Conversation
…fety - Add AVAudioSessionInterruptionNotification handler to restart engine after phone calls, Siri, or background interruptions - Add AVAudioEngineConfigurationChangeNotification handler to restart engine after headphone plug/unplug or route changes - Use MIN(numOutputChannels, actualChannels) in render callback to handle dynamic channel count changes safely (was using UInt8 loop) - Log init diagnostics (channel count, sample rate) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Allows updating node properties directly on the audio thread without re-rendering the entire graph. Essential for MIDI triggering, parameter automation, and any time-critical updates. - Native: builds SET_PROPERTY instruction batch (opcode 3) - TurboModule spec: setProperty(nodeHash, key, value) - JS export: setProperty() with JSDoc Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- getBundlePath(): returns app bundle resource path for loading bundled audio assets (pairs with existing getDocumentsDirectory) - getAudioInfo(): diagnostic method returning channels, sample rate, engine running status, and runtime readiness - sharedInstance: class method for native code to access the Elementary runtime outside the RN bridge (e.g. for real-time MIDI triggering) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
tamlyn
approved these changes
Mar 17, 2026
Owner
tamlyn
left a comment
tamlyn
left a comment
There was a problem hiding this comment.
Looks good, but the Android build is failing 🤔
Collaborator
Author
thanks. could you add harness for blocking merges without all green please? I am fixing this in the meantime. |
Implement the two missing abstract methods from NativeElementarySpec that were added to the JS spec and iOS but not wired up on Android, fixing the new-arch build. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…fety on Android - Add AudioManager.OnAudioFocusChangeListener to stop/restart device after phone calls, other media apps, or transient focus loss - Add BroadcastReceiver for ACTION_AUDIO_BECOMING_NOISY to handle headphone disconnect (equivalent to iOS AVAudioEngineConfigurationChangeNotification) - Clamp channel count in DeviceProxy::process() to prevent out-of-bounds if device reports more than 2 channels (matches iOS MIN safety) - Add getAudioInfo() diagnostic method returning channels, sampleRate, engineRunning, and runtimeReady - Add C++ device lifecycle control: stopDevice/startDevice/isDeviceRunning - Log init diagnostics (channel count, sample rate) - Clean up audio focus and receiver on host destroy Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Collaborator
Author
|
@tamlyn I still want to test this PR before we merge. Will try to run some tests with my android device. I already made several improvements on Android by actually testing on real device in various scenarios. But I also found an issue with audio sequencing drift over time. I think I have fix for it. |
- Add atomic mute flag to silence audio callback instantly on focus loss, eliminating audible glitch during stop - Reinitialize miniaudio device if ma_device_start fails after extended stop (device can go stale after prolonged background) - Re-request audio focus and restart device on host resume after permanent focus loss (AUDIOFOCUS_LOSS doesn't send GAIN callback) - Add diagnostic logging for device start/running state Tested on device: YouTube focus steal, headphone plug/unplug, extended background periods — all recover cleanly. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…ndroid) Added 30Hz polling timer that calls runtime->processQueuedEvents() and forwards events (el.snapshot, el.meter, el.scope, el.fft) to JS as "elementaryEvent". iOS: dispatch_source_t timer on main thread, RCTEventEmitter Android: Handler timer on main thread, RCTDeviceEventEmitter Both: listener tracking to skip polling when no JS listeners This enables audio-thread data (beat position, levels, spectrum) to reach JS for UI updates. Previously these events were queued but never consumed on either platform. iOS: - Elementary.h: eventPollTimer property - Elementary.mm: startEventPolling, stopEventPolling (dealloc), _hasEventListeners gating, elementaryEvent in supportedEvents Android: - ElementaryModule.kt: startEventPolling, stopEventPolling (onHostDestroy), hasEventListeners gating, JSON parsing, nativeProcessQueuedEvents external declaration - cpp-adapter.cpp: nativeProcessQueuedEvents JNI method returns JSON array of queued events
The Elementary runtime's applyInstructions (graph mutation) was called from the JS/RN thread while process() ran on the audio thread with no synchronization. This caused heap corruption (nanov2_guard_corruption_detected) during appendChild when allocating InletConnection vectors. Fix: Add std::mutex to guard runtime access on both iOS and Android. - Audio callback uses try_lock — outputs silence on contention (no blocking) - applyInstructions/setProperty/addSharedResource hold the lock briefly - Worst case: one silent audio block (~11ms) during graph rebuild Also: - Fix addListener/removeListeners to call [super ...] (suppresses 'no listeners registered' warnings from RCTEventEmitter parent) - Move _listenerCount/_hasEventListeners to proper ivars in header
addSharedResource and pruneSharedResources modify the SharedResourceMap
(std::unordered_map) without holding runtimeMutex, while applyInstructions
reads from it on the JS thread with the mutex held. This is undefined
behavior when loadAudioResource runs on a background thread (Kotlin
Thread{}). iOS already locks _runtimeMutex in Elementary.mm:308 — this
brings Android in line.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When AVAudioEngineConfigurationChangeNotification fires (headphone plug/unplug, Bluetooth connect, etc.), the audio subsystem is still mid-reconfiguration. Calling startAndReturnError: synchronously inside the notification handler causes AudioUnitInitialize to RPC the audio server, which deadlocks because the server is processing the same change. The RPC times out → _ReportRPCTimeout → abort() → SIGABRT. Fix: stop the engine immediately (it's inconsistent), then defer restart by 200ms via dispatch_after so the OS finishes releasing locks and settling the new audio route before reinitialization.
Collaborator
Author
|
@tamlyn been battle testing this in Midicircuit with a few beta versions out now and another app for practising bass guitar (unreleased). I think it's good so merging. Feel free to rollback if not happy. Thank you! |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
While integrating react-native-elementary with Midicircuit (which is/was fully native) I found a couple missing elements for assets loading, detecting hardware changes and midi trigger of notes. So I am adding those in here.