Re: serious inline math pb i386

[hjl at lucon dot org (H.J. Lu)]

> 
> 
>   In message <x8867h8ccxg.fsf@nbeckerpc.hns.com>you write:
>   > Yes! I have a test case!
>   > 
>   > --------------------------
>   > #include <math.h>
>   > 
>   > #define pc_ceiling_dB -4.9691
>   > #define pc_step_dB     0.5
>   > #define pc_loopdelay 1
>   > 
>   > float pc_ceiling;
>   > float pc_step;
>   > float pc_threshold_dB = 17;
>   > float pc_threshold;
>   > 
>   > main () {
>   >   float pc_command[pc_loopdelay];
>   >   int i_pc_loopdelay;
>   > 
>   >   for (i_pc_loopdelay = 0; i_pc_loopdelay < pc_loopdelay; i_pc_loopdelay++)
>   >     pc_command[i_pc_loopdelay] = 0;
>   >   pc_ceiling = pow (10.0, pc_ceiling_dB / 10.0);
>   >   pc_step = pow (10.0, pc_step_dB / 10.0);
>   >   pc_threshold = pow (10.0, pc_threshold_dB / 10.0);
>   > }
>   > ----------------------
>   > 
>   > When I compile with egcs-1.0.3 glibc-2.0.7(RH5.1) i686-pc-linux-gnu:
>   > 
>   > > g++ -o bug bug.cc -O3 -g
> It doesn't fail for me :(  Presumably because I'm not using glibc2.0.7.

You need both egcs-1.0.3 and glibc-2.0.7, which is mentioned in
the bug report.

-- 
H.J. Lu (hjl@gnu.org)