On Dec 22, 2007, at 10:32 AM, Doug McCorkle wrote:
> On May 3, 2007, at 6:16 PM, Patrick Hartling wrote:
>> Doug McCorkle wrote:
>>> I have a cluster configuration where I render a simulator
>>> viewport on the master node of the cluster to see the other
>>> in the system. In this simulator window I do not get the viewport
>>> outlines rendered for all the slave node displays. If I add these
>>> viewports to the master node display but disable them they render
>>> properly in the simulator window. Typically this is ok but with a
>>> large cluster of computers this can become cumbersome to duplicate
>>> all the viewports. Is this a bug or am I not setting up the
>>> viewport properly? Thanks for the help.
>> It is a side effect of the way that the cluster is configured. The
>> viewports only know about the projection surfaces for the display
>> windows on
>> the local machine. The display windows for the remote machines of the
>> cluster are never configured locally, and thus vrj::GlBasicSimulator
>> not know anything about them.
> With the clustering changes in the head version of vrjuggler is it
> possible for the vrjmaster to "know" about all the slaves' displays?
> Would it be possible for the vrjmaster to have a simulator window with
> all of the remote surface viewports outlined?
It appears to me that the same limitation exists in the 2.3/3.0 code.
What you want to do could probably be hacked in just the same as it
could have for 2.0 or 2.2, but it would not be easy or necessarily
Patrick L. Hartling
Senior Software Engineer, Priority 5