problem with variable scoping

[Tim Peters]

[Michael Dingler]
> I'd certainly like to hear them. I'd like to know how other languages
> handle this. Doesn't Perl do it the other way round, i.e. a _local_
> keyword?

Perl started life with dynamic scoping and grafted on sensible semantics
later <0.5 wink -- but "my" rather than "local" is almost always want you
want to use in Perl5>).

Python's local rules will make instant sense to the C part of your brain if
you consider that Python acts exactly like C does, except that there's no
local *declaration* kicking you in the face.

> last_line_empty = 0
> .
> .
> .
> def process(line):
>         ...
>         if last_line_empty:
>                 pass
>         else:
>                 print "<P>"
>         last_line_empty = 1

C:

int last_line_empty = 0;

void process(const char* line)
{
    int last_line_empty;
    if (last_line_empty)
        ;
    else
        printf("<P>\n");
    last_line_empty = 1;
}

In C too, the local last_line_empty shadows the global one, and the code is
nonsense as a result.  Because it has no type declarations, Python has to
use *some* rule to decide which names are local (and happens to use a rule
based on whether a name is bound in the function body), but having made that
decision acts the same way as C.  Well, not quite:  Python yells at you if
you reference an uninitialized local, while C entertains you in creative
ways <wink>.

it's-a-feature-and-a-dessert-topping-ly y'rs  - tim