Deepak Cherian

I'd like that but the code is inscrutable to me! :)

I keep misunderstanding this issue so typing this out to make sure I got it right. Writing out dimension names in square brackets ```python ds['a': 'x', 'b': 'z'].broadcast_like(da: ['x', 'y'])...

@mjwillson posted a nice summary in #10031 : > I had to summarize the overall problem, it's that behaviour of xarray.broadcast (and Dataset.broadcast_like etc) is not consistent with how actual...

Thanks for taking the time to file a bug report! > I totally fail to see how this process is deleting a level of the MultiIndex - or modifying the...

cc @Mikejmnez

The error is here: https://github.com/pydata/xarray/blob/e49cfc4f2009eb660411128eaf05fa3e714a0c83/xarray/core/groupby.py#L78 and is confusing because we check against a possibly stacked object https://github.com/pydata/xarray/blob/e49cfc4f2009eb660411128eaf05fa3e714a0c83/xarray/core/groupby.py#L679-L681 It may be simplest to reword the error message

@alippai it would be great if you could contribute a PR with these tests (xfailed if needed), suitably modified by @ilan-gold's feedback above.

@ilan-gold do you see a way to fix it? is it fixable?

Nice example! This is a very silly bug here: https://github.com/pydata/xarray/blob/172ba1ed6b4471aa941af886e0a3eded5c31f372/xarray/core/groupby.py#L180 ``` np.concatenate([[0], [1], [], [2]])# -> float64 np.concatenate([[0], [1], [2]])# -> int64 ``` Are you able to send a PR...

Nice profile! @jrbourbeau https://github.com/dask/dask/pull/10648 probably improves this by a lot. Can that be reviewed/merged please? edit: xref https://github.com/dask/dask/pull/10269