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/pull/6059#issuecomment-1050713711,https://api.github.com/repos/pydata/xarray/issues/6059,1050713711,IC_kwDOAMm_X84-oJ5v,10194086,2022-02-25T10:07:01Z,2022-02-25T10:16:09Z,MEMBER,"Turns out `align` does not entirely do what I thought it does. For e.g. `da1 * da2` there is an `align` call for the `DataArray` and then broadcasting happens on the level of the `Variable` - which we skip here. Therefore, I updated the PR ([5c251e0](https://github.com/pydata/xarray/pull/6059/commits/5c251e02ef1dbe5f8582ef1a2ec1843dfbeb9a13)) - to use `xr.broadcast` to ensure `da` and `weights` have the same shape.
Notes
1. The tests should fail because I updated the `skipna` logic and this was also ""wrong"" (inconsistent with the rest of xarray) in `xr.Dataset.quantile` - thus this needs #6303 first.
2. `broadcast` does an outer join which is kind of pointless here because it adds `NaN` values that are then removed again (#6304)
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1076265104
https://github.com/pydata/xarray/pull/6059#issuecomment-1047791529,https://api.github.com/repos/pydata/xarray/issues/6059,1047791529,IC_kwDOAMm_X84-dAep,10194086,2022-02-22T13:21:52Z,2022-02-22T13:21:52Z,MEMBER,"Thanks, for checking - I didn't anticipate the missing core dim... Seems the `align` may be required, but it should be possible to leave out the shape check. I'll have a closer look (also how `dot` does it - because this is what's used for the other weighted methods).","{""total_count"": 1, ""+1"": 1, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1076265104
https://github.com/pydata/xarray/pull/6059#issuecomment-1035104843,https://api.github.com/repos/pydata/xarray/issues/6059,1035104843,IC_kwDOAMm_X849snJL,10194086,2022-02-10T16:09:38Z,2022-02-10T16:09:38Z,MEMBER,"> We have local changes that expose `method` in the public API, and compare results against numpy for the case with uniform weights to confirm results are identical.
>
> Do you want us to push those changes here ?
If you are confident they work with the weights... numpy now recommends using median_unbiased or normal_unbiased, so would be nice if we have them.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1076265104
https://github.com/pydata/xarray/pull/6059#issuecomment-1032839546,https://api.github.com/repos/pydata/xarray/issues/6059,1032839546,IC_kwDOAMm_X849j-F6,10194086,2022-02-08T16:54:05Z,2022-02-08T16:54:05Z,MEMBER,"Good point, yes the keyword should be named `method` as well for consistency (but as far as I can see it is currently not exposed?).
Sorry for the slow pace - this PR merits another review. I can probably look at the code but I am not sure about the numerics. I think it's important to align with numpy but their code is not yet merged (AFAIK). Maybe @bzah knows if this would be consistent with numpys approach?","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1076265104