From: dontmailme AT iname DOT com (Steamer)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Possible truncation error
Date: Tue, 27 Jun 2000 19:09:37 GMT
Organization: always disorganized
Lines: 40
Message-ID: <3958fbdb.42786311@news.freeserve.net>
References: <_x665.15310$Za1 DOT 242989 AT newsc DOT telia DOT net>
NNTP-Posting-Host: modem-163.cleaner-wrasse.dialup.pol.co.uk
X-Trace: newsg4.svr.pol.co.uk 962132978 13907 62.136.246.163 (27 Jun 2000 19:09:38 GMT)
NNTP-Posting-Date: 27 Jun 2000 19:09:38 GMT
X-Complaints-To: abuse AT theplanet DOT net
X-Newsreader: Forte Free Agent 1.11/32.235
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Erik Berglund <erik2 DOT berglund AT telia DOT com> wrote:

> Hi, I tried the following program, which gives
> j=9999. I think the right result should be 10000,
> because a=0x40c3880000000000, which is exactly 10000.
> 
> #include <stdio.h>
> #include <math.h>
> int main() {
> double a;
> int j,*p;
> 
>    a = pow(100,2);

Although 'a' is a double, it is held for the time being in
the FPU, which is at 80-bit precision.  It's value is
0x400C9C3FFFFFFFFFFFFE, which is just less than 10000.

>    j = a;

This therefore sets j to 9999.

>    p = (int*)&a;
>    printf("%08x%08x\n",*(p+1),*p);
>    printf("%i\n",j);
>    return(0);
> }
> 
> If I insert a printf("\n") before j=a;, the result
> becomes 10000 instead, and I have no idea why.

This is because gcc couldn't leave the double in the FPU
during the function call, so it stores it in memory,
converting to 64-bit in the process (because it's a double).
This happens to make it 0x40c3880000000000, which (as you
pointed out) is exactly 10000.

Don't you just love floating point arithmetic?

S.