AssertionError from Series.append(DataFrame) with 1.0.0rc0

[TomAugspurger]

Code Sample, a copy-pastable example if possible

The behavior of the following changed with 1.0.0rc0

>>> import pandas as pd
>>> df = pd.DataFrame({"A": [1,2], "B": [3, 4]})
>>> df.A.append(df)

On 0.25.3, we have

     0    A    B
0  1.0  NaN  NaN
1  2.0  NaN  NaN
0  NaN  1.0  3.0
1  NaN  2.0  4.0

On 1.0.0rc0, we have

---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
<ipython-input-9-fde662a5b7c4> in <module>
----> 1 df.A.append(df)

~/sandbox/pandas/pandas/core/series.py in append(self, to_append, ignore_index, verify_integrity)
   2541         return self._ensure_type(
   2542             concat(
-> 2543                 to_concat, ignore_index=ignore_index, verify_integrity=verify_integrity
   2544             )
   2545         )

~/sandbox/pandas/pandas/core/base.py in _ensure_type(self, obj)
     93         Used by type checkers.
     94         """
---> 95         assert isinstance(obj, type(self)), type(obj)
     96         return obj
     97

AssertionError: <class 'pandas.core.frame.DataFrame'>

Problem description

I don't think the behavior of 0.25.3 is necessarily correct. I would expect either a TypeError from passing a DataFrame to Series.append, or something like

In [17]: r
Out[17]:
   A    B
0  1  3.0
1  2  4.0
0  1  NaN
1  2  NaN

where the name of the Series is aligned with the DataFrame. We clearly document that the elements passed to append should be Series, so IMO we should raise.

[simonjayhawkins]

We clearly document that the elements passed to append should be Series, so IMO we should raise.

I agree. I think for this issue, we should restrict to improving the error message to something like TypeError: to_append should be a Series or list/tuple of Series, got <type> since the docs state:

to_appendSeries or list/tuple of Series
   Series to append with self.

and a note in the whatsnew that this now raises

an enhancement to accept a DataFrame could be raised as a separate issue and could be discussed there.