`ffill` does not spread chunks when used with dask.array

[eserie]

What is your issue?

When using xarray with dask.array as in the following example:

import numpy as np
import xarray as xr
import dask
ds_ = xr.Dataset()
ds_["test"] = ("dim_0",), dask.array.from_array(
    [0, 1, np.nan, np.nan, np.nan], chunks=2
)

then the ffill method does not seem to propagate through the underlying chunks used by dask.array.
In the example above, this means that

 ds_.ffill("dim_0").compute().test.values

gives

>>> array([ 0., 1., 1., 1., nan])

where

ds_.compute().ffill("dim_0").test.values

gives

>>> array([0., 1., 1., 1., 1., 1.])

Is it an expected behavior ?

[mathause]

What version of xarray are you using? I think this should have been fixed in #5187 (i.e. in v0.18.0).

Edit: quickly checked and the newest version gets this right.

[eserie]

Hi @mathause, I'm in version 0.18.2.

[mathause]

Ok, I see there is another bugfix in v0.21.0 (#6118) - thus you will need at least this version for your example to work.

[eserie]

Ok, thanks, that solves the problem!

[mathause]

Ok, thanks, that solves the problem!

Great to hear!