fix(ci): inject CodSpeed V8 flags via vitest execArgv

[kelsos]

Summary

The Benchmarks (CodSpeed) job is failing with SyntaxError: Unexpected token '%' at @codspeed/core/src/optimization.ts:10 (the V8 native %OptimizeFunctionOnNextCall).

Root cause

@codspeed/vitest-plugin reads vitest's package.json to detect the major version and picks the v3 vs v4 execArgv injection path accordingly.

Our pnpm-workspace.yaml overrides alias vitest to @voidzero-dev/vite-plus-test@0.1.18, so the plugin sees major version 0 and falls back to the v3 path (poolOptions.forks.execArgv) — which vitest 4 ignores. The --allow-natives-syntax V8 flag never reaches the benchmark worker.

Fix

When CodSpeed is running (detected via CODSPEED_ENV which the action sets), resolve V8 flags from @codspeed/core ourselves and inject them into vitest 4's top-level execArgv.

Considered alternatives

• NODE_OPTIONS=--allow-natives-syntax on the CI step — Node's NODE_OPTIONS allowlist explicitly blocks --allow-natives-syntax. Fails at runtime with --allow-natives-syntax is not allowed in NODE_OPTIONS.
• Swap bench script to vitest bench --run — bypasses vp but loses vite-plus's config resolution. Workable fallback if the execArgv approach regresses.

Credit

The codspeed-hq bot correctly identified this root cause in a draft PR (#3). We closed that PR and re-authored the same fix here to keep contributors human.

Test plan

• Benchmarks (CodSpeed) CI job completes successfully
• CodSpeed posts a PR comment with benchmark measurements
• All other required checks pass (typecheck, tests, lint, coverage, build, commitlint)
• After merge, first run on main establishes the CodSpeed baseline

[codecov]

Welcome to Codecov 🎉

Once you merge this PR into your default branch, you're all set! Codecov will compare coverage reports and display results in all future pull requests.

Thanks for integrating Codecov - We've got you covered ☂️

[codspeed-hq]

Congrats! CodSpeed is installed 🎉

🆕 20 new benchmarks were detected.

You will start to see performance impacts in the reports once the benchmarks are run from your default branch.

Detected benchmarks

• plainfp Arrays.filter: 1.3 ms
• pipe(x, f, g, h, i): 24.6 µs
• flatMap 5 deep, ok path: 18.9 µs
• manual: chained const: 12.8 µs
• native [].filter: 1.3 ms
• native [].flatMap: 5.3 ms
• plainfp Arrays.flatMap: 5.3 ms
• map over err (short-circuit): 13.5 µs
• Result.map chain (5 deep, all ok): 17.5 µs
• plainfp Arrays.map (small): 35.8 µs
• map over ok: 37.8 µs
• manual: i(h(g(f(x)))): 12.9 µs
• native [].map (small): 22.9 µs
• data-first map(xs, fn): 779.3 µs
• plainfp Arrays.map (medium): 1 ms
• curried map(fn)(xs) — closure reused: 628.2 µs
• native [].map (medium): 1 ms
• curried map(fn)(xs) — closure allocated per call: 1 ms
• try/catch with throw (1 level): 69.2 µs
• try/catch chain (5 deep, no throw): 12 µs

---