Files
| Type | Filename | Size |
|---|---|---|
| DIR | evergreen | — |
| DIR | vita | — |
| FILE | ONE_TO_ONE_PLAN.md | 2437 B |
| FILE | ONE_TO_ONE_ROADMAP.md | 1580 B |
| FILE | README.md | 1487 B |
README
Cobalt / Starboard Vita research track
Target: actual YouTube-on-TV runtime, not a reimplementation of proprietary application binaries.
Upstream: https://github.com/youtube/cobalt Porting guide: https://developers.google.com/youtube/cobalt/docs/starboard/porting
Evidence: - Cobalt uses Starboard to isolate OS/platform APIs, so a Vita port requires a full Starboard target, graphics, audio, threads, networking, memory, media codecs and remote-input integration. - Current Cobalt documentation targets Starboard 18 for Cobalt 27; it requires a C++17 compiler and GN/Ninja/gclient build dependencies. - Google publishes Linux, Android and RDK targets, not a Vita target. A new Vita target cannot be assumed functional. - A classic (pre-Chromium) version is a candidate for reducing runtime cost, but must be validated for compatibility with the actual YouTube web app in 2026; no assumption of compatibility. - User's exact goal: fidelity to the official Nintendo Switch YouTube interface and function. Original console executable is ARM64/Horizon OS and cannot be loaded using an Android ARM32 so-loader.
Next experiments: 1. Pick and pin a Cobalt revision (older and modern), measure source+compiled footprint. 2. Build platform ABI stub + NPLB hello-world test with a relevant V8/JS component if feasible. 3. Confirm YouTube-on-TV can load, media API and auth limits. Do not ship proprietary code. 4. Compare Vita3K behavior to real PS Vita, particularly hardware decoder.