Abstract

A tool for compiler testing is implemented. The tool is driven by a BNF grammar extended by a device called control flags to incorporate those parts of a language that are not controlled by context-free rules (for instance, use of declared names). The tool produced two sets of program strings such that each production in the grammar is used at least once. One set is produced by Purdom's algorithm. The other is produced by a new algorithm designed by the author which builds a tree and then produces a set of strings from it for compiler testing. The program strings produced from the new algorithm differ less in length than those from Purdom's algorithm.

Library of Congress Subject Headings

Compiling (Electronic computers)--Testing; Computer programs--Testing

Publication Date

1987

Document Type

Thesis

Department, Program, or Center

Computer Science (GCCIS)

Advisor

Lutz, Peter

Advisor/Committee Member

Biles, John

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: QA76.76.C65 C493 1987

Campus

RIT – Main Campus

Download

Share

COinS