How to avoid boilerplate when usign super(…) in Python 2.6+?-Collection of common programming errors

tl;dr

As the OP said “Is it possible in Python 2.6+ without overt hacks?”, the answer is: No

Long version

You can make a simple decorator that will call the next parent with this method. The problem is that you will not have control on the arguments you want to pass.

Edit: This will not work for a subclass already using autosuper because it’ll chose the wrong class and make an infinite loop.

def autosuper(fn):
    def f(self, *args, **kw):
        cl = super(type(self), self)
        getattr(cl, fn.__name__)(*args, **kw)
        return fn(self, *args, **kw)
    return f

How could this be done? Python 3.x do have a super function that takes no arguments!

Unfortunally, the Python 3.x’s super is a class and at the same time a keyword, because just the presence of its name will change the current environment to unveil a variable named __class__ that is the right class you need to use!

If you check the frame inside a function declared in a class, there’s no __class__ variable and the co_freevars attribute of the frame’s f_code attribute is empty. When you write the name super (do not need to call it), the __class__ string will appear in co_freevars meaning it comes from another closure. Also, if you try to access the __class__ variable without using super, it’ll use the LOAD_DEFER bytecode for this same reason instead of LOAD_GLOBAL like would be normal to every undefined name.

This is so crazy that you cannot just do hyper = super and call this new hyper variable without arguments (that is exactly the same object as super).

As I cannot compete with this much of black magic inside the Python Interpreter, and because the autosuper decorator is not declared inside a class (so it can never access the __class__ variable even if that was possible in Python 2.x), I will not try to write a new decorator and will leave this answer here as a warn for other people who want to do that.

It is probably possible to make some hackeries to find the right class to use, but I will not dig that far. Things to consider:

• When the decorator is applied, the class do not exist yet, so this should be done when the decorated function is called.
• The function being decorated is not yet an unbound method (that were removed anyway from Py3k) so you cannot check the im_class attribute.
• The frame does not seem to hold any information of the class used to make this call (unless of course the __class__ variable do exist and it is possible to get a reference to it)
• This answer provided by OP is also quite broken because it makes a lot of bad assumptions and has problems with decorated functions.