Hi Per, You can modify the scene graph in different ways, and at different points in the frame, some are easier to fit within the context of multi-threaded viewer and others aren't. Nodes and viewer level objects it generally safe to modifying the outside the viewer.frame() call. You're example should be just fine.
It's only drawable levels and contents of StateSet that are used in seperate threads that one has to be extra careful about when using the DrawThreadPerContext and CullThreadPerCameraDrawThreadPerContext threading models as these let the next frame move on while the draw threads are still dispatching the drawable and state data to OpenGL. If you do wish to modify the drawable and stateset then you simply set their DataVariance field to DYNAMIC and the draw traversal holds back the next frame till all these dynamic objects are dispatched. This topic has been discussed lots of the years so having a look through the archive will provide a good insight into what is and what isn't safe to do. Robert. On Tue, Apr 12, 2011 at 1:15 PM, Per Nordqvist <[email protected]> wrote: > On a related note, it says in Rui Wang's new OSG 3.0 book that it is > dangerous to > do scene graph updates in the frame loop due to multithreading issues. > (Chapter 8, p193) > > Snippet from the specific example: > > while ( !viewer.done() ) > { > ... > rttcamera->setViewMatrixAsLookat(...); // unsafe!? > viewer.frame(); > } > > > Instead, Rui is recommending callbacks. > So is the above unsafe, and is this true for all kinds of scene graph updates? > > Regards, > > /Per > _______________________________________________ > osg-users mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
