[Python-Dev] Re: closure semantics

[Delaney, Timothy C (Timothy)]

> From: John Williams [mailto:jrw at pobox.com]
> 
> I can think of two reasonable possibilities--either it refers to the 
> innermost possible variable, or the compiler rejects this 
> case outright. 
>   Either way the problem is easy to solve by renaming one of 
> the variables.

Going on the principle of least surprise, I have to say that I think explicitly naming the scope in which a variable is to be used is the best approach.

My concern with the other proposal is that introducing code between scopes could silently change the semantics of a piece of code.

I'll use the 'outer' proposal since it's the shortest and least confusing to me ...

    def func1()

        x = 1

        def func2()

            def func3()
                outer x
                x += 2

            return func3

        return func2()

    print func1()

should print:

    3

Now, if we change it to:

    def func1()

        x = 1

        def func2()

            x = 2

            def func3()
                outer x
                x += 2

            return func3

        return func2()

    print func1()

it would now print:

    4

OTOH, specifying the scope prevents this type of error:

    def func1()

        x = 1

        def func2()

            def func3()
                global x in func1
                x += 2

            return func3

        return func2()

    print func1()

and

    def func1()

        x = 1

        def func2()

            x = 2

            def func3()
                global x in func1
                x += 2

            return func3

        return func2()

    print func1()

should both print

    3

'global x in func1' is also a *lot* easier to explain.

I think these two points should weigh heavily in any decision. I think the need to rename the target scope is of lesser importance.

Tim Delaney