c++/7369: weird results with the statement "#define PI 4.*atan(1.)"

[pel@ucla.edu]

>Number:         7369
>Category:       c++
>Synopsis:       weird results with the statement "#define PI 4.*atan(1.)"
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Jul 21 12:06:01 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     Peter Latham
>Release:        2.96
>Organization:
>Environment:
Red Hat Linux 7.3
>Description:
If PI is defined via the statement

#define PI 4.*atan(1.)

then 1/PI and 1/3.14159 are different.
>How-To-Repeat:
compile using "g++ main.c". then type "a.out", which
prints PI, 1/PI and 1/3.14159. the result is

   3.141593 0.196350 0.318310

If for some strange reasont the file didn't attach properly,
the source code is:


#include <stdio.h>
#include <math.h>

#define PI 4.*atan(1.)
//#define PI 3.14159

void main(int argc, char** argv)
{
        fprintf(stdout, "%f %f %f\n", PI, 1/PI, 1/3.14159);
}



   
>Fix:
the workaround is to not use atan in the define statement,

#define PI=3.14159

this produces the correct output.

>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: application/octet-stream; name="main.c"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="main.c"

CiNpbmNsdWRlIDxzdGRpby5oPgojaW5jbHVkZSA8bWF0aC5oPgoKI2RlZmluZSBQSSA0LiphdGFu
KDEuKQovLyNkZWZpbmUgUEkgMy4xNDE1OQoKdm9pZCBtYWluKGludCBhcmdjLCBjaGFyKiogYXJn
dikKewoJZnByaW50ZihzdGRvdXQsICIlZiAlZiAlZlxuIiwgUEksIDEvUEksIDEvMy4xNDE1OSk7
Cn0K