Zod 4.4.3 · LilScript port · @itslil/zod
Zod,
compiled.
The zod@4.4.3 classic API, rewritten in LilScript. Internals can be
mangled via config; only the public API stays named. Every size row is the same
official v4 bundle, then Oxc or Terser, mangling on and off. LilScript
rows compare closer-world (mangle on) and normal (mangle off).
npm install @itslil/zod
Official rows are an esbuild bundle of zod@4.4.3 v4, then Oxc
or Terser. LilScript rows compare closer-world (mangle on) and normal (mangle off).
Sizes are lilscript-codec raw / gzip-9 / Brotli-11. Ratios are lane / Oxc
closer-world (lower is smaller).
Measured in Chromium and Node
Parse,
not inferred.
Same 48-object batch, 400 rounds per sample. Quiet median after discarding the first 3.
Output must match official zod@4.4.3. Ratio is lane / official (lower is faster).
| Lane | Chromium | vs official | Node | vs official |
|---|
The compiled runtime, in the browser
Live
parse.
Left is JSON. Right is .parse() from @itslil/zod or official
zod@4.4.3. Both must produce the same object.
The receipts
Raw, gzip,
and Brotli.
Official 4.4.3 v4, then Oxc and Terser with mangling on and off.
LilScript is not post-minified. Closer-world mangles internals; normal does not.
The public API stays named in both. Ratios are lane / official Oxc closer-world.
Brotli-11
The wire number. LilScript closer-world is the published core.
| Lane | Brotli-11 | vs Oxc |
|---|
gzip-9
Same artifacts, gzip-9.
| Lane | gzip-9 | vs Oxc |
|---|
Raw
Uncompressed bytes.
| Lane | Raw | vs Oxc |
|---|
All three
One row per lane, every codec, with the Brotli ratio against Oxc.
| Lane | Raw | gzip-9 | Brotli-11 | vs Oxc raw / gzip / Brotli |
|---|
Why two LilScript lanes. Closer-world turns mangling on
(lilscript.dev.toml). Normal turns it off
(lilscript.nomangle.toml). public_aggregate_abi = "named"
in both, so z.string().parse stays the developer API.
Zero runtime dependencies
Install the
current port.
npm install @itslil/zod