Implementing attribute grammars by computational models

Abstract

We describe the implementation of attribute grammars with the knowledge-based system NUT. NUT uses intuitionistic propositional calculus for automatic synthesis of programs [2]. Penjam [3] discussed the connection between two formalisms - computational models, used in NUT system, and attribute grammars. We have developed this relation further. The system we are presenting is the extension of NUT and it allows automatic transformation from AG-representation to equivalent computational model (we call that the language model). NUT can then be used for automatic synthesis of tree-decoration algorithms for both - dynamic and static tree evaluation. For latter the algorithm synthesised is the multi-visit tree decoration algorithm.