AudioDeviceManager::setAudioDeviceSetup sends changes early

General JUCE discussion
1

Here is the beginning of setAudioDeviceSetup:

String AudioDeviceManager::setAudioDeviceSetup (const AudioDeviceSetup& newSetup,
                                                const bool treatAsChosenDevice)
{

    jassert (&newSetup != &currentSetup);    // this will have no effect

    if (newSetup == currentSetup && currentAudioDevice != nullptr)
        return String();

    if (! (newSetup == currentSetup))
        sendChangeMessage();

Is there a reason why sendChangeMessage() is called so early in the function?

I've set a class up to listen to my AudioDeviceManager instance so that I can display some information on the selected input and output channels, but when I call:

    void changeListenerCallback(ChangeBroadcaster *source)
    {
      AudioDeviceManager::AudioDeviceSetup setup;
      source->getAudioDeviceSetup(setup);
      DBG(setup.inputChannels.toString(2));

the changes that I have applied to my AudioDeviceSetup before calling setAudioDeviceSetup are not there anymore.

Am I using it wrong?

2

The message is asynchronous - it doesn't matter where it gets sent within the function, the result is the same.

3

Well the callback is asyncronous so it doesn't matter when the sendChangeMessage() is posted, it will never callback untill after
AudioDeviceManager::setAudioDeviceSetup has finished. Your setup probably isn't being applied for some reason. I'd step through it, mabe putting a watchpoint on the setup to see if it gets changed correctly.

4

I found this ancient thread because I just made the same observation :grinning_face_with_smiling_eyes:

But:

the callback is asyncronous so it doesn’t matter when the sendChangeMessage() is posted, it will never callback untill after
AudioDeviceManager::setAudioDeviceSetup has finished

IFAICT, this in only true if setAudioDeviceSetup is called on the message thread, right? But this is not guaranteed; setAudioDeviceSetup does not make any checks for this and it is not mentioned in the docs.

If setAudioDeviceSetup would be called from another thread, there would be a race, wouldn’t it?