Bruno Haible <[EMAIL PROTECTED]> writes:
> the diagnosis is
> clear: Bug in <float.h>. Same as on BeOS. We have a workaround in the
> 'float' module. Probably we need to add FreeBSD 6.1 here...
How about checking for this via a compile-time test? That way we don't
need to maintain a list of which software systems have the bug. Here
is a proposed patch (I don't have easy access to BeOS or FreeBSD, so
I can't easily test it).
2007-11-20 Paul Eggert <[EMAIL PROTECTED]>
Check for float.h double vs long double bug with a compile-time test.
This works around a bug in FreeBSD 6.1 that affected 'seq' in coreutils.
Problem reported by Jim Meyering in
<http://lists.gnu.org/archive/html/bug-gnulib/2007-11/msg00120.html>.
* lib/float.in.h: Do not depend on __i386__ and __BEOS__, since we
now check for the bug at comfigure-time and include this file only
if the bug is present.
* m4/float_h.m4 (gl_FLOAT_H): Check for the bug at compile-time
rather than relying on $host_os.
diff --git a/lib/float.in.h b/lib/float.in.h
index 986719b..773ba35 100644
--- a/lib/float.in.h
+++ b/lib/float.in.h
@@ -24,7 +24,6 @@
#define _GL_FLOAT_H
/* 'long double' properties. */
-#if defined __i386__ && defined __BEOS__
/* Number of mantissa units, in base FLT_RADIX. */
# undef LDBL_MANT_DIG
# define LDBL_MANT_DIG 64
@@ -52,7 +51,6 @@
/* Maximum e such that 10^e is in the range of representable finite numbers.
*/
# undef LDBL_MAX_10_EXP
# define LDBL_MAX_10_EXP 4932
-#endif
#endif /* _GL_FLOAT_H */
#endif /* _GL_FLOAT_H */
diff --git a/m4/float_h.m4 b/m4/float_h.m4
index 1b1ad10..c7f6828 100644
--- a/m4/float_h.m4
+++ b/m4/float_h.m4
@@ -1,4 +1,4 @@
-# float_h.m4 serial 2
+# float_h.m4 serial 3
dnl Copyright (C) 2007 Free Software Foundation, Inc.
dnl This file is free software; the Free Software Foundation
dnl gives unlimited permission to copy and/or distribute it,
@@ -7,13 +7,28 @@ dnl with or without modifications, as long as this notice is
preserved.
AC_DEFUN([gl_FLOAT_H],
[
AC_REQUIRE([AC_PROG_CC])
- AC_REQUIRE([AC_CANONICAL_HOST])
FLOAT_H=
- case "$host_os" in
- beos*)
- FLOAT_H=float.h
- gl_CHECK_NEXT_HEADERS([float.h])
- ;;
- esac
+ AC_CACHE_CHECK([for working float.h],
+ [gl_cv_header_working_float_h],
+ [AC_COMPILE_IFELSE(
+ [[/* Check for the bug that requires a float.h workaround, namely, that
+ float.h claims that long double and double are both IEEE double,
+ but actually, long double has a 64-bit fraction. */
+ #include <float.h>
+ int check_LDBL_MANT_DIG[
+ (! (FLT_RADIX == 2
+ && DBL_MANT_DIG == 53
+ && (0.1L - 0.1
+ == -5.5497598704101758215756490244530141353607177734375e-18L)
+ && LDBL_MANT_DIG != 64)
+ ? 1 : -1)
+ ];
+ ]],
+ [gl_cv_header_working_float_h=yes],
+ [gl_cv_header_working_float_h=no])])
+ if test $gl_cv_header_working_float_h != yes; then
+ FLOAT_H=float.h
+ gl_CHECK_NEXT_HEADERS([float.h])
+ fi
AC_SUBST([FLOAT_H])
])
