Defining seats

[David Zeuthen]

On Thu, 2007-01-25 at 15:14 -0500, William Jon McCann wrote:
> Hi,
> 
> On 1/18/07, William Jon McCann <mccann at jhu.edu> wrote:
> ...
> > Another question we need to answer is:
> > "Where and how are Sessions associated with a Seat?"
> >
> > At the moment, all locally attached sessions (ie. not XDMCP) are
> > assigned to Seat1.  So, we haven't really tried to answer this yet.
> 
> Here's an idea that is essentially an amalgam of the techniques used
> in menu spec, fdi files, and pam-console's console.perms.
> 
> What if we just statically define each seat and use matching criteria
> to add new sessions to them?  The definition file might look something
> like the attached file.  Possible problems with this are a) what if a
> session matches the criteria for more than one seat b) what if a
> session doesn't match any seats.
> 
> Not sure if this is a good idea yet.  What do you think?

So this is about determine what seat a given session originates from?

Well, it strikes me that, for local connections, one *should* be able to
ask the X server about what video device(s) it's using and then,
assuming we have info.seat property as described here,

 http://lists.freedesktop.org/archives/hal/2007-January/007111.html

then figure out what Seat this Session belongs to. Of course, it's
probably hard to get at that but if we had that everything would be
dandy right?

(of course, one would have to ensure that info.seat only contain one
Seat for a video device; have info.seat=Seat1,Seat2 for a video device
just don't make sense)

      David