issue_comments

all CI pass. Thanks for the quick fixes, @fujiisoup

thanks, that fixed the build on my own RTD setup, but the main setup is quite a bit slower...

Edit: it passed on the main setup, too. Let's merge on green.

Thanks, @keewis , for the clarification.

It was a bug in the documentation page but not in rolling.construct.

It should raise an error in this case, because for 2d rolling we need 2 dimension names, python rolling_da = r.construct(x="x_win", y='y_win' , stride=2) I corrected the documentation and error message.

normally you'd look at the output of the RTD build, but somehow just the official setup times out instead of printing the log so I put in a support request on RTD.

I can reproduce the failure using: ```python In [1]: import xarray as xr ...: import numpy as np ...:
...: arr = xr.DataArray(np.arange(0, 7.5, 0.5).reshape(3, 5), dims=("x", "y")) ...: display(arr) ...: r = arr.rolling(x=2, y=3, min_periods=2) ...: display(r) ...:
...: rolling_da = r.construct("window_dim", stride=2) ...: display(rolling_da) <xarray.DataArray (x: 3, y: 5)> array([[0. , 0.5, 1. , 1.5, 2. ], [2.5, 3. , 3.5, 4. , 4.5], [5. , 5.5, 6. , 6.5, 7. ]]) Dimensions without coordinates: x, y DataArrayRolling [x->2,y->3]

ValueError Traceback (most recent call last) <ipython-input-1-f9a10c0cfaeb> in <module> 7 display(r) 8 ----> 9 rolling_da = r.construct("window_dim", stride=2) 10 display(rolling_da) 11 rolling_da.mean("window_dim", skipna=False)

.../xarray/core/rolling.py in construct(self, window_dim, stride, fill_value, **window_dim_kwargs) 277 window_dim = {d: window_dim_kwargs[d] for d in self.dim} 278 --> 279 window_dim = self._mapping_to_list( 280 window_dim, allow_default=False, allow_allsame=False 281 )

.../xarray/core/rolling.py in _mapping_to_list(self, arg, default, allow_default, allow_allsame) 157 return [arg] 158 else: --> 159 raise ValueError("Mapping argument is necessary.") 160 161

ValueError: Mapping argument is necessary. `` so the bug is inRolling.construct`.

This looks good to me. Unfortunately, now that the repr is fixed, we run into a separate issue, triggered by https://github.com/pydata/xarray/blob/7e1fbf87b898851d26de3eb3ab328637d61437bc/doc/computation.rst#L236-L237

Edit: what do you think about using this to fix str-typed window_dims (while it works I'm not sure if that actually makes sense, though): ```diff diff --git a/xarray/core/rolling.py b/xarray/core/rolling.py index 62d14fad..a5f83250 100644 --- a/xarray/core/rolling.py +++ b/xarray/core/rolling.py @@ -275,6 +275,8 @@ class DataArrayRolling(Rolling): "Either window_dim or window_dim_kwargs need to be specified." ) window_dim = {d: window_dim_kwargs[d] for d in self.dim} + elif isinstance(window_dim, str): + window_dim = {d: window_dim for d in self.dim}

     window_dim = self._mapping_to_list(
         window_dim, allow_default=False, allow_allsame=False

``` Edit2: that should be the last docs-only failure related to "rolling", I can build the docs locally with that change.

Thanks @keewis for checking. I'm not sure what causes the error in rolling_da.mean("window_dim", skipna=False)

self._mapping_to_list should handle this problem. How can I get the details of this error? I just saw the time-out error...