Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site ism780c.UUCP
Path: utzoo!watmath!clyde!cbosgd!ukma!psuvm.bitnet!psuvax1!burdvax!sdcrdcf!ism780c!jim
From: jim@ism780c.UUCP (Jim Balter)
Newsgroups: net.micro.amiga
Subject: accuracy disregarded
Message-ID: <630@ism780c.UUCP>
Date: Mon, 17-Feb-86 01:15:33 EST
Article-I.D.: ism780c.630
Posted: Mon Feb 17 01:15:33 1986
Date-Received: Wed, 19-Feb-86 00:49:19 EST
Reply-To: jim@ism780c.UUCP (Jim Balter)
Organization: Interactive Systems Corp., Santa Monica, CA
Lines: 23
Summary: bad values for pi, e, log10 in mathffp.h

Every once in a while, someone tries to give pi a new value.
Commodore-Amiga is the latest to join this disreputable group.

From mathffp.h (with Lattice 3.02):

#define PI      ((FLOAT) 3.1415192653857)

Aside from the extra `1' thrown in, the last three digits are off (but they
are beyond the range of (float)).
Anyone who introduces such obvious errors into something so tightly focused
and easily verified is likely to introduce even more, and more subtle, errors
in more demanding code.  I hope that the code of the programmer involved will
be given an extra review, and that future works of that programmer will be
given extra attention during quality control procedures.  Review?  Quality
control?  But it's only a microcomputer!  Uh huh.

FYI: values to 50 places of

pi              3.14159265358979323846264338327950288419716939937510
e               2.71828182845904523536028747135266249775724709369995
log(10.0)       2.30258509299404568401799145468436420760110148862877
--
-- Jim Balter ({sdcrdcf!ism780c,ima}!jim)