Re: [gentoo-user] Removing pulseaudio

[Yuri K. Shatroff]

sorry for breaking in... a very interesting discussion :)

On 20.04.2013 19:41, Alan McKinnon wrote:
 > ...
> But back to audio. My needs are simplistic - all sound goes through the
> laptop speakers and I need one global volume knob. When a headset is
> plugged into the 3.5mm jack, all sound goes there. For a mic, I have the
> internal one and whilst there is a 3.5mm jack for an external mic, I've
> never used it but I do expect it to work when plugged in and to
> disconnect the internal mic.

Absolutely true. I also can predict that ALSA config tools are at least 
no more complicated than any other sound system's, including PA.

> Folk like Canek have complex setups that would drive me insane. I'm more
> than happy to fiddle with all that on my HTPC and home audio system, but
> never on my laptop.
>
> There's the extremes. Now, how would we determine the % numbers of how
> real users really use real audio?

Probably there's no way. But at large, I believe it would not be a big 
error to say that 90% of linux users never need anything in excess of 
ALSA. Even that is, well, too optimistic for PA.

When I heard about some sort of problems with apps when starting them in 
a wrong order, or bugs in Flash, or in WINE, or wherever, and *the* 
sound server which is designed to fix those bugs, eh... Well, just 
remember a couple of years ago when there was OSS, and ALSA came in to 
fix the problems of OSS-aware software. And now one'd say: here's *the* 
sound server that solves all problems of ALSA-using software, ... so the 
question is how long will it take to create another sound-superserver 
which will take care of problems with *that* already-fixing-everything 
soundserver? Or should bugs probably be fixed in the bugged software?

Another aspect, to my mind, is that there's a misunderstanding what 
"sound server" is for. A software on its own should not need any kind of 
server, it should just input/output audio. ALSA is itself pretty well 
aware of what is input and what is output. If one needs something like 
playing one stream through e.g. mic and recording another stream through 
e.g. headphones, he'd just install whatever sound server he deems fit. 
But it's improper to have apps use the "sound server" interface 
directly, it's like a browser being forced to use MAC addresses instead 
of HTTP (or sockets). Why ever build apps with pulseaudio support (or 
any other stuff of that sort) if it is just a layer atop the sound 
system? And that's the problem with pulseaudio: it wants too much.
Another example, there's music composition software for windows which 
uses e.g. ASIO. But it's kinda stupid to require all windows audio 
software to support ASIO.
As for complex cases, there are some, certainly. But the rule is: don't 
oversimplify the complex, nor -- overcomplicate the simple. The latter 
seems the way to go for Lennart ...

-- 
Best wishes,
Yuri K. Shatroff