From mboxrd@z Thu Jan  1 00:00:00 1970
Content-Type: text/plain; charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 6.5 \(1508\))
From: Bakul Shah <bakul@bitblocks.com>
In-Reply-To: <c93098b57ad99db811eb93d220b25f46@kw.quanstro.net>
Date: Wed, 21 Aug 2013 10:42:57 -0700
Content-Transfer-Encoding: quoted-printable
Message-Id: <2FEB81A6-7D40-427D-BF9B-07D90872E307@bitblocks.com>
References: <04891ed00c35de2583c1da9a017dd7e2@hamnavoe.com>
	<c93098b57ad99db811eb93d220b25f46@kw.quanstro.net>
To: Fans of the OS Plan 9 from Bell Labs <9fans@9fans.net>
Subject: Re: [9fans] comparisons with NaN
Topicbox-Message-UUID: 7342a6f4-ead8-11e9-9d60-3106f5b1d025

On Aug 21, 2013, at 9:55 AM, erik quanstrom <quanstro@quanstro.net> =
wrote:

> On Wed Aug 21 12:09:26 EDT 2013, 9fans@hamnavoe.com wrote:
>>> at least in terms of passing floating point test suites
>>> (like python's) the NaN issue doesn't come up
>>=20
>> Actually it was a test suite that revealed the NaN errors.
>> I wouldn't think it's something anyone needs in normal
>> day-to-day computation, but sometimes boxes must be ticked.
>=20
> :-)  it is hard to imagine how this is useful.  it's not like
> =E2=88=91{i=E2=86=92=E2=88=9E}-0 is interesting.  at least =
=E2=88=8F{i=E2=86=92=E2=88=9E}-0 has an alternating
> sign.  (so does it converge with no limit?)
>=20
> the difference i have seen is a situation like
> 	atan2(-0, x)	=E2=89=A1 -=CF=80
> 	atan2(+0, x)	=E2=89=A1 pi, =E2=88=80 x<0.
>=20
> any ideas on how this is useful?


See comments by Stephen Canon in
=
http://stackoverflow.com/questions/1565164/what-is-the-rationale-for-all-c=
omparisons-returning-false-for-ieee754-nan-values

Try this:

#include <u.h>
#include <libc.h>

main(){
        double a, b;
        setfcr(0);
        a =3D 0.0;
        b =3D a/a;
        if(a <  b) print(" (a <  b)");
        if(a <=3D b) print(" (a <=3D b)");
        if(a =3D=3D b) print(" (a =3D=3D b)");
        if(a !=3D b) print(" (a !=3D b)");
        if(a >=3D b) print(" (a >=3D b)");
        if(a >  b) print(" (a >  b)");
        if(b <  a) print(" (b <  a)");
        if(b <=3D a) print(" (b <=3D a)");
        if(b =3D=3D a) print(" (b =3D=3D a)");
        if(b !=3D a) print(" (b !=3D a)");
        if(b >=3D a) print(" (b >=3D a)");
        if(b >  a) print(" (b >  a)");
        if(b !=3D b) print(" (b !=3D b)");
        if(b =3D=3D b) print(" (b =3D=3D b)");
        print("\n");
	return 0;
}

It falsely reports b =3D=3D b when b is NaN.=20=