Roadmap

High-level program timeline. Detailed phase tables live under each initiative.

Program phases

Phase	Name	Focus	Exit
P0	Document	This tree, pins, empty honest matrices	Docs published; nav live
P1	Oracle	Test262 harness + CI report on Linux	Pass rate published every CI run
P2	Language kernel	ES-1 … ES-4 (parse, values, eval, GC)	Core semantics Test262 subsets green
P3	Built-ins	ES-5 … ES-7	Built-in matrix rows green for in-scope set
P4	Modules + harden	ES-8 … ES-9	I1 exit bar → language half of `0.0.1`
P5	Exclusive surfaces	I2 + I3 implementation waves	Matrices climb; dogfood smokes for key rows
P6	Tag	`0.0.1`	I1 hard gate + I2/I3 inventories + honest status

P5 may overlap the end of P3/P4 for inventory-driven stubs, but supported exclusive rows still require a working engine.

I1 phase summary (ES Engine)

ID	Focus	Exit sketch
ES-0	Spec pin + Test262 harness	CI runs suite, reports X/Y
ES-1	Full modern syntax (lexer/parser)	Syntax fixtures + parse subset
ES-2	Values, property attributes, exotic paths	Core value semantics
ES-3	Evaluation model (envs, this, completions)	Control-flow / scope tests
ES-4	GC roots for long suite runs	No UAF / false free under suite
ES-5	Built-ins wave 1	Object, Function, Array, String, Number, Boolean, Math, Date, Error, JSON
ES-6	Built-ins wave 2	Promise, Symbol, Map/Set/Weak*, RegExp, TypedArray family
ES-7	Built-ins wave 3 + ES2025–2026 deltas	Proxy/Reflect, iterators, Iterator helpers, Set methods, …
ES-8	Module linking (cycles, TLA, import attributes as required)	Module Test262 green
ES-9	Hardening	Zero open blockers on in-scope Test262

Full detail: ES phases.

I2 phase summary (Bun-exclusive)

ID	Focus	Exit sketch
B-0	Inventory freeze + matrix	Every known Bun-only row listed
B-1	`Bun.file` / `Bun.write` / path helpers	File smokes
B-2	`Bun.serve` HTTP	Server smoke
B-3	`bun:sqlite` + SQL helpers	DB smoke
B-4	Hash / password / Glob / spawn cluster	Utility smokes
B-5	S3 + remaining long tail	Matrix depth
B-6	Dogfood scripts	Key rows `dogfooded`

Full detail: Bun roadmap.

I3 phase summary (Deno-exclusive)

ID	Focus	Exit sketch
D-0	Inventory freeze + matrix	Every known Deno-only row listed
D-1	Permissions model	Grant / deny / query
D-2	Core `Deno.*` I/O namespace	File/env/args smokes
D-3	Net / `Deno.serve` (Deno shape)	Server smoke
D-4	Deno KV	KV smoke
D-5	FFI / test / bench / long tail	Matrix depth
D-6	Dogfood scripts	Key rows `dogfooded`

Full detail: Deno roadmap.

Suggested calendar framing (not a promise)

Horizon	Intent
Near	P0–P1: docs + Test262 harness; ES-1 parser growth
Mid	P2–P3: language kernel + first built-in waves
Far	P4: language gate; P5 exclusive surfaces; P6 tag

A greenfield Zig ES engine that passes modern Test262 is a large effort. Phases exist so progress is continuous and claimable without faking parity.

Risks

Risk	Mitigation
Underestimating ES surface area	Matrix rows + Test262 oracles; no silent “done”
Confusing exclusive APIs with shared Node	Strict I2/I3 filters
Host pressure (DX/yappy) derailing language work	Embed stubs OK; language gate stays primary for `0.0.1`
Upstream Bun/Deno API churn	Version pins in Sources
Scope creep into browsers / multi-OS	Permanent product locks

After `0.0.1`

Post-tag work (not this program’s exit bar):

Shared Node / Deno / Bun overlap surfaces
Framework dogfood (Next, Nuxt, Nest, Solid — equal priority)
Embed depth for DX and yappy
PM/CAS offline graphs
Broader npm without adapters

See Relationship to the v1 plan.

Get started

v0.0.1-dev program

I1 — ES Engine

I2 — Bun-exclusive

I3 — Deno-exclusive

Reference

Roadmap

Roadmap

Program phases

I1 phase summary (ES Engine)

I2 phase summary (Bun-exclusive)

I3 phase summary (Deno-exclusive)

Suggested calendar framing (not a promise)

Risks

After `0.0.1`

​Roadmap

​Program phases

​I1 phase summary (ES Engine)

​I2 phase summary (Bun-exclusive)

​I3 phase summary (Deno-exclusive)

​Suggested calendar framing (not a promise)

​Risks

​After 0.0.1

Roadmap

Program phases

I1 phase summary (ES Engine)

I2 phase summary (Bun-exclusive)

I3 phase summary (Deno-exclusive)

Suggested calendar framing (not a promise)

Risks

After `0.0.1`