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/1840#issuecomment-359895525,https://api.github.com/repos/pydata/xarray/issues/1840,359895525,MDEyOklzc3VlQ29tbWVudDM1OTg5NTUyNQ==,2443309,2018-01-23T19:04:24Z,2018-01-23T19:04:24Z,MEMBER,This all looks good now. I merged in master and resolved a small conflict in the docs. I'll merge after the tests pass. ,"{""total_count"": 1, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 1, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,289853579
https://github.com/pydata/xarray/pull/1840#issuecomment-359131278,https://api.github.com/repos/pydata/xarray/issues/1840,359131278,MDEyOklzc3VlQ29tbWVudDM1OTEzMTI3OA==,1217238,2018-01-20T01:06:07Z,2018-01-20T01:06:07Z,MEMBER,"As mentioned in https://github.com/pydata/xarray/issues/1842, maybe we should also make a point not to upcast float32 input?

One possible concern with changing precision from float64 -> float32 is that some reduce operations like mean could become due to lower precision. So it's a good think @fujiisoup wrote https://github.com/pydata/xarray/pull/1841 so we can specify `dtype` in reductions :).

With regards to Hypothesis: I haven't used it, but it does seem very intriguing. I'm sure that it could turn up quite a few bugs in xarray. Would it make sense to add it as an optional dependency to the test suite? Even if we use Hypothesis to cover more edge cases, I think we will still want normal test coverage for most behavior.","{""total_count"": 1, ""+1"": 1, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,289853579