open_mfdataset error: var not present in all datasets or conflicting dimensions

[gewitterblitz]

Hi,

I have unsuccessfully tried loading multiple netcdf files using xr.open_mfdataset. The issue is that not all the netcdf files contain the desired variable in them. Secondly, the dimension cells does not have the same size across all netcdf files. Here's a few screenshots of individual datasets for reference (notice the variable dimension size along dim ='cells'):

Here's what I have tried so far:

def preprocessing(ds):
    if ds['cells'].size == 0:
        ds_new = ds.assign_coords({'cell':np.nan})
        return ds_new
    else:
        return ds
    
xr.open_mfdataset(files,combine='nested',concat_dim='cells',preprocess=preprocessing,data_vars='different')

ValueError: 'corrected_differential_reflectivity' is not present in all datasets.

xr.open_mfdataset(files,combine='nested',concat_dim='iops',preprocess=preprocessing)

ValueError: cannot reindex or align along dimension 'cells' because of conflicting dimension sizes: {1, 2, 3, 4}

[dcherian]

I think we should've fixed this on xarray 2023.02.0 #7400 . What version are you running?

cc @kmuehlbauer

[gewitterblitz]

Thanks, @dcherian, the new version works fine. My conda environment took forever to update xarray (and other dependencies) from version '2023.1.0'.