Linux pipewire integration (Audio + Microphone) #1973
Conversation
|
I haven't reviewed completely. But linux specific code should go in |
Please don't review it completely yet, its just in barely compilable state, not ready for any serious review of the code itself. |
|
Cool! Any way to help? |
no not really unless you wanna pull a night of just reading documentation on oboe(not the rust crate but the google software one) oh and you somehow figure out what we're doing wrong so you have to know Rust too |
Meister1593
force-pushed
the
pipewire-microphone
branch
3 times, most recently
from
69a43ad
to
725909f
Compare
zarik5
force-pushed
the
master
branch
2 times, most recently
from
d01478e
to
005c4c7
Compare
Changed chan size to proper, as well as audio format to f32 Uncommitted older changes Updated pipewire libs, adapted code to new pipewire lib version, issue with moving threads for pipewire Sink creation done, pipewire-rs and alvr compiled, no buffer copying yet Dropped unnecessary custom pw-rs, tokio. Code cleanup and move to single mainloop thread. Cfg cleanup Changed cfgs to not linux for android client compatibility Removed unnecessary pipewire code from client Renamed microphone output for linux pipewire, removed Audio Type Updated Cargo.lock Started implementing channels for pipewire to handle buffers Implemented basic buffer, almost completely non functional, code related to actual bufferring needs to be re-added Refactored flow, experimenting with byte types, order, etc Added comment about graceful shutdown Fixed steamvr shutdown. Experimenting with alternative ways to process streamed data
Meister1593
force-pushed
the
pipewire-microphone
branch
from
725909f
to
62cffbb
Compare
… - fixed large delay
|
Current progress: Both audio sink and microphone source are being created and working correctly! |
Meister1593
force-pushed
the
pipewire-microphone
branch
from
8d108d0
to
6e5a546
Compare
Meister1593
force-pushed
the
pipewire-microphone
branch
from
af70b11
to
bbe16df
Compare
|
I would say that the pipewire audio code can replace cpal for linux. |
|
I think a fair amount of people still run pulseaudio and for them audio would not work at all. So either we make pipewire a hard requirement or we keep the cpal code around for the time being. |
|
I've considered removing cpal, but i know that linux mint is still on pulse That said, they won't have microphone and they can technically switch to pipewire (and they should really) I'm 50/50 about this, but inclined to removing cpal from linux and going pipewire all the way, and point people to install pipewire. |
|
I looked at it and basically all big distros (arch, debian, ubuntu, fedora, centos) have been shipping pipewire as default for at least a year, and every other sane distro has support for it. Even Mint will ship it as default in the upcoming release. So I think forcing pipewire should be the thing. |
Then i think that settles it, i would remove cpal from linux completely |
Linux Mint will be moving to PipeWire from the next release - https://blog.linuxmint.com/?p=4660. Obviously there will still be those on earlier versions of LM that are still on PulseAudio though, although maybe it'll be the nudge those users need to upgrade (or install PipeWire separately). :) |
Meister1593
changed the title
|
In current state: Only thing left to do is remove on connect script from setup guide, maybe even make a migration so that it will be removed to not interfere with pipewire from this PR Opened PR for review, for suggestions, and will use it for the time being to find any inconsistencies or issues if there will be any |
Meister1593
changed the title
| @@ -231,7 +232,7 @@ impl DataSources { | |||
| match ws.read() { | |||
| Ok(tungstenite::Message::Text(json_string)) => { | |||
| if let Ok(event) = serde_json::from_str(&json_string) { | |||
| debug!("Server event received: {:?}", event); | |||
| // debug!("Server event received: {:?}", event); | |||
There was a problem hiding this comment.
Do we not want to keep this? Cleaning up the logging is a different issue
There was a problem hiding this comment.
Unintentional, it was spamming into logs too much, couldn't see anything else because of it, just commented for the dev time
alvr/audio/src/lib.rs
Outdated
| }; | ||
| } | ||
|
|
||
| #[cfg(target_os = "android")] |
There was a problem hiding this comment.
Might want to move this out into it's own file if it's android only
There was a problem hiding this comment.
Plus as far as I can tell this code is also invoked on the non-android clients
There was a problem hiding this comment.
https://github.com/alvr-org/ALVR/blob/master/alvr/client_core/src/connection.rs#L327
As far as I can tell this code is relevant for all clients, not just android. If that's actually the case then you should rename that file from android to client and feature gate it as "not windows and not linux"
| } | ||
|
|
||
| // returns (sink, source) | ||
| pub fn new_virtual_microphone_pair(config: MicrophoneDevicesConfig) -> Result<(Self, Self)> { |
There was a problem hiding this comment.
I'm not sure what to do here, since this isn't needed on the client so it could just always return an error, but idk if it wouldn't be used in a potential future macos port
There was a problem hiding this comment.
yeah i'm kind of divivded about this as well, it's never used and i don't know what to do with it
This is not done by any means, just put up for anyone to see how things are progressing
So far: main flow (pipewire source, getting data from one place to another) is kinda done, but i've yet to determine why only thing i get is pops into mic, it might be a multiple issues at once (buffer overload + trying to push all at once)
Other issues are listed in TODO's