📊 Quantitative test results for language: eng, year: 2023, size: 10K, paranoia level: 1:
🚀 Quantitative testing did not detect new false positives

I'm curious why you think this would have an impact on performance. Did you measure any difference?

I'm curious why you think this would have an impact on performance. Did you measure any difference?

I'm not sure it's measurable, I mean the difference is not significant at all. I tried that with msc_retest on these ways:

$ cat r933151_old.txt 
\b([^\s]+)\s*[(]

$ cat r933151_new.txt 
\b([^\s]+)\s*\(

Please note that there is no EOL at the end of lines!

Then here are the commands:

echo -n "array_diff (" | src/pcre4msc2 -j -n 100000 r933151_old.txt 
echo -n "array_diff (" | src/pcre4msc2 -j -n 100000 r933151_new.txt 
echo -n "array_diff (" | src/pcre4msc3 -n 100000 r933151_old.txt 
echo -n "array_diff (" | src/pcre4msc3 -n 100000 r933151_new.txt 

These runs test with v2's and v3's regex engine with old and new patterns against the mentioned rule's 1st regression test. In case of v2 the regex engine (PCRE2 is the default) uses JIT. v3's regex code uses JIT by default.

The results are below:

v2/old pattern

Num of values: 100000
         Mean: 000.000000352
       Median: 000.000000324
          Min: 000.000000171
          Max: 000.000132326
        Range: 000.000132155
Std deviation: 000.000000519

v2/new pattern

Num of values: 100000
         Mean: 000.000000435
       Median: 000.000000347
          Min: 000.000000169
          Max: 000.000058875
        Range: 000.000058706
Std deviation: 000.000000507

v3/old pattern

Num of values: 100000
         Mean: 000.000000230
       Median: 000.000000218
          Min: 000.000000111
          Max: 000.000051658
        Range: 000.000051547
Std deviation: 000.000000281

v3/new pattern

Num of values: 100000
         Mean: 000.000000227
       Median: 000.000000213
          Min: 000.000000112
          Max: 000.000026295
        Range: 000.000026183
Std deviation: 000.000000194

I just leave this comment here to help anyone who wants to compare the regex patterns' performance, use msc_retest.

Hey, apologies i have a couple more prs coming in relation to the same regex (and 933151), which overall improves our performance between 48 to 112 times (i dont think the difference will be as pronounced for you guys, especially with pcre, however in rust regex not extracting from capture groups and just matching increases performance massively).
Specifically just this change improved our regex compile times very slightly, though not a noticeable amount.

Interesting, thanks for the explanation. Although I would have expected the engine to automatically optimise [(] to \( ;)