BUG?: div method with timedelta

[TomAugspurger]

Not sure if this is a bug or not.

import datetime

In [24]: s = pd.Series(pd.to_datetime(['2015-01-01', '2015-02-02']))

In [25]: s
Out[25]:
0   2015-01-01
1   2015-02-02
dtype: datetime64[ns]

In [26]: s.diff() / datetime.timedelta(1)
Out[26]:
0   NaN
1    32
dtype: float64

In [27]: s.diff().div(datetime.timedelta(1))
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-27-1197779902ae> in <module>()
----> 1 s.diff().div(datetime.timedelta(1))

C:\Miniconda\lib\site-packages\pandas\core\ops.pyc in flex_wrapper(self, other,
level, fill_value, axis)
    716                                level=level, fill_value=fill_value)
    717         else:
--> 718             return self._constructor(op(self.values, other),
    719                                      self.index).__finalize__(self)
    720

TypeError: ufunc true_divide cannot use operands with types dtype('<m8[ns]') and
 dtype('O')

Works with our Timedelta though, as suggested by the error message:

In [28]: s.diff().div(pd.Timedelta(1, unit='D'))
Out[28]:
0   NaN
1    32
dtype: float64

It at least seems strange that / worked while .div didn't.

This is the released 0.15.2, and python 2.

[jreback]

you can mark this as a bug / enhancement
timedelta generally need wrapping internally in TimeDelta in order to work