BUG: index sorting with strings & timestamps

[dniku]

Pandas 0.16.2, Python 3.4.

test = pd.DataFrame({
    pd.Timestamp('2012-01-01 00:00:00'): ['a', 'b'],
    pd.Timestamp('2012-01-02 00:00:00'): ['c', 'd'],
    'name': ['e', 'e'],
    'aaaa': ['f', 'g']
    })
print(test)
print(test.groupby('name').first())

Fails with:

  File "/usr/lib64/python3.4/site-packages/pandas/core/groupby.py", line 106, in f
    self._set_selection_from_grouper()
  File "/usr/lib64/python3.4/site-packages/pandas/core/groupby.py", line 489, in _set_selection_from_grouper
    self._group_selection = ax.difference(Index(groupers)).tolist()
  File "/usr/lib64/python3.4/site-packages/pandas/core/index.py", line 1506, in difference
    theDiff = sorted(set(self) - set(other))
  File "pandas/tslib.pyx", line 836, in pandas.tslib._Timestamp.__richcmp__ (pandas/tslib.c:15612)
TypeError: Cannot compare type 'Timestamp' with type 'str'

[jreback]

So the issue is below, its an Index set operation.

In [17]: Index([u'aaaa', Timestamp('2012-01-01 00:00:00'), Timestamp('2012-01-02 00:00:00'), u'name'], dtype='object').sort_values()
TypeError: Cannot compare type 'Timestamp' with type 'unicode'

You realize that this is a completely useless index. You should really not do this, mixing objects and strings.

I'll mark it as a bug, but not a high priority.

[mroeschke]

The original issue looks to work on master. Could use a test.

In [31]: test.groupby('name').first()
Out[31]:
     2012-01-01 00:00:00 2012-01-02 00:00:00 aaaa
name
e                      a                   c    f

In [32]: pd.__version__
Out[32]: '0.26.0.dev0+652.g30362ed82'