Previously the code did not ignore spur paths that had already been added into the list of potential paths. This could cause the search to return duplicate paths for certain graphs.