> Maybe the problem is that somewhere in the depths of is_perfect, a
> BipartiteGraph is being created without the intention to use it that
> way. Perhaps it is in self.complement(), or some other function that
> is implicitly assuming to use the current class to make a copy, when
> it should be making a copy as a different class. Maybe we just need to
> assume less when creating derived objects...
Nonon, it is just that if you now type g.complement() on a Bipartite
Graph in Sage, you will raise an exception ! My function fails only
when I call it on a bipartite Graph !
sage: g = graphs.CompleteBipartiteGraph(10,10)
sage: g.complement()
---------------------------------------------------------------------------
RuntimeError Traceback (most recent call last)
/home/ncohen/<ipython console> in <module>()
/home/ncohen/.Sage/local/lib/python2.6/site-packages/sage/graphs/generic_graph.py
in complement(self)
8497 for v in self:
8498 if not self.has_edge(u,v):
-> 8499 G.add_edge(u,v)
8500 return G
8501
/home/ncohen/.Sage/local/lib/python2.6/site-packages/sage/graphs/bipartite_graph.py
in add_edge(self, u, v, label)
690 # check for endpoints in different partitions
691 if self.left.issuperset((u,v)) or self.right.issuperset((u,v)):
--> 692 raise RuntimeError('Edge vertices must lie in
different partitions.')
693
694 # add the edge
RuntimeError: Edge vertices must lie in different partitions.
sage:
Obviously, we can not expect the function .complement() in
BipartiteGraph to return a BipartiteGraph (well, I know, if you graph
is the disjoint union of two edges, or a P_4, or ... ). I never
intended my message as a bug report, of course something needs to be
fixed there but the question is rather : what is to be done of this
BipartiteGraph class, and should it be a default one that users could
instantiate without knowing it ? I know where the bug comes from, but
I begin to know Sage's code for graphs, and I reviewed myself the
ticket about edge addition, but there is no way for a normal user to
think of it as something else than a bug in Sage, or a very bad idea
of class design... :-/
Nathann
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
