Re: Multi-head session management

[Havoc Pennington <hp redhat com>]

jacob berkman <jacob ximian com> writes: 
> why wasn't the --screen method chosen?  what was wrong with it?
> 

The argument was that the SM could adapt to missing displays, screens,
etc.  if it could access and modify an app's display. Which is
possible with the DISPLAY env variable and less hacky with Tom's 
session property idea.

Your suggestion to have apps fall back to :0.0 if they can't open the 
given screen or --display has two problems I can think of:

 - the SM could be smarter and also fall back to some other available
   screen/display instead of :0.0, perhaps you could give it a list of
   displays to use; apps can't really.  

 - maybe there are problems (even security issues) with connecting to
   a display the user didn't request

To me putting the responsibility for sorting out our "fallback" policy
in the SM has a lot of appeal, since we can then refine that policy 
without touching all the apps.

Havoc