forcing users to use osgDB::readRef*File() methods. The later is preferable as it closes a potential threading bug when using paging databases in conjunction
with the osgDB::Registry Object Cache. This threading bug occurs when one thread gets an object from the Cache via an osgDB::read*File() call where only
a pointer to the object is passed back, so taking a reference to the object is delayed till it gets reassigned to a ref_ptr<>, but at the same time another
thread calls a flush of the Object Cache deleting this object as it's referenceCount is now zero. Using osgDB::readREf*File() makes sure the a ref_ptr<> is
passed back and the referenceCount never goes to zero.
To ensure the OSG builds when OSG_PROVIDE_READFILE is to OFF the many cases of osgDB::read*File() usage had to be replaced with a ref_ptr<> osgDB::readRef*File()
usage. The avoid this change causing lots of other client code to be rewritten to handle the use of ref_ptr<> in place of C pointer I introduced a serious of
templte methods in various class to adapt ref_ptr<> to the underly C pointer to be passed to old OSG API's, example of this is found in include/osg/Group:
bool addChild(Node* child); // old method which can only be used with a Node*
tempalte<class T> bool addChild(const osg::ref_ptr<T>& child) { return addChild(child.get()); } // adapter template method
These changes together cover 149 modified files, so it's a large submission. This extent of changes are warrent to make use of the Object Cache
and multi-threaded loaded more robust.
git-svn-id: http://svn.openscenegraph.org/osg/OpenSceneGraph/trunk@15164 16af8721-9629-0410-8352-f15c8da7e697
set.
The optimization is based on the observation that matrix matrix multiplication
with a dense matrix 4x4 is 4^3 Operations whereas multiplication with a
transform, or scale matrix is only 4^2 operations. Which is a gain of a
*FACTOR*4* for these special cases.
The change implements these special cases, provides a unit test for these
implementation and converts uses of the expensiver dense matrix matrix
routine with the specialized versions.
Depending on the transform nodes in the scenegraph this change gives a
noticable improovement.
For example the osgforest code using the MatrixTransform is about 20% slower
than the same codepath using the PositionAttitudeTransform instead of the
MatrixTransform with this patch applied.
If I remember right, the sse type optimizations did *not* provide a factor 4
improovement. Also these changes are totally independent of any cpu or
instruction set architecture. So I would prefer to have this current kind of
change instead of some hand coded and cpu dependent assembly stuff. If we
need that hand tuned stuff, these can go on top of this changes which must
provide than hand optimized additional variants for the specialized versions
to give a even better result in the end.
An other change included here is a change to rotation matrix from quaterion
code. There is a sqrt call which couold be optimized away. Since we divide in
effect by sqrt(length)*sqrt(length) which is just length ...
"
