DataFrame.applymap(func) issue when func returns tuple

[lodagro]

In [82]: def func_tuple(x):
   ....:     return (x,x)
   ....:

In [83]: series = pandas.Series(['a', 'b', 'c', 'd'])

In [84]: df = pandas.DataFrame({'A': series})

In [85]: df
Out[85]:
   A
0  a
1  b
2  c
3  d

In [86]: df.applymap(func_tuple)
Out[86]:
   A
0  None
1  None
2  None
3  None

This works fine

In [87]: map(func_tuple, series)
Out[87]: [('a', 'a'), ('b', 'b'), ('c', 'c'), ('d', 'd')]

Let`s try with a list:

In [88]: def func_list(x):
   ....:     return [x,x]

In [89]: df.applymap(func_list)
Out[89]:
   A
0  ['a', 'a']
1  ['b', 'b']
2  ['c', 'c']
3  ['d', 'd']

Had a look at the pandas.DataFrame.applymap() code and this behavior is related to numpy.
I`m using numpy 1.5.1, i know that for other reason numpy 1.6.1 is preferred. Would more recent numpy version fix this?
I tried to install more recent version of numpy, but so far unsuccessful (various compilation errors).

In [103]: np.frompyfunc(func_tuple, 1, 1)(series)
Out[103]:
0    NaN
1    NaN
2    NaN
3    NaN

In [104]: np.frompyfunc(func_list, 1, 1)(series)
Out[104]:
0    ['a', 'a']
1    ['b', 'b']
2    ['c', 'c']
3    ['d', 'd']

[wesm]

This looks buggy enough to me...will see if I can sort out a fix here

[wesm]

Fixed in the above commit...turns out I'd written a frompyfunc replacement with type inference a few weeks ago that's pretty fast