This data as json
| sha | message | author_date | committer_date | raw_author | raw_committer | repo | author | committer |
|---|---|---|---|---|---|---|---|---|
| ffce4ec93c2e401a37dba1f0bf33dbfc648aeec9 | linear interp with NaNs in nd indexer (#4233) * Added test for nd interpolation with nan * Now ignoring NaNs in missing._localize When interpolating with an nd indexer that contains NaN's, the code previously threw a KeyError from the missing._localize function. This commit fixes this by swapping `np.min` and `np.max` with `np.nanmin` and `np.nanmax`, ignoring any NaN values. * Added `@requires_scipy` to test. Also updated what's new. * Added numpy>=1.18 checks with `LooseVersion` * Added checks for np.datetime64 type This means the PR now also works for numpy < 1.18, as long as index is not with datetime * Removed `raise ValueError` from previous commit It seems that np.min/max works in place of nanmin/nanmax for datetime types for numpy < 1.18, see https://github.com/pydata/xarray/pull/3924/files * Added datetime `NaT` test. Also added a test for `Dataset` to `test_interpolate_nd_with_nan`, and "Missing values are skipped." to the dosctring of `interp` and `interp_like` methods of `DataArray` and `Dataset`. | 2020-08-27T08:49:47Z | 2020-08-27T08:49:47Z | a675e5ffa13fe46de9fbdcaf2f82f2a4aab55a09 | cd792325681cbad9f663f2879d8b69f1edbb678f | 13221727 | 37403847 | 19864447 |