Doesn't Python's super() built-in violate DRY?

Question

There is obviously a reason for this but I am not experienced enough to recognise it.

This is the example given in the Python docs:

class C(B):
    def method(self, arg):
        super(C, self).method(arg) # Why do we have to mention 'C' again?

Is there not a way of 'knowing' the type from within the instance method that doesn't rely on duplicating the name of the class 'C'? Seems like the example could lead to a situation where the class name is updated but the type parameter in the super function call becomes stale?

That may be one reason for allowing `super().method(arg)` in Py3. — glglgl, Oct 25 '13 at 21:42
In Python 3 you can just say `super()`. It requires compile-time analysis that simply doesn't exist in Python 2. — Tim Peters, Oct 25 '13 at 21:43
You don't have to do this in Python 3 (where the type system has been overhauled), so I suspect it's an artifact of how classes were wedged into Python 2. — Codie CodeMonkey, Oct 25 '13 at 21:44
@martineau, it doesn't matter whether `C` has multiple bases. Note that *none* of the bases are passed to `super()`. — Tim Peters, Oct 25 '13 at 21:44
This is fixed in python 3, see http://stackoverflow.com/q/10482953/2647279 and http://stackoverflow.com/questions/2771904/usage-of-python-3-super — Bas Swinckels, Oct 25 '13 at 21:44
I love Python. This is one of the only mildly unintuitive things I've ever encountered in the language.. and it's fixed already ;-) — HorseloverFat, Oct 25 '13 at 21:50

score 2 · Answer 1 · answered Mar 12 '15 at 17:18

On the one hand, DRY is more about large sections of code being duplicated; on the other hand, yes, the class name in super() calls can become stale (it's happened to me).

In Python 3, super() has been revamped to not require the class name, so long as the function is defined inside the class -- so not a problem unless you are monkey-patching.

score 1 · Answer 2 · answered Mar 12 '15 at 16:59

The same way self is called manually in python super is a simple function, I mean, python functions doesn't know the execution context. This behaviour is practical for monkey patching for example.

For example, I took your C class from pypi and you update it every month as it's a useful class. I want to modify it's method behaviour so it'll update the comportment of other of my requirements using your C class:

from utils import do_magic
from a import C

def method(self, arg):
    super(C, self).method(arg)
    do_magic(self)

C.method = method

Doesn't Python's super() built-in violate DRY?

2 Answers2

Linked