REF: reshape.concat operate on arrays, not SingleBlockManagers

[jbrockmendel]

cc @TomAugspurger is there a better way to handle the assert_series_equal patch for the PandasArray tests?

[TomAugspurger]

Can you explain the issue your change to assert_series_equal is working around?

[TomAugspurger]

Are PandasDtype instances singletons? I wouldn't expect this to pass for PandasType('int') is PandasDtype('int')

[jbrockmendel]

Not a singleton, no

[jbrockmendel]

does this require action? AFAICT the fact that this is needed is an artifact of the way we are patching PandasArray._typ in these tests

[TomAugspurger]

I think so. Otherwise the implementation is incorrect since PandasArray.astype(PandasDtype("int64")) will raise

In [6]: arr = pd.core.arrays.PandasArray(np.array([1, 2]))

In [7]: arr.astype(pd.core.arrays.numpy_.PandasDtype("float64"))
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-7-eb4babcd3e70> in <module>
----> 1 arr.astype(pd.core.arrays.numpy_.PandasDtype("float64"))

~/sandbox/pandas/pandas/core/arrays/numpy_.py in astype(self, dtype, copy)
    440         if dtype is self.dtype:
    441             return self.copy() if copy else self
--> 442         return super().astype(dtype, copy=copy)
    443
    444     def to_numpy(

~/sandbox/pandas/pandas/core/arrays/base.py in astype(self, dtype, copy)
    432             NumPy ndarray with 'dtype' for its dtype.
    433         """
--> 434         return np.array(self, dtype=dtype, copy=copy)
    435
    436     def isna(self) -> ArrayLike:

TypeError: data type not understood

[jbrockmendel]

You're right, but that isnt any different from the status quo

[TomAugspurger]

I guess the difference is that we're implementing this now, but the implementation isn't correct.

If you don't see a straightforward was to resolve this now, then feel free to open a followup issue and merge.

[TomAugspurger]

I think that is_extension_dtype(other) and other.dtype == self.dtype might suffice.

[TomAugspurger]

To explain a bit more (at risk of over-explaining) a future reader of that exception is likely to be more confused about why this doesn't work than we are now.

[jbrockmendel]

updated to revert the astype change and xfail the affected tests

[jbrockmendel]

Can you explain the issue your change to assert_series_equal is working around?

Without the patched assert_series_equal, we get

_________________________________________________________________________ TestGetitem.test_loc_iloc_frame_single_dtype[object] _________________________________________________________________________

self = <pandas.tests.extension.test_numpy.TestGetitem object at 0x127f29f90>
data = <PandasArray>
[ (0,),  (1,),  (2,),  (3,),  (4,),  (5,),  (6,),  (7,),  (8,),  (9,), (10,),
 (11,), (12,), (13,), (14,...(87,),
 (88,), (89,), (90,), (91,), (92,), (93,), (94,), (95,), (96,), (97,), (98,),
 (99,)]
Length: 100, dtype: object

    def test_loc_iloc_frame_single_dtype(self, data):
        # GH#27110 bug in ExtensionBlock.iget caused df.iloc[n] to incorrectly
        #  return a scalar
        df = pd.DataFrame({"A": data})
        expected = pd.Series([data[2]], index=["A"], name=2, dtype=data.dtype)
    
        result = df.loc[2]
>       self.assert_series_equal(result, expected)

pandas/tests/extension/base/getitem.py:81: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

cls = <class 'pandas.tests.extension.test_numpy.TestGetitem'>, left = A    (2,)
Name: 2, dtype: object, right = A    (2,)
Name: 2, dtype: object, args = (), kwargs = {}

    @classmethod
    def assert_series_equal(cls, left, right, *args, **kwargs):
>       return tm.assert_series_equal(left, right, *args, **kwargs)
E       AssertionError: Attributes of Series are different
E       
E       Attribute "dtype" are different
E       [left]:  object
E       [right]: object

pandas/tests/extension/base/base.py:13: AssertionError

test_loc_iloc_frame_single_dtype is currently xfailed. The edit to PandasArray.astype was because this had caused one of the cases of test_loc_iloc_frame_single_dtype to XPASS. Removing the xfail led to 9 failures, 8 of which were fixed by the edit to PandasArray.astype.

I think the assert_series_equal issue has to do with the way we are patching PandasArray._typ in these tests, but i havent chased it down to be sure.

[jbrockmendel]

Updated to just xfail more selectively and avoid patching assert_series_equal

[TomAugspurger]

Can you check https://github.com/pandas-dev/pandas/pull/33125/files#r400169295?

[jreback]

lgtm. I think @TomAugspurger has some remaining comments