html_url,issue_url,id,node_id,user,created_at,updated_at,author_association,body,reactions,performed_via_github_app,issue
https://github.com/pydata/xarray/issues/1879#issuecomment-362942383,https://api.github.com/repos/pydata/xarray/issues/1879,362942383,MDEyOklzc3VlQ29tbWVudDM2Mjk0MjM4Mw==,1217238,2018-02-04T21:42:33Z,2018-02-04T21:42:33Z,MEMBER,"> What about toolz used by auto_combine?

`toolz` should probably be listed explicitly as an optional dependencies, but I didn't bother since it's a hard dependency of `dask.array`, and we only use `toolz` in cases where `dask.array` is also required. (In theory, dask.array could drop its use of toolz internally, but that's unlikely to happen anytime soon.)

>  For instance, given a feature with multiple backends, the dependencies for the preferred backend should be recommended, whilst the others should just be suggested.

I don't think the difference between recommended and suggested dependencies is relevant for xarray. We don't have any features with multiple backends where a single backend is strictly preferred.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,293830986