Appending with enlargment doesn't work for .ix in empty dataframe

[stathius]

Similar as #5226
Enlargment doesn't work for .ix on empty dataframe.
.loc works fine.
Affected version 0.15.2

df = pd.DataFrame(columns = ['a', 'b'])
df.ix[1] =[1, 1]

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-320-58384ba7155e> in <module>()
      1 df = pd.DataFrame(columns = ['a', 'b'])
----> 2 df.ix[1] =[1, 1]

/usr/local/pr/anaconda/ipyNotebook/lib/python2.7/site-packages/pandas/core/indexing.pyc in __setitem__(self, key, value)
    115                 indexer = self._convert_tuple(key, is_setter=True)
    116             else:
--> 117                 indexer = self._convert_to_indexer(key, is_setter=True)
    118 
    119         self._setitem_with_indexer(indexer, value)

/usr/local/pr/anaconda/ipyNotebook/lib/python2.7/site-packages/pandas/core/indexing.pyc in _convert_to_indexer(self, obj, axis, is_setter)
   1010                 if (obj >= self.obj.shape[axis] and
   1011                         not isinstance(labels, MultiIndex)):
-> 1012                     raise ValueError("cannot set by positional indexing with "
   1013                                      "enlargement")
   1014 

ValueError: cannot set by positional indexing with enlargement

Using .loc first:

df = pd.DataFrame(columns = ['a', 'b'])
df.loc[0] = [1, 1]
df.ix[1] =[1, 1]

everything is ok.

[jreback]

this is by definition positional indexing and is not allowed with enlargement. You are seeing fallback on .ix. A frame w/o an index at construction time has the default index (which is integer).

In [6]: df = pd.DataFrame(columns = ['a', 'b'],index=[2,3])

In [7]: df.ix[10] =[1, 1]

In [8]: df
Out[8]: 
     a   b
2  NaN NaN
3  NaN NaN
10   1   1