Tom Nicholas

> extrapolate functionality might get dropped. It was never consistently working and is more suited for actual physical boundaries Are we definitely doing this then? Because there are quite a...

> Are we definitely doing this then? Because there are quite a few xfailed tests that use extrapolate in grid_ufunc_refactor_project right now. Can I just remove them? We are removing...

From a code simplicity point of view it would be nice to find a way to do this generally (and optionally) within `apply_as_grid_ufunc`. EDIT: I think I'm agreeing with @dcherian...

> Should we try to start benchmarking the code somehow before merging @TomNicholas refactor? The only "benchmarking" I have done so far was checking that the new dask algorithm creates...

Also I think we should concentrate on making this work in the simplest way first, and then if that turns out not to be fast enough we divide into more...

https://github.com/xgcm/xgcm/pull/459/commits/a7c4a442aa6ed824485bfa621b9223160eda9acb means that the edge case of a vector input being passed (so that `isvector=True`) but an `other_component` is not supplied causes a `ValueError` to be raised. This isn't quite...

> Should we make a 'detailed' repr for devs? @jbusecke that's what `__repr__` is already supposed to be! The "conventional" thing to do is to have the `__str__` and `__repr__`...

Hi @rcaneill , thanks for raising your concerns here. As @rabernat says this is actually intended behaviour, though I can see how it is counterintuitive! The behaviour was implemented in...

Chiming in 4 years after the original opening to give the current situation (from an xarray dev's biased perspective). :sweat_smile: (@tlambert03 sent me here from [here](https://github.com/napari/napari/pull/3715#issuecomment-1189218372)) > There is a...

Yes Kevin and I also discussed support for multi-resolution data in Napari. Ndpyramid wraps xarray-DataTree, a prototype feature addition to xarray. I am the main person developing xarray-DataTree, so feel...