Date: Fri, 2 Dec 2011 14:33:43 +1100 (EST) From: Ian Smith <smithi@nimnet.asn.au> To: Alexander Best <arundel@freebsd.org> Cc: freebsd-wireless@freebsd.org Subject: Re: comparing floating points via "==" or "!=" Message-ID: <20111202142524.V3548@sola.nimnet.asn.au> In-Reply-To: <20111201000243.GB44877@freebsd.org> References: <20111130224422.GA36424@freebsd.org> <CADWvR2herXkcMaZXFKm_ALBXnVFDhRCNZ-9%2B-MwHVx5XJPG4AA@mail.gmail.com> <20111201000243.GB44877@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 1 Dec 2011, Alexander Best wrote:
> On Wed Nov 30 11, Igor Mozolevsky wrote:
> > On 30 November 2011 22:44, Alexander Best <arundel@freebsd.org> wrote:
> >
> > > i played a bit with the gcc -Wfloat-equal warning and noticed that inside
> > > sys/dev/ath/ath_rate/sample/tx_schedules.h, a lot of comparisons of floating
> > > points are happening ("=="). is there a better way to deal with this?
> >
> > as opposed to (abs(a-b) < tolerance)?
>
> fabs(3) actually, or some other way, e.g. described in:
>
> http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm
Bookmarked, thanks. Very good article, and I never knew IEEE floats
could be compared as integers! OT for wireless, but very useful for
some astronomical stuff I'm working on .. in Pascal, no less :)
cheers, Ian
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20111202142524.V3548>