It would be so much easier to use elapsed but elapsed could potentially panic is rare cases.

See: https://doc.rust-lang.org/std/time/struct.Instant.html#monotonicity

Otherwise this is pretty straight forward.
If anything fails getting expected_position_ms it will return 0 which will trigger a notify if either stream_position_ms or decoder_position_ms is > 1000.

If all goes well it's simply a matter of calculating the max delta of expected_position_ms and stream_position_ms and expected_position_ms and decoder_position_ms.
So if the decoder or the sample pipeline are off by more than 1 sec we notify.

…t it does

Collect is probably fine but for code that's this hot it's worth the couple extra lines to make certain there's only ever one allocation when it comes to the returned Vec.

release-dist-optimized inherits from `release`. Useful if you're distributing librespot at part of a project. The diffrences are:

panic = "abort", Makes librespot abort instead of unwind and hang on a panic.
Extremely useful when running librespot unattended as a system service for example to allow for auto-restarts.

codegen-units = 1 and lto = true, Take slighly longer to compile but produce more optimized binaries.

Do less calls into the worker.

Time impl's from f64 (as secs) so there's no need to manually calculate it beyond converting ms to sec.

If we grab the TimeBase in new we don't need to continually call decoder.codec_params().time_base everytime we want to convert ts to ms.

We don't need it since we now tell if we've skipped packets by calculating a max delta.

It's no longer necessary with the SymphoniaDecoder, in fact it causes the problem it used to prevent. Namely pops when you pause.

It would be so much easier to use elapsed but elapsed could potentially panic is rare cases.

See:

https://doc.rust-lang.org/std/time/struct.Instant.html#monotonicity

This prevents 2 active instances of librespot from effectively arguing over who the current active instance actually is. Without it events recurse.

Our interpolation_coefficients are already normalized.

In the case of a resampler bypass

It's best to deal with this after discovery is fixed.

sys.uptime() returns sec not mins so it should be 60 not 1.

… play

Since we are including the pipeline latency in the position we need to seek to the correct position when going from paused to play.

We can also drop the ALSA and PulseAudio buffers instead of draining them since their latency's are factored in.

Cap the output bandwidth to 92%.
Even at 48kHz it still translates to 100% source bandwidth.
This just provides a little bit of anti-alias filtering.
There is more then likely nothing there to filter,
but it doesn't hurt or cost us anything to make sure.

The resample_factor_reciprocal also happens to be our
anti-alias cutoff. In this case it represents the minimum
output bandwidth needed to fully represent the input.

All Windows calculated with pyfda (Python Filter Design Analysis Tool)
https://github.com/chipmuenk/pyfda
Window = Kaiser
beta = 8.6 (Similar to a Blackman Window)
fc = 22.5kHz
-86dB by 23kHz

This also gets rid of Linear Interpolation which leaves only Low and High both being Windowed Sinc.