Publikationer
Incremental Evaluation of Reference Attribute Grammars using Dynamic Dependency Tracking
Avdelning/ar:
Publiceringsår: 2012
Språk: Engelska
Publikation/Tidskrift/Serie: LU-CS-TR:2012-249
Fulltext:
Dokumenttyp: Rapport
Förlag: Department of Computer Science, Lund University
Sammanfattning
Reference attribute grammars (RAGs) have proven practical for gen-
erating production-quality compilers from declarative descriptions, as demonstrated by the JastAdd system. Recent results indicate their applicability also to generating semantic services in interactive editors. For use in editors, it is necessary to update the attribution after edit operations. Earlier algorithms based on statically scheduled incremental attribute evaluation are, however, not applicable to RAGs, as they do not account for the dynamic dependencies that reference attributes give rise to. In this report, we introduce a notion of consistency for RAG attributions, along with an algorithm for maintaining consistency after edit operations, based on dynamic dependency tracking. That is, we introduce a means to do incremental evaluation of RAGs using dynamic dependency tracking.
erating production-quality compilers from declarative descriptions, as demonstrated by the JastAdd system. Recent results indicate their applicability also to generating semantic services in interactive editors. For use in editors, it is necessary to update the attribution after edit operations. Earlier algorithms based on statically scheduled incremental attribute evaluation are, however, not applicable to RAGs, as they do not account for the dynamic dependencies that reference attributes give rise to. In this report, we introduce a notion of consistency for RAG attributions, along with an algorithm for maintaining consistency after edit operations, based on dynamic dependency tracking. That is, we introduce a means to do incremental evaluation of RAGs using dynamic dependency tracking.
Disputation
Nyckelord
- Technology and Engineering
Övrigt
Unpublished
98
- LUCAS
- ISSN: 1404-1200

