id,node_id,number,title,user,state,locked,assignee,milestone,comments,created_at,updated_at,closed_at,author_association,active_lock_reason,draft,pull_request,body,reactions,performed_via_github_app,state_reason,repo,type
359307319,MDU6SXNzdWUzNTkzMDczMTk=,2411,datetime parser different between 32bit and 64bit installations,2656596,closed,0,,,2,2018-09-12T04:03:11Z,2018-09-13T03:54:17Z,2018-09-13T03:54:17Z,NONE,,,,"I just noticed that my 64bit Python 3.6 installation with xarray does not parse the datetimes correctly from a netcdf, while the 32bit version seems to work fine. There seems to be some kind of offset.

Since I can't attach a netcdf to this issue, here's a link to an example netcdf file:
https://drive.google.com/open?id=1WhtWmN1ZBpbsfu7q7hGy2le-nujInfkI

This seems to happen for both 0.10.7 and 0.10.8.
The datetime values should be hourly.

Thanks!

```python
import xarray as xr

ms1 = xr.open_dataset(nc)
print(ms1.time)
```


","{""url"": ""https://api.github.com/repos/pydata/xarray/issues/2411/reactions"", ""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,completed,13221727,issue
270440308,MDU6SXNzdWUyNzA0NDAzMDg=,1681,Serializing attrs,2656596,closed,0,,,3,2017-11-01T20:24:48Z,2017-11-12T07:02:41Z,2017-11-12T07:02:41Z,NONE,,,,"I'm running Windows 10, Python 2.7, and xarray version 0.9.6

Using this dictionary for a Dataset attribute:
```python
from xarray import Dataset

crs = {u'ellps': u'GRS80',
 u'k': 0.9996,
 u'lat_0': 0,
 u'lon_0': 173,
 u'no_defs': True,
 u'proj': u'tmerc',
 u'units': u'm',
 u'x_0': 1600000,
 u'y_0': 10000000}

ds = Dataset()
ds.attrs['crs'] = crs
```

When attempting to serialize the dataset attribute, I get the 

```
TypeError ""Invalid value for attr: {u'lon_0': 173, u'k': 0.9996, u'ellps': u'GRS80', u'y_0': 10000000, u'no_defs': 'True', u'proj': u'tmerc', u'x_0': 1600000, u'units': u'm', u'lat_0': 0} must be a number string, ndarray or a list/tuple of numbers/strings for serialization to netCDF files""
```

This used to work in previous versions.

I then convert it to a list of tuples via crs.items(), and it successfully saves the netcdf.
But when I then open the netcdf back up, the list of tuples become one long concatenated string.
I'm guessing the core netcdf library has become more restrictive, but it would be nice to fully clarify what can be serialized or not.

Thanks","{""url"": ""https://api.github.com/repos/pydata/xarray/issues/1681/reactions"", ""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,completed,13221727,issue