Python 3.0 nonlocal statement

[Casey]

In PEP 3104 the nonlocal statement was proposed and accepted for
implementation in Python 3.0 for access to names in outer scopes. The
proposed syntax included an optional assignment or augmented
assignment to the outer name, such as:

nonlocal x += 1

This syntax doesn't appear to be supported in the 3.0 implementation.
My question is: was this intentional or was it missed in the initial
release? If it was intentional, is there any plan to support it in a
later 3.x release? I realize it is a very small convenience feature
but I have already come across a couple of cases where I use nested
functions where it does make the code seem a little cleaner.

Regards, Casey

 

[Matimus]

In PEP 3104 the nonlocal statement was proposed and accepted for
implementation in Python 3.0 for access to names in outer scopes.  The
proposed syntax included an optional assignment or augmented
assignment to the outer name, such as:

nonlocal x += 1

This syntax doesn't appear to be supported in the 3.0 implementation.
My question is: was this intentional or was it missed in the initial
release?  If it was intentional, is there any plan to support it in a
later 3.x release?  I realize it is a very small convenience feature
but I have already come across a couple of cases where I use nested
functions where it does make the code seem a little cleaner.

Regards, Casey

`nonlocal` should behave just like `global` does. It doesn't support
that syntax either. So, yes it was intentional. No, there probably is
no plan to support it in a later release.

Matt

 

[Casey]

`nonlocal` should behave just like `global` does. It doesn't support
that syntax either. So, yes it was intentional. No, there probably is
no plan to support it in a later release.

Matt

From my perspective, that's an unfortunate decision and I question the
rationale. First, nonlocal doesn't behave exactly like global (you
cannot declare a previously undefined variable nonlocal as you can
with global). Second, the PEP 3104 description explicitly mentions
that Guido favored adding this behavior to global; I would have
preferred that approach to not providing the extended assignment
support for nonlocal. Third, I believe that it adds some clarity to
the code. Seeing 'nonlocal x += 1' immediately tells me that I'm
incrementing a variable defined in an outer scope. Having 'nonlocal
x' on one line and 'x += 1' on another makes it a little less clear.

But I do appreciate the reply!

Casey

 

[Casey]

Matimus wrote in news:2a3d6700-85f0-4861-84c9-9f269791f044

Searching (AKA googling) for: nonlocal site:bugs.python.org
leads to:http://bugs.python.org/issue4199

Rob.
--http://www.victim-prime.dsl.pipex.com/

Doh. I looked at the PEP and the 3.0 BNF syntax and never thought to
search the online bugs.

Thanks, Rob!