Unexpected behavior on Series.irow (Series.iget) with non-unique index

[manuteleco]

Hi,

Series.irow(i) (and also Series.iget(i)), with i being an integer, returns more than one value on Series with non-unique indexes if the index at location i has duplicates. This is not the behavior I expected and it is also different from the behavior shown by DataFrame.irow(i), since it only returns the row located at position i regardless of any existing index duplicates. Also the documentation for Series.irow/Series.iget specifies that an "int" parameter should return a "scalar" value, so I guess this might be a bug.

Here is some sample code I've run using pandas '0.8.2.dev-f5a74d4':

In [52]: s = Series([0, 1, 2], index=[0, 1, 0])

In [53]: s.irow(0)  # here I was expecting 0
Out[53]: 
0    0
0    2

In [54]: s.irow(1)
Out[54]: 1

In [55]: s.irow(2)  # here I was expecting 2
Out[55]: 
0    0
0    2

and its DataFrame counterpart:

In [56]: df = DataFrame([[0], [1], [2]], index=[0, 1, 0])

In [57]: df.irow(0)
Out[57]: 
0    0
Name: 0

In [58]: df.irow(1)
Out[58]: 
0    1
Name: 1

In [59]: df.irow(2)
Out[59]: 
0    2
Name: 0

Thanks and regards.

[wesm]

Yes, it's a bug. Slated for 0.8.2