Introduction to Compiler Compilers

Abstract

Historically, the existence of compiler-compilers is a result of using syntax-directed compiling techniques in order to structure the compiler. Syntax becomes a language in which parts of the compiler may be written, and the concept is extended to semantics by including a compatible programming language, usually general purpose, which allows classical programming methods for those parts of the compiler not susceptible to treatment by syntax. We will divide the subject in two parts, dealing first with those compiler-compilers whose formal treatment is limited to context-free grammars, and subsequently with extensions of these formalisms by one method or another. This division is strictly arbitrary, but corresponds to a practical fact, which is that the former class can be used, and indeed are being used, in a commercial environment to produce efficient compilers more easily than by ad hoc methods. The second group includes a series of promising ideas which are at different stages of development, but for which it is as yet not possible to make a reasonable evaluation.