return type of predicate in select is too strict?

[ghost]

is there a reason to be so strict in the predicate return type?
notably python3 is much less cavalier about coercing things to bool.

In [75]: from pandas.util.testing import makeCustomDataframe as mkdf
    ...: mkdf(4,2,r_idx_nlevels=2).select(lambda x: bool(re.search("g1",x[0])))
Out[75]: 
C0              C_l0_g0 C_l0_g1
R0      R1                     
R_l0_g1 R_l1_g1    R1C0    R1C1

In [76]: from pandas.util.testing import makeCustomDataframe as mkdf
    ...: mkdf(4,2,r_idx_nlevels=2).select(lambda x: re.search("g1",x[0]))
    ...: 
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-76-c172e2763471> in <module>()
      1 from pandas.util.testing import makeCustomDataframe as mkdf
----> 2 mkdf(4,2,r_idx_nlevels=2).select(lambda x: re.search("g1",x[0]))
      3 

/home/user1/src/pandas/pandas/core/generic.pyc in select(self, crit, axis)
    316 
    317         if len(axis) > 0:
--> 318             new_axis = axis[np.asarray([crit(label) for label in axis])]
    319         else:
    320             new_axis = axis

/home/user1/src/pandas/pandas/core/index.pyc in __getitem__(self, key)
   1727             return tuple(retval)
   1728         else:
-> 1729             if com._is_bool_indexer(key):
   1730                 key = np.asarray(key)
   1731                 sortorder = self.sortorder

/home/user1/src/pandas/pandas/core/common.pyc in _is_bool_indexer(key)
    651         if not lib.is_bool_array(key):
    652             if isnull(key).any():
--> 653                 raise ValueError('cannot index with vector containing '
    654                                  'NA / NaN values')
    655             return False

ValueError: cannot index with vector containing NA / NaN values