Where are you looking? �The line I see (in astar_search_no_init) is:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> <br> � �MutableQueue Q(cost, index_in_heap, compare);<br> <br> which doesn't require the number of vertices in advance.</blockquote><div><br>So implicit graphs have been implemented since boost 1.39 ... Are you aware of any implementation notes, documentation, reference ? I'll give it a try tonight.<br> �</div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div class="im"><br> <br> <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> Notes :<br> I use astar_search_no_init, and property maps based on std::map so that put() can index it in a not-known-yet location.<br> Btw, my color map type is std::map<NodeID,boost::default_color_type > (plus the wrapper around it). I feel that using<br> "boost::default_color_type" is kind of a hack. Is it ?<br> </blockquote> <br></div> The reason is that the normal Boost A* search avoids processing the same vertex twice by using a color map. �If you want the implicit graph version that appears in most textbooks, you need a dummy color map that claims that all vertices are unvisited. �So it is somewhat a workaround to allow the same algorithm to work for both cases.<br> </blockquote><div><br>That was not exactly my question. The thing is, I discovered the type "boost::default_color_type" while watching internal boost headers I'm not really supposed to look at as an end-user. I feel like I should have used something like boost::color_map::type or whatever. This would be exactly the same from the compiler's point of view, but the code would be better.<br> More importantly, what are the logical steps I should have done to figure it out ?<br></div></div>