davidot

All of the `tco` tests test Tail Call Optimization which we dont do (yet). And these normally fail with the check in VM with the check at https://github.com/SerenityOS/serenity/blob/master/Userland/Libraries/LibJS/Runtime/VM.h#L92. However from...

New Unicode 15 tests have been merged in test262: ``` Summary: New Tests: +12 ❌ Diff Tests: -63 ✅ +63 ❌ ```

> Can we just add an 4 byte noop opcode, preferably just 0s, that gets emitted when padding is needed? Yes, that's exactly the solution I have tried some time...

> Ok, ci says, i screwed up sizes somewhere or something, Will investigate later Looks like the #12127 problem, quite an annoying one, you can probably just move around the...

> > it isn't clear if this is something that the spec provides or that it is a property of [[IsHTMLDDA]]. > > Neither, the changes introduced by the slot...

I found that this is a bit different and maybe a duplicate of https://github.com/tc39/test262/issues/674 ? This also has the flag `[caller]`, however as our engine does need to support the...

> As far as I could tell all of CI caught it Ah sorry I didn't check the CI before, good to know!

Checking `strtof` (it just does `strtod` and casts to float), I do see `got 24 / 5268191 total failures`. And this test set is not made with float in mind...

The changes should be done, just need to do the final benchmarks and rewrite some commit messages

Benchmark results:  Some remarks: For easy cases (generated with https://gist.github.com/davidot/22eed5fdd526d027ad4d2a4388b0aef4) ~the new parser is not faster in fact it seems a tiny bit slower. (These are marked in red)~...