FadeIn/FadeOut effects mess up the channel volume after playback is finished

[Casqade]

If a channel is done playing before fade_length is reached, its volume is never restored to fade_volume_reset. This messes up the volume for the next chunk to be played on the same channel.

Here, the volume is restored upon reaching fade_length. But here, the channel is done playing, its fading is set to MIX_NO_FADING and the volume is never restored. The problem is, in the next call to Mix_PlayChannel() on the same channel, the volume is stuck at whatever value was at the end of previous playback.

I believe there would be the same behaviour for expired channels.

Is this a bug? And what's the best way to deal with this problem when Mix_MasterVolume isn't available and the global volume is controlled by Mix_HaltChannel(-1); Mix_Volume(-1, volume)?

Potentional workaround 1

auto ms_to_fade = min(fade_length, chunk_ticks_left);
Mix_FadeOutChannel(channel, ms_to_fade);

In this case, the condition for restoring the volume will be fulfilled, but I don't know how to get or calculate chunk_ticks_left, as even SDL_Mixer doesn't know or expose how many ticks are left before chunk is done playing.

Potentional workaround 2

if ( Mix_Playing(channel) == true )
{
  auto volume = Mix_Volume(channel, -1);
  Mix_FadeOutChannel(channel, ms_to_fade);
}

// elsewhere
if ( Mix_Playing(channel) == false )
{
  Mix_Volume(channel, volume);
  Mix_PlayChannel(channel, chunk, 0);
}

This is only possible with dedicated channels, and it's a bit more error-prone, as the channel volume must be properly initialized before first playback.

[Casqade]

Here's how to reproduce:

Mix_Volume(0, 100);
auto volumeBeforeFadeOut = Mix_Volume(0, -1);

// sample.wav contains 1 second of white noise
Mix_Chunk* sound = Mix_LoadWAV( "sample.wav" );
Mix_PlayChannel(0, sound, 0);

using namespace std::chrono_literals;
std::this_thread::sleep_for(500ms);

Mix_FadeOutChannel(0, 1000);

// we can increase sleep time to 1100ms,
// it won't change anything
std::this_thread::sleep_for(600ms);

auto volumeAfterFadeOut = Mix_Volume(0, -1);

Mix_PlayChannel(0, sound, 0);

auto volumeForNextPlayback = Mix_Volume(0, -1);

std::cout << volumeBeforeFadeOut << "\n";
std::cout << volumeAfterFadeOut << "\n";
std::cout << volumeForNextPlayback << "\n";

// I suppose this should be true
assert(volumeBeforeFadeOut == volumeAfterFadeOut == volumeForNextPlayback);