Front page | perl.perl5.porters |
Postings from June 2001
On Tue, Jun 26, 2001 at 11:01:42PM +0100, Nicholas Clark wrote:
> On Mon, Jun 25, 2001 at 05:27:44PM -0500, Jarkko Hietaniemi wrote:
>
> >
> > tcsh ibmsp01e:/tmp/jhi/perl/t ; ./perl -I../lib x
> > not ok 13 # int(279964589018079/59) is 4745162525730, not 4745162525730
> > SV = PVNV(0x2005c5d8) at 0x20041f84
>
> > NV = 4745162525730
> > NV = 4745162525730
>
> This is crazy. They appear to be the same.
> What does C think? I believe that the translation is
>
> ________________________________________________________________
> #include <stdio.h>
>
> int main (void) {
> long double a = 279964589018079;
> long double y = 4745162525730;
> long double x;
>
> x = a / 59;
> (void) modfl(x, &x);
>
> printf ("x=%.64Lg\n", x);
> printf ("y=%.64Lg\n", y);
> if (x == y)
> puts ("x==y");
> else
> puts ("x!=y");
>
> return 0;
> }
> ________________________________________________________________
>
> which on all the platforms I can test on (ARMLinux, where sizeof(long
> double) == sizeof(double) and x86 FreeBSD, where I had to s/modfl/modf/)
> I got == on both.
x=4745162525730
y=4745162525730
x==y
though I had to add the LL suffix to the constant or face the warnings
"x.c", line 4.19: 1506-207 (W) Integer constant 279964589018079 out of range.
"x.c", line 5.19: 1506-207 (W) Integer constant 4745162525730 out of range.
I also have to compile with -qlongdouble or face a segmentation fault
(but that's okay since that's the whole point of the exercise,
the original failure happened only with -Duselongdouble).
I can add -qlonglong, too, but that has no visible effect in any direction.
> What are the values of HAS_MODFL and HAS_MODFL_POW32_BUG on the unhappy
> platform (AIX, isn't it?)
Yes, AIX, 4.3.10.
d_modfl='undef'
d_modfl_pow32_bug='undef'
The modfl bug testing program
#include <math.h>
#include <stdio.h>
int main() {
long double nv = 4294967303.15;
long double v, w;
v = modfl(nv, &w);
printf(" %Lf %Lf %Lf\n", nv, v, w);
return 0;
}
outputs
4294967303.150000 0.000000 0.150000
Again compiled with -qlongdouble or executiong will segfault.
modfl() seems to have no man page.
> Nicholas Clark
--
$jhi++; # http://www.iki.fi/jhi/
# There is this special biologist word we use for 'stable'.
# It is 'dead'. -- Jack Cohen
Thread Previous
|
Thread Next