You can hear it in the valley where live the lame and the blind.

[reddragdiva]

KDE comes with six or seven music players. I've been sticking with AmaroK as it doesn't piss me off as much as the others. Mind you, just getting sound to work sensibly on Leegnux is shitting me big time. Everything seems to want a different combination of ALSA, aRTs, EsounD, OSS, JACK, gstreamer, etc., etc., ad nauseam. UBUNTU IS SUPPOSED TO JUST FUCKING WORK YOU USELESS STREAK OF PISS.

Is there some magical thing I'm not doing, or do I have to wait for three or so more generations of all-'l33t all-inclusive new sound systems designed to utterly supersede the previous ones but failing to quite do so or work with your previous apps before someone finally cracks it?

(I'm not after specific help, it's the fact that anyone would have to ask for specific help that's unacceptable for something that claims it'll have a technophobe-ready distro by April. It's stupid and annoying crap that just doesn't happen on a Mac or even on fucking Windows and that the developer has no business making the end user even have to think about. The sound situation is actually worse than how goddamn weird default KDE apps look in the default GNOME Ubuntu. I'm more interested in the general state of the arse of Linux sound and if a solution is in sight before the heat death of the universe.)

Update: mjg59 points out that, obviously, supported stuff is supported and therefore KDE isn't. BUT IT STILL SUCKS! And until Rosegarden has that little Ubuntu logo next to it ...

(Mind you, given the fucking around Rosegarden requires anyway, I can hardly believe they bothered making a Debian package available.)

Comments

raaaaaaaaant

[sparksparkspark.livejournal.com]

Personally these are the things that I think are wrong with Linux sound at the minute:

Alsa is too complex (the whole concept of .asoundrc is newbie-incompatible although a nice structured interface to modifying the file might fix this) and the complexity is entirely bypassed by OSS emulation. If you have one app still using OSS it breaks any fancy dmix, resampling, etc, you have set up in .asoundrc

Jack is wonderful (more powerful than anything on Windows/Mac to the best of my knowledge) but is designed for pro-audio hacks and therefore doesn't really work as a system daemon (nothing security-related). It must be run by the user, which isn't newbie-compatible. This is a shame as it really opens up and makes intuitively visible and manipulable the dataflow from application to hardware, when you have a lot of apps running, and the low-latency aspect makes it excellent for users. Whether jackd should be a system daemon has been discussed in the mailing list but there is no consensus.

Getting Jack to work at all is relatively hard work because it is very sensitive to bugs in sound drivers / the kernel / hardware. Realtime LSM is still not part of the standard kernel (Linus doesn't like it) so this causes problems if the user doesnt want to run jack as root.

Esd, arts, etc are awful old hacks with none of the benefits and most of the problems of jack.

People are still releasing OSS software! Commercial projects seem most at fault (doom3, macromedia flash, etc). Also although a lot of old software has been ported, some "dead" projects still remain without support for something more sensible. LD_PRELOAD hacks that try and push OSS apps into using ALSA seem not to work half the time.

Sadly, what that "something more sensible" actually *is*, depends on the application / user in question so everyone has to make an XMMS-esque plugin library and bloat their app. At least in that case others can make plugins for various things they feel like using. Still a shit state of affairs though since a unified framework would be a lot more newbie-friendly.

Some ALSA drivers that I have used have been massively buggy (probably due to the devs having to write the drivers with no info from the manufacturer, and there being a multitude of similar-looking devices that react wildly differently when prodded in different ways (e.g. differences in number of output channels, presence of digital interfaces, etc)). Bug fixes (including the addition of newly released hardware) take a long time to make their way into distro stock kernels, newbies can't compile their own ALSA modules. Some bugs have weird work-arounds but these cant be included as part of general distro configurations, and again the newbie is thrown in the deep end.

I think the best thing to do on a linux desktop is to get a nice well-tested soundcard with hardware mixing (so many apps can use the same device simultaneously), get ALSA with the OSS emulation going, then use alsa/oss in all your apps (in many, you dont have a choice anyway). This drops the complication a lot. If you need low-latency or inter-app routing of sound then use jack. Unfortunately things like laptops can have weird hardware and expect software mixing, and that can be very frustrating in Linux. Especially when poorly written apps "freeze" because they can't get the sound device.

Re: raaaaaaaaant

[reddragdiva]

That looks like the detailed version of what I wrote.

FreeBSD is occasionally less worse, except you have to aRTs-wrap all non-KDE sound apps if you use KDE. And the bloody stupid way it won't detect and set up sound automatically, you have to hand-edit /boot/loader.conf to load the driver even in FBSD 6. But there's essentially one sound mechanism and it basically works.

But when even Windows is simpler, you know there's a real problem. And I'm really not convinced it's even largely down to closed hardware.