Steven D'Aprano wrote:
> So, based on this evidence, staticmethod() inside a class definition
> converts instance methods to functions. Outside a class definition,
> staticmethod() does one of two things: it either converts an instance
> method to a static method, or if the output is assigned to a class
> attribute, it leaves it as an instance method.
This is exactly why I'm concerned about augmenting staticmethod's
behavior. When people run into this, it should be the opportunity to
explain to them how descriptors work. Descriptors are hugely important
in the new object system, and even if we hide them by giving
staticmethod a __call__, we'll still run into problems when people try
to do:
class C(object):
@classmethod
def foo(cls):
print cls
bar = foo(None)
Then, not only do we have to explain how descriptors work, but we also
have to explain why staticmethod has a __call__, and classmethod doesn't.
(For anyone else out there reading who doesn't already know this, Steven
D'Aprano's comments are easily explained by noting that the __get__
method of staticmethod objects returns functions, and classes always
call the __get__ methods of descriptors when those descriptors are class
attributes:
>>> class C(object):
... @staticmethod
... def foo():
... pass
... print foo
...
<staticmethod object at 0x00E73950>
>>> print C.foo
<function foo at 0x00E80330>
>>> @staticmethod
... def foo():
... pass
...
>>> print foo
<staticmethod object at 0x00E73990>
>>> print foo.__get__(C, None)
<function foo at 0x00E80130>
Yes, you have to explain descriptors, but at the point that you start
trying to do funny things with staticmethods and classmethods, I think
you need to start learning about them anyway.)
All that said, you should probably just submit a patch and see what
happens. I'll make a brief comment on it along the above lines, but
since I'm not a committer, it's not really worth your time to try to
convince me. ;)
STeVe
--
http://mail.python.org/mailman/listinfo/python-list
