Defining seats

Fabian Steiner lists at fabis-site.net
Wed Jan 31 13:15:09 PST 2007


Hello!

William Jon McCann wrote:
> On 1/28/07, Fabian Steiner <lists at fabis-site.net> wrote:
>> First of all, I will probably have to ask a rather stupid question: How
>> do you define a "session"? Do you mean that a session can be regarded as
>> a user who is currently working on one display (among three or four in
>> multiseat environments)?
>
> I've tried to formulate a definition of the term here:
> http://gitweb.freedesktop.org/?p=ConsoleKit.git;a=blob_plain;f=README
>
Thanks, now I understand what you mean by the term "Session". In fact, 
your definition is quite similar to the one I have already presumed.

> If yes, why can't we simply associate display :0 with seat1 and seat1,
>> on the other hand, possesses a usb hub whose device permissions are set
>> to the user who owns display :0? Or does this setup involve any
>> shortcomings I did not think of?
>
> For the usage you described [1], perhaps the sysadmin can modify the
> standard seat configuration to be something like the file attached?
>
> What do you think?
>

Yes, this is how such a configuration file might look like. The 
x11-display string, however, shouldn't be :0.1, :0.2, etc. but :1, :2 
since the former style describes something like a Xinerama layout and 
not several independent displays.
In addition to that, I would also suggest adding the following match 
rule in order to ensure the possibility of generating parent/children 
relationships (i.e. to find out that my USB memory stick is plugged into 
that particular hub and not into my fellow's one):

          <Match name="usb-hub" 
type="string">/sys/devices/pci0000:00/0000:00:02.0/usb2/2-4</Match>

Unfortunately, I am not aware of the fact whether the sysfs address is 
appropriate for these purposes. Furthermore, I think that they will be 
removed in HAL in the near future ([1]).

Do you have any other suggestions?

Fabian


[1] http://article.gmane.org/gmane.comp.freedesktop.hal/6323/match=sysfs







More information about the hal mailing list