pulseaudio should use default alsa devices

Isn't the whole point of using PulseAudio to replace dmix?

Also this will not work if you have setup alsa to use the pulseaudio plugin by default.

Yes, but pulseaudio alsa plugin wasn't very stable when i tried it.
I only want pulseaudio to stream amarok audio to another computer where i have my speakers, so you can do what you want with this bug.

I've been having difficulties with using some alsa-native apps, such as the nonfree flash-plugin. The point is, I'm afraid many applications will be dependent on alsa for a good while. (Pro-audio stuff, won't probably switch to pulse in the near future, for instance)

Alsa and pulse needs to coexist.

The problem here, is that at least the HAL-autodetection overrides alsa-defaults, bypassing any dmix-setup. This must somehow be fixed for the cleanup-audio-jumble blueprint.

Things to keep in mind:

1) Non-Free Flash 9 will utilize libflashsupport if available. See https://code.launchpad.net/~crimsun/libflashsupport-pulse/devel for the minimum-necessary source changes to get current git into Ubuntu (copyright, etc., forthcoming). This addresses Ulrik's "difficulties with ... the nonfree flash-plugin."

2) Manipulating asoundrcs is an advanced topic. Many users do so ultimately because of an (arguably) ALSA limitation of attempting to cope with OEM quirks in a "non-intuitive" fashion. (In ALSA's defense, exposing universally coherent mixer elements is a /hard/ problem, particularly since it has to ensure backward compatibility, too.) We cannot expect a user to want to muck with asoundrcs. HAL is the way to go for the Ubuntu desktop, implying that we should favor /not/ exposing plug:dmix:# / plug:dsnoop:# / plug:asym:# in default.pa.

Of course, the "technical reason" for not exposing the above virtual devices is that they are index-unstable. There are myriad instances of the default ALSA device shifting due to PCI dev enumeration being non-deterministic. If we were to expose the above virtual devices, we would have to ensure that the ALSA configuration shipped with the appropriate slots argument(s) to snd.ko - not exactly pretty.

3) The PA daemon does release the audio device(s) after an idle period. If necessary, we should consider creating a whitelist for pasuspender (e.g., skype, q3a).

Wouldn't it be possible to pull the plug on pulseaudio? I mean ok - the beans are spilled - but couldn't you at least rework the packages so that pulseaudio can be removed again on a 8.04 system? And anyone who has an issue you'd be able to just tell them to remove the pulseaudio package. IMO this would solve most of the problems.
Anyhow - it would be good to have a page where all the different issues are lined up - I just browsed to a lot of bug descriptions with no real solution. Basicly it's said that a lot of things don't work (won't work?) but one should not change it to this or that. And I'm back to oss times where only one app could use the soundcard and one is shown the finger by the developers. Very unsatisfactory *ahem*