On Fri, Sep 10, 2010 at 01:44:31PM +0100, Colin Guthrie wrote: > The /etc/inid.d script should ONLY be used for system wide PA. We do not > even provide such a script in our upstream tarballs and generally > discourage using PA in system wide mode. That's good design from a traditional *nix perspective, assuming any system may have multiple users. I think where it collides with a distro like Ubuntu is that Ubuntu has two main use cases: servers where it may have many users but won't have a GUI and is unlikely to be involved with sound, and then desktops on notebooks where there's a polished GUI and generally only a single user. Canonical must figure that for single user systems system-wide mode isn't so bad. But then in the Ubuntu forums there are many, many reports of people going System > Preferences > Sound and then getting nothing but a message about waiting for the sound system to start. Obviously it works for many much of the time. But it also breaks for many too. It could be this fragility is just in running it system-wide. Maybe Canonical just shouldn't do that. Nonetheless, for now, your largest user base is probably on Ubuntu, and if it's not going to work so well in system-wide mode, either Canonical or Pulseaudio should work things out so it does, or else doesn't even try to work that way. > In the default, per-user setup that we ship, PA auto-spawns, so exiting > the daemon is generally not overly problematic (clients which support > reconnect will do so automatically). I'm all for building stuff from source, but sound is such a complex stack I'd worry about how much other stuff would need to be from source rather than from the distro's packages in order to keep this stable as other system components go through their upgrade cycles. Meanwhile is there a right way to restart the daemon on any level at all under Ubuntu without having to reboot? If the init.d script doesn't do it, and "pulseaudio&" at the command prompt doesn't do it, what's the invocation that would? Or does Ubuntu's having started it system-wide the first time around mean there's just no way to get there? > I take your point re the "exit" command tho'. I am tempted to agree with > you. I just never, ever type exit on any terminal out of habit (ctrl+D > is much quicker) and thus I've never really noticed this. Hmm. Stretch for control, hold it down, then press D - or just type "exit" from home position - same speed for me, and what with different keyboards moving ctrl around, often faster to type "exit." I only use ctrl-D in Python ; > Best, Whit