Nice result :) Btw, it's tempting to use std::function in the same way as you would use a function in Haskell, but it usually isn't a very good idea. In C++ st

Nice result :) Btw, it's tempting to use std::function in the same way as you would use a function in Haskell, but it usually isn't a very good idea. In C++ std::function is for when you want type erasure, usually because you wish to store a functor somewhere with a uniform type. Type erasure in C++ is similar to existential typing + a type class in Haskell. You could say that all Haskell closures are implicitly using type erasure and then the compiler has to work hard to optimize away the extra boxing/indirect calls. Here is another way to write map in C++: #include <iostream> #include <list> template<typename F, typename A, typename B = typename std::result_of<F(A)>::type> std::list<B> map(std::list<A> const & xs, F f) { std::list<B> ys; for (auto const & x : xs) ys.push_back(f(x)); return ys; } int main() { std::list<int> xs{1, 2, 3, 4, 5}; auto ys = map(xs, [](int i) { return i * 2; }); for (auto y : ys) std::cout << y << "\n"; } You can expect this code to run exactly as fast as if you had written the loop by hand instead of using the map template function.