DataTree.identical should consider checking that coordinates are defined at the same level

[shoyer]

What is your issue?

Consider the following two data trees:

one = DataTree.from_dict(
    {
        "/": xr.Dataset(coords={"a": 1}),
        "/b": xr.Dataset(coords={"a": 1}),
    }
)
two = DataTree.from_dict(
    {
        "/": xr.Dataset(coords={"a": 1}),
        "/b": xr.Dataset(),
    }
)

Currently, DataTree.identical() considers one and two to be identical, but the coordinate a is defined redundantly in the first tree.

For most purposes, these DataTree objects are identical, but are edge cases where it could be different, e.g., when wrting these trees to netCDF or Zarr.

It would be helpful if the .identical() checked for such discrepancies, e.g., to facilitate writing unit tests as in #9214.

[TomNicholas]

Can this be closed after the changes in #9555? That's what I understood from this comment.

[shoyer]

Yes, we can close. These trees are neither equal nor identical now.

…

On Thu, Oct 10, 2024 at 1:38 AM Tom Nicholas ***@***.***> wrote: Can this be closed after the changes in #9555 <#9555>? That's what I understood from this comment <https://github.com/pydata/xarray/pull/9588/files#r1791110146>. — Reply to this email directly, view it on GitHub <#9215 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJJFVQOMIX5S7UWTTVIY7TZ2VLXJAVCNFSM6AAAAABKPW2EHKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMBSHAYDMMBQGY> . You are receiving this because you authored the thread.Message ID: ***@***.***>