Joost Kraaijeveld wrote:
I want to sort a vector of Edges according to their remaining estimated distance to the finish point (an own implementation of a graph algorithm).
How do I sort this vector with edges:
// Returns the cost between start and finish, given a graph int CostEstimate( const Graph& aGraph, const Vertex& start, const Vertex& finish) ...
// Member function of a graph std::vector< Edge > v = ...; // Sort them according to the cost std::sort( v.begin(), v.end(), boost::lambda::bind( CostEstimate( (*this), boost::lambda::bind( &Edge::otherSide, &start, boost::lambda::_1), finish)) < boost::lambda::bind( CostEstimate( (*this), boost::lambda::bind( &Edge::otherSide, &start, boost::lambda::_2), finish)));
where otherSide is a const function of the class Edge which returns a "const Vertex&"
Compiling this results in an error:
error: invalid initialization of reference of type ‘const Vertex&’ from expression of type ‘const boost::lambda::lambda_functor<boost::lambda::lambda_functor_base<boost::lambda::action<3, boost::lambda::function_action<3, boost::lambda::detail::unspecified> >, boost::tuples::tuple<const Vertex& (Edge::* const)(const Vertex&)const, const Vertex* const, const boost::lambda::lambda_functor<boost::lambda::placeholder<1> >, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type> > >’
TIA
Hi, I am sure there is some way to extend the code you wrote in such a way that it compiles and works as it should. My question would be: Are you going to still understand it in, say, half a year? Four binds in one expression is certainly not unheard of, but personally I would do it with one bind and an old-fashioned function, see attached code. Hope I got your settings right. I was a bit confused by otherSide being a function of edge. This would mean that start is an Edge, while finish is a Vertex in the sort-code. But start and finish are both Vertexes in the CostEstimate function... To avoid such wondering, it would help if you could provide a small but complete program which shows the problem. Nothing more and nothing less. Regards, Roland