PKGBUILDs/community/agave/compat-round.patch

Description: Patched compat-round.cc to be more portable on systems with GNU libc
Author: Torsten Werner <twerner@debian.org>
Date: Thu, 5 Oct 2006 23:52:57 +0200
Index: agave-0.4.5/src/core/compat-round.cc
===================================================================
--- agave-0.4.5.orig/src/core/compat-round.cc	2006-10-01 03:25:58.000000000 +0530
+++ agave-0.4.5/src/core/compat-round.cc	2008-02-15 19:34:10.000000000 +0530
@@ -34,206 +34,9 @@
  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-#include <math.h>
-
-/* Symbolic constants to classify floating point numbers. */
-#define	COMPAT_FP_INFINITE	0x01
-#define	COMPAT_FP_NAN		0x02
-#define	COMPAT_FP_NORMAL	0x04
-#define	COMPAT_FP_SUBNORMAL	0x08
-#define	COMPAT_FP_ZERO		0x10
-
-#ifdef Linux
-#include <endian.h>
-#elif defined(__FreeBSD__)
-#include <machine/endian.h>
-#endif
-#define	compat_fpclassify(x) \
-    ((sizeof (x) == sizeof (float)) ? compat___fpclassifyf(x) \
-     : (sizeof (x) == sizeof (double)) ? compat___fpclassifyd(x) \
-     : compat___fpclassifyl(x))
-
-union compat_IEEEf2bits {
-    float	f;
-    struct {
-#if _BYTE_ORDER == _LITTLE_ENDIAN
-        unsigned int	man	:23;
-        unsigned int	exp	:8;
-        unsigned int	sign	:1;
-#else /* _BIG_ENDIAN */
-        unsigned int	sign	:1;
-        unsigned int	exp	:8;
-        unsigned int	man	:23;
-#endif
-    } bits;
-};
-
-union compat_IEEEd2bits {
-    double	d;
-    struct {
-#if _BYTE_ORDER == _LITTLE_ENDIAN
-        unsigned int	manl	:32;
-        unsigned int	manh	:20;
-        unsigned int	exp	:11;
-        unsigned int	sign	:1;
-#else /* _BIG_ENDIAN */
-        unsigned int	sign	:1;
-        unsigned int	exp	:11;
-        unsigned int	manh	:20;
-        unsigned int	manl	:32;
-#endif
-    } bits;
-};
-
-#ifdef __alpha__
-union compat_IEEEl2bits {
-    long double	e;
-    struct {
-        unsigned int	manl	:32;
-        unsigned int	manh	:20;
-        unsigned int	exp	:11;
-        unsigned int	sign	:1;
-    } bits;
-};
-#define	compat_mask_nbit_l(u)	((void)0)
-#elif __amd64__
-union compat_IEEEl2bits {
-    long double	e;
-    struct {
-        unsigned int	manl	:32;
-        unsigned int	manh	:32;
-        unsigned int	exp	:15;
-        unsigned int	sign	:1;
-        unsigned int	junkl	:16;
-        unsigned int	junkh	:32;
-    } bits;
-};
-#define	compat_mask_nbit_l(u)	((u).bits.manh &= 0x7fffffff)
-#elif __arm__
-union compat_IEEEl2bits {
-    long double	e;
-    struct {
-        unsigned int	manl	:32;
-        unsigned int	manh	:32;
-        unsigned int	exp	:15;
-        unsigned int	sign	:1;
-        unsigned int	junk	:16;
-    } bits;
-};
-#define	compat_mask_nbit_l(u)	((u).bits.manh &= 0x7fffffff)
-#elif __i386__
-union compat_IEEEl2bits {
-    long double	e;
-    struct {
-        unsigned int	manl	:32;
-        unsigned int	manh	:32;
-        unsigned int	exp	:15;
-        unsigned int	sign	:1;
-        unsigned int	junk	:16;
-    } bits;
-};
-#define	compat_mask_nbit_l(u)	((u).bits.manh &= 0x7fffffff)
-#elif __ia64__
-union compat_IEEEl2bits {
-    long double	e;
-    struct {
-#if _BYTE_ORDER == _LITTLE_ENDIAN
-        unsigned int	manl	:32;
-        unsigned int	manh	:32;
-        unsigned int	exp	:15;
-        unsigned int	sign	:1;
-        unsigned long	junk	:48;
-#else /* _BIG_ENDIAN */
-        unsigned long	junk	:48;
-        unsigned int	sign	:1;
-        unsigned int	exp	:15;
-        unsigned int	manh	:32;
-        unsigned int	manl	:32;
-#endif
-    } bits;
-};
-#define	compat_mask_nbit_l(u)	((u).bits.manh &= 0x7fffffff)
-#elif __powerpc__
-union compat_IEEEl2bits {
-    long double	e;
-    struct {
-        unsigned int		sign	:1;
-        unsigned int		exp	:15;
-        unsigned long long	manh	:48;
-        unsigned long long	manl	:64;
-    } bits;
-};
-#define	compat_mask_nbit_l(u)	((void)0)
-#elif __sparc64__
-union compat_IEEEl2bits {
-    long double	e;
-    struct {
-        unsigned int	sign	:1;
-        unsigned int	exp	:15;
-        unsigned long	manh	:48;
-        unsigned long	manl	:64;
-    } bits;
-};
-#define	compat_mask_nbit_l(u)	((void)0)
-#endif
-
-    static int
-compat___fpclassifyf(float f)
-{
-    union compat_IEEEf2bits u;
-
-    u.f = f;
-    if (u.bits.exp == 0) {
-        if (u.bits.man == 0)
-            return (COMPAT_FP_ZERO);
-        return (COMPAT_FP_SUBNORMAL);
-    }
-    if (u.bits.exp == 255) {
-        if (u.bits.man == 0)
-            return (COMPAT_FP_INFINITE);
-        return (COMPAT_FP_NAN);
-    }
-    return (COMPAT_FP_NORMAL);
-}
+/* modified by Torsten Werner <twerner@debian.org> for GNU libc */
 
-    static int
-compat___fpclassifyd(double d)
-{
-    union compat_IEEEd2bits u;
-
-    u.d = d;
-    if (u.bits.exp == 0) {
-        if ((u.bits.manl | u.bits.manh) == 0)
-            return (COMPAT_FP_ZERO);
-        return (COMPAT_FP_SUBNORMAL);
-    }
-    if (u.bits.exp == 2047) {
-        if ((u.bits.manl | u.bits.manh) == 0)
-            return (COMPAT_FP_INFINITE);
-        return (COMPAT_FP_NAN);
-    }
-    return (COMPAT_FP_NORMAL);
-}
-
-    static int
-compat___fpclassifyl(long double e)
-{
-    union compat_IEEEl2bits u;
-
-    u.e = e;
-    if (u.bits.exp == 0) {
-        if ((u.bits.manl | u.bits.manh) == 0)
-            return (COMPAT_FP_ZERO);
-        return (COMPAT_FP_SUBNORMAL);
-    }
-    compat_mask_nbit_l(u);		/* Mask normalization bit if applicable. */
-    if (u.bits.exp == 32767) {
-        if ((u.bits.manl | u.bits.manh) == 0)
-            return (COMPAT_FP_INFINITE);
-        return (COMPAT_FP_NAN);
-    }
-    return (COMPAT_FP_NORMAL);
-}
+#include <math.h>
 
     double
 compat_round(double x)
@@ -241,8 +44,8 @@
     double t;
     int i;
 
-    i = compat_fpclassify(x);
-    if (i == COMPAT_FP_INFINITE || i == COMPAT_FP_NAN)
+    i = fpclassify(x);
+    if (i == FP_INFINITE || i == FP_NAN)
         return (x);
 
     if (x >= 0.0) {