[ruby-core:65975] [ruby-trunk - Feature #10354] Optimize Integer#prime?

Issue #10354 has been updated by Marc-Andre Lafortune.


Dear Yugui,

Do you have an objection that I commit my fixes (which are really bug fixes) while you decide if you want to go in the direction of Nick's optimization?

Thanks


----------------------------------------
Feature #10354: Optimize Integer#prime?
https://bugs.ruby-lang.org/issues/10354#change-49711

* Author: Marc-Andre Lafortune
* Status: Open
* Priority: Normal
* Assignee: Yuki Sonoda
* Category: lib
* Target version: current: 2.2.0
----------------------------------------
Nick Slocum shows in https://github.com/ruby/ruby/pull/736 that Integer#prime? can be optimized quite a bit.

First, that's because there are some basic things to avoid in the current lib, like needlessly capturing blocks and there's a useless `loop do` too.

I'm attaching a patch that fixes many of these things.

Even after these fixes applied, Nick's version is still faster and I don't see why we would not use it for Fixnum#prime?

For Bignum#prime?, since division costs more, we can go slightly faster with the following implementation:

    class Integer
      # Returns true if +self+ is a prime number, else returns false.
      def prime?
        return true if self == 2
        return false if self % 2 == 0 || self % 3 == 0 || self < 2
        skip_division = true
        (5..(self**0.5).floor).step(2) do |i|
          return false if skip_division && self % i == 0
          skip_division = !skip_division
        end
        true
      end
    end


---Files--------------------------------
prime_opt.patch (1.55 KB)


-- 
https://bugs.ruby-lang.org/