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. ciao -- Marco Speranza <marcospera...@apache.org> Google Code: http://code.google.com/u/marco.speranza79/ Il giorno 29/feb/2012, alle ore 21:08, Thomas Neidhart ha scritto: > On 02/27/2012 07:35 PM, Claudio Squarcella wrote: >> Hello, >> >> if you want to postpone the edge visit to the right time (i.e. when the >> corresponding tail vertex is removed from the "waiting list") then I >> guess the waiting list itself should also keep track of the edge and/or >> the predecessor (vertex). The class PredecessorsList probably does that >> job fairly well. >> >> Rephrasing a bit: we are seeking a unified implementation for both >> standard "graph searches" (only concerned with vertices) and more >> general "graph visits" (e.g. the one needed for max flow algorithms). So >> yeah, let's use our brains for something cool :) > > Hi, > > I have followed the discussion and implemented the discussed changes > right away: > > - a unified graph search method using a wrapper class as James > suggested > - discoverEdge / finishEdge are called in the correct order according > to the chosen strategy (depth or breadth search) > > The call order is as follows: > > - discover the edge that lead to the vertex > unless we are at the start vertex of course > - call finishEdge, when we shall finish, skip the next steps > - if we shall expand the edge, discover the vertex > - find all connections from this vertex and add to the queue/stack > - finish vertex > > Proper unit tests are still missing, but will be added as a next step. > > Thomas > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
