id,node_id,number,state,locked,title,user,body,created_at,updated_at,closed_at,merged_at,merge_commit_sha,assignee,milestone,draft,head,base,author_association,auto_merge,repo,url,merged_by
434586489,MDExOlB1bGxSZXF1ZXN0NDM0NTg2NDg5,4155,closed,0,Implement interp for interpolating between chunks of data (dask),1005109,"In a project of mine I need to interpolate a dask-based xarray between chunk of data.

When using the current official `interp` function (xarray v0.15.1), the code:

```python
    datax = xr.DataArray(data=da.from_array(np.arange(0, 4), chunks=2),
                         coords={""x"": np.linspace(0, 1, 4)},
                         dims=""x"")
    datay = xr.DataArray(data=da.from_array(np.arange(0, 4), chunks=2),
                         coords={""y"": np.linspace(0, 1, 4)},
                         dims=""y"")
    data = datax * datay

    # both of these interp call fails
    res = datax.interp(x=np.linspace(0, 1))
    print(res.load())

    res = data.interp(x=np.linspace(0, 1), y=0.5)
    print(res.load())
```
fails with `NotImplementedError: Chunking along the dimension to be interpolated (0) is not yet supported.`, but succeed with this version

I also want to alert that my version does not work with ""advanced interpolation"" (as shown in the xarray documentation)
Also, my version cannot be used to make `interpolate_na` work with chunked data

<!-- Feel free to remove check-list items aren't relevant to your change -->

 - [x] Closes #4078
 - [x] Tests added
 - [x] Passes `isort -rc . && black . && mypy . && flake8`
 - [x] Fully documented, including `whats-new.rst` for all changes and `api.rst` for new API
",2020-06-15T14:42:32Z,2020-09-06T12:27:15Z,2020-08-11T23:15:49Z,2020-08-11T23:15:48Z,7daad4fce3bf8ad9b9bc8e7baa104c476437e68d,,,0,b60cddf176d0524ed0a09c3cbb9a5acb76449e76,a198218ddabe557adbb04311b3234ec8d20419e7,CONTRIBUTOR,,13221727,https://github.com/pydata/xarray/pull/4155,
787098153,PR_kwDOAMm_X84u6iop,6019,closed,0,Use complex nan by default when interpolating out of bounds,1005109,"- [X] Tests added
- [X] Passes `pre-commit run --all-files`
- [x] User visible changes (including notable bug fixes) are documented in `whats-new.rst`

When using the `da.interp` to interpolate outside of the bounds, by default,  `fill_value` is set to `np.nan` to set the values to NaN.
This is completely fine with real values, but with complex values this will in fact set the values to `np.nan + 0j` which can be a source of confusion and bugs.
This PR propose to detect if values are complex, and if so, to use `np.nan + np.nan*1j` as the default `fill_value`",2021-11-23T15:38:25Z,2021-11-28T04:40:06Z,2021-11-28T04:40:06Z,2021-11-28T04:40:06Z,fb01c72626a61310f874664cdb4d7b4c1b327bb3,,,0,93ef228399cdd9aff2c53c924e3b131bf8d9696f,dc68e0c997495f4a5966433ee602df548e9a0108,CONTRIBUTOR,,13221727,https://github.com/pydata/xarray/pull/6019,