r57789 (74cdd89) was gradually "improve"d by naruse through r57793 to
r57806, resulted in reverting the efect of r57789 while retaining its
complexity. I think the current situation is slightly worse than
before (same output complicated source code).
Here I introduce __builtin_add_overflow again, which (I think) is what
naruse wanted to do in r57793.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57807 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Looking at the source code, FIXABLE tends to be just before LOING2FIX
to check applicability of that operation. Why not try computing first
then check for overflow, which should be optimial.
I also tried the same thing for unsigned types but resulted in slower
execution. It seems RB_POSFIXABLE() is fast enough on modern CPUs.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57789 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
I was not aware of cases when LONG_LONG_MAX is undefined. Sorry.
Signed-off-by: Urabe, Shyouhei <shyouhei@ruby-lang.org>
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57786 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_insnhelper.c (opt_eq_func): method to dispatch is resolved by
only the receiver's class, not including the argument class.
even if basic operation is redefined, other class conditions
never meet. optimize Float and non-Float case, delegate to
rb_float_equal directly.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57785 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
We already check for __builtin_mul_overflow in configure but never
actually referred it before. Why not call it if available, because
that should render supposedly-optimial assembly outputs.
Optionally if __builtin_mul_overflow_p is available, which is the case
for recent GCC, use that to detect fixnum overflow. This is much
faster than the previous. On my machine generated assembly of
numeric.c:int_pow reduces from 480 to 448 bytes, according to nm(1).
Also on my machine, following script boosts from 7.819 to 6.929 sec.
time ./miniruby -e 'i=0; while i < 30_000_000 do i += 1; 7 ** 23; end'
Signed-off-by: Urabe, Shyouhei <shyouhei@ruby-lang.org>
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57784 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* eval_error.c (rb_threadptr_error_print): print backtrace and
error message in reverse order if STDERR is unchanged and a tty.
[Feature #8661]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57685 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* numeric.c (num_clone, num_dup): no longer raises TypeError,
returns the receiver instead as well as Integer and Float.
[ruby-core:79636] [Bug #13237]
* object.c (rb_immutable_obj_clone): immutable object clone with
freeze optional keyword argument.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57682 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* internal.h (rb_compile_error_str): remove declaration of removed
internal function at r54189.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57670 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* internal.h (rb_overflowed_fix_to_int): invert sign bit. should
not set LSB of fixnum value, which is always set, to MSB.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57509 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
NOTE:
(1) Fixnum's LSB is always 1.
It means you can always run `x - 1` without overflow.
(2) Of course `z = x + (y-1)` may overflow.
Now z's LSB is always 1, and the MSB of true result is also 1.
You can get true result in long as `(1<<63)|(z>>1)`,
and it equals to `(z<<63)|(z>>1)` == `ror(z)`.
GCC and Clang have __builtin_add_ovewflow:
* https://gcc.gnu.org/onlinedocs/gcc/Integer-Overflow-Builtins.html
* https://clang.llvm.org/docs/LanguageExtensions.html#checked-arithmetic-builtins
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57506 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* string.c (STR_IS_SHARED_M): new flag to mark shared mulitple times
(STR_SET_SHARED): set STR_IS_SHARED_M
(rb_str_tmp_frozen_acquire, rb_str_tmp_frozen_release): new functions
(str_new_frozen): set/unset STR_IS_SHARED_M as appropriate
* internal.h: declare new functions
* io.c (fwrite_arg, fwrite_do, fwrite_end): new
(io_fwrite): use new functions
Introduce rb_str_tmp_frozen_acquire and rb_str_tmp_frozen_release
to manage a hidden, frozen string. Reuse one bit of the embed
length for shared strings as STR_IS_SHARED_M to indicate a string
has been shared multiple times. In the common case, the string
is only shared once so the object slot can be reclaimed immediately.
minimum results in each 3 measurements. (time and size)
Execution time (sec)
name trunk built
io_copy_stream_write 0.682 0.254
io_copy_stream_write_socket 1.225 0.751
Speedup ratio: compare with the result of `trunk' (greater is better)
name built
io_copy_stream_write 2.680
io_copy_stream_write_socket 1.630
Memory usage (last size) (B)
name trunk built
io_copy_stream_write 95436800.000 6512640.000
io_copy_stream_write_socket 117628928.000 7127040.000
Memory consuming ratio (size) with the result of `trunk' (greater is better)
name built
io_copy_stream_write 14.654
io_copy_stream_write_socket 16.505
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57469 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* internal.h: Microsoft Visual C++ has never supported C99 yet,
even in 2017.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57462 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
17+ years passed since standardized in ISO, 8 years since we added
AC_HEADER_STDBOOL to configure.in. I'm quite confident that it's
already safe to use <stdbool.h>.
I understand that when we introduced AC_HEADER_STDBOOL, <stdbool.h>
was remain not included because C standard and SVR4 curses conflicted
miserably back then (#1). Though I believe such situation has been
fixed already(#2), I'm afraid of your operating system might ship a
proprietary curses that still conflicts with the standard. So to avoid
potential problem, we limit the inclusion to our internal use only.
#1 : 1997 version of SUSv2 said bool is "defined though typedef" in
<curses.h>, while C99 said bool is a macro, plus in C++ bool is a
keyword. AFASIK the curses library has never been a part of
POSIX.
#2 : In reality ncurses and NetBSD curses both just follow C99 to
include <stdbool.h> from <curses.h>. I think C99 is now widely
adopted.
----
* internal.h: #include <stdbool.h> if present. That is
believed to be the case for 99.9% systems that lives today.
Non-C99, non-C++ situations are intentionally left
undefined, advised by Motohiro Kosaki. If you have such
compiler, please fill the definition appropriately.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57460 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* eval.c (setup_exception): make unfrozen copy of special
exception before setting up a cause.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57415 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* file.c (rb_get_path_check_convert): refine the error message
when the path name contains null byte.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57336 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Patch by Eric Wong [ruby-core:78797].
I don't like the idea of making insns.def any bigger to support
a corner case, and "test_hash_aref_fstring_identity" shows
how contrived this is.
[ruby-core:78783] [Bug #12855]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57278 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* numeric.c (rb_int_fdiv_double): reduce first for more precise
result. [ruby-core:78886] [Bug #13078]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57227 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* string.c (rb_str_match_m_p): inverse of Regexp#match?. based on
the patch by Herwin Weststrate <herwin@snt.utwente.nl>.
[Fix GH-1483] [Feature #12898]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57053 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* complex.c (f_gt_p): optimize f_gt_p for specific types of arguments.
* internal.h (rb_int_gt, rb_float_gt, rb_rational_cmp): exported.
* numeric.c (rb_float_gt): rename from flo_gt and be exported.
* numeric.c (rb_int_gt): rename from int_gt and be exported.
* rational.c (rb_rational_cmp): rename from nurat_cmp and be exported.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56872 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* complex.c (f_negate): optimize for special numeric types.
* complex.c (nucomp_expt): use rb_int_uminus instead of f_negate for
fixnum value.
* internal.h (rb_float_uminus, rb_rational_uminus): exported.
* numeric.c (rb_float_uminus): rename from flo_uminus.
* rational.c (rb_rational_uminus): rename from nurat_negate, and add
assertion for the parameter.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56830 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* complex.c (numeric_polar): optimize for Integer, Float, and Rational.
* complex.c (numeric_arg): directly create the value of pi.
* complex.c (f_negative_p): optimize for Integer, Float, and Rational.
* rational.c (INT_NEGATIVE_P): move the definition into internal.h.
* internal.h (INT_NEGATIVE_P): ditto.
* numeric.c (rb_float_abs): rename from flo_abs and export to be used
from other source files..
* internal.h (rb_float_abs): ditto.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56809 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* rational.c (nurat_to_double): introduce to convert rational to double
without object allocation.
* rational.c (rb_rational_plus, nurat_{sub,mul,to_f}): rewrite by using
nurat_to_double.
* bignum.c (rb_big_fdiv_double): introduce to calculate fdiv and return
the result as a double value.
* bignum.c (big_fdiv{,_int,_float}): change the return types for
implementing rb_big_fdiv_double.
* bignum.c (rb_big_fdiv): rewrite by using rb_big_fdiv_double.
* numeric.c (rb_int_fdiv_double): introduce to calculate fdiv and return
the result as a double value.
* numeric.c (fix_fdiv_double): rewrite from fix_fdiv to return the
result as a double value.
* numeric.c (rb_int_fdiv): rewrite by using rb_int_fdiv_double.
* internal.h (rb_{big,int}_fdiv_double): exported.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56719 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
[Feature #12142]
See header of st.c for improvment details.
You can see all of code history here:
<https://github.com/vnmakarov/ruby/tree/hash_tables_with_open_addressing>
This improvement is discussed at
<https://bugs.ruby-lang.org/issues/12142>
with many people, especially with Yura Sokolov.
* st.c: improve st_table.
* include/ruby/st.h: ditto.
* internal.h, numeric.c, hash.c (rb_dbl_long_hash): extract a function.
* ext/-test-/st/foreach/foreach.c: catch up this change.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56650 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* numeric.c (flo_round, int_round): support round-to-nearest-even
semantics of IEEE 754 to match sprintf behavior, and add `half:`
optional keyword argument for the old behavior.
[ruby-core:76273] [Bug #12548]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56590 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* class.c (rb_undef_methods_from): undefine methods defined in
super from klass.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56482 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
a hash value of Object might be Bignum, but it causes many troubles
expecially the Object is used as a key of a hash. so I've gave up
to do so.
* array.c (rb_ary_hash): use above macro.
* bignum.c (rb_big_hash): ditto.
* hash.c (rb_obj_hash, rb_hash_hash): ditto.
* numeric.c (rb_dbl_hash): ditto.
* proc.c (proc_hash): ditto.
* re.c (rb_reg_hash, match_hash): ditto.
* string.c (rb_str_hash_m): ditto.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56340 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Fix compile error with some non-GCC compilers such as
Oracle Solaris Studio 12.3 on Solaris 10.
[Bug #12767] [ruby-dev:49807]
* internal.h (WARN_UNUSED_RESULT): ditto.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56172 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
actually check its availability rather to check GCC's version.
* configure.in (WARN_UNUSED_RESULT): moved to here.
* configure.in (RUBY_FUNC_ATTRIBUTE): change function declaration
to return int rather than void, because it makes no sense for a
warn_unused_result attributed function to return void.
Funny thing however is that it also makes no sense for noreturn
attributed function to return int. So there is a fundamental
conflict between them. While I tested this, I confirmed both
GCC 6 and Clang 3.8 prefers int over void to correctly detect
necessary attributes under this setup. Maybe subject to change
in future.
* internal.h (UNINITIALIZED_VAR): renamed to MAYBE_UNUSED, then
moved to configure.in for the same reason we move
WARN_UNUSED_RESULT.
* configure.in (MAYBE_UNUSED): moved to here.
* internal.h (__has_attribute): deleted, because it has no use now.
* string.c (rb_str_enumerate_lines): refactor macro rename.
* string.c (rb_str_enumerate_bytes): ditto.
* string.c (rb_str_enumerate_chars): ditto.
* string.c (rb_str_enumerate_codepoints): ditto.
* thread.c (do_select): ditto.
* vm_backtrace.c (rb_debug_inspector_open): ditto.
* vsnprintf.c (BSD_vfprintf): ditto.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56169 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* variable.c (rb_deprecate_constant): new function to deprecate a
constant by the name.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56119 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* internal.h (MEMO_V1_SET, MEMO_V2_SET): fix typos. use the macro
parameter, not the local variable.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56059 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
