issue_comments: 617280517

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/3774#issuecomment-617280517 https://api.github.com/repos/pydata/xarray/issues/3774 617280517 MDEyOklzc3VlQ29tbWVudDYxNzI4MDUxNw== 1217238 2020-04-21T16:49:55Z 2020-04-21T16:49:55Z MEMBER

I don't think it's a great idea to automatically turn scalar coords into 1d arrays. It's not uncommon to have datasets with a whole handful of scalar coordinates, which could potentially become very high dimensional due to this change.

I also don't like fallback logic that tries to do matching by coordinates with dimensions first, and then falls back to using scalars. These types of heuristics look very convenient first (and are very convenient much of the time) but then have a tendency to fail in unexpected/unpredictable ways.

The other choice for situations like this would be to encourage switching to combine_nested for use cases like this, e.g., ```

xr.combine_nested([data_0, data_1], concat_dim='trial') <xarray.Dataset> Dimensions: (time: 3, trial: 2) Coordinates: * time (time) int64 0 1 2 * trial (trial) int64 0 1 Data variables: temperature (trial, time) int64 10 20 30 50 60 70 ```

We could even put a reference to combine_nested in this error raised by combine_by_coords.

 
{
    "total_count": 2,
    "+1": 2,
    "-1": 0,
    "laugh": 0,
    "hooray": 0,
    "confused": 0,
    "heart": 0,
    "rocket": 0,
    "eyes": 0
}
   566490806