> Look at Uniqueness class for an example of a factory for common
> UniquenessFilters. I would implement it something like this (type in this
> mail client only):

> class NodePairUniqueness implements UniquenessFilter {
>     private final Set<Pair<Long,Long>> pairs = new
> HashSet<Pair<Long,Long>>();

>     public bollean checkFirst( TraversalBranch branch ) {
>         return true;
>     }

>     public boolean check( TraversalBranch branch ) {
>         Path position = branch.position();
>         Node nextToLast = null;
>         Node last = null;
>         Iterator<Node> nodes = position.nodes().iterator();
>         while ( nodes.hasNext() ) {
>             nextToLast = last;
>             last = nodes.next();
>         }
>         Pair<Long,Long> lastTwoNodes = Pair.of( nextToLast.getId(),
> last.getId() );
>         return pairs.add( lastTwoNodes );
>     }

> }

> I also realize that it would be grand with a Path#reversedNodes() or
> something this started from the end node and back to the start node. And
> memory-wise it would be more efficient with a custom PrimitiveLongPair
> class or similar instead of the Pair<Long,Long>.

> Den lördagen den 28:e april 2012 skrev hyperutila<derrickbrin...@gmail.com>:

> > I have a graph where there are sometimes multiple relationships
> > between the same two nodes. I want to perform a traversal wherein one
> > of those relationships can be visited once, but once one is visited,
> > none of the others can be.

> > In other words, I don't want to implement uniqueness of relationships,
> > if you were to define a relationship as a means of getting from node A
> > to node B. Once the traversal has gone from node A to node B at some
> > point, it would never do that again, even if there are other
> > relationships that connect those same two nodes. Does that make sense?
> > How do I do this? Can I implement my own version of Uniqueness in
> > order to specify this behavior?

> > Thanks!