[Bug math/564] New: "pow(1.0, nan)" should return nan

["shiba at mail2 dot accsnet dot ne dot jp" <sourceware-bugzilla at sources dot redhat dot com>]

"pow(1.0, nan)" returns "1.0", but should return "nan".
Because current behavior contradicts a specification written in
sysdeps/ieee754/ldbl-128/e_powl.c.  It says "Special cases: ... 3.  (anything)
** NAN is NAN."  The speification is appropriate, I believe.

This bug can be reproduced compiling with -lm and running following code:

#include <stdio.h>
#include <math.h>
main(){ double nan = 0.0 / 0.0; printf("%f\n", pow(1.0, nan)); }

It prints "1.000000", but should print "nan".

It occurs on Redhat9 (glibc-2.3.2-27.9.7, `uname -a`=Linux dhcppc0 2.4.20-31.9
#1 Tue Apr 13 18:04:23 EDT 2004 i686 i686 i386 GNU/Linux), and latest Debian
sarge (libc6 2.3.2.ds1-18, `uname -a`=Linux canna12 2.4.27-speakup #1 Thu Aug 19
21:46:14 CEST 2004 i686 GNU/Linux).
I've not tried latest CVS version of glibc, but I entered bug here because I
couldn't find any modification about pow() as far as I could, viewing codes
using cvsweb.

Note that this may (or may not) be IA-32 specific bug.

-- 
           Summary: "pow(1.0, nan)" should return nan
           Product: glibc
           Version: 2.3.2
            Status: NEW
          Severity: normal
          Priority: P2
         Component: math
        AssignedTo: aj at suse dot de
        ReportedBy: shiba at mail2 dot accsnet dot ne dot jp
                CC: glibc-bugs at sources dot redhat dot com
  GCC host triplet: i686-pc-linux-gnu


http://sources.redhat.com/bugzilla/show_bug.cgi?id=564

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.