issue_comments: 590925016

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-590925016 https://api.github.com/repos/pydata/xarray/issues/3774 590925016 MDEyOklzc3VlQ29tbWVudDU5MDkyNTAxNg== 35968931 2020-02-25T15:30:20Z 2020-02-25T15:30:20Z MEMBER

is _infer_concat_order_from_coords used in other places than combine_by_coords?

combine_by_coords can be called by open_mfdataset, but other than that no.

By the way, there is an additional problem to consider in addition to the above: ambiguity with multiple coordinates for the same dimension.

```python data0 = xr.Dataset({'temperature': ('time', [10,20,30])}, coords={'time': [0,1,2]}) data0.coords['trial'] = 0 data0.coords['day'] = 5

data1 = xr.Dataset({'temperature': ('time', [50,60,70])}, coords={'time': [0,1,2]}) data1.coords['trial'] = 1 data1.coord['day'] = 3 ```

In this example then combine_by_coords won't know which coordinate to use to order along the new dimension, should it be 'trial' or 'day'? They will give differently-ordered results.

We could pass extra optional arguments to deal with this, but for a "magic" function that doesn't seem desirable. It might be better to have some logic like:

1) Is there a dimension coordinate for this dimension? If yes just use that. 2) Otherwise, can the datasets be unambiguously ordered along the dim by looking at non-dimension coords? 3) If yes, do that, if no, throw informative error.

That would be backwards-compatible (currently all magic ordering is being done for dimension coords), and not make any arbitrary choices.

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