next up previous contents
Next: Index edge optimization Up: Meta-Optimizations for XGRS code Previous: Bitset optimization

Bidirectional edge optimization

If a functor provides bidirectional implementation (e.g. hgraph), can use both directions for code generation. During the computation of the edge-disjoing path cover of the rule test graphs, selects one of the directions. Thus bidirectional edge optimization completes bidirectional graph functor edges in order to find better paths for code generation.

Thus the order of a rule test graph can be reduced to 1, if enough bidirectional functors are used.



Uwe Assmann
1998-12-22