The xarray sharp bits

Two things about the integration surprise people. Both follow from how xarray itself works rather than from anything unxts.interop.xarray chose, which is why neither is going to be fixed.

Dimension coordinates cannot hold quantities

xarray backs every dimension coordinate (one named like its dimension, shown with a * in the repr) with a pandas.Index. Building that index coerces the data to a plain numpy array, so a dimension coordinate cannot hold a Quantity. This is inherent to xarray’s indexing model, not something unxts.interop.xarray can override, and it affects every duck-array unit library (including pint-xarray) the same way.

Assigning one raises rather than dropping the unit on the floor — Quantity.__array__ refuses to hand a dimensionful value to a consumer that cannot see its unit. quantify() handles this for you by leaving dimension coordinates plain.

Workaround: store the unitful values on a non-dimension coordinate, keeping a plain index on the dimension itself:

>>> import unxt as u
>>> import xarray as xr

>>> data = [10.0, 20.0, 30.0]
>>> quantities = u.Q([1.0, 2.0, 3.0], "m")

>>> # Dimension coordinate: refused, because the unit could not survive
>>> try:
...     xr.DataArray(data, dims=["x"], coords={"x": quantities})
... except Exception as e:
...     print(type(e).__name__)
UnitConversionError

>>> # Non-dimension coordinate: the Quantity (and its unit) is preserved
>>> da = xr.DataArray(data, dims=["i"],
...                   coords={"i": [0, 1, 2], "x": ("i", quantities)})
>>> da.coords["x"].data
Quantity(Array([1., 2., 3.], dtype=float32), unit='m')

Operations that drop units

A few xarray operations route through code paths that cannot preserve a Quantity:

• rolling / sliding-window reductions use numpy.lib.stride_tricks, which has no Array API (or jax.numpy) equivalent, so they are unsupported on JAX-backed data generally — not specific to units.

• interp delegates to scipy/numpy interpolation internally and returns a plain array (the same behavior as pint-xarray).

For these, dequantify, operate, then re-quantify, or work on .data with unxt/quaxed directly.