main.cpp

/* 
 Sample Results:

inserting: 6.18748
query without box: 0.0735919
query with box: 1.4e-07
POINT(4 4) - 4000
POINT(5 5) - 5000
POINT(6 6) - 6000
*/

#include <ctime>

#include <iostream>
#include <boost/foreach.hpp>

#include <boost/geometry.hpp>
#include <boost/geometry/geometries/point.hpp>
#include <boost/geometry/index/rtree.hpp>


namespace bg = boost::geometry;
namespace bgi = boost::geometry::index;

typedef bg::model::point<float, 2, bg::cs::cartesian> point;
typedef bg::model::box<point> box;
typedef std::pair<point, unsigned> value;

int main(int argc, char *argv[])
{
    bgi::rtree< value, bgi::quadratic<16> > rtree;

    auto startup = std::clock();
    const int N = 100;
    const int M = 10000000;

    // create some values
    for ( unsigned i = 0 ; i < M ; ++i )
    {
        point p = point(i, i);
        rtree.insert(std::make_pair(p, i*1000));
    }

    // search for nearest neighbours
    std::vector<value> returned_values;
    point sought(5, 5);
    box query_box(point(5 - 2, 5 - 2), point(5 + 2, 5 + 2));

    std::cout << "inserting: " << double(std::clock() - startup)/CLOCKS_PER_SEC << std::endl;
    startup = std::clock();
    
    for ( unsigned i = 0; i < N ; ++i )
    {      
      returned_values.clear();
      rtree.query(bgi::satisfies([&](value const& v) {return bg::distance(v.first, sought) < 2;}),
                  std::back_inserter(returned_values));
    }

    std::cout << "query without box: " << double(std::clock() - startup)/CLOCKS_PER_SEC/N << std::endl;
    startup = std::clock();

    for ( unsigned i = 0; i < N ; ++i )
    {      
      returned_values.clear();
      rtree.query(
        bgi::within(query_box) &&
        bgi::satisfies([&](value const& v) {return bg::distance(v.first, sought) < 2;}),
        std::back_inserter(returned_values));
    }
    
    std::cout << "query with box: " << double(std::clock() - startup)/CLOCKS_PER_SEC/N << std::endl;
      
    // print returned values
    BOOST_FOREACH(value const& v, returned_values)
      std::cout << bg::wkt<point>(v.first) << " - " << v.second << std::endl;

    
    return 0;
}

If you're curious about the difference in performance of both queries. The query using only the satisfies() predicate is much slower because without the within() predicate all values stored in the rtree are tested. Without spatial or nearest predicate the query should take more or less the same ammount of time as if the values were just stored in a STL container like std::vector and e.g. searched with std::find_if().

I have tried compiling this in boost 1.66.0, however I needed to change
https://gist.github.com/weidenrinde/89e835611721f0c1d907#file-main-cpp-L67
to the lambda accepting point, not value (and then in distance just v instead of v.first).