Linked by Kroc Camen on Fri 9th Apr 2010 10:29 UTC
Linux "To be clear about this article's intent, it's not to bash Microsoft, or Windows. Because to be fair, despite using Linux 95% of the time while I'm on the PC, I can find more faults with it than Windows. So, this article's goal is to highlight some of the major pluses of Linux, and also showcase where Windows could improve in the future, should Microsoft take heed of the suggestions."
Thread beginning with comment 418019
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[2]: Comment by tuma324
by sorpigal on Fri 9th Apr 2010 13:00 UTC in reply to "RE: Comment by tuma324"
sorpigal
Member since:
2005-11-02

- Audio. Linux audio is currently more broken than ever


This is both true and not true. At the moment things are *so close* to being perfect that I can smell the finish line, if I may mix my metaphors.

If distributions would set up everything to just use jack by default then 90% of everything would work correctly automatically. The other 10% is mostly the same 10% that fails to work with PulseAudio, too.

Perfection is on its way. I am not a PA fan but I understand that it has some advantages that users apparently want. As such my proposed ideal audio stack in Linux is

PulseAudio -> Jack -> ALSA

And stack everything else on top as PA recommends. Each application targets jack if it can, PA if it must, or a higher level library (e.g. libao). Anything targeting ALSA gets routed through jack for mixing.

This gives you a stack that is flexible and friendly. The only issues are PA being a resource hog, jack stability and the unfriendly fact that they both require everything to be run as the same user. All three of these problems have solutions that will arrive sooner or later. Meanwhile audio *does* work, it just has to be configured with care. This makes it like any number of issues Linux has had in the past--from X to wireless networking--which have gradually gone from horrid to Just Works.

Reply Parent Score: 5

RE[3]: Comment by tuma324
by rhavenn on Fri 9th Apr 2010 17:18 in reply to "RE[2]: Comment by tuma324"
rhavenn Member since:
2006-05-12

PulseAudio is a step backwards. It's a sound server just like ESD and aRTS were. It's solving problems that should be solved at the driver level and because ALSA either can't or refuses to "fix" their stuff the Linux guys just coded around it. DO a search in news groups for how much people hated aRTS or ESD and I don't see how PA is any different. The network sound stuff is cool, but the rest of the "features" it has should be in the driver or hardware level. For example, the FreeBSD OSS drivers just create virtual sound channels so my apps all share the same sound card without issue. I install Linux (Gentoo or Debian) and whammo, suddenly I can only use one sound source and when I'm streaming radio the sound channel is locked. Under FreeBSD I can stream audio, boot up Windows 7 inside VirtualBOX and play a Flash video there and then open Xine and play a video. Sure, the audio is "garbled", but all 3 play without issue.

The audio "problem" should be fixed with ALSA at the device level and not with some software hack or drop ALSA and go back to OSS. FreeBSD did it right. Linux slapped some patches together and then went on to the next "shiny object" without bothering to polish and refine the original.

Reply Parent Score: 1

RE[4]: Comment by tuma324
by zlynx on Fri 9th Apr 2010 17:47 in reply to "RE[3]: Comment by tuma324"
zlynx Member since:
2005-07-20

Not even Windows does audio at a "driver level" anymore. It's just stupid and leads to impossible to debug problems and crashes.

Suck it up and learn to live with a user-space audio server.

Reply Parent Score: 4

RE[4]: Comment by tuma324
by sorpigal on Fri 9th Apr 2010 18:07 in reply to "RE[3]: Comment by tuma324"
sorpigal Member since:
2005-11-02

I don't have to do a search to find people who hate artsd and esd, I only have to remember my own experiences. Both suck. Sound servers in general have always sucked. PA is ESD writ large, and it's true that in a lot of ways it sucks.

This is something that should have been solved in the kernel by alsa a long time ago. Or, it should have been solved by alsa in libalsa a long time ago. But, Linus doesn't want audio processing in the kernel (fine, I can get that argument).

And, alsa people don't think there's a problem... but since alsa's just-above-kernel layer is *already* in userspace...

The next level of solution is "just above alsa" and that is precisely where jack lives.

Incidentally, this problem actually 'should' be solved at the hardware layer for hardware that supports it, which is why I think alsa should have solved this in the kernel--where better to determine whether or not to pass work off to the hardware?--but that's all water under the bridge.

If you treat alsa as your "hardware layer" then the jack approach is immediately, obviously correct. Jack works, jack does $everything. The only complaint I've seen is that it's "too hard for ordinary users" which is *true* but I seem to recall seeing the same complaint about everything in Linux at one time or another. When I hear "too hard" I interpret this as meaning "the tools aren't good enough or simple enough yet" and "the distributions don't set good defaults yet."

I agree with you! Fix it at the alsa layer. That would be *correct*--but it's never going to happen as long as Linus and the current crop of alsa devs are alive. Forget it. Move on. Code around them.

Reply Parent Score: 3

RE[4]: Comment by tuma324
by _txf_ on Fri 9th Apr 2010 19:02 in reply to "RE[3]: Comment by tuma324"
_txf_ Member since:
2008-03-17

You are aware that windows (since vista) and osx have sound servers with userspace audio?

I agree that alsa should be fixed but there is a need for software mixing and pulse audio is currently the only modern sound server designed for users (no, jack does not count at least not in its present state).

to those oss4 proponents I would say oss is the reason that alsa came along in the first place.

Edited 2010-04-09 19:09 UTC

Reply Parent Score: 2