Re: Potential bug swarm v-1.2/gcc-2.8.1 (was Negative double wierdness)

[Marcus G. Daniels]

Thanks for the excellent bug report, Steve.

Turns out there was a bug in the version(s) of glibc distributed with
Redhat 5.  To fix it, you need to tweak /usr/include/__math.h so that
the pow definition looks like this:

__MATH_INLINE double pow (double __x, double __y);
__MATH_INLINE double
pow (double __x, double __y)
{
  register double __value, __exponent;
  long long int __p = (long long int) __y;


Unfortunately, stock gcc-2.8.1 has a bug that makes it impossible to
compile this inline function with -O2 (the current snapshot of gcc
also has the bug).  However, a no-threads configuration of egcs-1.0.3a
worked for me.  I don't know what the configuration of gcc/egcs is
with Redhat 5; what's needed is a version that isn't configured with
"--enable-threads=posix".

If you'd rather not modify your system, build with
"-O2 -fno-inline" instead of "-O2".

                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.
                  ==================================