perf(es/codegen,es/utils): Migrate to dragonbox_ecma for faster Number::toString

[magic-akari]

Description:

BREAKING CHANGE:

Related issue (if exists):

[changeset-bot]

🦋 Changeset detected

Latest commit: 3737ddd

The changes in this PR will be included in the next version bump.

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[socket-security]

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	@​swc/​counter@​0.1.3
	jest@​29.7.0 ⏵ 25.5.4	+1		+1
	@​typescript-eslint/​parser@​6.21.0
	husky@​7.0.4 ⏵ 8.0.3			+1
	@​typescript-eslint/​eslint-plugin@​6.21.0
	@​types/​node@​20.12.12 ⏵ 17.0.45
	@​types/​node@​20.12.12 ⏵ 20.14.2	+1		+1
	@​octokit/​rest@​18.12.0
	@​napi-rs/​cli@​3.2.0
	@​ast-grep/​napi@​0.40.3
	execa@​1.0.0 ⏵ 8.0.1	+1		+1
	@​octokit/​core@​5.2.0
	@​antfu/​ni@​0.21.12
	typescript@​5.3.3 ⏵ 4.9.5			+1
	lint-staged@​12.5.0 ⏵ 14.0.1	+1		+1
	prettier@​2.8.8 ⏵ 3.3.2	-6
	@​actions/​core@​1.10.1

View full report

[codspeed-hq]

CodSpeed Performance Report

Merging #11412 will not alter performance

_{Comparing magic-akari:perf/use-dragonbox-ecma (3737ddd) with main (147df2f)}

Summary

✅ 138 untouched

[github-actions]

Binary Sizes

File	Size
swc.linux-x64-gnu.node	28M (28529864 bytes)

Commit: c50e9b7

[Copilot]

Pull request overview

This PR migrates from the ryu-js library to the dragonbox_ecma library for converting f64 numbers to JavaScript strings, with the goal of improving performance.

Key Changes:

• Replaced ryu_js::Buffer with dragonbox_ecma::Buffer in number-to-string conversion code
• Updated dependencies in swc_ecma_utils and swc_ecma_codegen crates
• Added dragonbox_ecma version 0.1.0 to workspace dependencies

Reviewed changes

Copilot reviewed 5 out of 6 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
crates/swc_ecma_utils/src/number.rs	Updated ToJsString::to_js_string() to use dragonbox_ecma::Buffer instead of ryu_js::Buffer
crates/swc_ecma_codegen/src/lit.rs	Updated Print::print() for f64 to use dragonbox_ecma::Buffer instead of ryu_js::Buffer
crates/swc_ecma_utils/Cargo.toml	Replaced ryu-js dependency with dragonbox_ecma and reformatted dependency alignment
crates/swc_ecma_codegen/Cargo.toml	Replaced ryu-js dependency with dragonbox_ecma, reformatted dependencies, and moved swc_sourcemap to dev-dependencies
Cargo.toml	Added dragonbox_ecma = "0.1.0" to workspace dependencies
Cargo.lock	Updated lock file to include dragonbox_ecma and remove ryu-js references

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[kdy1]

Is it really faster in real world workload? I can't find any differences in codspeed

[magic-akari]

oxc-project/oxc#12821 shows a ~4% codegen improvement when introducing dragonbox, but that result was based on dragonbox v0.0.5. This PR uses v0.1.0 instead, so the results are not directly comparable.

The lack of a visible improvement could be due to a performance regression between versions or simply benchmark noise. I'll open a anther PR that introduces dragonbox at v0.0.5 and re-runs Codespeed for an apples-to-apples comparison.

[kdy1]

As of #11413 (comment)