MasterDuke17

See also https://github.com/perl6/nqp/issues/293

> Good catch. Wouldn' this also apply to ADD-PAIRS-TO-MIX ? Oh, looks like there are a couple other places where this could probably be done also. I'll give them a...

https://gist.github.com/MasterDuke17/8833c501a6840ab9b2f29ea66d3db120 Here are all nqp::s where there's no entry for them in ops.markdown.

Something is off with the test, https://github.com/Raku/nqp/blob/master/t/nqp/060-bigint.t#L32 explicitly has an `nqp::add_I` (and a lot of the other `*_I` ops too).

Ah, https://github.com/Raku/nqp/blob/master/t/docs/tests.t#L39 (and line 42) should be case-insensitive (or just include 'I').

Looks like the new list of 118 is: ``` add_i64 asyncconnect asynclisten asyncreadbytes asyncwritebytes atomicadd_i atomicbindattr atomicdec_i atomicinc_i atomicload atomicload_i atomicstore atomicstore_i atposref backtrace barrierfull bindkey_o bindlex bindlex_i bindlex_n bindlex_s...

Bisect log: https://gist.github.com/Whateverable/75374a41f9e5e933282ed340f9740a67 Bisects to https://github.com/rakudo/rakudo/commit/6f0153afa8603f329359b8794e6e21ea4f16e53b

I'll have to revisit the heaptrack reports now that we've removed the FSA. But the impact on `parse_int_frac_exp` is still noticeable, going from ~2.12s to ~2.08s.

> @MasterDuke17 sorry, what is the current state of this PR? It seems like quite a lot of work and it would really make a difference for performance. It needs...

Made obsolete by https://github.com/MoarVM/MoarVM/pull/1645