Distributed Multihead X and overall X concept - experiences and problems

Oliver Schäfer oliver.schaefer at desy.de
Mon May 12 04:20:57 PDT 2014


following up on my first mail, I have some more concrete 
ideas/questions. I hope to trigger discussion in this field of 
application of X. As I don't know the internal workings of X in detail, 
this may also be naive.

As I understand it, RandR shall replace Xinerama and take care of 
"putting together" all the physical displays to a big screen in a 
configurable way. Wouldn't it be straight forward to pass it existing 
X-Screens as a device or a setup tree? RandR could query those 
screen/device/monitor/server layout, ...- properties, could even query 
another RandR, running on a remote system and add that system of screens 
to its own configuration.

Xdmx in such a scenario would simply control which RandR instance 
(computer) is in charge of putting together the overall screen layout 
and make the necessary connections. It could (should) also be enabled to 
use a secure channel at will, so you don't have to invoke Xdmx 
separately via ssh with X-forwarding on the remote (and potentially less 
performant) computer. Instead it could run locally and use secure 
connections to the other X-servers.

As RandR is meant to reconfigure a running X-session, Xdmx wouldn't even 
necessarily need to start a new X-session with all the combined screens 
and a new Window manager anymore. Instead it would provide new screens 
to RandR (like a newly connected graphics card) and one could rearrange 
those with the usual Windowmanager tools or xorg.conf(.d) or command 
line. That way, also much less detail knowledge would be needed for 
setting up such systems and it becomes more plug&play.

I saw an email from 2008: 
It is about dmx-2 and sounded quite interesting. Is there any work 
ongoing with that project? Has it been renamed or simply abandoned?

Best regards,

More information about the xorg mailing list