Two strange errors in tests TestVariable using .copy()

[ArchangeGabriel]

Setup is as described in #3777.

The two failures are the following ones:

____________________ TestVariable.test_index_0d_not_a_time _____________________

self = <xarray.tests.test_units.TestVariable object at 0x7f3926b2ad00>

    def test_index_0d_not_a_time(self):
        d = np.datetime64("NaT", "ns")
        x = self.cls(["x"], [d])
>       self._assertIndexedLikeNDArray(x, d)

xarray/tests/test_variable.py:206: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

self = <xarray.tests.test_units.TestVariable object at 0x7f3926b2ad00>
variable = <xarray.Variable (x: 1)>
<Quantity(['NaT'], 'meter')>
expected_value0 = numpy.datetime64('NaT'), expected_dtype = None

    def _assertIndexedLikeNDArray(self, variable, expected_value0, expected_dtype=None):
        """Given a 1-dimensional variable, verify that the variable is indexed
        like a numpy.ndarray.
        """
        assert variable[0].shape == ()
        assert variable[0].ndim == 0
        assert variable[0].size == 1
        # test identity
>       assert variable.equals(variable.copy())
E       AssertionError: assert False
E        +  where False = <bound method Variable.equals of <xarray.Variable (x: 1)>\n<Quantity(['NaT'], 'meter')>>(<xarray.Variable (x: 1)>\n<Quantity(['NaT'], 'meter')>)
E        +    where <bound method Variable.equals of <xarray.Variable (x: 1)>\n<Quantity(['NaT'], 'meter')>> = <xarray.Variable (x: 1)>\n<Quantity(['NaT'], 'meter')>.equals
E        +    and   <xarray.Variable (x: 1)>\n<Quantity(['NaT'], 'meter')> = <bound method Variable.copy of <xarray.Variable (x: 1)>\n<Quantity(['NaT'], 'meter')>>()
E        +      where <bound method Variable.copy of <xarray.Variable (x: 1)>\n<Quantity(['NaT'], 'meter')>> = <xarray.Variable (x: 1)>\n<Quantity(['NaT'], 'meter')>.copy

xarray/tests/test_variable.py:151: AssertionError
_____________________ TestVariable.test_equals_all_dtypes ______________________

self = <xarray.tests.test_units.TestVariable object at 0x7f39282b4e50>

    def test_equals_all_dtypes(self):
        for v, _ in self.example_1d_objects():
            v2 = v.copy()
>           assert v.equals(v2)
E           AssertionError: assert False
E            +  where False = <bound method Variable.equals of <xarray.Variable (x: 3)>\n<Quantity(['2000-01-01T00:00:00.000000000' '2000-01-02T00:00:00.000000000'\n '2000-01-03T00:00:00.000000000'], 'meter')>>(<xarray.Variable (x: 3)>\n<Quantity(['2000-01-01T00:00:00.000000000' '2000-01-02T00:00:00.000000000'\n '2000-01-03T00:00:00.000000000'], 'meter')>)
E            +    where <bound method Variable.equals of <xarray.Variable (x: 3)>\n<Quantity(['2000-01-01T00:00:00.000000000' '2000-01-02T00:00:00.000000000'\n '2000-01-03T00:00:00.000000000'], 'meter')>> = <xarray.Variable (x: 3)>\n<Quantity(['2000-01-01T00:00:00.000000000' '2000-01-02T00:00:00.000000000'\n '2000-01-03T00:00:00.000000000'], 'meter')>.equals

xarray/tests/test_variable.py:386: AssertionError

I have no idea what’s wrong here.

[mathause]

Could you list the versions of the installed packages, i.e. the output of xr. show_versions()

[ArchangeGabriel]

Sure:

INSTALLED VERSIONS
------------------
commit: None
python: 3.8.1 (default, Jan 22 2020, 06:38:00) 
[GCC 9.2.0]
python-bits: 64
OS: Linux
OS-release: 5.5.3-arch1-1
machine: x86_64
processor: 
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8
libhdf5: 1.10.5
libnetcdf: 4.7.3

xarray: 0.15.0
pandas: 1.0.1
numpy: 1.18.1
scipy: 1.4.1
netCDF4: 1.5.3
pydap: None
h5netcdf: None
h5py: None
Nio: None
zarr: None
cftime: 1.1.0
nc_time_axis: None
PseudoNetCDF: None
rasterio: None
cfgrib: None
iris: None
bottleneck: 1.3.1
dask: None
distributed: None
matplotlib: 3.1.3
cartopy: None
seaborn: 0.10.0
numbagg: None
setuptools: 45.2.0
pip: None
conda: None
pytest: 5.3.5
IPython: None
sphinx: None

BTW, why does it report h5py? I did not see it in the list of optional dependencies. Same for distributed by the way.

[mathause]

@keewis could it be that the wrong version of pint is installed? Should pint be added to xr.show_versions?

[ArchangeGabriel]

I have pint 0.10.1, and 0.10 definitively changed a lot of things as also shown in #3266 (comment).

[max-sixty]

Should pint be added to xr.show_versions?

👍

[keewis]

since pint became (or will become, once #3594 is closed) a optional dependency, we will definitely have to add it to show_versions.

Note that none of them will fail with dask installed (which is why we didn't catch this with CI):
without dask:

d = np.datetime64("NaT", "ns")
v = xr.Variable(["x"], ureg.Quantity([d], ureg.m))
assert v.equals(v.copy())  # AssertionError

with dask, equals returns True and the tests pass.

[keewis]

equals returning False in the absence of dask is because of

xarray/xarray/core/duck_array_ops.py

Lines 74 to 81 in 52ee5df

	if hasattr(np, "isnat") and (
	dask_array is None or hasattr(dask_array_type, "__array_ufunc__")
	):
	# np.isnat is available since NumPy 1.13, so __array_ufunc__ is always
	# supported.
	isnat = np.isnat
	else:
	isnat = _dask_or_eager_func("isnull", eager_module=pd)

which uses np.isnat if dask is absent and pd.isnull if dask is present. Since dask_array_type is a tuple, hasattrwill always return False.

Wrapping a datetime in a quantity does not really make sense so I'll try to remove those tests.