Path: csiph.com!usenet.pasdenom.info!aioe.org!news.stack.nl!newsfeed.xs4all.nl!newsfeed2.news.xs4all.nl!xs4all!newsgate.cistron.nl!newsgate.news.xs4all.nl!post.news.xs4all.nl!not-for-mail
To: python-list@python.org
From: Terry Reedy <tjreedy@udel.edu>
Subject: Re: Local variable in a closure
Date: Sun, 18 Aug 2013 16:42:52 -0400
References: <107941d9-a981-4dd6-8460-336afc16f025@googlegroups.com> <kuq8hk$1hf$1@ger.gmane.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:17.0) Gecko/20130801 Thunderbird/17.0.8
In-Reply-To: <kuq8hk$1hf$1@ger.gmane.org>
Precedence: list
Newsgroups: comp.lang.python
Message-ID: <mailman.3.1376858584.19984.python-list@python.org>
Lines: 66
NNTP-Posting-Host: 2001:888:2000:d::a6
Xref: csiph.com comp.lang.python:52662

On 8/18/2013 6:44 AM, Dave Angel wrote:
> w.w.milner@googlemail.com wrote:
>
>> Is f local or not?
>> http://pastebin.com/AKDJrbDs
>
> Please have a little respect, and include the source in your message.
> You managed quite nicely to keep it small, but you put it in an obscure
> place that some people won't be able to reach, and that might not
> survive for the archives.
>
> def multiplier(f):
>      def times(n):
>          # is f local?
>          nonlocal f
>          f=f+1
>          # if not, why is it here?
>          print("Locals: ",locals())

Because nonlocal names are not in the global dict and the devs wanted 
globals() + locals() to report all accessible names, rather than add 
nonlocals() or leave them invisible.

>          return n*f
>      return times
>
> times2 = multiplier(2)
> print(times2(4)) # 3X4=12
> print(times2(4)) # 4X4=16
>
> Inside function times, the variable 'f' is a free variable, not a local.
>   You can prove that to yourself by adding a dis.dis(times) just before
> the "return times" statement.  Here's how it begins:
>
>    7           0 LOAD_DEREF               0 (f)
>                3 LOAD_CONST               1 (1)
>                6 BINARY_ADD
>                7 STORE_DEREF              0 (f)
>
> In the dis.dis listing, the LOAD_DEREF and STORE_DEREF opcodes are
> referring to free variables, the LOAD_FAST is referring to a local, and
> the LOAD_GLOBAL is referring to a global.
>
> The locals() function is just over-simplifying.  it's only a convenience
> function, not what I would consider part of the language,

I think this is a good way to look at it.

> and it wasn't apparently deemed necessary to have a separate function
 > for debugging free varaibles.

One should think of 'locals' as meaning 'non_globals', which was exactly 
true when there were no non-global, non-local names. When closures were 
first added, such names were only readable. There was a long debate over 
what term to use for the keyword that would allow rebinding the names in 
outer functions. 'Nonlocal' is, at best, the least bad of the options 
considered.

In standard usage, the terms 'free' and 'bound' are context dependent.
https://en.wikipedia.org/wiki/Free_variable
Within a function or class body, all global variables are non-local and 
free, just like 'nonlocals'.

-- 
Terry Jan Reedy