issue_comments: 456766782

Apologies about adding my thoughts after a bit of a gap on here...

My only hesitation here is on the name

I like the to_iris and from_iris method names suggested here. They're consistent with existing functionality available for DataArray, and I think it's reasonable to expect that as DataArrays map to cubes, so also for Datasets and CubeLists. As it's possible for both Datasets and CubeLists to contain only a single object there could be some extra logic to return a DataArray or cube respectively in such a case, but I think that would add needless complexity as it's not hard from a user perspective to get back to the single item from the Dataset or CubeList.

We have some special logic already for figuring out names in DataArray.from_iris

Does this logic include handling multiple cubes of the same name in a single CubeList? Iris will quite happily handle this, but I guess the name:DataArray mapping in Xarray requires unique names. For example:

```python

names = [c.name() for c in cubes] print(names) ['air_pressure', 'air_pressure', 'air_pressure_at_sea_level', 'air_temperature', 'air_temperature', 'air_temperature', 'air_temperature', 'air_temperature', 'dew_point_temperature', 'geopotential_height', 'relative_humidity', 'relative_humidity', 'specific_humidity', 'surface_air_pressure', 'upward_air_velocity', 'x_wind', 'x_wind', 'x_wind', 'y_wind', 'y_wind', 'y_wind'] ```

The differences between these cubes is one or more of: * one cube describes the phenomenon at the surface, and * another cube the phenomenon on height or pressure levels, or * another cube describes the phenomenon after statistical processing, and so on.

If such a case isn't currently handled, it could be handled by using this differing metadata to modify the name used for the key; for example air_temperature --> air_temperature__maximum_1_hr – so long as returning to a CubeList will also return the names to their originals.