Matching-based incremental evaluators for hierarchical attribute grammar dialects

Abstract

Although attribute grammars have been very effective for defining individual modules of language translators, they have been rather ineffective for specifying large program-transformational systems. Recently, several new attribute grammar “dialects” have been developed that support the modular specification of these systems by allowing modules, each described by an attribute grammar, to be composed to form a complete system. Acceptance of these new hierarchical attribute grammar dialects requires the availability of efficient batch and incremental evaluators for hierarchical specifications. This paper addresses the problem of developing efficient incremental evaluators for hierarchical specifications. A matching-based approach is taken in order to exploit existing optimal change propagation algorithms for nonhierarchical attribute grammars. A sequence of four new matching algorithms is presented, each increasing the number of previously computed attribute values that are made available for reuse during the incremental update.