issue_comments: 276695798

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/1084#issuecomment-276695798 https://api.github.com/repos/pydata/xarray/issues/1084 276695798 MDEyOklzc3VlQ29tbWVudDI3NjY5NTc5OA== 6628425 2017-02-01T15:57:29Z 2017-02-01T15:57:29Z MEMBER

@shoyer @jhamman I will defer to both of you on this issue. In light of recent discussion, what do you think is the preferred approach? It seems like three avenues have been discussed (here and in pandas-dev/pandas#15258):

  1. Subclass pandas.Index (as I've started on above)
  2. Subclass pandas.DatetimeIndex
  3. Create custom calendars for use in PeriodIndex (do we care about the semantic difference between "Periods" and "Datetimes"?)

It's not immediately obvious to me that avenues 2 and 3 would be clean and straightforward, but if there is a way to easily adapt them for custom calendars — even unusual ones like "all-leap" (29-day Februaries every year) and "360-day" (30-day Februaries every year) calendars, which cannot be fully represented by ordinary datetime.datetime objects — I'd certainly be open to looking into it more (surely the less code we would need to add / test the better).

Are we back to the drawing board, or should we continue along the path of avenue 1?

 
{
    "total_count": 0,
    "+1": 0,
    "-1": 0,
    "laugh": 0,
    "hooray": 0,
    "confused": 0,
    "heart": 0,
    "rocket": 0,
    "eyes": 0
}
   187591179