inplace operation for Series causes reappearance popped column

[s-wakaba]

Hi, Thanks for all efforts for the great tool!

I've found a strange behavior like this...

In [1]: a = pd.DataFrame([[1,2,3],[4,5,6]], columns=['A','B','C'], index=['X','Y'])

In [2]: a
Out[2]: 
   A  B  C
X  1  2  3
Y  4  5  6

In [3]: b = a.pop('B')

In [4]: a
Out[4]: 
   A  C
X  1  3
Y  4  6

In [5]: b
Out[5]: 
X    2
Y    5
Name: B, dtype: int64

In [6]: b += 1

In [7]: a # Why does a popped column "B" appear again?
Out[7]: 
   A  C  B
X  1  3  3
Y  4  6  6

In [8]: b
Out[8]: 
X    3
Y    6
Name: B, dtype: int64

Is it a bug of pandas?
This behavior can be avoided by using b = b+1 instead of implace operations.

[jreback]

I think the cacher is somehow getting updated here.