[GraphHopper] Making efficient use of DijkstraOneToMany

[Peter K]

Hi Chris,

> I'm looking at using DijkstraOneToMany to get path costs (just the sum
> of the edge weights, not the whole path) from a single origin to all
> nodes closer than some fixed cost limit.
>
> I know about findEndNode(), but am not sure if I need to walk the
> graph of nodes myself, calling findEndNode() and weight() on each
> node, or if there is a way to walk exhaustively with only a limit as
> the termination point.
>
> Ideally, I'd have something like findEndNodesAndCosts() that would
> return a map of node ID to path cost for all nodes less than the cost
> cutoff set with DijkstraOneToMany.limit(). What would be the suggested
> implementation of this?

You already have all you need, I think. At the end of a call ala
findEndNode(node, -1) you need to iterate through the changed nodes only:

for(int i = 0; i < changedNodes.size(); i++) {
   int nodeIndex = changedNodes.get(i);
   // info: changedNodes could contain duplicates
   map.put(nodeIndex, weights[nodeIndex])
}

Regards,
Peter.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/graphhopper/attachments/20130520/26fca73b/attachment-0001.html>