Re: yet another Java oddity

[Mike Miller]

On Thu, Feb 7, 2013 at 9:24 AM, John W. Eaton wrote:
> On 02/07/2013 09:15 AM, Mike Miller wrote:
>
>> The tests that are failing for me are chop, axis, betainv, normpdf, and
>> dec2base. I looked into them enough that I believe all of them are
>> failing because the value returned by std::pow is different when Java is
>> running.

EDIT: I contradicted myself here. I believe the chop, betainv, and
normpdf tests are failing due to std::pow. Dec2base fails because of
the subtraction example I showed above truncating the number of
significant bits. I am not sure about axis.

> It certainly does seem strange that std::pow would return a different value
> after Java is loaded.  I don't know how that could happen.  Can you use gdb
> and set a breakpoint in std::pow and then step through it to see what
> happens in both cases?

Yes, I had done this yesterday and didn't want to flood with too many
details at once. Here's a transcript.

octave:1> 10^-5

Breakpoint 1, __pow (x=10, y=-5) at w_pow.c:29
29      {
(gdb) n
34              z=__ieee754_pow(x,y);
(gdb) s
__ieee754_pow () at ../sysdeps/i386/fpu/e_pow.S:66
66              fldl    12(%esp)        // y
(gdb) finish
Run till exit from #0  __ieee754_pow () at ../sysdeps/i386/fpu/e_pow.S:66
__pow (x=10, y=-5) at w_pow.c:35
35              if(_LIB_VERSION == _IEEE_|| __isnan(y)) return z;
(gdb) finish
Run till exit from #0  __pow (x=10, y=-5) at w_pow.c:35
xpow (a=10, b=-5) at ../../default/libinterp/interp-core/xpow.cc:95
95        return retval;
Value returned is $1 = 1.0000000000000001e-05
(gdb) cont
Continuing.
ans =  1.0000e-05
octave:2> javaObject ("java.lang.String");
[New Thread 0xa4ba8b70 (LWP 660)]
[New Thread 0xa4b28b70 (LWP 661)]
[New Thread 0xa4aa8b70 (LWP 662)]
[New Thread 0xa4a28b70 (LWP 663)]
[New Thread 0xa466eb70 (LWP 664)]
[New Thread 0xa45eeb70 (LWP 665)]
[New Thread 0xa4975b70 (LWP 666)]
[New Thread 0xa459eb70 (LWP 667)]
[New Thread 0xa454eb70 (LWP 668)]
[New Thread 0xa44ceb70 (LWP 669)]
[New Thread 0xa444eb70 (LWP 670)]
[New Thread 0xa43feb70 (LWP 671)]
octave:3> 10^-5

Breakpoint 1, __pow (x=10, y=-5) at w_pow.c:29
29      {
(gdb) n
34              z=__ieee754_pow(x,y);
(gdb) s
__ieee754_pow () at ../sysdeps/i386/fpu/e_pow.S:66
66              fldl    12(%esp)        // y
(gdb) finish
Run till exit from #0  __ieee754_pow () at ../sysdeps/i386/fpu/e_pow.S:66
__pow (x=10, y=-5) at w_pow.c:35
35              if(_LIB_VERSION == _IEEE_|| __isnan(y)) return z;
(gdb) finish
Run till exit from #0  __pow (x=10, y=-5) at w_pow.c:35
xpow (a=10, b=-5) at ../../default/libinterp/interp-core/xpow.cc:95
95        return retval;
Value returned is $2 = 1.0000000000000006e-05
(gdb) cont
Continuing.
ans =  1.0000e-05

Looks to me like the same call stack with different return values.

-- 
mike