Deepak Cherian

Thanks for testing! Yes, these are both important features, but we'll need to make progress on https://github.com/pydata/xarray/issues/8005 first, which is finally happening in https://github.com/pydata/xarray/pull/10137, https://github.com/pydata/xarray/pull/9671

@keewis do you know what's going on here?

Thanks @keewis . I was only looking for input on `pint`. The rest just appeared in the latest run!

Thanks @snowman2 If you're interested, see #3497 for the inverse problem of using pandas attrs when constructing Xarray objects (in a future PR) :)

They were thinking of removing it at one point: https://github.com/pandas-dev/pandas/issues/52166, also https://github.com/dask/dask/issues/11146 perhaps we should punt until someone really really wants it?

@Illviljan can you comment on why this approach is useful if we can't advertise array_api compliance based on [upstream input](https://github.com/data-apis/array-api/issues/698#issuecomment-1800078933)?

> I'm not yet convinced it is impossible for us to not support both. Sure, but the way to proceed would be to build consensus around this in an issue...

> It's hard to change peoples mind sometimes, so I like to have some working examples for discussion. Don't you have enough already? > Still, there is duck_array_ops.py that does...

generally expanding `duck_array_ops` to use the `array_api` more widely would be a great improvement :) still not sure about handling namedarray at that level though

I think that one should be merged in with some kind of gallery that should also include the ecosystem book in this repo. I created a new one so we...