Skip to content

CHANGELOG.md

Latest commit

 

History

History
52 lines (41 loc) · 3.5 KB

CHANGELOG.md

File metadata and controls

52 lines (41 loc) · 3.5 KB

Changelog

All notable changes to this project are documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.1.0] - 2026-05-17

Added

  • Full TypeScript rewrite.
  • New codec-token API: t.bool, t.i8/u8/i16/u16/i32/u32, t.f32/f64, t.i64/u64, t.string, t.shortBytes, t.bytes.
  • Little-endian variants under t.le.*.
  • struct(schema) factory that compiles a schema once and reuses it across encode/decode/sizeOf calls.
  • Functional encode(value, schema) and decode(buf, schema).
  • Infer<S> type helper that derives the TypeScript shape of decoded values from a schema.
  • DataStructError with codes (VALUE_OUT_OF_RANGE, STRING_TOO_LONG, BYTES_TOO_LONG, ARRAY_TOO_LONG, BUFFER_UNDERFLOW, SCHEMA_MISMATCH, INVALID_SCHEMA), path (e.g. $.skills[1].description), and offset (for decode errors).
  • Dual ESM + CommonJS build with separate .d.mts / .d.cts type declarations.
  • Vitest test suite (wire-format goldens, roundtrip, error paths) with v8 coverage.
  • tinybench-based benchmark suite.
  • GitHub Actions: CI matrix (Node 20/22/24 on Linux + Node 22 on macOS/Windows), on-demand benchmark workflow, tag-triggered release with npm provenance.
  • Dependabot, PR template, CODEOWNERS.

Changed

  • Module format: package is now "type": "module"; entry points are dist/index.mjs (ESM) and dist/index.cjs (CJS).
  • Minimum Node.js version: >=20.9.
  • Encoder now allocates the output Buffer exactly once via a two-pass measure/write strategy (no Buffer.concat).
  • Decoded shortBytes / bytes are returned as independent copies of the source memory rather than aliased views.
  • Replaced deprecated new Buffer(...) calls with Buffer.allocUnsafe (for output) and TextEncoder/TextDecoder for string codecs.
  • Build: tsup now emits minified output. dist/index.mjs shrinks from ~15 KB to 7.5 KB (-50%); behaviour is unchanged.

Removed

  • Legacy exports DataTypes, DataReader, DataWriter (clean break — see README migration table).
  • grunt, grunt-simple-mocha, jit-grunt, grunt-contrib-jshint, chai, benchmark.

Wire format

  • Byte-identical with 0.0.x for the corresponding new codecs. The previous test suite's golden buffers are preserved as goldens in test/wire-format.test.ts.

Performance

  • Field-iteration path strings (${path}.${key}, ${path}[${i}]) are no longer allocated on the happy path. struct / array codecs now wrap each child call in try { ... } catch (e) { rethrowWithPrefix(e, segment) } and build the error path only when something throws. Measured on Node 22:
    • nested encode (compiled): -33%
    • list of list encode (compiled): -59%
    • list of list decode (compiled): -45%
    • hero encode / decode (compiled): -17% each

Security

  • Decoded structs are now created with a null prototype, neutralising prototype-pollution risk if a schema is built from untrusted input that contains a __proto__ key.
  • String decoder now uses TextDecoder in fatal: true mode and throws SCHEMA_MISMATCH on malformed UTF-8 (previously such bytes were silently replaced with U+FFFD).
  • struct.encode now asserts that write() emits exactly as many bytes as measure() reported. Defends against codec bugs leaking uninitialised memory from Buffer.allocUnsafe.
  • CI now runs npm audit --audit-level=high --omit=dev and fails on high/critical advisories in production dependencies.