deprecate open_rasterio in favour of a rioxarray entrypoint?

[dcherian]

Now that our backends work is well underway, it seems like a good time to discuss deprecating the open_rasterio function and outsourcing it to rioxarray (which would then implement support for the new entrypoint).

From @jhamman's comment

There's also the question of where the rasterio backend should live once #1970 is implemented. I think we can make a pretty strong argument that it should move out of xarray and into a 3rd party package. Perhaps rasterio itself, or, more likely something like rio-xarray (cc @snowman2). See #3166 for a demo of how this may work.

As you can see, there are a few big questions up in the air right now. Work on #1970 will begin this summer so now is a great time to game out the various options for the rasterio backend.

I am in favour of doing this. I see no reason why there should be two slightly divergent versions of the same function, and (anecdotally) the rioxarray version seems to be more full-featured and regularly improved.

@pydata/xarray What do you think?

If we agree to do this (and rioxarray agrees), we could put this as a "deliverable" in our NASA proposal. This work would be totally relevant to that call.

Related: #4142

cc @snowman2 @scottyhq @JessicaS11 @WeatherGod

[snowman2]

Looks like entrypoint support was added in #4577. What is the timeline for the next release?

and rioxarray agrees

Is rioxarray agreeing to allow the addition of entrypoint support? If so, 👍

[keewis]

What is the timeline for the next release?

There's no timeline, yet. Since we just released 0.16.2, maybe in about 2-3 months?

[snowman2]

@alexamici added the entrypoint here; corteva/rioxarray#281. It seems xarray 0.18+ is required for it to work. Any timeline for the 0.18 release?

[max-sixty]

I think we could release 0.17.1 fairly soon — nothing is pending but it's also fairly easy to do a release.

Do we want to deprecate open_rasterio and point people towards rioxarray? We can do that now.

[alexamici]

@max-sixty I'm +1 on deprecating xarray flavour of open_rasterio and direct users to rioxarray.

BTW, I would rather prefer the next release to be 0.18.0 as the backend changes are both large and user visible (open_dataset now accept **kwargs that are added to **backend_kwargs).

[max-sixty]

OK great @alexamici

[keewis]

if we deprecate xr.open_rasterio we should also do the same for xr.tutorial.open_rasterio and get xr.tutorial.open_dataset("RGB.byte") to work.

[keewis]

Since it comes up pretty often: should we go through with deprecating open_rasterio and the builtin rasterio backend?