What happened?

There used to be a xarray.core.rolling, and the documentation suggests it still exists, but with xarray-2022.6.0 I get: module xarray.core has no attribute rolling. It works with xarray-2022.3.0.

What did you expect to happen?

Shouldn't we be able to access xarray.core.rolling?

Minimal Complete Verifiable Example

```Python import xarray as xr

xr.core.rolling

Traceback (most recent call last):

File "<stdin>", line 1, in <module>

AttributeError: module 'xarray.core' has no attribute 'rolling'

```

MVCE confirmation

• [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray.
• [X] Complete example — the example is self-contained, including all data and the text of any traceback.
• [X] Verifiable example — the example copy & pastes into an IPython prompt or Binder notebook, returning the result.
• [X] New issue — a search of GitHub Issues suggests this is not a duplicate.

Relevant log output

No response

Anything else we need to know?

No response

Environment

MCVE Code Sample

```python import xarray as xr

da = xr.DataArray(1) da.attrs['foo'] = 'bar' xr.where(da==0, -1, da).attrs

shows: {}

```

Expected Output

{'foo': 'bar'}

Problem Description

I would expect the attributes to remain in the data array.

Versions

Code Sample

```python import numpy as np import xarray as xr import matplotlib.pyplot as plt %matplotlib inline

a = np.ones((4000, 4000), dtype=np.uint8) a[:1000, :1000] = 0 lat = np.array([0-i for i in range(a.shape[0])]) lon = np.array([0+i for i in range(a.shape[1])]) da = xr.DataArray(a, coords=[lat, lon], dims=['lat', 'lon'])

da.plot.pcolormesh() ```

Problem description

The code above shows an empty plot in a notebook (but not in the console).

Expected Output

If I replace da.plot.pcolormesh() with da.plot.imshow() or plt.pcolormesh(a), it works fine. Also, if I decrease the size of the array, it works.

Output of xr.show_versions()

I would like to concatenate many DataArray whose dimension coordinates are almost aligned, allowing for some offset in the coordinates:

python import xarray as xr da1 = xr.DataArray([[0, 1], [2, 3]], coords=[[0, 1], [0, 1]], dims=['x', 'y']) da2 = xr.DataArray([[4, 5], [6, 7]], coords=[[1.1, 2.1], [1.1, 2.1]], dims=['x', 'y']) da = xr.concat([da1, da2], 'z', method='nearest', tolerance=0.2) # doesn't exist yet da

And I would get:

``` <xarray.DataArray (z: 2, x: 3, y: 3)> array([[[ 0., 1., nan], [ 2., 3., nan], [nan, nan, nan]],

   [[nan, nan, nan],
    [nan,  4.,  5.],
    [nan,  6.,  7.]]])

Coordinates: * x (x) float64 0 1 2.1 * y (y) float64 0 1 2.1 Dimensions without coordinates: z ```

@jhamman suggested to use reindex_like in this StackOverflow question, but it doesn't produce the union of coordinates, so I cannot chain them like this:

python da2 = da2.reindex_like(da1, method='nearest', tolerance=0.2)

Is there another work around? Do you think it would be worth having this feature in concat?

When appending to a zarr store, we need to set append_dim=dim_name but also mode='a'. Any reason to also specify the writing mode? I think it should automatically be set to 'a' if append_dim is not None.

MCVE Code Sample

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

In [2]: da = xr.DataArray([1, 2], [('time', pd.date_range('2000-01-01', '2000-01-02', periods=2))]) ...: da
Out[2]: <xarray.DataArray (time: 2)> array([1, 2]) Coordinates: * time (time) datetime64[ns] 2000-01-01 2000-01-02

In [3]: da.interp(time=da.time-np.timedelta64(1, 'D'))
Out[3]: <xarray.DataArray (time: 2)> array([nan, 1.]) Coordinates: * time (time) datetime64[ns] 2000-01-01 2000-01-02 ```

Problem Description

The data has been interpolated, but not the time coordinate.

Expected Output

When we explicitly get the values of the time coordinate, it works fine:

python In [4]: da.interp(time=da.time.values-np.timedelta64(1, 'D')) Out[4]: <xarray.DataArray (time: 2)> array([nan, 1.]) Coordinates: * time (time) datetime64[ns] 1999-12-31 2000-01-01

Output of xr.show_versions()

For instance in http://xarray.pydata.org/en/stable/generated/xarray.apply_ufunc.html, there is no space between the parameters and their type description, e.g. funccallable instead of func : callable.