Re: Pulseaudio

["Gustavo J. A. M. Carneiro" <gjc inescporto pt>]

On Ter, 2007-10-09 at 21:15 +1000, Jeff Waugh wrote:
> <quote who="Gustavo J. A. M. Carneiro">
> 
> > Why be forced to use a userspace mixing program when hardware mixing would
> > work equally well (or better) in most situations?
> 
> Because the vast majority of audio hardware available today does not *do*
> hardware mixing, *and* PulseAudio provides many more features and benefits
> to users than just audio mixing, particularly with regards to programmatic
> volume control and plug-n-play hardware integration. Audio mixing is *NOT*
> the #1 feature opportunity here by a long shot!

Simple audio mixing should be the #1 feature; it is by far the most
important one!

While the other features are all nice, introducing userspace daemon
software mixing could lead to the following potential regressions:

   1- Increased latency: just context switching from one process to
another one and then back to the kernel for doing the hardware
interfacing might introduce some latency in the audio.  The daemon
itself could introduce more latency if it is not carefully coded;

   2- Stability: I lost count on how many times my whole desktop
complely "froze", only to find out later that the esound daemon was
deadlocked and was responsible for blocking every gtk/gnome application.

I am not saying Pulse Audio has these problems.  I simply don't know
[1]. I'm just saying that we would avoid these problems for sure if PA
was less intrusive and allowed direct hardware access by ALSA clients.
Routing all sound through a userspace daemon has its risks, and I would
rather have "simple" sound working flawlessly, even if at the cost of
volume control and PNP.

[1] Yes, in that sense this email can be classified as FUD, but I don't
mind :) 

-- 
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert