Series name information is lost when doing DataFrame.apply()

[lbeltrame]

On both axes, the generated Series on which the function has to work lack the "name" attribute (either the row name or th column name).

[lbeltrame]

I tested this on a non-mixed type DataFrame, I'm unsure in the case of mixed-types.

[lodagro]

see als mailing list discussion

[wesm]

it's a done deal

[lbeltrame]

Unfortunately, there are still some issues. When doing axis=1, the name is preserved only in the first iteration (or not preserved at all in some cases): I checked by using a function that only prints out the Series out and returns the same series and after the first iteration, name information is lost.

The behavior is more consistent with axis=0.

In [17]: df = pandas.DataFrame(index=["A", "B", "C"],
                               data={"Foo": [1,2,3],  "Bar": [5,6,7], 
                                     "Baz":[8, 9, 10]})

In [18]: def foo(x):
    print x.name
    return x
   ....: 

In [19]: df.apply(foo, axis=1)
None
None
None
Out[19]: 
   Bar  Baz  Foo
A    5    8    1
B    6    9    2
C    7   10    3

In [20]: df.apply(foo, axis=0)
Bar
Baz
Foo
Out[20]: 
   Bar  Baz  Foo
A    5    8    1
B    6    9    2
C    7   10    3

[wesm]

You're right-- I only fixed it in the reduction code. I'll take another look

[wesm]

OK, fixed this and added tests. looks OK now