Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!accelerator.eng.ohio-state.edu!kaa.eng.ohio-state.edu!rob
From: rob@kaa.eng.ohio-state.edu (Rob Carriere)
Newsgroups: comp.lang.c
Subject: Re: Array indexing vs. pointers...
Message-ID: <733@accelerator.eng.ohio-state.edu>
Date: 9 Oct 88 00:52:25 GMT
References: <836@proxftl.UUCP> <3105@hubcap.UUCP> <1700@dataio.Data-IO.COM> <1706@dataio.Data-IO.COM> <8273@alice.UUCP>
Sender: news@accelerator.eng.ohio-state.edu
Reply-To: rob@kaa.eng.ohio-state.edu (Rob Carriere)
Organization: Ohio State Univ, College of Engineering
Lines: 18

In article <8273@alice.UUCP> ark@alice.UUCP (Andrew Koenig) writes:
>In particular, when
>I convert 100 degrees Celsius to Fahrenheit [using * and /], I get 212, not
>211.9999999999999 or 212.0000000000001.  The other two ways of
>doing it do not guarantee that.

Probably.  However, as thermometers more accurate than about +-.1K are
fairly rare, what's the diff?  You can't rely on the number being more
precise than about +-.05 degrees F anyway, due to the inaccuracy of
your input.  This is exactly the point the original poster was trying
to make; if *you* know the last couple of bits don't matter, you might
as well take advantage of that fact.  If on the other hand, this
measurement was made using a Star Trek tricorder (which is accurate to
12 decimal places) and had to be reported to Spock, *then* you'd
better be careful that you don't destroy any bits of accuracy, and use
your trick (and double precision).

Rob Carriere