perf(parser): inline identifier handler into byte handlers

[overlookmotel]

Inlining identifier_name_handler into the byte handlers for a-z, A-Z, _ and $ produces a modest performance boost, as it's a hot path.

[overlookmotel]

Current dependencies on/for this PR:

• main
  • PR perf(parser): inline identifier handler into byte handlers #2416 👈

This stack of pull requests is managed by Graphite.

[overlookmotel]

Will have to check if this bloats binary size too much. It may not be worth it.

How do I go about that? I was expecting the bloat CI job to run, but just realised it's not triggered on PRs (and maybe isn't for that purpose anyway).

[codspeed-hq]

CodSpeed Performance Report

Merging #2416 will not alter performance

_{Comparing 02-15-perf_parser_inline_identifier_handler_into_byte_handlers (ce74089) with main (3ff3495)}

Summary

✅ 27 untouched benchmarks

[Boshen]

but just realised it's not triggered on PRs (and maybe isn't for that purpose anyway).

You need to click the top right "run workflow" from the UI https://github.com/oxc-project/oxc/actions/workflows/bloat.yml

It'll display the result on the action page once it's done.

e.g. https://github.com/oxc-project/oxc/actions/runs/7930840687

[Boshen]

I think this optimization is a bit overkill if you look at the absolute change.

[overlookmotel]

Fair enough!

It might be worth revisiting once SIMD is implemented, as then identifier_name_handler will shrink considerably and cost of inlining it would drop, but that's a way off...