DEPR: kind kwarg in _maybe_cast_slice_bound

[jbrockmendel]

• closes #xxxx
• tests added / passed
• Ensure all linting tests pass, see here for how to run them
• whatsnew entry

When I tried just removing it that broke some dask tests.

[jreback]

sure

[tlvu]

Hi, would this change possibly introduce this regression (we have pandas-1.3.0)?

14:18:02  /opt/conda/envs/birdy/lib/python3.7/site-packages/pandas/core/indexes/base.py in slice_indexer(self, start, end, step, kind)
14:18:02     5684         slice(1, 3, None)
14:18:02     5685         """
14:18:02  -> 5686         start_slice, end_slice = self.slice_locs(start, end, step=step)
14:18:02     5687 
14:18:02     5688         # return a slice
14:18:02  
14:18:02  /opt/conda/envs/birdy/lib/python3.7/site-packages/pandas/core/indexes/base.py in slice_locs(self, start, end, step, kind)
14:18:02     5886         start_slice = None
14:18:02     5887         if start is not None:
14:18:02  -> 5888             start_slice = self.get_slice_bound(start, "left")
14:18:02     5889         if start_slice is None:
14:18:02     5890             start_slice = 0
14:18:02  
14:18:02  /opt/conda/envs/birdy/lib/python3.7/site-packages/pandas/core/indexes/base.py in get_slice_bound(self, label, side, kind)
14:18:02     5796         # For datetime indices label may be a string that has to be converted
14:18:02     5797         # to datetime boundary according to its resolution.
14:18:02  -> 5798         label = self._maybe_cast_slice_bound(label, side)
14:18:02     5799 
14:18:02     5800         # we need to look up the label
14:18:02  
14:18:02  TypeError: _maybe_cast_slice_bound() missing 1 required positional argument: 'kind'

[jbrockmendel]

Are you sure that traceback is produced using 1.3.0? 'kind' is a keyword argument with a default value, so i dont see how that traceback could be produced

[tlvu]

@jbrockmendel yes it is pandas-1.3.0 from our docker image https://hub.docker.com/repository/docker/pavics/workflow-tests tag preview.

$ docker run -it --rm --entrypoint bash pavics/workflow-tests:preview

$ conda init bash
$ bash
$ conda activate birdy
$ conda env export |grep pandas
  - geopandas=0.9.0=pyhd8ed1ab_1  
  - geopandas-base=0.9.0=pyhd8ed1ab_1                      
  - intake-geopandas=0.2.4=pyhd8ed1ab_0                                
  - pandas=1.3.0=py37h219a48f_0

This is the full error from our notebook https://github.com/bird-house/finch/blob/0346b3aa5096b88ef5bf03339a8c67ca7e4d6fdd/docs/source/notebooks/dap_subset.ipynb

18:18:02  _________ finch-master/docs/source/notebooks/dap_subset.ipynb::Cell 2 __________
18:18:02  Notebook cell execution failed
18:18:02  Cell 2: Cell execution caused an exception
18:18:02  
18:18:02  Input:
18:18:02  #NBVAL_IGNORE_OUTPUT
18:18:02  # Use the `remap_label_indexers` function to convert coordinates to *positional* indexes.
18:18:02  import datetime as dt
18:18:02  coords = dict(lat=45, lon=290)
18:18:02  index, _ = xr.core.coordinates.remap_label_indexers(ds, coords, method="nearest")
18:18:02  
18:18:02  # The `nearest` method cannot be used with slices, so we do another call for the time period.
18:18:02  ti, _ = xr.core.coordinates.remap_label_indexers(ds, dict(time=slice("2060-01-01", "2064-12-30")))
18:18:02  
18:18:02  # Merge the spatial and temporal indices
18:18:02  index.update(ti)
18:18:02  index
18:18:02  
18:18:02  Traceback:
18:18:02  
18:18:02  ---------------------------------------------------------------------------
18:18:02  TypeError                                 Traceback (most recent call last)
18:18:02  <ipython-input-3-d0a5be34f657> in <module>
18:18:02        6 
18:18:02        7 # The `nearest` method cannot be used with slices, so we do another call for the time period.
18:18:02  ----> 8 ti, _ = xr.core.coordinates.remap_label_indexers(ds, dict(time=slice("2060-01-01", "2064-12-30")))
18:18:02        9 
18:18:02       10 # Merge the spatial and temporal indices
18:18:02  
18:18:02  /opt/conda/envs/birdy/lib/python3.7/site-packages/xarray/core/coordinates.py in remap_label_indexers(obj, indexers, method, tolerance, **indexers_kwargs)
18:18:02      420 
18:18:02      421     pos_indexers, new_indexes = indexing.remap_label_indexers(
18:18:02  --> 422         obj, v_indexers, method=method, tolerance=tolerance
18:18:02      423     )
18:18:02      424     # attach indexer's coordinate to pos_indexers
18:18:02  
18:18:02  /opt/conda/envs/birdy/lib/python3.7/site-packages/xarray/core/indexing.py in remap_label_indexers(data_obj, indexers, method, tolerance)
18:18:02      272             coords_dtype = data_obj.coords[dim].dtype
18:18:02      273             label = maybe_cast_to_coords_dtype(label, coords_dtype)
18:18:02  --> 274             idxr, new_idx = convert_label_indexer(index, label, dim, method, tolerance)
18:18:02      275             pos_indexers[dim] = idxr
18:18:02      276             if new_idx is not None:
18:18:02  
18:18:02  /opt/conda/envs/birdy/lib/python3.7/site-packages/xarray/core/indexing.py in convert_label_indexer(index, label, index_name, method, tolerance)
18:18:02      122             _sanitize_slice_element(label.start),
18:18:02      123             _sanitize_slice_element(label.stop),
18:18:02  --> 124             _sanitize_slice_element(label.step),
18:18:02      125         )
18:18:02      126         if not isinstance(indexer, slice):
18:18:02  
18:18:02  /opt/conda/envs/birdy/lib/python3.7/site-packages/pandas/core/indexes/base.py in slice_indexer(self, start, end, step, kind)
18:18:02     5684         slice(1, 3, None)
18:18:02     5685         """
18:18:02  -> 5686         start_slice, end_slice = self.slice_locs(start, end, step=step)
18:18:02     5687 
18:18:02     5688         # return a slice
18:18:02  
18:18:02  /opt/conda/envs/birdy/lib/python3.7/site-packages/pandas/core/indexes/base.py in slice_locs(self, start, end, step, kind)
18:18:02     5886         start_slice = None
18:18:02     5887         if start is not None:
18:18:02  -> 5888             start_slice = self.get_slice_bound(start, "left")
18:18:02     5889         if start_slice is None:
18:18:02     5890             start_slice = 0
18:18:02  
18:18:02  /opt/conda/envs/birdy/lib/python3.7/site-packages/pandas/core/indexes/base.py in get_slice_bound(self, label, side, kind)
18:18:02     5796         # For datetime indices label may be a string that has to be converted
18:18:02     5797         # to datetime boundary according to its resolution.
18:18:02  -> 5798         label = self._maybe_cast_slice_bound(label, side)
18:18:02     5799 
18:18:02     5800         # we need to look up the label
18:18:02  
18:18:02  TypeError: _maybe_cast_slice_bound() missing 1 required positional argument: 'kind'

[jbrockmendel]

What kind of Index do you have here? maybe its a custom subclass that overrides _maybe_cast_slice_bound and has kind as a positional argument? All of our Index subclasses have kind as a keyword argument

[tlvu]

What kind of Index do you have here? maybe its a custom subclass that overrides _maybe_cast_slice_bound and has kind as a positional argument? All of our Index subclasses have kind as a keyword argument

You are right, xarray has an override !

$ grep -R  _maybe_cast_slice_bound . | grep def
./lib/python3.7/site-packages/xarray/coding/cftimeindex.py:    def _maybe_cast_slice_bound(self, label, side, kind):

And they seem to already made the matching change as well pydata/xarray#5359 but did not cut a release containing the change yet !