On 02/29/2012 11:07 PM, Marco Speranza wrote: > Cool good job... > > Only one think.. I ran the tests and I experienced one failure: > > ==== > Results : > > Failed tests: > findMaxFlowAndVerify(org.apache.commons.graph.flow.EdmondsKarpTestCase): > expected:<3> but was:<5> > > Tests run: 109, Failures: 1, Errors: 0, Skipped: 1 > > [INFO] > ------------------------------------------------------------------------ > [INFO] BUILD FAILURE > [INFO] > ------------------------------------------------------------------------ > ==== > > the Edmonds and Karp algo uses the Breadth First Search, maybe there are some > problem in bsf visit.
ah, thanks for pointing that out. I debugged the Edmonds algorithm, and I have an idea where this problem may come from: In the old BFS implementation, the discoverEdge method in the visitor was even called for nodes that have been already visited, which is not the case anymore. From my understanding the new behavior is correct, or am I missing something? Thomas --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
