A comprehensive and algorithmical sequence point model for C programming language

Abstract

Trade-off between performance and portability forces the International Standard for C to define the order of evaluation of an expression only in part and to leave occurrence of side effects within an expression unspecified. Furthermore use of ordinary English in the Standard to specify the semantics of an expression makes it very difficult to strictly discern validity of a given expression with side effects, where the validity decided by a rule written in terms of the sequence point, the concept indicating certain points within a program to guarantee occurrence of side effects. This paper proposes a new sequence point model that is based on a tagged AST for comprehension and implementation to decide whether an expression with side effects conforms to the Standard or not, and SEA (Simple Expression Analyzer), the experimental implementation of the proposed model shows that the model works well with real expressions used in various programs.