Hi, I spent a bit of time today on an ALSA-level sound mixing solution, i.e. this bug: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=130593 Basically my solution was to allocate the IPC keys at the start of the X session, and generate an ALSA config file on login which uses those keys. This avoids issues with possible IPC key collisions and is more secure. There's a little program alsa-launch which does this. It monitors your X connection and destroys the IPC keys and such when you log out. I had to patch alsa-lib to let it read a config from a file specified an environment variable. alsa-launch is part of alsa-utils now. I'm sure some people will complain this makes alsa-utils dependent on X; I guess we'll need an alsa-utils-x11 package. Also, I needed to add alsa-launch to the X startup; however there was no programmatic way to do this in the current xinitrc setup. So I refactored the X init scripts so that packages can drop a file in /etc/X11/xinit/xinitrc.d/ which adds to the DESKTOP_SERVICES environment variable. I then modified the dbus package to use this new variable. All the new packages are here: http://people.redhat.com/walters/alsa/ These are based off the new ALSA 1.0.7 packages I uploaded to rawhide today. I'd like to get feedback on this and see how well it works for people; I've just done some light tests with ogg123, rhythmbox, and xmms. Particularly interesting is recording and complex ALSA applications. Oh, and multiple sound cards. If there are no objections I'd like to feed this all into rawhide in a few days after it's gotten a bit of testing here and see how things go. Hopefully for FC4 the idea of configuring applications for different sound output backends will totally disappear. Longer term of course I'd like to not have to generate an ALSA config file in /tmp; this kind of thing just makes sense inside the alsa-lib code itself. But this was just a quick one-day hack to see how well asym could work as the default. Comments appreciated.