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/615#issuecomment-146988731,https://api.github.com/repos/pydata/xarray/issues/615,146988731,MDEyOklzc3VlQ29tbWVudDE0Njk4ODczMQ==,1217238,2015-10-09T21:21:41Z,2015-10-09T21:21:41Z,MEMBER,"`pandas.Timedelta` has a `to_timedelta64()` method that you also use to do the coercion.

The reason why this broke is that pandas used to return `numpy.timedelta64` objects, but then added the `Timedelta` type (a few versions ago) and switched over to that.

I agree that it would be nice to support this sort of conversion automatically. I opened a new issue for that (#616).
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,110726841
https://github.com/pydata/xarray/issues/615#issuecomment-146976549,https://api.github.com/repos/pydata/xarray/issues/615,146976549,MDEyOklzc3VlQ29tbWVudDE0Njk3NjU0OQ==,291576,2015-10-09T20:15:49Z,2015-10-09T20:15:49Z,CONTRIBUTOR,"hmm, good point. I wish I knew why I ended up using `pd.to_timedelta()` in the first place. Did numpy not support converting timedelta objects at one point?
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,110726841
https://github.com/pydata/xarray/issues/615#issuecomment-146975622,https://api.github.com/repos/pydata/xarray/issues/615,146975622,MDEyOklzc3VlQ29tbWVudDE0Njk3NTYyMg==,2443309,2015-10-09T20:10:45Z,2015-10-09T20:10:45Z,MEMBER,"Do you have to use `pd.to_timedelta()`?  This seems to work:

``` Python
a = xray.Dataset({'time': [datetime(2000, 1, 1)]})
a['time'] -= np.timedelta64(timedelta(hours=6))
```
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,110726841