Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!lll-winken!unixhub!shelby!csli!francis
From: francis@csli.Stanford.EDU (Dave Francis)
Newsgroups: comp.sys.mac.programmer
Subject: Floating Point accuracy in Think C
Message-ID: <16594@csli.Stanford.EDU>
Date: 28 Nov 90 20:34:05 GMT
Reply-To: francis@csli.stanford.edu (Dave Francis)
Organization: Center for the Study of Language and Information, Stanford U.
Lines: 28

Hello netters!

I have a roundoff problem when using floating point numbers in Think C 4.0.2.

I've written a function which takes a floating point number and returns
a string that can be used for displaying with QuickDraw.

The problem is that when I pass the value to be converted Think C changes
it to a less accurate number so when I generate the string I don't get exactly
what was passed in. Here's a example:

The number input is 2.314 but inside the routine it becomes 2.3139996765 or
something like that (I think becuase Think C passes floating point as
extended). Anyway when I generate the string the return is "2.313" if the
precision is set to 3 (the function also takes a precision parameter)
It gets worse if I pass a value which has less significant digits than the
precision. So the value 3.2 passed in with a precision parameter of 3
in some cases gets returned as "3.199", which is unexceptable for display
purposes if the user e ntered 3.2.

Is there a way to stop this loss of accuracy when passing floating point
values? Why is there  a loss of accuracy anyway? What should I do? Are
there any rounding functions out there or better float->string converters?


Help!!

Dave Francis  Sapphire Design Systems