Roadmap
These are acceptance gates, not release dates. The current work establishes contracts and tooling; it does not claim M0 is already complete.
| Milestone | Outcome | Exit evidence |
|---|---|---|
| M0 · A safe place | Exact-version Hyprland adapter, small helper, original static room, free-flight, hello/snapshot and emergency exit | Load/unload repeatedly; safe render-phase assertions; daemon failure returns desktop; no external asset dependency |
| M1 · A usable desktop | Planar live windows, capture leases, pointer/keyboard policy, reference shell, full control IPC | Hidden-workspace video stays live; app close mid-drag recovers; shell reconnect and output removal work; real DRM/hardware-cursor tests |
| M2 · An authorable framework | World tools, atomic world swaps, sandboxed process extensions, first provider contracts | A new original world validates without core edits; denied grants stay denied; extension crash/update/rollback works; licensing/provenance checked |
| M3 · An ecosystem | Independent shell, outside contributors, compatibility matrix, distribution and performance discipline | Second shell passes conformance; external world/provider needs no core patch; published frame/capture budgets and accessible interaction review |
What is deliberately later
Section titled “What is deliberately later”Third-party motion-intent streaming, portable widgets, custom GPU buffer transport, multi-seat routing and signed marketplace distribution need focused contracts. v0 does not reserve a vague implementation promise in their place. Native third-party plugins in the compositor are outside the public extension model.
Next contributions
Section titled “Next contributions”Review IPC and world format first. Build the smallest original room. Prove one live window and one reliable escape. Keep evidence close to each implementation PR; update milestone status only when the exit criteria actually pass.