add native dsd support

[ljufa]

No description provided.

[roderickvd]

Really nice add. I know it's still a draft, but I took the opportunity to a quick review anyway. Also don't forget to add a changelog entry - worth it 😄

[roderickvd]

This should probably be fill_typed!(DsdU8)

[ljufa]

now when I removed structs I am not sure what to use here. any suggestion?

[roderickvd]

Something like this?

const DSD_SILENCE_BYTE: u8 = 0x69;

SampleFormat::DsdU8 | SampleFormat::DsdU16 | SampleFormat::DsdU32 => buffer.fill(DSD_SILENCE_BYTE),

[ljufa]

Really nice add. I know it's still a draft, but I took the opportunity to a quick review anyway. Also don't forget to add a changelog entry - worth it 😄

Thank you for you review! I am not alsa expert but I've tested it in my rsplayer repo.
I hope that change is not too specific for my use case and it will be generaly useful.

[roderickvd]

Thanks for the update. In addition to the included review comments, I see the following:

1. Please add a changelog entry.
2. In alsa/mod.rs we need to expand the check to correctly fill buffers with DSD silence:

if sample_format.is_uint() || sample_format.is_dsd() {
    fill_with_equilibrium(&mut silence_template, sample_format);
}

3. Then in examples/record_wav.rs, bail out when the input is DSD:

fn sample_format(format: cpal::SampleFormat) -> hound::SampleFormat {
    if format.is_float() {
        hound::SampleFormat::Float
    } else if format.is_dsd() {
        panic!("DSD formats cannot be written to WAV files");
        // Or handle gracefully
    } else {
        hound::SampleFormat::Int
    }
}

[roderickvd]

Something like this?

const DSD_SILENCE_BYTE: u8 = 0x69;

SampleFormat::DsdU8 | SampleFormat::DsdU16 | SampleFormat::DsdU32 => buffer.fill(DSD_SILENCE_BYTE),

[roderickvd]

As a matter of style we could reduce this to SampleFormat::DsdU8 | SampleFormat::DsdU16 | SampleFormat::DsdU32 => 1.

[roderickvd]

For consistency, change into lowercase dsdu8.

[roderickvd]

For consistency I suggest /// DSD 1-bit stream in u8 container (8 samples per byte) with 0x69 being the silence pattern.

[ljufa]

All suggestions applied. Thank you for your patience!

[roderickvd]

No problem. Some hopefully final remarks.

[roderickvd]

DSD is not unsigned integer; these lines should be removed here.

[roderickvd]

I know I suggested this before but how about:

"DSD 1-bit stream in u16 container (16 bits = 16 DSD samples) with 0x69 being the silence byte pattern."

[roderickvd]

Add support for native DSD playback