While exploring the spec for #6039, I noticed they give some quirks in notes. We don't quite match some of them:

22.2.2.3 Disjunction https://tc39.es/ecma262/#sec-disjunction

For /a|ab/.exec("abc") we return "ab" instead of "a" Fixed by #6057

For /((a)|(ab))((c)|(bc))/.exec("abc") we return ["abc", "ab", "ab", "c", "c", undefined, undefined] instead of ["abc", "a", "a", undefined, "bc", undefined, "bc"]. Matches fixed by #6057. Order fixed by #6090

22.2.2.5.1 RepeatMatcher https://tc39.es/ecma262/#sec-runtime-semantics-repeatmatcher-abstract-operation

/a[a-z]{2,4}?/.exec("abcdefghi") fails to parse. It should return "abc" Fixed by #6209

For /(aa|aabaac|ba|b|c)*/.exec("aabaac") we return [ "aabaac", "aabaac" ] instead of ["aaba", "ba"] Fixed by #6057

For /(z)((a+)?(b+)?(c))*/.exec("zaacbbbcac") we return [ "zaacbbbcac", "z", "ac", "a", "bbb", "c" ] instead of ["zaacbbbcac", "z", "ac", "a", undefined, "c"] Fixed at some point

For /(a*)*/.exec("b") we return [ "", "" ] instead of ["", undefined]

For /(a*)b\1+/.exec("baaaac") we return null instead of ["b", ""] Fixed by #6057

22.2.2.8.2 Canonicalize ( ch ) https://tc39.es/ecma262/#sec-runtime-semantics-canonicalize-ch

For /(?=(a+))a*b\1/.exec("baaabac") we return [ "aa", "aa" ] [ "b", "" ] instead of ["aba", "a"] Fixed at some point.

For /(.*?)a(?!(a+)b\2c)\2(.*)/.exec("baaabaac") we return [ "baaabaac", "b", "aa", "baac" ] crash instead of ["baaabaac", "ba", undefined, "abaac"] Fixed by cd83325