What is your issue?

I'm pretty sure my PR https://github.com/pydata/xarray/pull/6542 is the culprit. I never figured out how to get around the build timeout with these docs edits.

If you all are on top of this then no worries - feel free to close. Just wanted to point in the right direction so you don't need to go hunting.

Is your feature request related to a problem?

xr.Dataset.where and xr.DataArray.where currently do not allow providing both other and drop=True, stating that they are mutually exclusive. Conceptually, this doesn't strike me as true. Drop does not drop all points where cond is False, only those where an entire coordinate label evaluates to False. This is most important when trying to avoid type promotion, such as in the below example, where an integer FillValue might be preferred over dtypes.NA.

```python In [2]: da = xr.DataArray(np.arange(16).reshape(4, 4), dims=['x', 'y'])

In [3]: da Out[3]: <xarray.DataArray (x: 4, y: 4)> array([[ 0, 1, 2, 3], [ 4, 5, 6, 7], [ 8, 9, 10, 11], [12, 13, 14, 15]]) Dimensions without coordinates: x, y If *`other`* is not provided, the array is promoted to `float64` to accommodate the default,`dtypes.NA`:python In [4]: da.where(da > 6, drop=True) Out[4]: <xarray.DataArray (x: 3, y: 4)> array([[nan, nan, nan, 7.], [ 8., 9., 10., 11.], [12., 13., 14., 15.]]) Dimensions without coordinates: x, y However, the combination of *`other`* and *`drop=True`* is not allowed:python In [5]: da.where(da > 6, -1, drop=True)

ValueError Traceback (most recent call last) Input In [5], in <module> ----> 1 da.where(da > 6, -1, drop=True)

File ~/miniconda3/envs/rhodium-env/lib/python3.10/site-packages/xarray/core/common.py:1268, in DataWithCoords.where(self, cond, other, drop) 1266 if drop: 1267 if other is not dtypes.NA: -> 1268 raise ValueError("cannot set other if drop=True") 1270 if not isinstance(cond, (Dataset, DataArray)): 1271 raise TypeError( 1272 f"cond argument is {cond!r} but must be a {Dataset!r} or {DataArray!r}" 1273 )

ValueError: cannot set other if drop=True ```

Describe the solution you'd like

Current implementation

The current behavior is enforced within the block handling the drop argument (currently https://github.com/pydata/xarray/blob/main/xarray/core/common.py#L1266-L1268):

python if drop: if other is not dtypes.NA: raise ValueError("cannot set `other` if drop=True")

Proposed fix

I just removed the above if statement on a fork, and the example now works!

```python

import xarray as xr, numpy as np da = xr.DataArray(np.arange(16).reshape(4, 4), dims=['x', 'y']) da <xarray.DataArray (x: 4, y: 4)> array([[ 0, 1, 2, 3], [ 4, 5, 6, 7], [ 8, 9, 10, 11], [12, 13, 14, 15]]) Dimensions without coordinates: x, y da.where(da > 6, drop=True) <xarray.DataArray (x: 3, y: 4)> array([[nan, nan, nan, 7.], [ 8., 9., 10., 11.], [12., 13., 14., 15.]]) Dimensions without coordinates: x, y da.where(da > 6, -1, drop=True) <xarray.DataArray (x: 3, y: 4)> array([[-1, -1, -1, 7], [ 8, 9, 10, 11], [12, 13, 14, 15]]) Dimensions without coordinates: x, y ``` Making this change fails a single test in https://github.com/pydata/xarray/blob/main/xarray/tests/test_dataset.py#L4548-L4549, which explicitly checks for this behavior:

python with pytest.raises(ValueError, match=r"cannot set"): ds.where(ds > 1, other=0, drop=True) This could easily be reworked to check for valid handling of both arguments.

Describe alternatives you've considered

No response

Additional context

I haven't yet investigated what would happen with chunked, sparse, or other complex arrays, or if it's compatible with trees and other things on the roadmap. It's possible this breaks things I'm not imagining. Currently, where(cond, other) and where(cond, drop=True) are well-tested, flexible operations, and I don't see why allowing their union would break anything, but I'll wait to hear from the experts on that front!

I'm definitely open to creating a pull request (and have the simple implementation I've outlined here ready to go).

The use case is for netCDF files stored on s3 or other generic cloud storage

```python import requests, xarray as xr fp = 'http://nasanex.s3.amazonaws.com/NEX-GDDP/BCSD/rcp45/day/atmos/tasmax/r1i1p1/v1.0/tasmax_day_BCSD_rcp45_r1i1p1_MPI-ESM-LR_2029.nc'

data = requests.get(fp, stream=True) ds = xr.open_dataset(data.content) # raises TypeError: embedded NUL character ```

Ideal would be integration with the (hopefully) soon-to-be implemented dask.distributed features discussed in #798.

series.to_xarray() fails when MultiIndex not sorted in xarray 0.15.1

Summary

It seems that series.to_xarray() fails (returns incorrect data) in xarray 0.15.1 when the dataframe's MultiIndex dimensions are not sorted

Demonstration

xarray should be able to handle MultiIndices with unsorted dimensions. Using a fresh conda environment with xarray 0.14.1:

```python $ conda run -n py37xr14 python test.py

df alpha B A num 0 1 4 1 2 5 2 3 6

df.stack('alpha') num alpha 0 B 1 A 4 1 B 2 A 5 2 B 3 A 6 dtype: int64

df.stack('alpha').to_xarray() <xarray.DataArray (num: 3, alpha: 2)> array([[1, 4], [2, 5], [3, 6]]) Coordinates: * num (num) int64 0 1 2 * alpha (alpha) object 'B' 'A' ```

This fails in xarray 0.15.1 - note the data is not merely reordered - the data in column 'B' now has the incorrect values 4, 5, 6 rather than 1, 2, 3:

```python $ conda run -n py37xr15 python test.py

df alpha B A num 0 1 4 1 2 5 2 3 6

df.stack('alpha') num alpha 0 B 1 A 4 1 B 2 A 5 2 B 3 A 6 dtype: int64

df.stack('alpha').to_xarray() <xarray.DataArray (num: 3, alpha: 2)> array([[4, 1], [5, 2], [6, 3]]) Coordinates: * num (num) int64 0 1 2 * alpha (alpha) object 'B' 'A' ```

Test setup & environment info

Spelling error in value error raised on index-based assignment with incorrect shape

Very easy one here:

```python In [1]: import xarray as xr, pandas as pd, numpy as np

In [2]: da = xr.DataArray(np.random.random((2, 3)), dims=['x','y'])

In [3]: da Out[3]: <xarray.DataArray (x: 2, y: 3)> array([[0.882927, 0.604024, 0.316146], [0.06342 , 0.503182, 0.297988]]) Dimensions without coordinates: x, y

In [4]: da[0, 1] = [1, 2]

ValueError Traceback (most recent call last) <ipython-input-4-1fbe1d206e00> in <module>() ----> 1 da[0, 1] = [1, 2]

~/miniconda2/envs/xarray-dev/lib/python3.6/site-packages/xarray/core/dataarray.py in setitem(self, key, value) 486 key = {k: v.variable if isinstance(v, DataArray) else v 487 for k, v in self._item_key_to_dict(key).items()} --> 488 self.variable[key] = value 489 490 def delitem(self, key):

~/miniconda2/envs/xarray-dev/lib/python3.6/site-packages/xarray/core/variable.py in setitem(self, key, value) 682 'shape mismatch: value array of shape %s could not be' 683 'broadcast to indexing result with %s dimensions' --> 684 % (value.shape, len(dims))) 685 if value.ndim == 0: 686 value = Variable((), value)

ValueError: shape mismatch: value array of shape (2,) could not bebroadcast to indexing result with 0 dimensions

In [5]: xr.show_versions() /Users/delgadom/miniconda2/envs/xarray-dev/lib/python3.6/site-packages/h5py/init.py:36: FutureWarning: Conversion of the second argument of issubdtype from float to np.floating is deprecated. In future, it will be treated as np.float64 == np.dtype(float).type. from ._conv import register_converters as _register_converters

INSTALLED VERSIONS

commit: None python: 3.6.4.final.0 python-bits: 64 OS: Darwin OS-release: 17.4.0 machine: x86_64 processor: i386 byteorder: little LC_ALL: None LANG: en_US.UTF-8 LOCALE: en_US.UTF-8

xarray: 0.10.2 pandas: 0.22.0 numpy: 1.14.2 scipy: 1.0.0 netCDF4: 1.3.1 h5netcdf: 0.5.0 h5py: 2.7.1 Nio: None zarr: 2.2.0 bottleneck: 1.2.1 cyordereddict: None dask: 0.17.2 distributed: 1.21.4 matplotlib: 2.2.2 cartopy: 0.15.1 seaborn: 0.8.1 setuptools: 38.5.1 pip: 9.0.1 conda: None pytest: 3.4.1 IPython: 6.2.1 sphinx: 1.6.6

```

Problem description

The error message in variable.py#L682 seems to be missing an end-of-line space. Happy to create a PR.

Using both xarray 0.9.6 and current (0.9.6-16-gb201ff7), da.sel(drop=True, **indexers) and da.isel(drop=True, **indexers) return unexpectedly different results.

Setup:

```python In [1]: import xarray as xr, pandas as pd, numpy as np

In [2]: years = pd.Index( ...: pd.date_range('1981-01-01', '2100-01-01', freq='A', closed='left'), ...: name='time') ...: ages = pd.Index(['age0', 'age1', 'age2', 'age3'], name='age')

In [3]: arr = xr.DataArray( ...: np.random.random((len(years), 4)), dims=('time', 'age'), ...: coords={'time': years, 'age': ages})

In [4]: arr Out[4]: <xarray.DataArray (time: 119, age: 4)> array([[ 0.755194, 0.1316 , 0.283485, 0.616929], [ 0.01667 , 0.907853, 0.667366, 0.146755], [ 0.338319, 0.782972, 0.367624, 0.390907], ..., [ 0.453521, 0.807693, 0.094811, 0.603297], [ 0.405114, 0.821691, 0.633314, 0.259406], [ 0.41722 , 0.012957, 0.329089, 0.774966]]) Coordinates: * age (age) object 'age0' 'age1' 'age2' 'age3' * time (time) datetime64[ns] 1981-12-31 1982-12-31 1983-12-31 ... ```

I would expect the following operations to return identical results:

```python In [5]: arr.sel(time='2012', drop=True) Out[5]: <xarray.DataArray (time: 1, age: 4)> array([[ 0.086045, 0.467905, 0.101005, 0.503311]]) Coordinates: * age (age) object 'age0' 'age1' 'age2' 'age3' Dimensions without coordinates: time

In [6]: arr.isel(time=31, drop=True) Out[6]: <xarray.DataArray (age: 4)> array([ 0.086045, 0.467905, 0.101005, 0.503311]) Coordinates: * age (age) object 'age0' 'age1' 'age2' 'age3' `` Note thattimeis still seen as a dimension in in the.sel` results.

The same behavior is seen for Dataset.sel(drop=true):

```python In [7]: ds = xr.Dataset({'arr': arr})

In [8]: ds.sel(time='2012', drop=True) Out[8]: <xarray.Dataset> Dimensions: (age: 4, time: 1) Coordinates: * age (age) object 'age0' 'age1' 'age2' 'age3' Dimensions without coordinates: time Data variables: arr (time, age) float64 0.08604 0.4679 0.101 0.5033

In [9]: ds.isel(time=31, drop=True) Out[9]: <xarray.Dataset> Dimensions: (age: 4) Coordinates: * age (age) object 'age0' 'age1' 'age2' 'age3' Data variables: arr (age) float64 0.08604 0.4679 0.101 0.5033 ```