John Skaller wrote:
> I do not see that. The current algorithm is recursive descent
> is it not? That can also be used with a trivial modification
> to test graph equivalence, by simply keeping a list of visited
> nodes and not revisiting them. 

If you want the equivalence to be invariant by unfolding, you need to
keep a set of pairs of nodes, not just a set of nodes.

-- Alain