Directly mixing a non-UHJ stereo signal into a UHJ stereo signal results in a
non-UHJ stereo signal. Such a mix can't be properly decoded anymore.
An option can probably be added for users that intend to listen to UHJ output
undecoded and let a stereo sound come through as-is on their speakers, but it
probably shouldn't be the default for cases where the output may be decoded
back.
MinGW-w64 generates bad code when accessing extern thread_local objects.
Wrapper functions are used to ensure it only accesses them from the same place
they're defined. This unfortunately adds a bit of overhead for what should be a
relatively simple thing.
These functions are inlined for non-MinGW targets, avoiding the overhead on
non-affected targets.
It can be initialized once with the device's speaker distance since it won't
change in between resets, then copied into the voice where it can be adjusted
as needed.
With the increased use of PipeWire, which offers JACK compatibility, the JACK
backend has a higher potential to be auto-selected for users. However, due to
the backend's inability to auto-configure output to the device format, this
creates a less optimal out-of-the-box experience.
Unfortunately this also means the JACK backend won't be used automatically when
a real JACK server is running along with PulseAudio. While not ideal, this is
probably the better of the two options, to have the user explicitly configure
the library to use JACK when they really want to use it, instead of unwantingly
selecting JACK because of PipeWire.
It messes with 5.1 sources using direct channels, and the surround channels are
supposed to map to the side labels. Individual backends can deal with the
channel order/label differences, as they already do to a degree.
Rather than stopping voices/sources when the device becomes disconnected, the
context can be set to leave them alone. As a consequence, their state will
remain as playing and they'll keep their last known sample offset indefinately.
For applications mindful of this behavior, it will allow resetting or reopening
the device to reconnect and automatically resume where it left off.
The inverted atomic flag replaces test_and_set+clear with test_and_clear+set,
essentially inverting the flag status. This makes more logical sense for
flagging dirty state, which is less confusing than flagging clean state. The
one caveat is ATOMIC_FLAG_INIT (or default construction in C++20) initializes
the state to true rather than false.
This uses a bit more memory (each voice needs to hold buffers for the
deinterleaved samples of each channel, instead of just one buffer for the
current channel being mixed on the device), but it will allow for handling
formats that need or prefer their channels decoded together.
