Path: utzoo!attcan!uunet!snorkelwacker!tut.cis.ohio-state.edu!rutgers!news-server.csri.toronto.edu!dgp.toronto.edu!rodney
From: rodney@dgp.toronto.edu (Rodney Hoinkes)
Newsgroups: comp.sys.sgi
Subject: MINFLOAT
Summary: can anyone explain this...
Message-ID: <1990Jul31.121947.25092@jarvis.csri.toronto.edu>
Date: 31 Jul 90 16:19:47 GMT
Distribution: na
Organization: CSRI, University of Toronto
Lines: 55


	I have been working with a routine in which I need to
	determine whether a variable is out of the bounds of a
	float (NaN etc.), so I have been trying to check it
	against MINFLOAT and MAXFLOAT as defined in values.h.

	If I check 

	if(val > MAXFLOAT) ... this works fine for NaN
							problems

	BUT!

	if val = 0.0, and I do the following

	if(val < MINFLOAT) ... it tells me that val IS
							less than MINFLOAT

	The test program below shows this simply:


#include <stdio.h>
#include <values.h>

#define ZERO	0.0

main() {

	printf("MINFLOAT %g ZERO %g\n",MINFLOAT,ZERO);

	if(ZERO < MINFLOAT)
		printf("ZERO is less than MINFLOAT\n");
	else	
		printf("ZERO is greater than MINFLOAT\n");

}

	compiled with cc test.c -lc_s -o test

	I get the following output:

MINFLOAT 1.4013e-45 ZERO 0
ZERO is less than MINFLOAT

	Can anyone explain?  If I printf %f they both return 0.00000
	or whatever.

	--

	Rodney Hoinkes
	Centre for Landscape Research
	University of Toronto
	CANADA

	rodney@dgp.toronto.edu