> On Fri, Dec 18, 2009 at 8:07 PM, Robert Bradshaw
> <[email protected]> wrote:
>> On Dec 18, 2009, at 2:41 PM, mark florisson wrote:
>>
>>> Hello,
>>>
>>> Currently, bint Cython types get converted to int Python objects
>>> when accessed in Python code as attributes of an object, and I think
>>> it would be more consistent and logical to have it return actual
>>> bool objects (for instance if you return a bint from a function or
>>> method it gets converted to a bool).
>>
>> I agree, that's a bug.
>>
>>> So instead of storing a bint as an int, it could be stored as a
>>> char, and the T_BOOL macro name could be used in the PyMemberDef to
>>> have it automatically converted to True/False. I have included a
>>> patch that does this, but this obviously causes overflows for code
>>> that assigns values that don't fit in a char. I'm not sure if that's
>>> an issue, if it is, could we perhaps generate get/set methods that
>>> return a bool object when accessed from python code?
>>
>>
>> For consistency, at least as things are now, we should store it as an
>> int. However, we could generate custom get/set methods as we do with
>> attributes that are typed as builtin or extension types.
>>
>
> Or use custom get/set for everything... IIRC, we discussed about this
> in the past.
>

Heh -- yeah, I'm not writing the patch this time around :-)

Dag Sverre

_______________________________________________
Cython-dev mailing list
[email protected]
http://codespeak.net/mailman/listinfo/cython-dev

Reply via email to