On 04/11/2012 20:54, Matt Calabrese wrote:
On Sun, Nov 4, 2012 at 2:23 PM, Mathias Gaunard < mathias.gaunard@ens-lyon.org> wrote:
Personally, I've found it more useful to just not define result_of<F(args)>::type if F(args) cannot be called.
The only problem with that is that you have to rely on the creator of a function object to abide by this convention. "can_be_called" (callable might be a simpler name and is what was used with C++0x concepts) would work regardless of whether or not result_of is defined.
result_of does not have to be implemented in terms of result<Sig>::type/result_type. template<class T, class R = void> struct enable_if_type { typedef R type; }; template<class Sig, class Enable = void> struct result_of {}; template<class F, class... Args> struct result_of<F(Args...) , typename enable_if_type< decltype( std::declval<F>() (std::declval<Args>()...) ) >::type > { typedef decltype(std::declval<F>()(std::declval<Args>()...)) type; }; Of course, if you don't have decltype, you can't do that and must use the result_type or result templates. In C++03, you'll be able to deduce whether the expression in valid, but not what the result type is.