Project

General

Profile

Feature #12788

Improve String lstrip and strip performance

Added by asterite (Ary Borenszweig) about 1 year ago. Updated about 1 year ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Target version:
-
[ruby-core:77392]

Description

This adds a fast path for lstrip_offset in the case of single byte optimizable strings. This optimization already exists in rstrip_offset but I noticed it was missing in rstrip.

I made sure to only remove ascii space, not '\0', to preserve backwards compatibility. (rstrip seems to remove '\0', but not lstrip)

A benchmark:

s = " " * 10_000_000
time = Time.now
p s.lstrip
p s.strip
puts Time.now - time

s = " " * 100_000_000
time = Time.now
p s.lstrip
p s.strip
puts Time.now - time

Times before: 0.195159 and 1.918969
Times after: 0.013684 and 0.143966

So a bit more than 10x faster.

By the way, Akira Matsuda-san gave a really nice closing keynote in RubyConf Brasil yesterday that encouraged all of us to use Ruby and contribute to it, so here is my first patch :-)

string_lstrip_improvement.patch (685 Bytes) string_lstrip_improvement.patch asterite (Ary Borenszweig), 09/26/2016 01:02 AM

Associated revisions

Revision 56250
Added by nobu (Nobuyoshi Nakada) about 1 year ago

string.c: fast path of lstrip_offset

  • string.c (lstrip_offset): add a fast path in the case of single byte optimizable strings, as well as rstrip_offset. [Feature #12788]

Revision 56250
Added by nobu (Nobuyoshi Nakada) about 1 year ago

string.c: fast path of lstrip_offset

  • string.c (lstrip_offset): add a fast path in the case of single byte optimizable strings, as well as rstrip_offset. [Feature #12788]

History

#1 Updated by nobu (Nobuyoshi Nakada) about 1 year ago

  • Status changed from Open to Closed

Applied in changeset r56250.


string.c: fast path of lstrip_offset

  • string.c (lstrip_offset): add a fast path in the case of single byte optimizable strings, as well as rstrip_offset. [Feature #12788]

#2 [ruby-core:77405] Updated by asterite (Ary Borenszweig) about 1 year ago

Thank you Nobu-san for reviewing and accepting this. Matsuda-san also mentioned a "patch monster" in his talk, I think now I know what he means :-)

#3 [ruby-core:77406] Updated by matsuda (Akira Matsuda) about 1 year ago

Ary Borenszweig wrote:

Thank you Nobu-san for reviewing and accepting this. Matsuda-san also mentioned a "patch monster" in his talk, I think now I know what he means :-)

🤘🏻

Also available in: Atom PDF