BUG: "ValueError: The truth value ... is ambiguous" when using .loc setitem

[kernc]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

>>> import pandas as pd
>>> df = pd.DataFrame({'a':[1]})
>>> s = pd.Series({'label': df})  # This works
>>> s.loc['new_label'] = df       # This broken

Issue Description

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-5-58528b0c1243> in <module>
----> 1 s.loc['new_label'] = df

.../site-packages/pandas/core/indexing.py in __setitem__(self, key, value)
    817 
    818         iloc = self if self.name == "iloc" else self.obj.iloc
--> 819         iloc._setitem_with_indexer(indexer, value, self.name)
    820 
    821     def _validate_key(self, key, axis: int):

.../site-packages/pandas/core/indexing.py in _setitem_with_indexer(self, indexer, value, name)
   1784 
   1785             if missing:
-> 1786                 self._setitem_with_indexer_missing(indexer, value)
   1787                 return
   1788 

.../site-packages/pandas/core/indexing.py in _setitem_with_indexer_missing(self, indexer, value)
   2114                 value = na_value_for_dtype(self.obj.dtype, compat=False)
   2115                 new_dtype = maybe_promote(self.obj.dtype, value)[0]
-> 2116             elif isna(value):
   2117                 new_dtype = None
   2118             elif not self.obj.empty and not is_object_dtype(self.obj.dtype):

.../site-packages/pandas/core/generic.py in __nonzero__(self)
   1524     @final
   1525     def __nonzero__(self) -> NoReturn:
-> 1526         raise ValueError(
   1527             f"The truth value of a {type(self).__name__} is ambiguous. "
   1528             "Use a.empty, a.bool(), a.item(), a.any() or a.all()."

ValueError: The truth value of a DataFrame is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().

The offending line isna(value) was added in bd9a6f0#diff-07282871a235456baecd3a55a43339b0da374ca270228b5530f5449aabdca2e1R2099 by @phofl. Would love to see milestone:1.5 on this regression. 😅

Thanks!

Expected Behavior

No crash.

Installed Versions

pandas-1.5.0rc0

[phofl]

Hi, thanks for your report and especially for already pointing to the correct commit. This helps a lot!

I'll look into this. That said, support for nested data is not great, we don't have many test cases covering this.

Just to avoid misunderstandings:

label           a
0  1
new_label       a
0  1
dtype: object

this would be your expected output?

[kernc]

this would be your expected output?

Simply a dtype=object series with two str-indexed objects, yes. I don't think pandas should ever be looking too deeply into or choosing to flatten nested objects by default.

Thanks!