@@ -259,18 +259,16 @@ along a particular dimension, an exception is raised when you try to access
259259 arrays in a dataset share the same chunking alignment. Neither of these
260260 are currently done.
261261
262- NumPy ufuncs like ``np.sin `` currently only work on eagerly evaluated arrays
263- (this will change with the next major NumPy release). We have provided
264- replacements that also work on all xarray objects, including those that store
265- lazy Dask arrays, in the :ref: `xarray.ufuncs <api.ufuncs >` module:
262+ NumPy ufuncs like ``np.sin `` transparently work on all xarray objects, including those
263+ that store lazy Dask arrays:
266264
267265.. ipython :: python
268266
269- import xarray.ufuncs as xu
267+ import numpy as np
270268
271- xu .sin(rechunked)
269+ np .sin(rechunked)
272270
273- To access Dask arrays directly, use the new
271+ To access Dask arrays directly, use the
274272:py:attr: `DataArray.data <xarray.DataArray.data> ` attribute. This attribute exposes
275273array data either as a Dask array or as a NumPy array, depending on whether it has been
276274loaded into Dask or not:
@@ -281,8 +279,8 @@ loaded into Dask or not:
281279
282280 .. note ::
283281
284- In the future, we may extend ``.data `` to support other "computable" array
285- backends beyond Dask and NumPy (e.g., to support sparse arrays).
282+ ``.data `` is also used to expose other "computable" array backends beyond Dask and
283+ NumPy (e.g. sparse and pint arrays).
286284
287285.. _dask.automatic-parallelization :
288286
0 commit comments