Octane regexp.js does not complete with the right check sum.

[nbp]

I was trying to look at the performance of this library, by running the octane regexp benchmark.

After doing some modification[1] to the benchmark, such as it use the RegExp constructor instead of the literal RegExp notation. I noticed that at least the first 2 regexps which are involving \s are not working as expected and find more matches than YARR.

[1] sed -ie '
s~= /([^"])/([ig]);= new RegExp("\1", "\2"); ;
sExec(/([^"])/([ig]), Exec(new RegExp("\1", "\2"), ~ ;
T; :b;
snew RegExp("([^"]*)~new RegExp("\1.backslash. ;
t b; s~.backslash.~\~g' regexp.js

[jviereck]

Thanks for opening this issue!

Could you please make the JS you test/execute available to me? My sed understanding is very limited and I am not sure if the issue you describe is due to a bug in RegExp.JS or due to a bug in the sed commands. Once I can get hold of the JS you execute, I am more than happy to check why the library has more matches than YARR.

[jviereck]

Looking at the octane benchmark, I think the problem with the \s RegExp comes due to this lines:

var re2 = /^\s*|\s*$/g;
[...]
sum += s2[i].replace(re2, '').length;

The RegExp is used as argument to the String.prototype.replace function. The problem is, that passing a RegExpJS object to the String.prototype.replace function does not work.

Could this be the root of the problem?

[nbp]

Yes, this sounds like this might be the root of the problem, in which case I guess can replace String.prototype.replace such as it check for RegExpJS.

About the sed expression, this is my mistake, I replaced a non-printable character by the "...." sequence, you can replace this sequence by ":!:!:" and this should work fine.

https://github.com/nbp/regexp.js/tree/octane-benchmarkgit@github.com:nbp/regexp.js.git