Comment 46 for bug 198453

I fail to understand why the PulseAudio task was rejected, and the alsa-lib one, retained. Allow me to explain, from the perspective of avoiding regressions from previous, supported Ubuntu releases, why setting the default pcm/ctl device (using an asoundrc) to pulse is a HORRIBLE idea for Ubuntu (not Edubuntu) 8.04 LTS (or its initial point release).

Existing ALSA-only apps must continue to work via ALSA. By that statement, I imply that, for an LTS, ALSA apps lacking a native PulseAudio output plugin should not be routed through PulseAudio. Not only do certain high profile ALSA-native apps (e.g., Skype, Flash, Audacity) break when routed thusly, but having them break in an LTS is not supportable. For 8.10, sure, I wholly support breaking "the few" to force a migration toward PulseAudio, but please let's not make things suck any harder for 8.04. Furthermore, we now have a situation, thanks to changes landed in the archive due to bug 192888, where a race condition clearly exists between two high profile applications (PulseAudio and Flash, both competing for exclusive and shared, respectively, ALSA device access). Audio will be inaudible nondeterministically depending which of those two applications grabs the device and in what mode.

Clearly I am not pleased with either of these "solutions":

1) having PulseAudio grab hw:X exclusively (the current state), having ALSA-native apps routed through the pulse pcm plugin (only the current state for Edubuntu), and having libflashsupport (not the current state, and known to wreak havoc on many a Firefox 3.0b5 install with flashplugin-nonfree) installed;

2) having PulseAudio use dmix & dsnoop (not the current state, and discouraged) but allowing existing ALSA-only apps to continue working.

We're a couple days from release, so it seems too late to roll in any new changes. I will state unequivocally, however, that our current state is a mish-mash of (1) and (2) and upon release will result - heck, already has resulted - in quite some concern.