mpt, I think there's some confusion. Lars and you want a system-wide volume. John wants a per-user volume (i.e. like we have today, just with the greeter reflecting the user's volume).
With per-user volume, in your example when B logs in, B will have a muted sound. That's the point of changing the greeter volume when B is selected.
I have implemented this in the attached branch inside indicator-sound (it uses AccountsService to synchronize user volume like we do for other settings -- keyboard layout, preferred desktop session, etc).
mpt, I think there's some confusion. Lars and you want a system-wide volume. John wants a per-user volume (i.e. like we have today, just with the greeter reflecting the user's volume).
With per-user volume, in your example when B logs in, B will have a muted sound. That's the point of changing the greeter volume when B is selected.
I have implemented this in the attached branch inside indicator-sound (it uses AccountsService to synchronize user volume like we do for other settings -- keyboard layout, preferred desktop session, etc).