Skip to content

feat: mobile voice integration + APN relay#19545

Open
R44VC0RP wants to merge 33 commits intodevfrom
opencode-remote-voice
Open

feat: mobile voice integration + APN relay#19545
R44VC0RP wants to merge 33 commits intodevfrom
opencode-remote-voice

Conversation

@R44VC0RP
Copy link
Copy Markdown
Collaborator

image

porter-deployment-app bot and others added 15 commits March 28, 2026 13:59
Co-authored-by: porter-deployment-app[bot] <87230664+porter-deployment-app[bot]@users.noreply.github.com>
Add in-button model download progress plus a model reset control in mobile-voice, and switch APN relay defaults to apn.dev.opencode.ai in serve and docs.
Retry sends on BadEnvironmentKeyInToken with the opposite APNs environment, persist the corrected env, and add request/send logs for register/unregister/event delivery debugging.
- Updated package dependencies in bun.lock and package.json for mobile-voice and opencode.
- Added expo-camera and improved camera permission handling in mobile-voice.
- Introduced QR code generation for relay setup in opencode serve command.
- Enhanced server management and logging in DictationScreen component.
Replace Expo icon/adaptive icon assets with the provided image and simplify the server dropdown so adding a server is done by scanning the setup QR code only.
Persist scanned servers across reloads, smooth server/session UI states, and make recording feel immediate. Add session-aware push notification title/body metadata from the OpenCode server.
@R44VC0RP R44VC0RP added the beta label Mar 29, 2026
Include serverID in relay event payloads and prefer server+session matching in mobile notification handling so taps reliably open the correct context and stale state is refreshed.
Allow serve to publish preferred host/domain entries in QR payloads and make mobile choose the first reachable host by QR order so preferred addresses like .ts.net are selected consistently.
Generate mobilevoice deep links in serve QR output and let mobile parse both raw payloads and pair query links, while keeping advertised-host ordering and removing QR name overrides.
Document package-specific React Native best practices and add lint warnings so state, effect, and complexity issues surface earlier during mobile-voice work.
Replace the onboarding step ternary chain with a typed step config so the screen is easier to read and lint can highlight the remaining hotspots more clearly.
Move server session response parsing into a typed helper so the mobile screen no longer relies on inline any-based mapping in the refresh path.
Replace the hydration state effect with useSyncExternalStore so the web color-scheme hook keeps its static-render fallback without triggering the set-state-in-effect lint warning.
Mark intentional async work in the mobile screen with the void operator so lint can distinguish real promise bugs from deliberate fire-and-forget behavior.
Rewrite a focused cluster of nested ternaries in the mobile screen into straight-line derived logic so the render state is easier to read without changing behavior.
Extract server/session and monitoring workflows into focused hooks so DictationScreen no longer owns every network and notification path. Add a dedicated mobile typecheck config so TypeScript checks pass without breaking Expo export resolution.
@opencode-agent
Copy link
Copy Markdown
Contributor

⚠️ Blocking Beta Release

This PR cannot be merged into the beta branch due to: Merge conflicts with dev branch

Please resolve this issue to include this PR in the next beta release.

Add mobile permission approval flow, simplify dictation settings into toggles, and remove oversized Whisper models while syncing the iOS project with the current runtime configuration.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant