Correctness

Wakaru takes correctness and JavaScript semantics seriously. The goal is a trustworthy tool, not another code beautifier. Wakaru is not perfect, so we measure instead of promising. Two harnesses run continuously against real toolchain output and count every miss.

Test262 round trips

Test262 is the official ECMAScript conformance suite. The round-trip harness takes a Test262 case, compiles or minifies it with a real toolchain, decompiles the result with Wakaru, and runs all three versions through the same harness: the original, the transformed input, and Wakaru's output. A pass means the decompiled code preserves the exact expected behavior.

Current state: 62,061 passing round trips, with zero Wakaru correctness failures. The canonical matrix contains 66,729 runnable inputs. The 4,668 cases Wakaru cannot support are classified and excluded, not counted as passes. The numbers live in test262-stats.json.

Pattern recovery

Reproduction matrices compile known inputs through real Babel, TypeScript, SWC, esbuild, and Terser version combinations, then verify that Wakaru recovers the original construct. Current rate: 97.6% across 2,023 test shapes. Per-matrix rates live in stats.json.

What this promises, and what it does not

The measurements cover the constructs and toolchains in the matrices. That is strong evidence, not a formal equivalence proof. When behavioral fidelity matters most, use --level minimal, and treat decompiled output as something to verify rather than something verified.

If you want the exact rules, the full semantic contract lives in rewrite-assumptions.md: every named assumption, and which levels may rely on it.

And if you ever observe decompiled code behaving differently from its input, that is a bug we want. Open an issue with the input, reduced if you can.