Adjust after pydata#8183

xarray/core/rolling_exp.py

@@ -131,42 +131,15 @@ def mean(self, keep_attrs: bool | None = None) -> T_DataWithCoords:
 
         dim_order = self.obj.dims
 
-        from xarray.core.dataset import Dataset
-
-        if isinstance(self.obj, Dataset):
-            new_ds = self.obj.copy()
-            if not keep_attrs:
-                new_ds.attrs = {}
-            for name, var in self.obj.variables.items():
-                if self.dim in self.obj[name].dims:
-                    new_ds[name] = apply_ufunc(
-                        move_exp_nanmean,
-                        var,
-                        input_core_dims=[[self.dim]],
-                        kwargs=dict(alpha=self.alpha, axis=-1),
-                        output_core_dims=[[self.dim]],
-                        # I'm not sure whether this line should be here (can look later,
-                        # docs aren't _that_ clear vs. `input_core_dims`, or I'm being
-                        # slow at understanding them.)
-                        exclude_dims={self.dim},
-                        keep_attrs=keep_attrs,
-                    )
-                # Currently we test for attrs being removed on all arrays, even those
-                # without the dim, not sure if we should be doing this or not (we can
-                # indent a line and then only remove attrs on arrays we compute on)
-                if not keep_attrs:
-                    new_ds[name].attrs = {}
-            return new_ds.transpose(*dim_order)
-        else:
-            return apply_ufunc(
-                move_exp_nanmean,
-                self.obj,
-                input_core_dims=[[self.dim]],
-                kwargs=dict(alpha=self.alpha, axis=-1),
-                output_core_dims=[[self.dim]],
-                exclude_dims={self.dim},
-                keep_attrs=keep_attrs,
-            ).transpose(*dim_order)
+        return apply_ufunc(
+            move_exp_nanmean,
+            self.obj,
+            input_core_dims=[[self.dim]],
+            kwargs=dict(alpha=self.alpha, axis=-1),
+            output_core_dims=[[self.dim]],
+            exclude_dims={self.dim},
+            keep_attrs=keep_attrs,
+        ).transpose(*dim_order)
 
     def sum(self, keep_attrs: bool | None = None) -> T_DataWithCoords:
         """