The bug affected only the calculation of arctan 37/4, and of no other angle . Had the program been put into commercial servic e with the bug still present, it would have disabled the automatic landing system every time the plane flew in a precis e southeasterly direction . Of course, this bug was caught by testing, not by a correctness proof, but those of us who work i n correctness of practical programs are confident that correctnes s techniques will, in the long run, prove a better way of attackin g this kind of problem .
Unfortunately, ACM prohibits us from displaying non-influential references for this paper.
To see the full reference list, please visit http://dl.acm.org/citation.cfm?id=1010759.