Mark Dickinson wrote: > Here's a curiosity: after > > def my_hex(x): > return hex(x) > > one might expect hex and my_hex to be interchangeable > in most situations. But (with both Python 2.x and 3.x) > I get: > >>>> def my_hex(x): return hex(x) > ... >>>> class T(object): f = hex > ... >>>> class T2(object): f = my_hex > ... >>>> T().f(12345) > '0x3039' >>>> T2().f(12345) > Traceback (most recent call last): > File "<stdin>", line 1, in <module> > TypeError: my_hex() takes exactly 1 argument (2 given) > [36412 refs] > > Anyone know what the precise rules that lead to this > behaviour are, or where they're documented?
To work properly as a method a callable needs a __get__() method implementing the binding mechanism. Functions written in C generally don't have that. >>> class A(object): ... def __call__(self, *args): ... print "called with", args ... >>> class B(object): ... a = A() ... >>> B().a() called with () Now let's turn A into a descriptor: >>> def __get__(self, inst, cls=None): ... print inst, cls ... if inst is not None: ... def bound(*args): ... self(inst, *args) ... return bound ... return self ... >>> A.__get__ = __get__ >>> B().a() <__main__.B object at 0x2ae49971b890> <class '__main__.B'> called with (<__main__.B object at 0x2ae49971b890>,) I don't know if there is something official, I google for http://users.rcn.com/python/download/Descriptor.htm or "descrintro" every time I need a refresher. Peter -- http://mail.python.org/mailman/listinfo/python-list
