Dear Jeremy, et al, I am trying to implement a shortest paths algorithm on a directed graph. As I have unit edge weights, I would rather use breadth_first_search rather than dijkstra. // Is this a good candidate for addition? template struct vertex_marker : public boost::base_visitor > { typedef Tag event_filter; vertex_marker(Map p_map,int p_val) : m_map(p_map),m_val(p_val) { } template void operator()(Vertex v,const Graph&) { boost::put(m_map,v,m_val); } Map m_map; int m_val; }; template vertex_marker mark_vertices(Map p_map, Tag, int p_v) { return vertex_marker(p_map,p_v); } boost::breadth_first_search( mygraph, rootnode, boost::visitor(boost::make_bfs_visitor( std::make_pair( boost::record_distances( boost::get(boost::vertex_distance,mygraph), boost::on_tree_edge()) , std::make_pair( mark_vertices( boost::get(boost::vertex_distance,mygraph), boost::on_initialize_vertex(), std::numeric_limits<int>::max()) , mark_vertices( boost::get(boost::vertex_distance,mygraph), boost::on_start_vertex(), 0) ) ) )) ); The problem is that on_start_vertex does not seem to be implemented for breadth_first_search, but it is for depth_first. For dijkstra, the start node seems to be automaticly zeroed. I guess I could mod Am I missing something really obvious here or is there an inconsistency across the algorithms? Shouldn't on_start_vertex be used for all of the graph algorithms? Also, I can't seem to find documentation on neighbor_bfs.hpp Thanks, Joel