knows three kinds of fixpoint detection for rule groups: direct fixpoint check, central neighbor set comparison, and local neighbor set comparison. These detection methods have different runtime costs. Not all of them are apt for all XGRS.
The first method is probably the fastest. It can be used if the functor functions which are used to perform graph manipulations, give back a change flag, if something has changed. Then functor functions for edge addition are queried if they have changed something. Direct fixpoint checking is also chosen, if no rule of the rule group is edge additive, i.e. if all rules manipulate nodes.
Central neighbor set comparison runs a bit slower. It compares the neighbor sets of order loop nodes before and after a fixpoint loop. It memorizes the old values of neighbor sets of order loop nodes by copying or assignment.
If an assigned edge of a rule group does not start at an order loop node, local neighbor set comparison is performed. This is probably the slowest mode: It copies/memorizes old values of neighbor sets each time the source node of the assigned edge is traversed.
Option flags can be used to override the default modes (section 3.6.1).