Hi again. After a quick debug session through the optimizer (thats one long file 4103 lines and I think every single line is executed during an optimization pass) , anyway, Its hard to say what really goes wrong, but there IS check for that there are descriptions within a node (in the call to isOperationPermissibleForObjectImplementation) but during the FLATTEN_STATIC_TRANSFORMS pass it is not called.
I think that this comes from the fact that during this pass the scene is traversed all the way down to a geometry, collecting the static transforms on the way and during a back-traversal of the parents of the Geometry they are actually marked for removal, and there is no call to isOperationPermissibleForObjectImplementation during that back-traversal. Later on, during other optimization passes such as REMOVE_LOADED_PROXY_NODES etc. the call to isOperationPermissibleForObjectImplementation is performed, but then the descriptions in "my" node is alreadygone... So without a solution, this seems to be the problem. I wouldn't dare sticking my head into the optimizer code right now without a life-long devotion to it ;-) /Anders On 2/15/07, Robert Osfield < [EMAIL PROTECTED]> wrote:
Hi Andres, On 2/15/07, Anders Backman < [EMAIL PROTECTED]> wrote: > Well then it is optimized away during the load, and not in the optimizer... > Because the call to osgDB::readNodeFile() returns a scene with the node (the > pointer adress is the same, with the same name, same type etc, but with no > description attached). > > I would expect the optimizer to remove any STATIC transforms etc. after it > has actually traversed the scene, not before? > > Or is the Optimizer incorporated into the loader nowadays? I have just done an osgconv on your file, and the descriptions remain in place when the Optimizer is switch off, but are optimized away along with the static MatrixTransform they are attached too. Obviously the Optimizer isn't checking for presence of descriptions when deciding whether it can optimizer away a node. It does check for a number of parameters such as callbaks, so adding a test for descriptions should be easy. Robert. _______________________________________________ osg-users mailing list osg-users@openscenegraph.net http://openscenegraph.net/mailman/listinfo/osg-users http://www.openscenegraph.org/
-- ________________________________________________________________ Anders Backman Email: [EMAIL PROTECTED] HPC2N/VRlab Phone: +46 (0)90-786 9936 Umea university Cellular: +46 (0)70-392 64 67 S-901 87 UMEA SWEDEN Fax: +46 90-786 6126 http://www.cs.umu.se/~andersb<http://www.cs.umu.se/%7Eandersb>
_______________________________________________ osg-users mailing list osg-users@openscenegraph.net http://openscenegraph.net/mailman/listinfo/osg-users http://www.openscenegraph.org/
