Re: [Swig-devel] Python 3.0 support branch merged

[Haoyu B. <div...@gm...>]

On Sun, Sep 14, 2008 at 2:09 AM, Richard Boulton
<ri...@le...> wrote:
> Haoyu Bai wrote:
>>>
>>> Please add to the documentation any limitations, eg overloaded methods
>>> don't
>>> get annotated. This particular limitation is quite a biggy. Any plans to
>>> fix?
>>>
>>
>> Okey, I documented this limitation yet. For fixing this, I can't
>> figure out how right now. All functions overloaded the same name will
>> share the same proxy function, so there's no way to distinguish them.
>> Also, there's no built-in support of overloading in Python.
>
> Perhaps we could supply partial annotations in such cases.  It's quite often
> for the type of some parameters to be common between the different
> overloaded forms.  So, for example, if we had the four functions:
>
> void foo();
> void foo(const char * name);
> void foo(const char * name, int number);
> void foo(const char * name, float number);
>
> A reasonable annotated wrapper might be:
>
> def foo(name : "const char *", number):
>
>
> Here, the first parameter is always of type "const char *" (or absent), so
> we can unambiguously annotate it.  The second parameter, when present, has
> two different types, so we can't annotate it.
>
> I'm not sure if such annotations will actually be particularly useful
> (though it would allow an editor which understood the annotations to do some
> useful hinting, I suppose), but this seems like the most plausible and
> consistent behaviour.
>
> --
> Richard
>

That's a better solution, I will have a try.

-- 
Haoyu Bai