Conflict tracing in LALR(1) analysis

Abstract

Tracing of conflicts manually is not so easy in LALR(1) analysis; It involves troublesome tracing of LR(0) states and/or examination of productions in the grammar. DeRemer and Penello suggested an automatic tracing method, where information called trace which shows sources of conflicts, is produced in a predefined form. In the original method of computing traces, a relation needed in computing lookahead sets is used to find out the items which make the conflict symbol by included in the lookahead set of a reducible item. In this thesis, the trace computation method is re-examined under new formalism of Park et al., and a more efficient method based on the formalism is proposed. Both the old and new method are implemented on KAIST Parser Generating System (KPGS), and experimental results for the comparison of the two methods are presented.