[#118784] [Ruby master Feature#20664] Add `before` and `until` options to Enumerator.produce — "knu (Akinori MUSHA) via ruby-core" <ruby-core@...>

Issue #20664 has been reported by knu (Akinori MUSHA).

12 messages 2024/08/03

[#118791] [Ruby master Bug#20666] Segmentation fault instead of LoadError exception — "ErezGeva2@... (Erez Geva) via ruby-core" <ruby-core@...>

Issue #20666 has been reported by ErezGeva2@gmail.com (Erez Geva).

9 messages 2024/08/04

[#118811] [Ruby master Feature#20669] Add error classes to differentiate Marshal ArgumentErrors — "olleolleolle (Olle Jonsson) via ruby-core" <ruby-core@...>

Issue #20669 has been reported by olleolleolle (Olle Jonsson).

7 messages 2024/08/08

[#118844] [Ruby master Feature#20676] Pathnames aren't Comparable — "gmcgibbon (Gannon McGibbon) via ruby-core" <ruby-core@...>

SXNzdWUgIzIwNjc2IGhhcyBiZWVuIHJlcG9ydGVkIGJ5IGdtY2dpYmJvbiAoR2Fubm9uIE1jR2li

8 messages 2024/08/13

[#118879] [Ruby master Bug#20682] Slave PTY output is lost after a child process exits in macOS — "ono-max (Naoto Ono) via ruby-core" <ruby-core@...>

Issue #20682 has been reported by ono-max (Naoto Ono).

9 messages 2024/08/19

[#118932] [Ruby master Bug#20693] Dir.tmpdir should perform a real access check before warning about writability — "kjtsanaktsidis (KJ Tsanaktsidis) via ruby-core" <ruby-core@...>

Issue #20693 has been reported by kjtsanaktsidis (KJ Tsanaktsidis).

9 messages 2024/08/22

[#118979] [Ruby master Feature#20705] Should "0.E-9" be a valid float value? — "kou (Kouhei Sutou) via ruby-core" <ruby-core@...>

Issue #20705 has been reported by kou (Kouhei Sutou).

11 messages 2024/08/29

[#118983] [Ruby master Bug#20706] Can't build Ruby on macOS Sonoma and Sequoia due to: ignoring duplicate libraries, archive member '/' not a mach-o file in libruby.3.3-static.a — "wkoszek (Adam Koszek) via ruby-core" <ruby-core@...>

Issue #20706 has been reported by wkoszek (Adam Koszek).

7 messages 2024/08/29

[ruby-core:118998] [Ruby master Feature#20709] Improve String#rindex performance on OSX

From: "zack.ref@... (Zack Deveau) via ruby-core" <ruby-core@...>
Date: 2024-08-31 18:05:13 UTC
List: ruby-core #118998
Issue #20709 has been reported by zack.ref@gmail.com (Zack Deveau).

----------------------------------------
Feature #20709: Improve String#rindex performance on OSX
https://bugs.ruby-lang.org/issues/20709

* Author: zack.ref@gmail.com (Zack Deveau)
* Status: Open
----------------------------------------
`String#rindex` is much slower on OSX than on Linux hosts. This appears due to the lack of a `memrchr` implementation on OSX. The [fallback solution](https://github.com/ruby/ruby/blob/e69945fc57efba5c8895c21235e109145865952d/string.c#L4348-L4402) currently performs a `memcmp` on each character in the search string looking for a substring match.

```c
    while (s) {
        if (memcmp(s, t, slen) == 0) {
            return s - sbeg;
        }
        if (s <= sbeg) break;
        s = rb_enc_prev_char(sbeg, s, e, enc);
    }
```


```
ruby-master | 10 char substring (long) near the end of a search string | osx

search string enc: UTF-8
search string size: 8010

                        user     system      total        real
index (long substring)  0.000075   0.000000   0.000075 (  0.000074)
rindex                  0.000445   0.000000   0.000445 (  0.000448)
```

Proposed improvement is available on this pull request:



This patch proposes using an implementation of `memrchr` made for the interpreter on hosts that lack a native implementation. Below are benchmark results which demonstrate that this approach either improves or maintains performance of `String#rindex` for all cases we could come up with.

```c
static void*
rb_memrchr(const char *search_str, int chr, long search_len)
{
    const char *ptr = search_str + search_len;
    do {
       if (*(--ptr) == chr) return (void *)ptr;
    } while (ptr >= search_str);

    return ((void *)0);
}

```

Special thanks to @jhawthorn who provided some helpful feedback on this solution during development!

# Benchmarks

Each of the following were made with a variant of the following general purpose benchmark. Note that `String#index` speed can be [impacted by the size of the provided substring](https://github.com/ruby/ruby/blob/669d1f79d800d554918c7d5647b73a6431920ea8/re.c#L272-L274) (Denoted as long/short substring in the results, assume long substring when none is specified).
```
require "benchmark"

substring = "98279827AA"
the_string = ""
800.times{ the_string.concat(('a'..'z').to_a.shuffle[0,8].join)}
the_string.concat(substring)
200.times{ the_string.concat(('a'..'z').to_a.shuffle[0,8].join)}
puts the_string.encoding
puts the_string.size

Benchmark.bm do |x|
  x.report(:index) { 100.times { the_string.index(substring) } }
  x.report(:rindex) { 100.times { the_string.rindex(substring) } }
end
```

```
ruby-master | substring near the end | osx
https://github.com/ruby/ruby/blob/669d1f79d800d554918c7d5647b73a6431920ea8/re.c#L272-L274

UTF-8

       user     system      total        real
index  0.000075   0.000000   0.000075 (  0.000074)
rindex  0.000445   0.000000   0.000445 (  0.000448)
------------------------------------------------------
ruby-patched | substring near the end | osx
https://github.com/ruby/ruby/blob/669d1f79d800d554918c7d5647b73a6431920ea8/re.c#L272-L274

UTF-8
                        user     system      total        real
index (short substring) 0.000400   0.000001   0.000401 (  0.000399)
index (long substring)  0.000080   0.000000   0.000080 (  0.000080)
rindex                  0.000057   0.000001   0.000058 (  0.000057)
```


```
ruby-master | substring in the middle | osx

UTF-8
       user     system      total        real
index  0.000031   0.000000   0.000031 (  0.000030)
rindex  0.001086   0.000002   0.001088 (  0.001088)
------------------------------------------------------
ruby-patched | substring in the middle | osx

UTF-8
       user     system      total        real
index  0.000075   0.000000   0.000075 (  0.000075)
rindex  0.000130   0.000000   0.000130 (  0.000133)
```

```
ruby-master | substring in the middle with a short substring | osx
https://github.com/ruby/ruby/blob/669d1f79d800d554918c7d5647b73a6431920ea8/re.c#L272-L274

UTF-8
                        user     system      total        real
index (short substring) 0.000243   0.000004   0.000247 (  0.000246)
rindex                  0.001176   0.000005   0.001181 (  0.001182)
------------------------------------------------------
ruby-patched | substring in the middle with a short substring | osx
https://github.com/ruby/ruby/blob/669d1f79d800d554918c7d5647b73a6431920ea8/re.c#L272-L274

UTF-8
                         user     system      total        real
index (short substring)  0.000255   0.000005   0.000260 (  0.000259)
rindex                   0.000130   0.000000   0.000130 (  0.000133)
```

```
ruby-master | substring at the end | osx

UTF-8
       user     system      total        real
index  0.000049   0.000001   0.000050 (  0.000048)
rindex  0.000008   0.000001   0.000009 (  0.000007)
------------------------------------------------------
ruby-patched | substring at the end | osx

UTF-8
       user     system      total        real
index  0.000075   0.000000   0.000075 (  0.000074)
rindex  0.000006   0.000000   0.000006 (  0.000006)
```


```
ruby-master | substring at the start | osx

UTF-8
       user     system      total        real
index  0.000008   0.000001   0.000009 (  0.000007)
rindex  0.002444   0.000008   0.002452 (  0.002452)
------------------------------------------------------
ruby-patched | substring at the start | osx

UTF-8
       user     system      total        real
index  0.000015   0.000000   0.000015 (  0.000015)
rindex  0.000280   0.000001   0.000281 (  0.000280)
```



-- 
https://bugs.ruby-lang.org/
 ______________________________________________
 ruby-core mailing list -- ruby-core@ml.ruby-lang.org
 To unsubscribe send an email to ruby-core-leave@ml.ruby-lang.org
 ruby-core info -- https://ml.ruby-lang.org/mailman3/lists/ruby-core.ml.ruby-lang.org/


In This Thread

Prev Next