mirror of
https://github.com/ruby/ruby.git
synced 2026-08-03 13:08:59 +08:00
e73e4f2d4c
(https://github.com/ruby/strscan/pull/205) See also: https://bugs.ruby-lang.org/issues/21943 This is semantically equivalent to `scanner[specifier]&.to_i(base)` but this is faster than `scanner[specifier]&.to_i(base)` because `integer_at` doesn't create a temporary String when possible. This PR also includes a benchmark for them: ```console $ ruby -v -S benchmark-driver benchmark/integer_at.yaml ruby 4.1.0dev (2026-05-01T19:25:51Z master https://github.com/ruby/strscan/commit/f2845eab29) +PRISM [x86_64-linux] Warming up -------------------------------------- [].to_i 24.272M i/s - 25.109M times in 1.034481s (41.20ns/i, 32clocks/i) integer_at 61.188M i/s - 62.491M times in 1.021289s (16.34ns/i, 62clocks/i) Calculating ------------------------------------- [].to_i 26.831M i/s - 72.816M times in 2.713883s (37.27ns/i, 169clocks/i) integer_at 81.331M i/s - 183.564M times in 2.256998s (12.30ns/i, 43clocks/i) Comparison: integer_at: 81331225.5 i/s [].to_i: 26831046.3 i/s - 3.03x slower ``` In this environment, `integer_at` is 3.03x faster than `[].to_i`. https://github.com/ruby/strscan/commit/8a60879b2d Co-authored-by: jinroq <jinroq@gmail.com>
15 lines
522 B
Ruby
15 lines
522 B
Ruby
# frozen_string_literal: true
|
|
require 'mkmf'
|
|
if RUBY_ENGINE == 'ruby'
|
|
$INCFLAGS << " -I$(top_srcdir)" if $extmk
|
|
have_func("onig_region_memsize(NULL)")
|
|
have_func("rb_reg_onig_match", "ruby/re.h")
|
|
have_func("rb_deprecate_constant")
|
|
have_func("rb_int_parse_cstr", "ruby.h") # RUBY_VERSION >= 2.5
|
|
have_func("rb_gc_location", "ruby.h") # RUBY_VERSION >= 2.7
|
|
have_const("RUBY_TYPED_EMBEDDABLE", "ruby.h") # RUBY_VERSION >= 3.3
|
|
create_makefile 'strscan'
|
|
else
|
|
File.write('Makefile', dummy_makefile("").join)
|
|
end
|