__dict__ strangeness

[Georg Brandl]

Hi,

can someone please tell me that this is correct and why:
[color=blue][color=green][color=darkred]
>>> class C(object):[/color][/color][/color]
.... pass
....[color=blue][color=green][color=darkred]
>>> c = C()
>>> c.a = 1
>>> c.__dict__[/color][/color][/color]
{'a': 1}[color=blue][color=green][color=darkred]
>>> c.__dict__ = {}
>>> c.a[/color][/color][/color]
Traceback (most recent call last):
File "<stdin>", line 1, in ?
AttributeError: 'C' object has no attribute 'a'[color=blue][color=green][color=darkred]
>>>
>>> class D(object):[/color][/color][/color]
.... __dict__ = {}
....[color=blue][color=green][color=darkred]
>>> d = D()
>>> d.a = 1
>>> d.__dict__[/color][/color][/color]
{}[color=blue][color=green][color=darkred]
>>> d.__dict__ = {}
>>> d.a[/color][/color][/color]
1

Thanks,
Georg

[Alan Franzoni]

Re: __dict__ strangeness

Georg Brandl on comp.lang.pytho n said:
[color=blue][color=green][color=darkred]
>>>> d.__dict__[/color][/color]
> {}[/color]

Which Python version, on which system? I can see the properly inserted
attribute in __dict__, both with old style and new style classes.


--
Alan Franzoni <alan.franzoni. xyz@gmail.com>
-
Togli .xyz dalla mia email per contattarmi.
Rremove .xyz from my address in order to contact me.
-
GPG Key Fingerprint:
5C77 9DC3 BD5B 3A28 E7BC 921A 0255 42AA FE06 8F3E

[Georg Brandl]

Re: __dict__ strangeness

Alan Franzoni wrote:[color=blue]
> Georg Brandl on comp.lang.pytho n said:
>[color=green][color=darkred]
>>>>> d.__dict__[/color]
>> {}[/color]
>
> Which Python version, on which system? I can see the properly inserted
> attribute in __dict__, both with old style and new style classes.[/color]

It's 2.4.2, on Linux. The 2.5 SVN trunk has the same symptom.

Georg

[Alex Martelli]

Re: __dict__ strangeness

Georg Brandl <g.brandl-nospam@gmx.net> wrote:
[color=blue]
> can someone please tell me that this is correct and why:[/color]

IMHO, it is not correct: it is a Python bug (and it would be nice to fix
it in 2.5).
[color=blue][color=green][color=darkred]
> >>> class C(object):[/color][/color]
> ... pass
> ...[color=green][color=darkred]
> >>> c = C()
> >>> c.a = 1
> >>> c.__dict__[/color][/color]
> {'a': 1}[color=green][color=darkred]
> >>> c.__dict__ = {}
> >>> c.a[/color][/color]
> Traceback (most recent call last):
> File "<stdin>", line 1, in ?
> AttributeError: 'C' object has no attribute 'a'[/color]

So far so good, I think we agree;-).
[color=blue][color=green][color=darkred]
> >>> class D(object):[/color][/color]
> ... __dict__ = {}
> ...[color=green][color=darkred]
> >>> d = D()
> >>> d.a = 1
> >>> d.__dict__[/color][/color]
> {}[color=green][color=darkred]
> >>> d.__dict__ = {}
> >>> d.a[/color][/color]
> 1[/color]

Yep, that's the bug, fully reproducible in 2.3 and 2.4. FWIW, mucking
around with gc.getreferrers (with a more uniquely identifiable value for
d.a;-) shows a dictionary "somewhere" with keys 'a' and '__dict__'...


Alex

[Georg Brandl]

Re: __dict__ strangeness

[moving to python-dev]

Alex Martelli wrote:[color=blue]
> Georg Brandl <g.brandl-nospam@gmx.net> wrote:
>[color=green]
>> can someone please tell me that this is correct and why:[/color]
>
> IMHO, it is not correct: it is a Python bug (and it would be nice to fix
> it in 2.5).[/color]

Fine. Credits go to Michal Kwiatkowski for discovering that in bug #1448042
which I closed out of ignorance ;)
[color=blue][color=green][color=darkred]
>> >>> class C(object):[/color]
>> ... pass
>> ...[color=darkred]
>> >>> c = C()
>> >>> c.a = 1
>> >>> c.__dict__[/color]
>> {'a': 1}[color=darkred]
>> >>> c.__dict__ = {}
>> >>> c.a[/color]
>> Traceback (most recent call last):
>> File "<stdin>", line 1, in ?
>> AttributeError: 'C' object has no attribute 'a'[/color]
>
> So far so good, I think we agree;-).[/color]
Yes.
[color=blue][color=green][color=darkred]
>> >>> class D(object):[/color]
>> ... __dict__ = {}
>> ...[color=darkred]
>> >>> d = D()
>> >>> d.a = 1
>> >>> d.__dict__[/color]
>> {}[color=darkred]
>> >>> d.__dict__ = {}
>> >>> d.a[/color]
>> 1[/color]
>
> Yep, that's the bug, fully reproducible in 2.3 and 2.4. FWIW, mucking
> around with gc.getreferrers (with a more uniquely identifiable value for
> d.a;-) shows a dictionary "somewhere" with keys 'a' and '__dict__'...[/color]

Georg

[Alex Martelli]

Re: __dict__ strangeness

Georg Brandl <g.brandl-nospam@gmx.net> wrote:
[color=blue]
> [moving to python-dev]
>
> Alex Martelli wrote:[color=green]
> > Georg Brandl <g.brandl-nospam@gmx.net> wrote:
> >[color=darkred]
> >> can someone please tell me that this is correct and why:[/color]
> >
> > IMHO, it is not correct: it is a Python bug (and it would be nice to fix
> > it in 2.5).[/color]
>
> Fine. Credits go to Michal Kwiatkowski for discovering that in bug #1448042
> which I closed out of ignorance ;)[/color]

Yep, I remember Michal posting about the bug here, and I suggested that
he open it as an actual bug on SF;-)... [[I'm supposed to be finishing
up the 2nd edition of the Nutshell, so I can't really do much on Python
myself, these days...]]


Alex

[Ziga Seilnacht]

Re: __dict__ strangeness


Georg Brandl wrote:[color=blue]
> Hi,
>
> can someone please tell me that this is correct and why:
>[color=green][color=darkred]
> >>> class C(object):[/color][/color]
> ... pass
> ...[color=green][color=darkred]
> >>> c = C()
> >>> c.a = 1
> >>> c.__dict__[/color][/color]
> {'a': 1}[color=green][color=darkred]
> >>> c.__dict__ = {}
> >>> c.a[/color][/color]
> Traceback (most recent call last):
> File "<stdin>", line 1, in ?
> AttributeError: 'C' object has no attribute 'a'[color=green][color=darkred]
> >>>
> >>> class D(object):[/color][/color]
> ... __dict__ = {}
> ...[color=green][color=darkred]
> >>> d = D()
> >>> d.a = 1
> >>> d.__dict__[/color][/color]
> {}[color=green][color=darkred]
> >>> d.__dict__ = {}
> >>> d.a[/color][/color]
> 1
>
> Thanks,
> Georg[/color]

Here is another example that might help:
[color=blue][color=green][color=darkred]
>>> class E(object):[/color][/color][/color]
.... __dict__ = {'a': 1}
....[color=blue][color=green][color=darkred]
>>> e = E()
>>> e.__dict__[/color][/color][/color]
{'a': 1}[color=blue][color=green][color=darkred]
>>> E.__dict__[/color][/color][/color]
<dictproxy object at 0x00A81930>[color=blue][color=green][color=darkred]
>>> E.__dict__['a'][/color][/color][/color]
Traceback (most recent call last):
File "<stdin>", line 1, in ?
KeyError: 'a'[color=blue][color=green][color=darkred]
>>> E.__dict__['__dict__'][/color][/color][/color]
{'a': 1}

Ziga

[Alan Franzoni]

Re: __dict__ strangeness

Georg Brandl on comp.lang.pytho n said:
[color=blue]
> It's 2.4.2, on Linux. The 2.5 SVN trunk has the same symptom.[/color]

Yes, I confirm this. I think I had done something strange on my system
(probably misspelled something).

--
Alan Franzoni <alan.franzoni. xyz@gmail.com>
-
Togli .xyz dalla mia email per contattarmi.
Rremove .xyz from my address in order to contact me.
-
GPG Key Fingerprint:
5C77 9DC3 BD5B 3A28 E7BC 921A 0255 42AA FE06 8F3E