Another Language Change for Debate

Fri Jan 11 22:10:21 EST 2002

Matt Mueller wrote:
> After using some staticmethod() (in python2.2), it popped out at
> me how support for staticmethod/classmethod could be added without
> new keywords:
> 
> class MyClass:
>     def self.foo(x):
>         #instance method
>     def bar(y):
>         #staticmethod
>     def class.baz(z):
>         #classmethod
> 
> (Of course, since this would break ALL old code, adding new
> keywords would be probably be a better idea. ;)

I much prefer a different approach.

    class MyClass:
        def staticmethod bar(y):
            blah()
            stuff += stuff
        def classmethod baz(z):
            return 1

This generalizes in an obvious way:
    def <name1> <name2>(<formals>): <body>
turns into
    def <name2>(<formals>): <body>
    <name2> = <name1>(<name2>)
which would make it possible, e.g., to define
a function called "memo" which memoizes its argument
and say (typical useless example follows):
    def memo fib(n):
        if n<2: return n
        return fib(n-2)+fib(n-1)
and get only O(n) calls to "fib" on doing fib(n).

Perhaps there's an obvious reason why this doesn't
work, but I can't see what it might be right now.
Would it be difficult for the parser?

-- 
Gareth McCaughan  Gareth.McCaughan at pobox.com
.sig under construc