Path: utzoo!utgpu!watserv1!watmath!maytag!vlsi!watale!sunee!avadekar
From: avadekar@sunee.waterloo.edu (Ashok Vadekar)
Newsgroups: comp.sys.amiga
Subject: Re: scanf() and double precision numbers
Message-ID: <250@sunee.waterloo.edu>
Date: 30 Jan 90 18:22:10 GMT
References: <73.25c20097@intersil.uucp> <1990Jan28.130653.26684@helios.physics.utoronto.ca>
Reply-To: avadekar@sunee.waterloo.edu (Ashok Vadekar)
Organization: U. of Waterloo, Ontario
Lines: 43

In article <73.25c20097@intersil.uucp> hamilton@intersil.uucp (Fred Hamilton) writes:
>
>Is the scanf() function supposed to be able to handle double precision
>floating points?  Here's why I ask:

>#include <stdio.h>
>void main()
>{
>	float number;
> 	printf("Enter a number: ");
>	scanf("%f",&number);
>	printf("\nNumber was %f\n",number);
>}

[...]

>-- 
>Fred Hamilton                  Any views, comments, or ideas expressed here
>Harris Semiconductor           are entirely my own.  Even good ones.
>Santa Clara, CA

The problem you've had is, as already pointed out, the %f vs. %lf specifier.
However, I have found (I think) another related problem:

double x;

	do {
		printf("Enter Value: ");
		scanf("%lf",x);
		printf("Value: %G\n",x);
		} while (x);

Try this with values in the range 2E-7 to 6E-7, and you will find
that the printf is returning a value of zero.  The correct results happen
for values less than (about) 1.000000000000000027E-7 and greater than
5.000000000000001E-7 (I havn't checked negative numbers), but fails for
values in this range.  I havn't checked if this is also true using
%lf or %le, nor have I found if the error is related to the ffp/ieee
math library choice.

Ashok Vadekar
University of Waterloo			In Search of Signature
Waterloo, Ontario