At a loss on python scoping.

Shiyao Ma i at
Tue Mar 26 12:14:22 CET 2013

After read Dave's answer, I think I confused LEGB with attribute lookup.
So, a.r has nothing to do with LEGB.

On Tue, Mar 26, 2013 at 7:03 PM, Shiyao Ma <i at> wrote:

> Thx, really a nice and detailed explanation.
> On Tue, Mar 26, 2013 at 6:07 PM, Dave Angel <davea at> wrote:
>> On 03/26/2013 02:17 AM, Shiyao Ma wrote:
>>> Hi,
>>> suppose I have a file like this:
>>> class A:
>>>      r = 5
>>>      def func(self, s):
>>>          self.s = s
>>> a = A()
>>> print(a.r)    # this should print 5, but where does py store the name of
>>> r
>>> a.func(3)
>>> print(a.s)    # this should print 3, also where does py store this name.
>>> what's the underlying difference between the above example?
>> I don't think this is a scoping question at all.  These references are
>> fully qualified, so scoping doesn't enter in.
>> The class A has a dictionary containing the names of r and func.  These
>> are class attributes.  Each instance has a dictionary which will contain
>> the name s AFTER the A.func() is called.  Ideally such an attribute will be
>> assigned in the __init__() method, in which case every instance will have s
>> in its dictionary.
>> When you use a.qqq  the attribute qqq is searched for in the instance
>> dictionary and, if not found, in the class dictionary.  If still not found,
>> in the parent classes' dictionary(s).
>> You can use dir(A) and dir(a) to look at these dictionaries, but it shows
>> you the combination of them, so it's not as clear.  In other words, dir(a)
>> shows you both dictionaries, merged.  (Seems to me dir also sometimes
>> censors some of the names, but that's a vague memory. It's never left out
>> anything I cared about, so maybe it's things like single-underscore names,
>> or maybe just a poor memory.)
>> --
>> DaveA
>> --
> --
> My gpg pubring is available via: gpg --keyserver 307CF736
> More on:

My gpg pubring is available via: gpg --keyserver 307CF736

More on:
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Python-list mailing list