[macruby-changes] [4096] MacRuby/trunk
source_changes at macosforge.org
source_changes at macosforge.org
Thu May 13 18:58:31 PDT 2010
Revision: 4096
http://trac.macosforge.org/projects/ruby/changeset/4096
Author: martinlagardette at apple.com
Date: 2010-05-13 18:58:28 -0700 (Thu, 13 May 2010)
Log Message:
-----------
Improve core/math passing rate (backported ruby19 changes)
Modified Paths:
--------------
MacRuby/trunk/math.c
MacRuby/trunk/spec/frozen/tags/macruby/core/math/atanh_tags.txt
Removed Paths:
-------------
MacRuby/trunk/spec/frozen/tags/macruby/core/math/gamma_tags.txt
MacRuby/trunk/spec/frozen/tags/macruby/core/math/lgamma_tags.txt
Modified: MacRuby/trunk/math.c
===================================================================
--- MacRuby/trunk/math.c 2010-05-14 01:14:58 UTC (rev 4095)
+++ MacRuby/trunk/math.c 2010-05-14 01:58:28 UTC (rev 4096)
@@ -14,7 +14,12 @@
#include <errno.h>
VALUE rb_mMath;
+VALUE rb_eMathDomainError;
+#define numberof(array) (int)(sizeof(array) / sizeof((array)[0]))
+#define domain_error(msg) \
+ rb_raise(rb_eMathDomainError, "Numerical argument is out of domain - " #msg);
+
static VALUE
to_flo(VALUE x)
{
@@ -31,32 +36,12 @@
return rb_convert_type(x, T_FLOAT, "Float", "to_f");
}
-#define Need_Float(x) (x) = to_flo(x)
+#define Need_Float(x) do {if (TYPE(x) != T_FLOAT) {(x) = to_flo(x);}} while(0)
#define Need_Float2(x,y) do {\
Need_Float(x);\
Need_Float(y);\
} while (0)
-static void
-domain_check(double x, const char *msg)
-{
- while(1) {
- if (errno) {
- rb_sys_fail(msg);
- }
- if (isnan(x)) {
-#if defined(EDOM)
- errno = EDOM;
-#elif defined(ERANGE)
- errno = ERANGE;
-#endif
- continue;
- }
- break;
- }
-}
-
-
/*
* call-seq:
* Math.atan2(y, x) => float
@@ -69,8 +54,17 @@
VALUE
math_atan2(VALUE obj, SEL sel, VALUE y, VALUE x)
{
+ double dx, dy;
Need_Float2(y, x);
- return DOUBLE2NUM(atan2(RFLOAT_VALUE(y), RFLOAT_VALUE(x)));
+ dx = RFLOAT_VALUE(x);
+ dy = RFLOAT_VALUE(y);
+ if (dx == 0.0 && dy == 0.0) {
+ domain_error("atan2");
+ }
+ if (isinf(dx) && isinf(dy)) {
+ domain_error("atan2");
+ }
+ return DBL2NUM(atan2(dy, dx));
}
@@ -131,13 +125,16 @@
static VALUE
math_acos(VALUE obj, SEL sel, VALUE x)
{
- double d;
+ double d0, d;
Need_Float(x);
- errno = 0;
- d = acos(RFLOAT_VALUE(x));
- domain_check(d, "acos");
- return DOUBLE2NUM(d);
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (d0 < -1.0 || 1.0 < d0) {
+ domain_error("acos");
+ }
+ d = acos(d0);
+ return DBL2NUM(d);
}
/*
@@ -150,13 +147,16 @@
static VALUE
math_asin(VALUE obj, SEL sel, VALUE x)
{
- double d;
+ double d0, d;
Need_Float(x);
- errno = 0;
- d = asin(RFLOAT_VALUE(x));
- domain_check(d, "asin");
- return DOUBLE2NUM(d);
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (d0 < -1.0 || 1.0 < d0) {
+ domain_error("asin");
+ }
+ d = asin(d0);
+ return DBL2NUM(d);
}
/*
@@ -252,13 +252,16 @@
static VALUE
math_acosh(VALUE obj, SEL sel, VALUE x)
{
- double d;
+ double d0, d;
Need_Float(x);
- errno = 0;
- d = acosh(RFLOAT_VALUE(x));
- domain_check(d, "acosh");
- return DOUBLE2NUM(d);
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (d0 < 1.0) {
+ domain_error("acosh");
+ }
+ d = acosh(d0);
+ return DBL2NUM(d);
}
/*
@@ -285,13 +288,19 @@
static VALUE
math_atanh(VALUE obj, SEL sel, VALUE x)
{
- double d;
+ double d0, d;
Need_Float(x);
- errno = 0;
- d = atanh(RFLOAT_VALUE(x));
- domain_check(d, "atanh");
- return DOUBLE2NUM(d);
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (d0 < -1.0 || +1.0 < d0) {
+ domain_error("atanh");
+ }
+ /* check for pole error */
+ if (d0 == -1.0) return DBL2NUM(-INFINITY);
+ if (d0 == +1.0) return DBL2NUM(+INFINITY);
+ d = atanh(d0);
+ return DBL2NUM(d);
}
/*
@@ -308,15 +317,6 @@
return DOUBLE2NUM(exp(RFLOAT_VALUE(x)));
}
-#if defined __CYGWIN__
-# include <cygwin/version.h>
-# if CYGWIN_VERSION_DLL_MAJOR < 1005
-# define nan(x) nan()
-# endif
-# define log(x) ((x) < 0.0 ? nan("") : log(x))
-# define log10(x) ((x) < 0.0 ? nan("") : log10(x))
-#endif
-
/*
* call-seq:
* Math.log(numeric) => float
@@ -331,18 +331,25 @@
math_log(VALUE rcv, SEL sel, int argc, VALUE *argv)
{
VALUE x, base;
- double d;
+ double d0, d;
rb_scan_args(argc, argv, "11", &x, &base);
Need_Float(x);
- errno = 0;
- d = log(RFLOAT_VALUE(x));
- if (!NIL_P(base)) {
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (d0 < 0.0) {
+ domain_error("log");
+ }
+ /* check for pole error */
+ if (d0 == 0.0) {
+ return DBL2NUM(-INFINITY);
+ }
+ d = log(d0);
+ if (argc == 2) {
Need_Float(base);
d /= log(RFLOAT_VALUE(base));
}
- domain_check(d, "log");
- return DOUBLE2NUM(d);
+ return DBL2NUM(d);
}
#ifndef log2
@@ -367,15 +374,20 @@
static VALUE
math_log2(VALUE obj, SEL sel, VALUE x)
{
- double d;
+ double d0, d;
Need_Float(x);
- errno = 0;
- d = log2(RFLOAT_VALUE(x));
- if (errno) {
- rb_sys_fail("log2");
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (d0 < 0.0) {
+ domain_error("log2");
}
- return DOUBLE2NUM(d);
+ /* check for pole error */
+ if (d0 == 0.0) {
+ return DBL2NUM(-INFINITY);
+ }
+ d = log2(d0);
+ return DBL2NUM(d);
}
/*
@@ -388,13 +400,20 @@
static VALUE
math_log10(VALUE obj, SEL sel, VALUE x)
{
- double d;
+ double d0, d;
Need_Float(x);
- errno = 0;
- d = log10(RFLOAT_VALUE(x));
- domain_check(d, "log10");
- return DOUBLE2NUM(d);
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (d0 < 0.0) {
+ domain_error("log10");
+ }
+ /* check for pole error */
+ if (d0 == 0.0) {
+ return DBL2NUM(-INFINITY);
+ }
+ d = log10(d0);
+ return DBL2NUM(d);
}
/*
@@ -407,13 +426,19 @@
VALUE
math_sqrt(VALUE obj, SEL sel, VALUE x)
{
- double d;
+ double d0, d;
Need_Float(x);
- errno = 0;
- d = sqrt(RFLOAT_VALUE(x));
- domain_check(d, "sqrt");
- return DOUBLE2NUM(d);
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (d0 < 0.0) {
+ domain_error("sqrt");
+ }
+ if (d0 == 0.0) {
+ return DBL2NUM(0.0);
+ }
+ d = sqrt(d0);
+ return DBL2NUM(d);
}
/*
@@ -522,50 +547,91 @@
*
* Calculates the gamma function of x.
*
- * Note that gamma(n) is same as fact(n-1) for integer n >= 0.
- * However gamma(n) returns float and possibly has error in calculation.
+ * Note that gamma(n) is same as fact(n-1) for integer n > 0.
+ * However gamma(n) returns float and can be an approximation.
*
* def fact(n) (1..n).inject(1) {|r,i| r*i } end
- * 0.upto(25) {|i| p [i, Math.gamma(i+1), fact(i)] }
- * =>
- * [0, 1.0, 1]
- * [1, 1.0, 1]
- * [2, 2.0, 2]
- * [3, 6.0, 6]
- * [4, 24.0, 24]
- * [5, 120.0, 120]
- * [6, 720.0, 720]
- * [7, 5040.0, 5040]
- * [8, 40320.0, 40320]
- * [9, 362880.0, 362880]
- * [10, 3628800.0, 3628800]
- * [11, 39916800.0, 39916800]
- * [12, 479001599.999999, 479001600]
- * [13, 6227020800.00001, 6227020800]
- * [14, 87178291199.9998, 87178291200]
- * [15, 1307674368000.0, 1307674368000]
- * [16, 20922789888000.0, 20922789888000]
- * [17, 3.55687428096001e+14, 355687428096000]
- * [18, 6.40237370572799e+15, 6402373705728000]
- * [19, 1.21645100408832e+17, 121645100408832000]
- * [20, 2.43290200817664e+18, 2432902008176640000]
- * [21, 5.10909421717094e+19, 51090942171709440000]
- * [22, 1.12400072777761e+21, 1124000727777607680000]
- * [23, 2.58520167388851e+22, 25852016738884976640000]
- * [24, 6.20448401733239e+23, 620448401733239439360000]
- * [25, 1.5511210043331e+25, 15511210043330985984000000]
+ * 1.upto(26) {|i| p [i, Math.gamma(i), fact(i-1)] }
+ * #=> [1, 1.0, 1]
+ * # [2, 1.0, 1]
+ * # [3, 2.0, 2]
+ * # [4, 6.0, 6]
+ * # [5, 24.0, 24]
+ * # [6, 120.0, 120]
+ * # [7, 720.0, 720]
+ * # [8, 5040.0, 5040]
+ * # [9, 40320.0, 40320]
+ * # [10, 362880.0, 362880]
+ * # [11, 3628800.0, 3628800]
+ * # [12, 39916800.0, 39916800]
+ * # [13, 479001600.0, 479001600]
+ * # [14, 6227020800.0, 6227020800]
+ * # [15, 87178291200.0, 87178291200]
+ * # [16, 1307674368000.0, 1307674368000]
+ * # [17, 20922789888000.0, 20922789888000]
+ * # [18, 355687428096000.0, 355687428096000]
+ * # [19, 6.402373705728e+15, 6402373705728000]
+ * # [20, 1.21645100408832e+17, 121645100408832000]
+ * # [21, 2.43290200817664e+18, 2432902008176640000]
+ * # [22, 5.109094217170944e+19, 51090942171709440000]
+ * # [23, 1.1240007277776077e+21, 1124000727777607680000]
+ * # [24, 2.5852016738885062e+22, 25852016738884976640000]
+ * # [25, 6.204484017332391e+23, 620448401733239439360000]
+ * # [26, 1.5511210043330954e+25, 15511210043330985984000000]
*
*/
static VALUE
math_gamma(VALUE obj, SEL sel, VALUE x)
{
- double d;
+ static const double fact_table[] = {
+ /* fact(0) */ 1.0,
+ /* fact(1) */ 1.0,
+ /* fact(2) */ 2.0,
+ /* fact(3) */ 6.0,
+ /* fact(4) */ 24.0,
+ /* fact(5) */ 120.0,
+ /* fact(6) */ 720.0,
+ /* fact(7) */ 5040.0,
+ /* fact(8) */ 40320.0,
+ /* fact(9) */ 362880.0,
+ /* fact(10) */ 3628800.0,
+ /* fact(11) */ 39916800.0,
+ /* fact(12) */ 479001600.0,
+ /* fact(13) */ 6227020800.0,
+ /* fact(14) */ 87178291200.0,
+ /* fact(15) */ 1307674368000.0,
+ /* fact(16) */ 20922789888000.0,
+ /* fact(17) */ 355687428096000.0,
+ /* fact(18) */ 6402373705728000.0,
+ /* fact(19) */ 121645100408832000.0,
+ /* fact(20) */ 2432902008176640000.0,
+ /* fact(21) */ 51090942171709440000.0,
+ /* fact(22) */ 1124000727777607680000.0,
+ /* fact(23)=25852016738884976640000 needs 56bit mantissa which is
+ * impossible to represent exactly in IEEE 754 double which have
+ * 53bit mantissa. */
+ };
+ double d0, d;
+ double intpart, fracpart;
Need_Float(x);
- errno = 0;
- d = tgamma(RFLOAT_VALUE(x));
- domain_check(d, "gamma");
- return DOUBLE2NUM(d);
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (isinf(d0) && signbit(d0)) {
+ domain_error("gamma");
+ }
+ fracpart = modf(d0, &intpart);
+ if (fracpart == 0.0) {
+ if (intpart < 0) {
+ domain_error("gamma");
+ }
+ if (0 < intpart &&
+ intpart - 1 < (double)numberof(fact_table)) {
+ return DBL2NUM(fact_table[(int)intpart - 1]);
+ }
+ }
+ d = tgamma(d0);
+ return DBL2NUM(d);
}
/*
@@ -585,15 +651,22 @@
static VALUE
math_lgamma(VALUE obj, SEL sel, VALUE x)
{
- double d;
- int sign;
+ double d0, d;
+ int sign = 1;
VALUE v;
Need_Float(x);
- errno = 0;
- d = lgamma_r(RFLOAT_VALUE(x), &sign);
- domain_check(d, "lgamma");
- v = DOUBLE2NUM(d);
+ d0 = RFLOAT_VALUE(x);
+ /* check for domain error */
+ if (isinf(d0)) {
+ if (signbit(d0)) {
+ domain_error("lgamma");
+ }
+ return rb_assoc_new(DBL2NUM(INFINITY), INT2FIX(1));
+ }
+ d = lgamma_r(d0, &sign);
+ v = DBL2NUM(d);
return rb_assoc_new(v, INT2FIX(sign));
+
}
/*
@@ -608,6 +681,7 @@
Init_Math(void)
{
rb_mMath = rb_define_module("Math");
+ rb_eMathDomainError = rb_define_class_under(rb_mMath, "DomainError", rb_eStandardError);
#ifdef M_PI
rb_define_const(rb_mMath, "PI", DOUBLE2NUM(M_PI));
Modified: MacRuby/trunk/spec/frozen/tags/macruby/core/math/atanh_tags.txt
===================================================================
--- MacRuby/trunk/spec/frozen/tags/macruby/core/math/atanh_tags.txt 2010-05-14 01:14:58 UTC (rev 4095)
+++ MacRuby/trunk/spec/frozen/tags/macruby/core/math/atanh_tags.txt 2010-05-14 01:58:28 UTC (rev 4096)
@@ -1,5 +1,3 @@
-fails:Math.atanh raises an Errno::EDOM if x = 1.0
-fails:Math.atanh raises an Errno::EDOM if x = -1.0
fails:Math.atanh raises an Math::DomainError for arguments greater than 1.0
fails:Math.atanh raises an Math::DomainError for arguments less than -1.0
fails:Math#atanh raises an Math::DomainError for arguments greater than 1.0
Deleted: MacRuby/trunk/spec/frozen/tags/macruby/core/math/gamma_tags.txt
===================================================================
--- MacRuby/trunk/spec/frozen/tags/macruby/core/math/gamma_tags.txt 2010-05-14 01:14:58 UTC (rev 4095)
+++ MacRuby/trunk/spec/frozen/tags/macruby/core/math/gamma_tags.txt 2010-05-14 01:58:28 UTC (rev 4096)
@@ -1,12 +0,0 @@
-fails:Math.gamma returns NaN given NaN
-fails:Math.gamma returns 14! given 15
-fails:Math.gamma returns 15! given 16
-fails:Math.gamma returns 16! given 17
-fails:Math.gamma returns 17! given 18
-fails:Math.gamma returns 18! given 19
-fails:Math.gamma returns 19! given 20
-fails:Math.gamma returns 21! given 22
-fails:Math.gamma returns 22! given 23
-fails:Math.gamma raises Math::DomainError given -1
-fails:Math.gamma raises Math::DomainError given negative infinity
-fails:Math.gamma returns 20! given 21
Deleted: MacRuby/trunk/spec/frozen/tags/macruby/core/math/lgamma_tags.txt
===================================================================
--- MacRuby/trunk/spec/frozen/tags/macruby/core/math/lgamma_tags.txt 2010-05-14 01:14:58 UTC (rev 4095)
+++ MacRuby/trunk/spec/frozen/tags/macruby/core/math/lgamma_tags.txt 2010-05-14 01:58:28 UTC (rev 4096)
@@ -1,9 +0,0 @@
-fails:Math.lgamma returns an array of 2 elements
-fails:Math.lgamma returns known values
-fails:Math.lgamma returns correct values given +/- infinity
-fails:Math.lgamma returns correct value given NaN
-fails:Math.lgamma returns [Infinity, 1] when passed 0
-fails:Math.lgamma returns [Infinity, 1] when passed -1
-fails:Math.lgamma raises Math::DomainError when passed -Infinity
-fails:Math.lgamma returns [Infinity, 1] when passed Infinity
-fails:Math.lgamma returns [NaN, 1] when passed NaN
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.macosforge.org/pipermail/macruby-changes/attachments/20100513/c4598411/attachment-0001.html>
More information about the macruby-changes
mailing list