tsbockman
9da546b2a6
core.bitop bsr(), bsf(), and popcnt() now support ulong inputs even on a 32-bit arch.
...
This also fixes a few small issues:
* In std.numeric, `Ulong union` didn't consider endianness, and
* in `roundDownToPowerOf2()`, this expression: (1 << bsr(num)) is always 32-bit,
whereas it should, at a minimum, match the bit-ness of num.
* ilogb() supports integer inputs, but this was undocumented.
2016-01-13 13:54:27 -08:00
tsbockman
4b690ef72f
Added a slower generic fallback method, in case real.mant_dig > 64.
2015-12-05 15:39:48 -08:00
tsbockman
7929a45c1f
Optimized slightly and made the unittests more generic in case real.mant_dig < 64.
2015-12-05 15:38:31 -08:00
tsbockman
01860999a9
Added static assert documenting the maximum precision for which this fix is valid.
2015-12-05 15:38:31 -08:00
tsbockman
49d4720e5c
Fix Issue 14786 - std.math.pow sometimes gets the sign of the result wrong.
2015-12-05 15:38:31 -08:00
Infiltrator
91c38b4bfd
fix issue 9717 - std.math.round rounds away from zero instead of to the nearest even integer
2015-12-03 03:56:37 +11:00
Iain Buclaw
edba3c09f9
Adjust expected result of test to match ARM and others
2015-11-19 10:12:40 +01:00
Walter Bright
cf22992cbb
Merge pull request #3768 from CyberShadow/pull-20151024-051829
...
HTML fixes
2015-10-24 15:05:00 -07:00
Vladimir Panteleev
86cf380007
HTML fixes
2015-10-24 06:19:23 +00:00
Dragos Carp
d698887729
Remove obsolete TypeTuple references
...
Replace following names:
std.typetuple -> std.meta
TypeTuple -> AliasSeq
ParameterTypeTuple -> Parameters
FieldTypeTuple -> Fields
std.traits requires more work than search/replace and is left unchanged.
2015-10-13 20:37:44 +02:00
Andrei Alexandrescu
8b974d2b49
Undo inadvertent changes to std/math.d and std/traits.d, delete std/typed_allocator.d
2015-10-02 07:35:09 -04:00
Martin
c1f00fe3a2
Fix compile error in std.math.exp(real) for non-x86 platforms
2015-09-26 14:35:51 +02:00
Dmitry Olshansky
f2220c2139
Merge pull request #3022 from ivan-timokhin/totalOrder
...
Total ordering for floating-point values.
2015-09-02 12:39:55 +03:00
Daniel Murphy
8cbacba491
Forward std.math intrinsics to core.math
2015-08-30 17:34:07 +10:00
Martin Nowak
14a19c2645
Merge remote-tracking branch 'upstream/stable' into merge_stable
2015-08-05 01:34:36 +02:00
Martin Nowak
bd8ed8c13c
fix Issue 14842 - [REG 2.068-b2] approxEqual does not work with integers
...
- promote any comparision with integers to real
2015-08-04 23:57:41 +02:00
Johan Engelen
321f8d7493
Add integer ilogb
2015-08-02 23:46:31 +02:00
Johan Engelen
98638abfc4
Add a D implementation for std.math.ilogb, templated for different floating point types.
2015-07-29 19:10:07 +02:00
Martin Nowak
5ca7ddad60
Merge remote-tracking branch 'upstream/stable' into merge_stable
...
Conflicts:
posix.mak
Manually edited:
std/datetime.d (rename toNormalizedPath -> asNormalizedPath)
2015-07-24 01:28:58 +02:00
Iain Buclaw
73493bbb49
Fix-up coefficients and smallest denormal value for ieeeDouble in exp()
2015-07-23 08:02:11 +02:00
Ivan Timokhin
9f5aa69050
Remove unnecessary empty line
2015-07-16 12:05:30 +03:00
Ivan Timokhin
baf1482936
Relax unittests
2015-07-16 11:40:12 +03:00
Ivan Timokhin
5786fe47a1
Extend testing
2015-07-16 00:59:48 +03:00
Ivan Timokhin
9755253926
Additional braces
2015-07-16 00:53:12 +03:00
Ivan Timokhin
e1a2115bbe
Change arrays into tuples to force loop unrolling
2015-07-16 00:51:32 +03:00
Ivan Timokhin
96227bde13
Rewrite to use union repainting instead of pointer casts
2015-07-16 00:01:03 +03:00
Ivan Timokhin
54edaa1351
Change parameter types to const to avoid template bloat.
2015-07-15 23:21:13 +03:00
Ivan Timokhin
87a4b107d1
Fix pointer conversion in 'real' branch
2015-07-15 23:14:52 +03:00
Ivan Timokhin
e1a80d7cf1
Relax return value requirements in the ddoc comment
2015-07-15 23:13:16 +03:00
Ivan Timokhin
24129bcbbb
Rewrite test runner for consistency.
2015-07-15 22:46:17 +03:00
Ivan Timokhin
68762b746e
Got rid of if-else nesting.
2015-07-15 22:46:16 +03:00
Ivan Timokhin
843d14f913
Styling.
...
Removed ?: chains and braces around single statements.
2015-07-15 22:46:16 +03:00
Ivan Timokhin
1de499f0ff
Changed to do three-way comparison.
2015-07-15 22:46:16 +03:00
Ivan Timokhin
cb57d23089
Total ordering for floating-point values.
2015-07-15 22:46:16 +03:00
Dmitry Olshansky
db4d872d6e
Merge pull request #3011 from 9il/float-traits
...
std.math: №1, change protection
2015-07-15 21:31:02 +03:00
Martin Nowak
eaeeccd4ac
workaround Issue 14718 - float parsing depends on platform strtold
...
- comment out the affected unittests for now
2015-06-21 16:36:37 +02:00
Andrei Alexandrescu
a0ccc918d6
Merge pull request #3351 from rainers/trail_whitespace
...
remove trailing whitespace, detab, tolf
2015-05-31 19:12:25 -07:00
Rainer Schuetze
adab223837
remove trailing whitespace, detab, tolf
2015-05-31 11:24:01 +02:00
Daniel Murphy
ebc90dc149
Merge pull request #2657 from joakim-noah/separate_glibc
...
Separate linux kernel, glibc, and bionic APIs where appropriate
2015-05-30 08:22:30 -06:00
Robert burner Schadek
a3fc53e601
math: alias syntax update
2015-05-28 18:52:01 +02:00
Joakim
3352c90c2e
Separate linux kernel, glibc, and bionic APIs where appropriate
2015-05-27 21:54:52 -05:00
Clement Courbet
e5613663dc
Make isNan and isInfinite available at compile time.
2015-05-22 14:34:28 +02:00
David Nadlinger
f640df12e4
Merge pull request #3298 from smolt/arm-fpscr
...
ARM FPSCR bit corrections
2015-05-20 15:42:49 +02:00
Dan Olson
cc8a3e1e0d
ARM FPSCR bit corrections
...
Correct IeeeFlags exception masks and FloatingPointControl
RoundingModes for ARM.
2015-05-19 23:24:15 -07:00
Walter Bright
e1f1838a52
Merge pull request #3287 from BBasile/fix-misleading-math-isSomething
...
fix several misleading ddocs and return types in the function isXXX()
2015-05-19 20:23:51 -07:00
Martin
70e5e942a0
std.math: proper support for 64-bit real in exp()
2015-05-17 20:30:43 +02:00
Martin
a514ac9686
std.math: refactor feqrel() incl. subnormal fix
...
For subnormal diffs, the old code used the whole ushort containing
the exponent bits as exponent, without masking and shifting. This
seems to have worked for 80-bit reals so far and the single unit-
test was only enabled for x87 reals.
2015-05-17 20:30:43 +02:00
Daniel Murphy
972c9485b6
Merge pull request #3293 from redstar/mathsqrt
...
Add additional tests for std.math.sqrt()
2015-05-17 23:48:11 +10:00
kai
d29d5d4130
Add additional tests for std.math.sqrt()
...
These tests spotted an implementation issue of LDC.
Provided by KevinBrogan.
2015-05-17 14:18:46 +02:00
kai
d3c6757286
Add missing EXPBIAS value for 128bit floats.
...
Found this source for the value: http://en.wikipedia.org/wiki/Quadruple-precision_floating-point_format
2015-05-17 13:51:20 +02:00
