mirror of
https://github.com/ruby/ruby.git
synced 2022-11-09 12:17:21 -05:00
* ext/bigdecimal/extconf.rb, ext/bigdecimal/bigdecimal.h (labs, llabs): support environments missing labs and llabs.
* ext/bigdecimal/bigdecimal.h (vabs): added. * ext/bigdecimal/extconf.rb, ext/bigdecimal/bigdecimal.h, ext/bigdecimal/bigdecimal.c, test/bigdecimal/test_bigdecimal.rb (TestBigDecimal#test_new): replace U_LONG, S_LONG, S_INT, and U_INT with appropreate standard or ruby-provided types. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@28992 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This commit is contained in:
parent
01e3f613d7
commit
d86da0fcf1
5 changed files with 347 additions and 315 deletions
12
ChangeLog
12
ChangeLog
|
@ -1,3 +1,15 @@
|
|||
Sat Aug 15 00:22:00 2010 Kenta Murata <mrkn@mrkn.jp>
|
||||
|
||||
* ext/bigdecimal/extconf.rb, ext/bigdecimal/bigdecimal.h
|
||||
(labs, llabs): support environments missing labs and llabs.
|
||||
|
||||
* ext/bigdecimal/bigdecimal.h (vabs): added.
|
||||
|
||||
* ext/bigdecimal/extconf.rb, ext/bigdecimal/bigdecimal.h,
|
||||
ext/bigdecimal/bigdecimal.c, test/bigdecimal/test_bigdecimal.rb
|
||||
(TestBigDecimal#test_new): replace U_LONG, S_LONG, S_INT, and U_INT
|
||||
with appropreate standard or ruby-provided types.
|
||||
|
||||
Sun Aug 15 00:15:45 2010 Tanaka Akira <akr@fsij.org>
|
||||
|
||||
* ext/pathname/pathname.c (path_utime): Pathname#utime translated
|
||||
|
|
File diff suppressed because it is too large
Load diff
|
@ -23,6 +23,35 @@
|
|||
extern "C" {
|
||||
#endif
|
||||
|
||||
#ifndef HAVE_LABS
|
||||
static inline long
|
||||
labs(long const x)
|
||||
{
|
||||
if (x < 0) return -x;
|
||||
return x;
|
||||
}
|
||||
#endif
|
||||
|
||||
#ifndef HAVE_LLABS
|
||||
static inline long long
|
||||
llabs(long long const x)
|
||||
{
|
||||
if (x < 0) return -x;
|
||||
return x;
|
||||
}
|
||||
#endif
|
||||
|
||||
#ifdef vabs
|
||||
# undef vabs
|
||||
#endif
|
||||
#if SIZEOF_VALUE <= SIZEOF_INT
|
||||
# define vabs abs
|
||||
#elif SIZEOF_VALUE <= SIZEOF_LONG
|
||||
# define vabs labs
|
||||
#elif SIZEOF_VALUE <= SIZEOF_LONG_LONG
|
||||
# define vabs llabs
|
||||
#endif
|
||||
|
||||
extern VALUE rb_cBigDecimal;
|
||||
|
||||
#if 0 || SIZEOF_BDIGITS >= 16
|
||||
|
@ -57,11 +86,6 @@ extern VALUE rb_cBigDecimal;
|
|||
*/
|
||||
#define VP_EXPORT static
|
||||
|
||||
#define U_LONG unsigned long
|
||||
#define S_LONG long
|
||||
#define U_INT unsigned int
|
||||
#define S_INT int
|
||||
|
||||
/* Exception codes */
|
||||
#define VP_EXCEPTION_ALL ((unsigned short)0x00FF)
|
||||
#define VP_EXCEPTION_INFINITY ((unsigned short)0x0001)
|
||||
|
@ -98,13 +122,13 @@ extern VALUE rb_cBigDecimal;
|
|||
*/
|
||||
typedef struct {
|
||||
VALUE obj; /* Back pointer(VALUE) for Ruby object. */
|
||||
U_LONG MaxPrec; /* Maximum precision size */
|
||||
size_t MaxPrec; /* Maximum precision size */
|
||||
/* This is the actual size of pfrac[] */
|
||||
/*(frac[0] to frac[MaxPrec] are available). */
|
||||
U_LONG Prec; /* Current precision size. */
|
||||
size_t Prec; /* Current precision size. */
|
||||
/* This indicates how much the. */
|
||||
/* the array frac[] is actually used. */
|
||||
S_INT exponent;/* Exponent part. */
|
||||
SIGNED_VALUE exponent; /* Exponent part. */
|
||||
short sign; /* Attributes of the value. */
|
||||
/*
|
||||
* ==0 : NaN
|
||||
|
@ -126,9 +150,9 @@ typedef struct {
|
|||
*/
|
||||
|
||||
VP_EXPORT Real *
|
||||
VpNewRbClass(U_LONG mx, char *str, VALUE klass);
|
||||
VpNewRbClass(size_t mx, char *str, VALUE klass);
|
||||
|
||||
VP_EXPORT Real *VpCreateRbObject(U_LONG mx,const char *str);
|
||||
VP_EXPORT Real *VpCreateRbObject(size_t mx,const char *str);
|
||||
|
||||
static inline BDIGIT
|
||||
rmpd_base_value(void) { return RMPD_BASE; }
|
||||
|
@ -148,8 +172,8 @@ VP_EXPORT double VpGetDoubleNegInf(void);
|
|||
VP_EXPORT double VpGetDoubleNegZero(void);
|
||||
|
||||
/* These 2 functions added at v1.1.7 */
|
||||
VP_EXPORT U_LONG VpGetPrecLimit(void);
|
||||
VP_EXPORT U_LONG VpSetPrecLimit(U_LONG n);
|
||||
VP_EXPORT size_t VpGetPrecLimit(void);
|
||||
VP_EXPORT size_t VpSetPrecLimit(size_t n);
|
||||
|
||||
/* Round mode */
|
||||
VP_EXPORT int VpIsRoundMode(unsigned long n);
|
||||
|
@ -160,33 +184,33 @@ VP_EXPORT int VpException(unsigned short f,const char *str,int always);
|
|||
#if 0 /* unused */
|
||||
VP_EXPORT int VpIsNegDoubleZero(double v);
|
||||
#endif
|
||||
VP_EXPORT U_LONG VpNumOfChars(Real *vp,const char *pszFmt);
|
||||
VP_EXPORT size_t VpNumOfChars(Real *vp,const char *pszFmt);
|
||||
VP_EXPORT size_t VpInit(BDIGIT BaseVal);
|
||||
VP_EXPORT void *VpMemAlloc(U_LONG mb);
|
||||
VP_EXPORT void *VpMemAlloc(size_t mb);
|
||||
VP_EXPORT void VpFree(Real *pv);
|
||||
VP_EXPORT Real *VpAlloc(U_LONG mx, const char *szVal);
|
||||
VP_EXPORT U_LONG VpAsgn(Real *c,Real *a,int isw);
|
||||
VP_EXPORT U_LONG VpAddSub(Real *c,Real *a,Real *b,int operation);
|
||||
VP_EXPORT U_LONG VpMult(Real *c,Real *a,Real *b);
|
||||
VP_EXPORT U_LONG VpDivd(Real *c,Real *r,Real *a,Real *b);
|
||||
VP_EXPORT Real *VpAlloc(size_t mx, const char *szVal);
|
||||
VP_EXPORT size_t VpAsgn(Real *c, Real *a, int isw);
|
||||
VP_EXPORT size_t VpAddSub(Real *c,Real *a,Real *b,int operation);
|
||||
VP_EXPORT size_t VpMult(Real *c,Real *a,Real *b);
|
||||
VP_EXPORT size_t VpDivd(Real *c,Real *r,Real *a,Real *b);
|
||||
VP_EXPORT int VpComp(Real *a,Real *b);
|
||||
VP_EXPORT S_LONG VpExponent10(Real *a);
|
||||
VP_EXPORT ssize_t VpExponent10(Real *a);
|
||||
VP_EXPORT void VpSzMantissa(Real *a,char *psz);
|
||||
VP_EXPORT int VpToSpecialString(Real *a,char *psz,int fPlus);
|
||||
VP_EXPORT void VpToString(Real *a,char *psz,int fFmt,int fPlus);
|
||||
VP_EXPORT void VpToFString(Real *a,char *psz,int fFmt,int fPlus);
|
||||
VP_EXPORT int VpCtoV(Real *a,const char *int_chr,U_LONG ni,const char *frac,U_LONG nf,const char *exp_chr,U_LONG ne);
|
||||
VP_EXPORT int VpVtoD(double *d,S_LONG *e,Real *m);
|
||||
VP_EXPORT void VpToString(Real *a, char *psz, size_t fFmt, int fPlus);
|
||||
VP_EXPORT void VpToFString(Real *a, char *psz, size_t fFmt, int fPlus);
|
||||
VP_EXPORT int VpCtoV(Real *a, const char *int_chr, size_t ni, const char *frac, size_t nf, const char *exp_chr, size_t ne);
|
||||
VP_EXPORT int VpVtoD(double *d, ssize_t *e, Real *m);
|
||||
VP_EXPORT void VpDtoV(Real *m,double d);
|
||||
#if 0 /* unused */
|
||||
VP_EXPORT void VpItoV(Real *m,S_INT ival);
|
||||
#endif
|
||||
VP_EXPORT int VpSqrt(Real *y,Real *x);
|
||||
VP_EXPORT int VpActiveRound(Real *y,Real *x,int f,S_LONG il);
|
||||
VP_EXPORT int VpMidRound(Real *y, int f, S_LONG nf);
|
||||
VP_EXPORT int VpLeftRound(Real *y, int f, S_LONG nf);
|
||||
VP_EXPORT void VpFrac(Real *y,Real *x);
|
||||
VP_EXPORT int VpPower(Real *y,Real *x,S_INT n);
|
||||
VP_EXPORT int VpActiveRound(Real *y, Real *x, int f, ssize_t il);
|
||||
VP_EXPORT int VpMidRound(Real *y, int f, ssize_t nf);
|
||||
VP_EXPORT int VpLeftRound(Real *y, int f, ssize_t nf);
|
||||
VP_EXPORT void VpFrac(Real *y, Real *x);
|
||||
VP_EXPORT int VpPower(Real *y, Real *x, SIGNED_VALUE n);
|
||||
|
||||
/* VP constants */
|
||||
VP_EXPORT Real *VpOne(void);
|
||||
|
@ -209,7 +233,7 @@ VP_EXPORT Real *VpOne(void);
|
|||
/* VpGetSign(a) returns 1,-1 if a>0,a<0 respectively */
|
||||
#define VpGetSign(a) (((a)->sign>0)?1:(-1))
|
||||
/* Change sign of a to a>0,a<0 if s = 1,-1 respectively */
|
||||
#define VpChangeSign(a,s) {if((s)>0) (a)->sign=(short)Abs((S_LONG)(a)->sign);else (a)->sign=-(short)Abs((S_LONG)(a)->sign);}
|
||||
#define VpChangeSign(a,s) {if((s)>0) (a)->sign=(short)Abs((ssize_t)(a)->sign);else (a)->sign=-(short)Abs((ssize_t)(a)->sign);}
|
||||
/* Sets sign of a to a>0,a<0 if s = 1,-1 respectively */
|
||||
#define VpSetSign(a,s) {if((s)>0) (a)->sign=(short)VP_SIGN_POSITIVE_FINITE;else (a)->sign=(short)VP_SIGN_NEGATIVE_FINITE;}
|
||||
|
||||
|
@ -247,5 +271,4 @@ VP_EXPORT int VPrint(FILE *fp,const char *cntl_chr,Real *a);
|
|||
#if defined(__cplusplus)
|
||||
} /* extern "C" { */
|
||||
#endif
|
||||
|
||||
#endif /* RUBY_BIG_DECIMAL_H */
|
||||
|
|
|
@ -1,3 +1,6 @@
|
|||
require 'mkmf'
|
||||
|
||||
have_func("labs", "stdlib.h")
|
||||
have_func("llabs", "stdlib.h")
|
||||
|
||||
create_makefile('bigdecimal')
|
||||
|
|
|
@ -25,7 +25,7 @@ class TestBigDecimal < Test::Unit::TestCase
|
|||
assert_equal( 1, BigDecimal.new("Infinity").infinite?)
|
||||
assert_equal(-1, BigDecimal.new("-Infinity").infinite?)
|
||||
assert_equal(true, BigDecimal.new("NaN").nan?)
|
||||
assert_equal( 1, BigDecimal.new("1E11111111111").infinite?)
|
||||
assert_equal( 1, BigDecimal.new("1E1111111111111111111").infinite?)
|
||||
end
|
||||
|
||||
def _test_mode(type)
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue