Tried Pulse Audio Again--No Good For A11y

[Lennart Poettering]

On Mon, 22.09.08 14:00, Janina Sajka (janina at rednote.net) wrote:

> 
> I've been meaning to comment on your multi-seat setup ...
> 
> Nils Philippsen writes:
> > On Tue, 2008-09-16 at 17:28 -0400, Janina Sajka wrote:
> > > Matthias Clasen writes:
> > > > On Tue, 2008-09-16 at 14:34 -0400, Janina Sajka wrote:
> > [...]
> > > > 
> > > > > 5.)	While I paplay, I try to go Ctrl-Alt-F1. While I'm not prevented
> > > > > from doing so, paplay believes it should pause playing while I'm away
> > > > > from the gui tty. Now, who's the genius that figured out this "feature?"
> > > > 
> > > > Insults won't help your cause. 
> > > Well, my apologies if this offends you. Is it supposed to work that way,
> > > though? Is there actually a use case for that behavior? Or is this just
> > > some incidental artifact?
> > 
> > There is a use case for this, namely if you have several "seats" with
> > different logged in users: in this case one probably only wants the
> > sounds of the "active session" being played and others muted. Just in
> > case you ask, we're using that setup at home, my wife and I are logged
> > in all the time an switch between users via the applet or
> > <Ctrl><Alt>F7/F9.
> > 
> I don't believe it necessarily follows that technology should enforce
> who does, and who does not have audio in this circumstance.
> 
> Consider an office environment--perhaps it's similar to what you and
> your wife have. There are several desks with chairs. Perhaps these are
> in an open room. Perhaps there are dividers between the desks to provide
> a small measure of privacy.
> 
> Each desk in this scenario is undoubtedly equipped with a telephone. No
> one would think that the occupant of Desk B should be denied telephone
> calls just because the occupant of Desk A is on the phone. Technology is
> not required to enforce appropriate social behavior in this
> circumstance. The humans who work in this environment have learned
> appropriate behaviors in the past century.
> 
> Why is the computer desktop so different? What compells us to require
> that technology enforce no audio for B while A is provided audio?
> Perhaps the users want to listen to voice mail attachments received via
> email. Even four or five feet separation is sufficient to do so. We
> don't need technology to do for us such things as we humans clearly know
> how to negotiate for ourselves.
> 
> Indeed, it seems to me precluding multiseat audio is one way of saying
> that Fedora does not support telephone answering environments--airline
> reservations, bank customer support, etc., etc.
> 
> What am I missing here?

I am pretty sure the airlines and the banks are quite happy that we
are closing the security vulnerarbility that without this feature
allows people to eavesdrop on your audio, without you knowing.

To suspend audio for inactive sessions and only allow audio for active
sessions fixes a big security hole. And it's not just we who fixed
this hole like this. Apple for example does it too. And usually Apple
is the gold standard of user-friendliness, right?

Allowing multiple different users audio device access at the same is a
security nightmare. It has been with ALSA dmix. And it is even more so
in PA. 

Far down on my todo list is adding some kind of handover logic between
multiple PA instances, so that we can add fading of audio when we
switch sessions. This would also allow us to continue playback from
inactive sessions if the now active user is OK with that. But this is
complex, security-sensitive and not a priority. So don't expect any
quick results.

Lennart

-- 
Lennart Poettering                        Red Hat, Inc.
lennart [at] poettering [dot] net         ICQ# 11060553
http://0pointer.net/lennart/           GnuPG 0x1A015CC4