PR #258 introduces future-proofing functionality but also has some minor caveats.
Have a look at these points:
1: Null dereference after error log (SDLAudio::GetSampleRate)
After logging the error when drv is null, execution falls through and dereferences it. Add return 0; after the Trace::Error call in SDLAudio.cpp
2: AudioSettings uninitialized members
bufferSize, preBufferCount, and sampleRate have no default values. Any stack-constructed AudioSettings with a missed field is undefined behavior. Add in-class initializers to the struct.
3: Audio::GetSampleRate() base class returns hardcoded 44100
WavFileWriter and FxPrinter now call Audio::GetInstance()->GetSampleRate() expecting the real negotiated rate. Non-SDL backends (JACK, RTAudio) inherit the default and silently return 44100 regardless of what the driver actually opened. Either make it pure virtual or have those backends override it.
PR #258 introduces future-proofing functionality but also has some minor caveats.
Have a look at these points:
1: Null dereference after error log (SDLAudio::GetSampleRate)
After logging the error when drv is null, execution falls through and dereferences it. Add return 0; after the Trace::Error call in SDLAudio.cpp
2: AudioSettings uninitialized members
bufferSize, preBufferCount, and sampleRate have no default values. Any stack-constructed AudioSettings with a missed field is undefined behavior. Add in-class initializers to the struct.
3: Audio::GetSampleRate() base class returns hardcoded 44100
WavFileWriter and FxPrinter now call Audio::GetInstance()->GetSampleRate() expecting the real negotiated rate. Non-SDL backends (JACK, RTAudio) inherit the default and silently return 44100 regardless of what the driver actually opened. Either make it pure virtual or have those backends override it.