UxDataset.isel() silently fails to slice when providing grid dim not in the dataset #1713

Version

v2026.8.0

How did you install UXarray?

Source

What happened?

Calling UxDataset.isel() to select along a grid dimension other than the grid dimension where the data is currently located successfully returns a UxDataset with the grid sliced properly, but the data not sliced at all. For example, for data with "n_face" dimension, slicing at n_edge=7 as mentioned in the docstring (added in #1684) leads to a result with uxgrid having only the two faces touching edge 7, but with "n_face" data dimension untouched.

I confirmed that this bug was not introduced by #1684; it exists on main even before that commit was merged. The behavior is inconsistent with UxDataArray.isel(), which instead also slices the "n_face" data dimension appropriately to keep the uxgrid and data in synch.

What did you expect to happen?

I expected the data to sliced appropriately such that the result's data contains precisely the edges/nodes/faces corresponding to the result's uxgrid.

Can you provide a MCVE to repoduce the bug?

import uxarray as ux
ds = ux.tutorial.open_dataset("outCSne30-vortex")
assert "n_face" in ds.dims
result = ds.isel(n_edge=7)
print((result.sizes["n_face"], result.uxgrid.n_face))
# expected (2, 2), got (5400, 2).

# Note: when using UxDataArray, there is no issue. This works just fine:
arr = ds["psi"]
result = arr.isel(n_edge=7)
print((result.sizes["n_face"], result.uxgrid.n_face))
# prints (2, 2) as expected.