>>>>> "S" == Sean Russell <ser / efn.org> writes:

S> My guess is that the faster machine is flooding the Ruby stack with 
S> objects, and this is causing the segfault.  This is reasonable, but 
S> shouldn't Ruby throw raise an exception rather than segfaulting, if this is 
S> the case?

 Well, not really sure but the problem seems to be with @prev (instance
 variable of Node ?). The stack level is too deep when ruby try to mark the
 object. Actually ruby try to detect this problem only when it run
 rb_call0() and not when it try to mark an object.

Guy Decoux