BUG: clip_lower/-upper don't accept a sequence starting from 0.12.0

[immerrr]

Here's what I mean:

In [4]: pd.Series(np.arange(7))
Out[4]: 
0    0
1    1
2    2
3    3
4    4
5    5
6    6
dtype: int64

# Let's put some limit on weekends
In [5]: _4.clip_upper([3,3,3,3,3, 4,4])

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-5-3268c7bee292> in <module>()
----> 1 _4.clip_upper([3,3,3,3,3, 4,4])
/home/dshpektorov/sources/pandas/pandas/core/generic.pyc in clip_upper(self, threshold)
   2636         clipped : same type as input
   2637         """
-> 2638         if isnull(threshold):
   2639             raise ValueError("Cannot use an NA value as a clip threshold")
   2640 

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

[jreback]

thats just a bug,
should be isnull().any()

but can easily do:

In [1]: s = pd.Series(np.arange(7))

In [3]: y = Series([3,3,3,3,3,4,4])

In [5]: s.where(s<y,y)
Out[5]: 
0    0
1    1
2    2
3    3
4    3
5    4
6    4
dtype: int64

[dsm054]

What's the right idiom for allowing both scalars and vectors to pass here? isnull(3).any() won't work because False doesn't have an any method and calling as(any)array might make a copy we don't want (not actually sure about that). Could do use hasattr, or try/except, or even do an is check on False/True.

Basically, is there already a scalar/vector idiom for this?

[immerrr]

I'd also like to note that pandas's df.any() aggregates across only one axis, while ndarray.any() aggregates the whole array regardless of dimensionality and shapes.