Contribute

Contribute

To use native code, bindings from Dart to C/C++ are needed. To avoid writing these manually, they are generated from the header file (src/ffi_gen_tmp.h) using package:ffigen and temporarily stored in lib/flutter_soloud_FFIGEN.dart. You can generate the bindings by running dart run ffigen.

Since I needed to modify the generated .dart file, I followed this workflow:

1. Copy the C function declarations to be generated into src/ffi_gen_tmp.h.
2. Run dart run ffigen --config ffigen.yaml in teminal.
3. The file lib/flutter_soloud_FFIGEN.dart will be generated.
4. Copy the relevant code for the new functions from lib/flutter_soloud_FFIGEN.dart into lib/flutter_soloud_bindings_ffi.dart.

Project structure

This plugin uses the following structure:

• lib: Contains the Dart code that defines the API of the plugin relative to all platforms.
• src: Contains the native source code. Linux, Android and Windows have their own CmakeFile.txt file in their own subdir to build the code into a dynamic library.
• src/soloud: Contains the SoLoud sources of my fork
• web: Contains the scripts to build the plugin module on the web platform.
• xiph: Contains the script to build ogg and opus libraries for all the platforms.

The flutter_soloud plugin utilizes a forked repository of SoLoud, where the miniaudio audio backend is used by default.

Debugging

I have provided the necessary settings in the .vscode directory for debugging native C++ code on both Linux and Windows. To debug on Android, please use Android Studio and open the project located in the example/android directory. On Mac and iOS, please use XCode.

Logging

When debugging the package using the example/ app, you might want to change the logging level to something more granular. For example, in main():

// Capture even the finest log messages.
Logger.root.level = Level.ALL;

Linux

If you encounter any glitches, they might be caused by PulseAudio. To troubleshoot this issue, you can try disabling PulseAudio within the linux/src.cmake file. Look for the line add_definitions(-DMA_NO_PULSEAUDIO) and uncomment it (now it is the default behavior).

Android

The default audio backend is miniaudio, which will automatically select the appropriate audio backend based on your Android version:

• AAudio with Android 11.0 and newer.
• OpenSL|ES for older Android versions.

Web

In the web directory, there is a compile_wasm.sh script that generates the .js and .wasm files for the native C code located in the src dir. Run it after installing emscripten. There is also a compile_worker_and_init_module.sh to compile the web worker needed by native code to communicate with Dart and the init_module.dart which initializes the WASM module. The default emscripten Module name is Module_soloud instead of the default Module to prevent some other WASM plugins from conflicting.

The generated files are already provided, but if it is needed to modify C/C++ code or the web/worker.dart code, the scripts must be run to reflect the changes.

The compile_wasm.sh script uses the -O3 code optimization flag. To see a better errors logs, use -O0 -g -s ASSERTIONS=1 in compile_wasm.sh.

The AudioIsolate has been removed and all the logic has been implemented natively. Events like voice ended are sent from C back to Dart. However, since it is not possible to call Dart from a native thread (the audio thread), a new web worker is created using the WASM EM_ASM directive. This allows sending the voice ended event back to Dart via the worker.

Here a sketch to show the step used:

#1. This function is called while initializing the player with FlutterSoLoudWeb.setDartEventCallbacks(). It creates a Web Worker in the WASM Module using the compiled web/worker.dart. After calling this, the WASM Module will have a new variable called Module_soloud.wasmWorker which will be used in Dart to receive messages. By doing this it will be easy to use the Worker to send messages from within the CPP code.

#2. This function, like #1, uses EM_ASM to inline JS. This JS code uses the Module_soloud.wasmWorker created in #1 to send a message.

#3. This is the JS used and created in #1. Every messages sent by #2 are managed here and sent to #4.

#4. Here when the event message has been received, a new event is added to a Stream. This Stream is listened by the SoLoud API.

#5. Here we listen to the event messages coming from the WorkerController stream. Currently, only the "voice ended" event is supported. The Stream is listened in SoLoud._initializeNativeCallbacks().