Gnome (and Linux) Sound Issues

[Kent Frazier]

I have some questions regarding the way sound is handled on Ubuntu. 
These should apply to both Hoary (what I am running) and Warty (I
remember having many of the same issues).

Now I have finally learned enough about the way sound is handled that
I can do most of what I want to under Ubuntu.  For instance, I finally
figured out that the reason I couldn't hear sound from Flash stuff in
Firefox is because it does not play well with ESD.  When I kill ESD
(if you are having this problem, try 'killall esd') before starting
Firefox, the Flash plugin plays sound fine, but if I don't, it is
completely silent (now that I think about it, I don't think this
problem showed up under Warty, but rather after the Hoary upgrade).

Another issue with ESD.  I use Audacity to handle the sound samples
for my thesis (in other words, I have to be able to use the program). 
If ESD is running when I start Audacity, it gives me an error message
and tells me that I will not be able to play or record audio.  This
problem disappears if I kill ESD first.

This is not a huge hassle, as I can kill ESD, use the program, then
restart ESD when I am done.  After I do this, however, Gnome no longer
plays sounds for system events.  Is there something I need to restart
after restarting ESD in order to make them work again?  If so, what do
I need to restart and what is the command to do so?

Is polypaudio going to improve this situation at all?  It seems like
ESD causes a lot of incompatibility problems with programs, and I am
hoping that a new sound server would fix these issues, but many of
them seem like they would still occur unless polypaudio behaves
differently.

I read about the dmix plugin for ALSA.  Is there a reason that Ubuntu
is using a sound server instead of this?  From what I understand about
dmix, it would allow for software mixing to occur at a lower level, so
all the applications could just be set up to point at ALSA (many more
programs support ALSA than ESD) and then the dmix plugin takes care of
the software mixing.  Perhaps my understanding of this is a little
shaky, but it seems like an ideal solution.  Are there problems with
this approach that make it less than ideal?

There is a wiki page
(http://www.ubuntulinux.org/wiki/DrainingTheLinuxAudioSwamp) regarding
some of these issues, but it seems to be mostly brainstorming for the
future.  I am glad that the developers are looking at alternate
solutions, because the current situation is guaranteed to put a sour
taste in the mouth of fresh Windows-converts.  In Windows, sound just
works (normally).  We need it to be that way in Linux too.

Kent