[oliverames]

This is a fork-only experiment targeting codex/experiment-base. Keep it isolated from main and the existing upstream PR heads. The switch is off by default.

What and why

After a failed iOS audio-session activation, the current frontend retries on a foreground transition but ignores route changes. DM_IOS_AUDIO_ROUTE_RECOVERY=1 logs the actual output/sample rate/buffer duration and lets recognized route events retry that failed activation while foreground and uninterrupted. A successful retry recreates outputs; normal output/watchdog handling controls audible restart. Healthy routes retain RemoteIO handling, and category notifications cannot recurse into setup. There is no polling/retry timer.

Checked

• 27 actual event-policy checks; production iOS build; documentation checks pass.
• Newt, iOS 27.2: 9 production-observer checks with activation responses and output resets mocked. These establish observer decisions, not AudioUnit/callback or audible recovery.
• Actual CarPlay/headphone/Bluetooth changes, long-session sound and the cause of the previously observed progressive glitches remain unverified. No real audio-route gameplay run.

Docs

• Cause, opt-in and physical limits documented; symptom index regenerated.
• No game files, keys or extracted game data.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-29T01:38:24.026480Z	0d439c2	PR opened
🔒 Security Review	✅ Completed 2026-10-29T01:39:50.650058Z	0d439c2	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[oliverames]

Closing: experiment evidence and testing are tracked in the upstream nxnative PRs, so the fork-only experiment copy comes down. Keeping the branch here for reference.