Javascript regexes are "getting better" feature-wise, but performance-wise not much have improved. The the a?a?a? example regex in Ross Cox article "Regular Expression Matching Can Be Simple And Fast" from 2007 is still getting exponentially slower and slower with the number of "a?"-repetitions (https://swtch.com/~rsc/regexp/regexp1.html)

for (let i=20; i<30; i++) { let str = "a".repeat(i); let regex = new RegExp("a?".repeat(i) + "a".repeat(i)); let t0 = performance.now(); str.match(regex); let t1 = performance.now(); console.log(i + " took " + Math.round(t1 - t0) + " milliseconds."); }

20 took 7 milliseconds. 21 took 14 milliseconds. 22 took 27 milliseconds. 23 took 52 milliseconds. 24 took 102 milliseconds. 25 took 224 milliseconds. 26 took 421 milliseconds. 27 took 814 milliseconds. 28 took 1604 milliseconds. 29 took 3470 milliseconds.

Tested today on latest Chrome on a MacBook Pro 2,6GHz Intel Core i7.

Here comes an idea on what could be done to improve performance for regexes that are actually regular, and to still keep support for non-regular advanced regexes that e.g. contain back-references:

Add a step immediately after parsing the regex, check if it is regular and can be handled by an engine like RE2, otherwise handle it using the normal regex engine.

An even more exotic idea would be to let multiple regex engines execute in parallell, and return the result for the one that finishes first.