type(None)()

On 8/08/12 04:14:01, Steven D'Aprano wrote:
> NoneType raises an error if you try to create a second instance. bool
> just returns one of the two singletons (doubletons?) again.
>
> py> type(None)()
> Traceback (most recent call last):
> File "<stdin>", line 1, in <module>
> TypeError: cannot create 'NoneType' instances

Why is that?

Why doesn't it just return an existing instance of the type,
like bool, int, str and other built-in non-mutable types do?

> py> type(False)() is False
> True


Just wondering,

-- HansM

On 2012-08-16 14:47, Hans Mulder wrote:
> On 8/08/12 04:14:01, Steven D'Aprano wrote:
>> NoneType raises an error if you try to create a second instance. bool
>> just returns one of the two singletons (doubletons?) again.
>>
>> py> type(None)()
>> Traceback (most recent call last):
>> File "<stdin>", line 1, in <module>
>> TypeError: cannot create 'NoneType' instances
> Why is that?
Because None is a singleton. It is the only instance of its class. This
is very useful because it allows you to write conditions like this:

if obj is None:
do_something()

On Thu, Aug 16, 2012 at 10:47 PM, Hans Mulder <hansmu@xs4all.nl> wrote:
> Why doesn't it just return an existing instance of the type,
> like bool, int, str and other built-in non-mutable types do?
>
>> py> type(False)() is False
>> True

With int and str, it's only an optimization, and not guaranteed to happen.

>>> a=int("1234")
>>> a is int("1234")
False

>>> a=str(1234)
>>> a is str(1234)
False

But with bool, it's required, as a means of "casting to boolean". With
True/False/None, it's normal to compare them with is:

>>> a=bool("1")
>>> a is bool("2")
True

So bool() has to return one of those two actual objects, and not an equivalent.

(Note: All examples done in CPython 3.2's IDLE on Windows. Other
environments, Pythons, versions, etc, may affect exactly what these
show.)

ChrisA

On Thu, Aug 16, 2012 at 6:47 AM, Hans Mulder <hansmu@xs4all.nl> wrote:
> On 8/08/12 04:14:01, Steven D'Aprano wrote:
>> NoneType raises an error if you try to create a second instance. bool
>> just returns one of the two singletons (doubletons?) again.
>>
>> py> type(None)()
>> Traceback (most recent call last):
>> File "<stdin>", line 1, in <module>
>> TypeError: cannot create 'NoneType' instances
>
> Why is that?
>
> Why doesn't it just return an existing instance of the type,
> like bool, int, str and other built-in non-mutable types do?

Because unlike those other types there is no use case for that. It's
simpler to raise an error.

On Thu, 16 Aug 2012 14:47:47 +0200, Hans Mulder wrote:

> On 8/08/12 04:14:01, Steven D'Aprano wrote:
>> NoneType raises an error if you try to create a second instance. bool
>> just returns one of the two singletons (doubletons?) again.
>>
>> py> type(None)()
>> Traceback (most recent call last):
>> File "<stdin>", line 1, in <module>
>> TypeError: cannot create 'NoneType' instances
>
> Why is that?
>
> Why doesn't it just return an existing instance of the type, like bool,
> int, str and other built-in non-mutable types do?

bool must return an instance, because it is designed to cast objects to a
boolean. Since (by design) True and False are singletons (doubletons?),
bool(x) will always return a pre-existing instance.

Other built-in immutable types do not promise to do that. For example:

py> a = float(42)
py> b = float(42)
py> a is b
False

Sometimes int and str will cache their instances, but this is an
implementation detail subject to change without notice from version to
version.

None, NotImplemented and Ellipsis are singletons, but unlikely bool,
there is no common use-case for having their types return the singleton
instance. The standard design pattern for singletons is to raise an
exception if you try to create an instance, so they do. However, this
behaviour really only makes sense for singletons that hold state. (If
they hold state, you might be tempted to change that state, not realising
that you are changing a singleton and not a second instance.)

In my opinion, this is a PITA for None and better behaviour would be to
return the pre-existing NoneType instance, but I didn't design the
language.


--
Steven

On 8/16/12 2:56 PM, Ian Kelly wrote:
> On Thu, Aug 16, 2012 at 6:47 AM, Hans Mulder <hansmu@xs4all.nl> wrote:
>> On 8/08/12 04:14:01, Steven D'Aprano wrote:
>>> NoneType raises an error if you try to create a second instance. bool
>>> just returns one of the two singletons (doubletons?) again.
>>>
>>> py> type(None)()
>>> Traceback (most recent call last):
>>> File "<stdin>", line 1, in <module>
>>> TypeError: cannot create 'NoneType' instances
>>
>> Why is that?
>>
>> Why doesn't it just return an existing instance of the type,
>> like bool, int, str and other built-in non-mutable types do?
>
> Because unlike those other types there is no use case for that. It's
> simpler to raise an error.

What are the use cases for the empty-argument versions of bool(), int(),
float(), and str()?

--
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
that is made terrible by our own mad attempt to interpret it as though it had
an underlying truth."
-- Umberto Eco

On 16/08/2012 15:56, Robert Kern wrote:
> On 8/16/12 2:56 PM, Ian Kelly wrote:
>> On Thu, Aug 16, 2012 at 6:47 AM, Hans Mulder <hansmu@xs4all.nl> wrote:
>>> On 8/08/12 04:14:01, Steven D'Aprano wrote:
>>>> NoneType raises an error if you try to create a second instance. bool
>>>> just returns one of the two singletons (doubletons?) again.
>>>>
>>>> py> type(None)()
>>>> Traceback (most recent call last):
>>>> File "<stdin>", line 1, in <module>
>>>> TypeError: cannot create 'NoneType' instances
>>>
>>> Why is that?
>>>
>>> Why doesn't it just return an existing instance of the type,
>>> like bool, int, str and other built-in non-mutable types do?
>>
>> Because unlike those other types there is no use case for that. It's
>> simpler to raise an error.
>
> What are the use cases for the empty-argument versions of bool(), int(),
> float(), and str()?
>
They can be used with defaultdict. For example:

counts = defaultdict(int)
for i in items:
counts[i] += 1

Of course, an alternative would be:

counts = defaultdict(lambda: 0)