Priority controlled incremental attribute evaluation in attributed graph grammars

Abstract

We show how incremental attribute change propagation can be efficiently done on Graph Grammars. The order of evaluation is controlled by priority numbers assigned to each attribute vertex. The priority numbers are assigned in linear time as directed by static grammar analysis. We also develop a powerful set-theoretic based embedding description format, and introduce the idea of roles as a mechanism to aid in graph grammar engineering. Roles are also a powerful aid to static analysis of grammar properties. A key to our approach is the uniformity gained by collapsing the substrate (structure) graph and the attribute dependency graph into a single directed graph structure. Related groups of attributes are simultaneously replaced by cluster rewrites.