Parsing in a hostile world

Abstract

The paper describes a formalization and development effort aimed at automating the production of parsers for languages for which the common classifications (LALR (F. DeRemer and T.J. Pennello, 1982; B.B. Kristensen and O.L. Madsen, 1981), LL), and even the common division of labor between lexical and syntactical analysis, do not apply. This effort, implemented in tools named DURA and lexyt, aims at real-world applications of parsing technology, applied to software renovation prior to plain compilers, rather than theoretical soundness only. A number of industrial results, around a tool named RainCode, are also described.