running-test262

The fleet has multiple parsers + runtimes that conform to ECMA262 or to a TC39 proposal:

• ultrathink/packages/acorn/ — the JS parser, multiple lang ports (cpp/go/rust/wasm).
• ultrathink/packages/test262-parser-runner/ — the canonical runner package.
• socket-btm/packages/temporal-infra/ — Temporal-proposal C++ port.
• (Future) @ultrathink/acorn standalone package once it ships.

Every one of them ships its own scripts/test262-*.mts runner + an allowlist or unsupported-features config. Running test262 by hand (downloading the suite, scanning the metadata blocks, running each test) is the wrong shape — the runners already encode the suite-traversal, the per-feature skip logic, the harness setup, and the result-aggregation. Always reach for the existing runner.

When to use

• "Run the spec tests" / "check test262 conformance" / "are we passing the suite?"
• "This failing test262 case keeps tripping the parser — help debug."
• "Compare this parser against the reference implementation on test262."
• Asked to add a runner for a new language port — use the existing runners as the template; never start fresh.

Canonical runners per repo