Pattern-based tree attribution

摘要

Attribute grammars have been used for many language-oriented tasks, including the formal description of semantics and the implementation of compilation tasks from simple type checking through code generation. Despite their successful use, attribute grammars have some disadvantages, including the monolithic nature of the grammar and the fixed factoring of all attribute descriptions by a single set of grammar productions. Attribute pattern sets provide a more expressive attribution system by using pattern matching, instead of grammar productions, to perform case analysis. Attribute pattern sets can be implemented in terms of attribute grammars in a way that maintains the dependency structure of the attribute system, making it straightforward to convert many of the practical results from attribute grammar theory to similar results for attribute pattern sets.