odinthenerd / tmp Goto Github PK

Hi Odin, I hope you have a great day, I'm really impress by some of your work.

It's not really an issue, but since I'm stuck at work with GCC 4.3 -std=gnu++0x, I did some experimentation to adapt your code to this environment. To my surprise, I did manage to replace a lot of using statement to their typedef counterpart :
template<typename T,typename C = identity_> struct dispatch<1,is_<T,C>> { template<typename A> using f = typename dispatch... f<std::is_same<T,A>>;};
become
template<typename T,typename C = identity_> struct dispatch<1,is_<T,C>> { template<typename A> struct f {typedef typename dispatch... f<std::is_same<T,A>::type>::type type; };};

This add a lot of boilerplate but I was able to make it compatible with GCC 4.3 since it have variadic template. The only difference is that listify_ is now a specialization of dispatch implemented with lift_<list,C>, and I must provide typename C = listify_<> as a default template metaclosure, but it get the job done. ALso most constexpr function like find_dispatch() must be reimplemented as a uint_<(calculation ? ...)>::value.

I only remade simple construct until now and I'm seeing some progress, but I may not see the big picture on where I'll hit a wall due to some properties of using statement.
Fusion-based features were an important part of your talks, and I'm not even scratching the surface of those properties which I doubt could work at compile time under my conditions, but
it's worth a try.

My issue is about the development of this : Do you foresee problem using this method and are you interested in seeing a merge request ? I also don't know if you are rewriting things up since the number of commits is low, as you would have yet again found another template metatrick.

Thanks and I really enjoy your conference,
Remi123

there are many fast track algorithms which only work up to a small fast track size