concat fails with attrs that are dictionaries with ndarrays

[griffinmilsap]

Code Sample

import numpy as np
import xarray as xr

arrs = [
    xr.DataArray( [ [1], [2] ], 
                 dims = [ 'x', 'y' ], 
                 attrs = { 'meta': { 'bar': np.array( [ 10, 20, 30 ] ) } } ),
    xr.DataArray( [ [3], [4] ],
                 dims = [ 'x', 'y' ],
                 attrs = { 'meta': { 'bar': np.array( [ 10, 20, 30 ] ) } } )
]
print( arrs[0] )
print( arrs[1] )
print( xr.concat( arrs, dim = 'y' ) )

Fails with the following error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-117-9d4c21b40356> in <module>
      9 print( arrs[0] )
     10 print( arrs[1] )
---> 11 print( xr.concat( arrs, dim = 'y' ) )

/usr/local/lib/python3.6/dist-packages/xarray/core/combine.py in concat(objs, dim, data_vars, coords, compat, positions, indexers, mode, concat_over)
    118         raise TypeError('can only concatenate xarray Dataset and DataArray '
    119                         'objects, got %s' % type(first_obj))
--> 120     return f(objs, dim, data_vars, coords, compat, positions)
    121 
    122 

/usr/local/lib/python3.6/dist-packages/xarray/core/combine.py in _dataarray_concat(arrays, dim, data_vars, coords, compat, positions)
    337 
    338     ds = _dataset_concat(datasets, dim, data_vars, coords, compat,
--> 339                          positions)
    340     return arrays[0]._from_temp_dataset(ds, name)
    341 

/usr/local/lib/python3.6/dist-packages/xarray/core/combine.py in _dataset_concat(datasets, dim, data_vars, coords, compat, positions)
    303         if k in concat_over:
    304             vars = ensure_common_dims([ds.variables[k] for ds in datasets])
--> 305             combined = concat_vars(vars, dim, positions)
    306             insert_result_variable(k, combined)
    307 

/usr/local/lib/python3.6/dist-packages/xarray/core/variable.py in concat(variables, dim, positions, shortcut)
   1964         return IndexVariable.concat(variables, dim, positions, shortcut)
   1965     else:
-> 1966         return Variable.concat(variables, dim, positions, shortcut)
   1967 
   1968 

/usr/local/lib/python3.6/dist-packages/xarray/core/variable.py in concat(cls, variables, dim, positions, shortcut)
   1417                 if var.dims != first_var.dims:
   1418                     raise ValueError('inconsistent dimensions')
-> 1419                 utils.remove_incompatible_items(attrs, var.attrs)
   1420 
   1421         return cls(dims, data, attrs, encoding)

/usr/local/lib/python3.6/dist-packages/xarray/core/utils.py in remove_incompatible_items(first_dict, second_dict, compat)
    174         if (k not in second_dict or
    175             (k in second_dict and
--> 176                 not compat(first_dict[k], second_dict[k]))):
    177             del first_dict[k]
    178 

/usr/local/lib/python3.6/dist-packages/xarray/core/utils.py in equivalent(first, second)
    122     else:
    123         return ((first is second) or
--> 124                 (first == second) or
    125                 (pd.isnull(first) and pd.isnull(second)))
    126 

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Problem description

This is a problem because the following code actually executes properly

import numpy as np
import xarray as xr
arrs = [
    xr.DataArray( [ [1], [2] ], 
                 dims = [ 'x', 'y' ], 
                 attrs = { 'meta': np.array( [ 10, 20, 30 ] ) } ),
    xr.DataArray( [ [3], [4] ],
                 dims = [ 'x', 'y' ],
                 attrs = { 'meta': np.array( [ 10, 20, 30 ] ) } )
]
print( arrs[0] )
print( arrs[1] )
print( xr.concat( arrs, dim = 'y' ) )

<xarray.DataArray (x: 2, y: 1)>
array([[1],
       [2]])
Dimensions without coordinates: x, y
Attributes:
    meta:     [10 20 30]
<xarray.DataArray (x: 2, y: 1)>
array([[3],
       [4]])
Dimensions without coordinates: x, y
Attributes:
    meta:     [10 20 30]
<xarray.DataArray (x: 2, y: 2)>
array([[1, 3],
       [2, 4]])
Dimensions without coordinates: x, y
Attributes:
    meta:     [10 20 30]

Equivalence for an array within a nested dictionary as an attribute is evaluated differently than an array attribute, which is non-intuitive. This bug is related to #2060 but is additionally pointing out a difference in evaluation for more complex attributes.

Expected Output

The output of the code sample should concatenate successfully with the nested dictionary attribute, or a more easily interpretable error should be thrown telling me I'm dumb for using dictionaries in attributes. (See #2060)

Output of xr.show_versions()

Details

INSTALLED VERSIONS

commit: None
python: 3.6.6.final.0
python-bits: 64
OS: Linux
OS-release: 4.15.0-23-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8

xarray: 0.10.9
pandas: 0.23.4
numpy: 1.15.3
scipy: 1.1.0
netCDF4: None
h5netcdf: None
h5py: None
Nio: None
zarr: None
cftime: None
PseudonetCDF: None
rasterio: None
iris: None
bottleneck: None
cyordereddict: None
dask: None
distributed: None
matplotlib: 3.0.0
cartopy: None
seaborn: None
setuptools: 40.4.3
pip: 9.0.1
conda: None
pytest: None
IPython: 7.0.1
sphinx: None

[shoyer]

See my comment on #2060. I think we should probably drop this attribute check all together.

Any interest in putting together a PR?