Statement local namespaces summary (was Re: python3: 'where' keyword)

[Andrey Tatarinov]

Nick Coghlan wrote:
>> Semantics
>> ---------
>> The code::
>>
>> <statement> with:
>>    <suite>
>>
>> translates to::
>>
>> def unique_name():
>>     <suite>
>>     <statement>
>> unique_name()
>>
> 
> Bleh. Not only was my proposed grammar change wrong, my suggested 
> semantics are wrong, too.
> 
> Raise your hand if you can see the problem with applying the above 
> semantics to the property descriptor example.
> 
> So I think the semantics will need to be more along the lines of 
> "pollute the namespace but mangle the names so they're unique, and the 
> programmer can *act* like the names are statement local".
> 
> This will be much nicer in terms of run-time performance, but getting 
> the locals() builtin to behave sensibly may be a challenge.

afair you told yourself that

var = <statement> where:
     <suite>

translates to:

def unique_name():
     <suite>
     return <statement>
var = unique_name()

in this case class gets unique_name() function? is it that bad?

anyway I'd prefer to change semantics deeper. adding new statement-only 
scope and adding our suite-definitions there.