Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10 5/3/83; site utcsrgv.UUCP Path: utzoo!utcsrgv!peterr From: peterr@utcsrgv.UUCP (Peter Rowley) Newsgroups: net.lang Subject: Dangerous compiler bug anecdote Message-ID: <3338@utcsrgv.UUCP> Date: Wed, 22-Feb-84 00:43:21 EST Article-I.D.: utcsrgv.3338 Posted: Wed Feb 22 00:43:21 1984 Date-Received: Wed, 22-Feb-84 02:44:31 EST Organization: CSRG, University of Toronto Lines: 12 Quoted without comment, from ACM SIGSOFT SOFTWARE ENGINEERING NOTES, Vol 9 No 1 Jan 84, Page 11 " ... A stress analysis program was written in FORTRAN for vendor X's machine. It was then ported to vendor Y's machine without recertification. A pressure containment vessel was then designed using this program running on vendor Y's machine. The pressure containment vessel was built, instrumented with stress gauges, etc., and tested under pressure. It exploded, hurting or killing several technicians. The problem was that the FORTRAN compiler for vendor Y's machine produced incorrect code. " (excerpt from a letter to the editor by David Benson, Wash. State U.)