names: - other[k] = obj.drop(coord_names) + if not isinstance(other, Dataset): + other = OrderedDict(other) I am slightly wondering this behavior may change after we finally support #884 <#884>, i.e. we will not include coordinates in .key(). — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#2162 (review)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABKS1gLkp5aD26C7wH18rk_zdfk9h5Kdks5t3Mj0gaJpZM4UFEGE> .

[fujiisoup]

I just missed this line.
Thanks.
LGTM.