std :: ptr_fun模板化类和结构的参数难度

Question

I'm trying to work my way through using libkdtree++, trying to implement RRT, though I'm finding some trouble understanding how to use this library. Following the examples , I try to define the outline of my RRT class as such:

#pragma once

#include "coupling_tree.h"
#include "kdtree++/kdtree.hpp"
#include <deque>
#include <iostream>
#include <vector>
#include <limits>
#include <functional>
#include <set>


namespace trajectory {


    template<int c_dim> struct Point {

        std::set<const void*> registered;
        Eigen::VectorXd p;


        Point(Eigen::VectorXd point) :
            p(point)
        {
            assert(point.size() == c_dim);
        }



        ~Point()
        {
            bool unreg_ok = (registered.find(this) != registered.end());
            assert(unreg_ok);
            registered.erase(this);
        }


        double distance_to(Point const & x) const
        {
            double dist = 0;
            for (int i = 0; i < c_dim; ++i)
                dist += (p[i] - x[i])*(p[i] - x[i]);
            return std::sqrt(dist);
        }

        inline double operator[](size_t const N) const { return p(N); }



    };

    template<int c_dim> double tac(Point<c_dim> p, size_t k) { return p[k]; }


    template<int plant_dim, int c_dim>
    class RRT { //TODO: Should this be abstract so we can quickly implement lots of RRT variants?

        ///////TYPEDEFS
        typedef Point<c_dim> point;
        typedef KDTree::KDTree<c_dim, point, std::pointer_to_binary_function<point, size_t, double> > kd_tree;



        ////////////VARIABLES

    private:
        kd_tree tree;


        ////////////////////




    public:





    protected:


    private:

        const int getNumDim() const {
            return plant_dim;
        }


    };



}

This yields the following errors:

Severity    Code    Description Project File    Line    Suppression State
Error   C2664   'std::pointer_to_unary_function<trajectory::Point<5>,std::size_t,size_t (__cdecl *)(trajectory::Point<5>)> std::ptr_fun<trajectory::Point<5>,std::size_t>(_Result (__cdecl *)(_Arg))': cannot convert argument 1 from 'double (__cdecl *)(trajectory::Point<c_dim>,std::size_t)' to 'size_t (__cdecl *)(trajectory::Point<5>)'  test_RRT    C:\ResearchCode\robot-new\robot\projects\RRT\include\RRT.h  83  
Error   C2512   'std::pointer_to_binary_function<trajectory::Point<5>,std::size_t,double,_Result (__cdecl *)(_Arg1,std::_Arg2)>': no appropriate default constructor available  test_RRT    C:\ResearchCode\robot-new\robot\externals\libkdtree\kdtree++\kdtree.hpp 126 

I'm getting very lost in the typing here and what the complaint specifically is, especially since I'm new to ptr_fun being used in this way. Can someone explain the error, and the fix?

Answer 1

So, you're code is incomplete, as it's missing the vital implementation that causes the trigger. Look at the warning

cannot convert argument 1 from 'double (__cdecl *)(trajectory::Point<c_dim>,std::size_t)' to 'size_t (__cdecl *)(trajectory::Point<5>)'

So you're trying to call something that expects a size_t as an argument with a double.

One of the things I can think of is std::pointer_to_binary_function<point, size_t, double> . If you use this for double distance_to(Point const & x) it will not work for instance, as distance expects two Point-type inputs.

edit: So look at the line

typedef KDTree::KDTree<c_dim, point, std::pointer_to_binary_function<point, size_t, double> > kd_tree;

I cannot see where you use this type, but the type itself is based on a templated type, that has three parameters. I am not sure what is expected, but you set template argument 1 to c_dim, argument 2) to Point and argument 3 to... an undefined binary function pointer?!

You should look-up more information on std::pointer_to_binary_function, as you will see that you need to supply the constructor with the function you wish to point to. Ie

typedef KDTree::KDTree<c_dim, point, std::pointer_to_binary_function<point, size_t, double>(tac) > kd_tree;

I cannot test the code here, I'm afraid.