Maxim Yegorushkin
10/31/2008 9:44:00 AM
On Oct 31, 2:39 am, Hill <zhubi...@gmail.com> wrote:
> This is an exercise on TCPL:
> Write a b i n d e r 3 () that binds the second and third arguments of
> a threeargument function to produce a unary predicate. Give an example
> where b i n d e r 3 () is a useful function.
> And i got a solution from << C++ solutions>>:
>
> template<typename FO>
> struct binder2_3
> {
> typedef typename FO::result_type result_type;
> typedef typename FO::first_argument first_argument;
> binder2_3(const FO& fo,
> typename FO::second_argument& a2,
> typename FO::third_argument& a3)
> :fo_(fo), a2_(a2), a3_(a3){}
> result_type operator()(first_argument a1){
> return fo_(a1, a2_, a3_);
> }
> private:
> FO fo_;
> const typename FO::second_argument a2_;
> const typename FO::third_argument a3_;};
>
> template<typename FO, typename P2, typename P3> inline
> binder2_3<FO> binder3(const FO& fo, const P2& a2, const P3& a3){
> return binder2_3<FO>(fo, a2, a3);
>
> }
>
> I can't understand the followingn sentence:
> typedef typename FO::result_type result_type;
> It require FO has a member named result_type ?
It requires FO to have a type member named result_type: typedef,
struct, class or union.
> This is a too restrict rule.
> Could someone give an explaination?
This is because your binder object wraps the call to the underlying
functor. The binder has to return the result of the functor call,
therefore to be able to declare the type of the return value of
binder::operator() it needs to know the type of the return value of
the underlying functor.
--
Max