A parser-generating system for constructing compressed compilers
- 1 November 1973
- journal article
- Published by Association for Computing Machinery (ACM) in Communications of the ACM
- Vol. 16 (11) , 669-676
- https://doi.org/10.1145/355611.362536
Abstract
This paper describes a parser-generating system (PGS) currently in use on the CDC-6500 computer at Purdue University. The PGS is a Fortran-coded program that accepts a translation grammar as input and constructs from it a compact, machine-coded compiler. In the input translation grammar, each BNF syntactic rule corresponds to a (possibly empty) “code generator” realizable as an assembly language, Fortran or Algol, subroutine that is called whenever that syntactic rule is applied in the parse of a program. Typical one-pass compilers constructed by the PGS translate source programs at speeds approaching 14,000 cards per minute. For an XPL compiler, the parser program and its tables currently occupy 288 words of 60-bit core memory of which 140 words are parsing table entries and 82 words are links to code generators.Keywords
This publication has 4 references indexed in Scilit:
- Empirical comparison of LR(k) and precedence parsersACM SIGPLAN Notices, 1970
- A technique for generating almost optimal Floyd-Evans productions for precedence grammarsCommunications of the ACM, 1970
- EULERCommunications of the ACM, 1966
- A Descriptive Language for Symbol ManipulationJournal of the ACM, 1961