An Introduction To Attribute Grammars

Beside the syntax, semantic is a very important part of programming languages. Without a semantic a program would no longer be a program but a lifeless sequence of characters which is part of the language. The dynamic semantic properties of a program are determined during execution at runtime. Typically they directly depend on the input values of the program. By contrast the static semantic properties of a program are computed at compile-time when it is translated from source code representation into some kind of goal representation. The static semantic of a programming language is used to determine those properties. The concept of attributed context-free grammars (Attribute Grammars) addresses this aspect. It enables a user to create complete specifications of static semantics. This article gives a short introduction on the topic of attributed context-free grammars. It shows how to define such a construct and introduces the concept of synthesised and inherited attributes which can be associated to grammar symbols. Furthermore it will be shown how semantic rules on attributes can be defined. The final part of the paper gives an introduction into different types of attribute grammars and attribute evaluators.