issue_comments

Thanks, that looks interesting, although sounds like it's addressing a slightly different problem; I'm not so much interested in validation of external inputs, more just in having some basic datatypes that can be used to specify dims/shape/coords as templates for DataArray / Datasets internally within my codebase.

Some things I'm looking for but don't appear to be supported: * Support for specifying coords * __hash__, __eq__ etc for the Schema objects * Convenient APIs to alter and combine these Schemas in similar ways to what can be done with DataArray / Datasets themselves, e.g. adding/removing dimensions, broadcasting against eachother etc -- perhaps mirroring APIs like expand_dims that can be used on DataArray / Dataset themself, to the extent this makes sense.

Feels to me that it would make sense to have these basic datatypes inside xarray, perhaps with something like xarray-schema providing extra validation helpers etc on top of them? But just my 2 cents :)

Re xr.broadcast(ds)[0].sum(dims) -- Thanks, that's neat and may be useful as a workaround, but it looks like it could incur significant extra CPU and RAM costs (tiling all variables to the full size in memory before summing over the tiled values)? Or is there some clever optimisation under the hood which would avoid this?

I also only wanted it to (behave as though it) broadcast the dims that are summed over, but this looks like it will broadcast all dims including those not summed over?

Overall I think it'd be better to have an option on sum (like missing_dim='broadcast' as suggested in #6749), rather than documenting a partial workaround like this, given the caveats attached to the workaround and that (to me at least) the broadcasting sum is more in keeping with the usual mathematical semantics of 'sum' than what 'sum' currently does.

+1 for a check_dim_order option to .equals, assert_equal that can be disabled. (Ideally I think the default would be not to check dim order, but that ship has sailed now).

Or failing that, it would at least be nice to have xarray.testing.assert_equal_modulo_dim_order etc. When writing tests I usually don't care about dimension order and it's frustrating to have to manually do e.g. xarray.testing.assert_allclose(a, b.transpose(a.dims)).

As pointed out, most of the xarray API is dimension-order-invariant and so it's odd to have no supported way to do comparisons in a dimension-order-invariant way.

For now I am working around this using:

python orig_tight_layout = mpl.figure.Figure.tight_layout mpl.figure.Figure.tight_layout = lambda *a: None try: xarray.plot.pcolormesh(...args...) finally: mpl.figure.Figure.tight_layout = orig_tight_layout

obviously you don't get a tight layout this way, but the proportions are a lot better.