Ryan Abernathey

> I've always found it slightly annoying that xgcm separates the metadata (Grid) and the data (Dataset). The accessor approach in #197 could mitigate this.

> 1. For conservative regridding on irregularly spaced depth, you'll need the API to read center depth and interfaces depth This is already part of xgcm's data model, which is...

> I think it is preferable to have one single algorithm that accomplishes everything. I'm not sure I agree with this 100%. There is a tradeoff between the complexity of...

This discussion has exploded and it's amazing to see all the ideas and enthusiasm. Thanks everyone for contributing. I want to turn the discussion towards the code itself. The main...

This is the same core issues as #518

The ` --> right` indicates the "default" shift. I would favor just removing this part from the repr.

Thanks for opening this issue @rcaneill. I agree this needs some discussion. In my recollection, the idea is that cumsum should be the inverse of diff. So whatever order the...

Hi @untoreh! Sorry no one has replied to your issue yet. It has been a very busy period for most Zarr devs. It looks like we don't support datetimes in...

In order to integrate the notebook into the docs, I highly recommend [myst-nb](https://myst-nb.readthedocs.io/en/latest/). We use it on all our documentation sites. It supports all of the fancy sphinx syntax in...

Very good issue. There is a lot to unpack here. But I just wanted to point you to https://github.com/dask/s3fs/issues/285 -- this could likely be the source of some slowness in...