Re: lib/regex_internal.h on 1.4.11

[Bruno Haible]

Eric Blake wrote:
> It looks like you failed to check .m4 tests.  I'm applying the patch
> below.

Yes, I missed to change the .m4 tests. Thank you for the fix.

> Meanwhile, I have a question about tests/test-signbit.c - your
> patch changed most occurrences of -0.0L to -zerol, to work around one HP
> bug, but there is another comment that states:
> 
> /* We cannot use the expression '-zerol' here, because on HP-UX/hppa it
>    evaluates to 0.0L, not -0.0L.  */
> long double minus_zerol = -0.0L;
> 
> and still uses minus_zerol throughout the file.  So which is it, or are we
> really stuck with no portable way to represent negative zero across the
> spectrum of HP compilers/architectures?

Indeed, we're in a maze of twisty little HP cc compiler bugs:
  <http://lists.gnu.org/archive/html/bug-gnulib/2008-04/msg00179.html>
  <http://lists.gnu.org/archive/html/bug-gnulib/2008-08/msg00000.html>

It requires more creativity to write a negative 'long double' zero that
all compilers will understand. The updated test program

============================ fool.c ==================================
#include <float.h>
#include <math.h>
#include <stdio.h>
#include <string.h>
long double x = 1.0L;
long double y = -0.0L;
long double z = 0.0L;
int different_from_plus_zero (long double arg)
{
  return memcmp (&arg, &z, sizeof (z)) != 0;
}
int main()
{
  printf ("%Lf\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
#define x 1.0L
#define y -0.0L
  printf ("%Lf\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
#undef y
#define y -z
  printf ("%Lf\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
#undef y
#define y (-0.1L * LDBL_MIN)
  printf ("%Lf\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
#undef y
#define y (-0.1L / LDBL_MAX)
  printf ("%Lf\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
  return 0;
}
====================================================================

yields the following output on HP-UX 11:

-inf
1
1
1
-inf
1
1
1
inf
0
0
0
-inf
1
1
1
-inf
1
1
1

Jonathan, could you please run this on your HP-UX 10.20 machine? What is
the result? Also, for completeness, what result do you get for

============================ food.c ==================================
#include <float.h>
#include <math.h>
#include <stdio.h>
#include <string.h>
double x = 1.0;
double y = -0.0;
double z = 0.0;
int different_from_plus_zero (double arg)
{
  return memcmp (&arg, &z, sizeof (z)) != 0;
}
int main()
{
  printf ("%f\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
#define x 1.0
#define y -0.0
  printf ("%f\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
#undef y
#define y -z
  printf ("%f\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
#undef y
#define y (-0.1 * DBL_MIN)
  printf ("%f\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
#undef y
#define y (-0.1 / DBL_MAX)
  printf ("%f\n", x / y);
  printf ("%d\n", x / y < 0);
  printf ("%d\n", signbit (x/y) != 0);
  printf ("%d\n", different_from_plus_zero (y));
  return 0;
}
====================================================================

?

Bruno