Direct loading via xr.open_dataset() sets ds.encoding['unlimited_dims']
But currently, ncdata.xarray.to_xarray does not do this, meaning that unlimited dims get lost.
( Maybe surprisingly though, this does not affect Dataset.identical() checking : encodings do not count there )
NOTE:
if fixed, remove reference to this in docs
Direct loading via
xr.open_dataset()setsds.encoding['unlimited_dims']But currently,
ncdata.xarray.to_xarraydoes not do this, meaning that unlimited dims get lost.( Maybe surprisingly though, this does not affect
Dataset.identical()checking : encodings do not count there )NOTE:
if fixed, remove reference to this in docs