issue_comments: 234434065

This data as json

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/907#issuecomment-234434065 https://api.github.com/repos/pydata/xarray/issues/907 234434065 MDEyOklzc3VlQ29tbWVudDIzNDQzNDA2NQ== 1217238 2016-07-22T01:50:06Z 2016-07-22T01:50:22Z MEMBER

This is a tricky one to fix. Stack is not the only operation that can convert string arrays to dtype=object -- reindex and align also do this.

The challenge is that NumPy does not support variable width strings or strings with missing values. Hence, we end up with dtype=object arrays.

In this particular case, we could preserve the string dtype by keeping track of the original dtype of the arrays at go into a MultiIndex as metadata, similar to how we currently use PandasIndexAdapter to preserve the original dtype on arrays converted into non-multi indexes.

 
{
    "total_count": 0,
    "+1": 0,
    "-1": 0,
    "laugh": 0,
    "hooray": 0,
    "confused": 0,
    "heart": 0,
    "rocket": 0,
    "eyes": 0
}
   166441031