Re: [luabind] [SPAM] [Bayesian][testmode] Re: MetroWerks Template/Function Overloading: How do you kick this...

[Stephan K. <ste...@ya...>]

The is_user_defined overload problem is the same as the get_const_holder
overload problem, afaics.

We got around it by including our custom overloads *before* including any
of the Luabind headers (which would pull in the "is_user_defined(...)"
overload).

Sorry for not having been clear enough.

On Mo, 3.12.2007, 01:14, Peter Suwara wrote:
> Hi Stephen,
>  I am not sure how you got around the is_user_defined overload in the case
> you presented here. Was you example a case where you got around the
> is_user_defined overload, or another other similar overload issue
> regerading
> smart_ptr's?
>
> In any case, I am going to try and include all of the headers before
> POLICY.HPP and see if that actually provides the correct overload.
>
> Cheers,
> Peter
>
> ----- Original Message -----
> From: "Stephan Kaiser" <ste...@ya...>
> To: <lua...@li...>
> Sent: Saturday, December 01, 2007 2:15 AM
> Subject: [SPAM] [Bayesian][testmode] Re: [luabind] MetroWerks
> Template/Function Overloading: How do you kick this...
>
>
>> The problem is not limited to Metrowerks and is due to a
>> misunderstanding
>> of overload resolution.
>>
>> The problem is also the same for smart_ptr overloads, btw.
>>
>> A description of the issue can be found in our sources:
>> http://yake.svn.sourceforge.net/viewvc/yake/trunk/yake/src/bindings.lua/detail/task.lua.cpp?view=markup
>> at line 30
>>
>> Here's the gist:
>> This is actually a bug in Luabind as "it assumes that overloaded
>> functions
>> are bound at instantiation time of templates, but overloaded functions
>> are
>> bound at definition time, so the luabind" code "only sees the
>> luabind::get_const_holder(...) catch all handler.
>>
>> Our workaround is to include the headers with overloads before including
>> the Luabind headers.
>>
>> Hope this helps.
>>
>>
>> On Fr, 30.11.2007, 08:43, Peter Suwara wrote:
>>> We are running to major issues regarding function overloading for the
>>> is_user_defined types.
>>>
>>> This is what is happening:
>>> BOOST_STATIC_ASSERT(luabind::detail::is_user_defined<cString>::value)
>>> is
>>> always failing, even though we have a file that does the following:
>>>
>>> namespace luabind {
>>>
>>> namespace converters {
>>>
>>> yes_t is_user_defined(by_value<cString>);
>>>
>>> yes_t is_user_defined(by_const_reference<cString>);
>>>
>>> }
>>> }
>>>
>>> Which means that the following code will fail when attempting to find
>>> an
>>> appropriate policy, it will pass through and not find any policy to
>>> match
>>> the user defined type:
>>>
>>> template<class T, class Direction>
>>>
>>> struct apply
>>>
>>> : eval_if<is_user_defined<T>, user_defined_converter<Direction>,
>>> eval_if<...
>>>
>>> Now, in order to fix this situation we include a template
>>> specialisation
>>> that does the following:
>>>
>>> template<>
>>>
>>> struct is_user_defined<cString> {
>>>
>>> enum {value=true};
>>>
>>> };
>>>
>>>
>>>
>>> This creates an interdependency with our codebase and LuaBind. Now does
>>> anyone know why the yes_t function overload fails on MetroWerks and are
>>> there any other ways to force user defined types for LUA.
>>>
>>> Thanks,
>>>
>>> Peter
>>> -------------------------------------------------------------------------
>>> SF.Net email is sponsored by: The Future of Linux Business White Paper
>>> from Novell.  From the desktop to the data center, Linux is going
>>> mainstream.  Let it simplify your IT future.
>>> http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4_______________________________________________
>>> luabind-user mailing list
>>> lua...@li...
>>> https://lists.sourceforge.net/lists/listinfo/luabind-user
>>>
>>
>>
>>
>> -------------------------------------------------------------------------
>> SF.Net email is sponsored by: The Future of Linux Business White Paper
>> from Novell.  From the desktop to the data center, Linux is going
>> mainstream.  Let it simplify your IT future.
>> http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
>> _______________________________________________
>> luabind-user mailing list
>> lua...@li...
>> https://lists.sourceforge.net/lists/listinfo/luabind-user
>>
>
>
> -------------------------------------------------------------------------
> SF.Net email is sponsored by: The Future of Linux Business White Paper
> from Novell.  From the desktop to the data center, Linux is going
> mainstream.  Let it simplify your IT future.
> http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
> _______________________________________________
> luabind-user mailing list
> lua...@li...
> https://lists.sourceforge.net/lists/listinfo/luabind-user
>