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/branches/OpenSceneGraph-3.4@15165 16af8721-9629-0410-8352-f15c8da7e697
I believe the offending lines are in the osg::Geometry copy constructor:
if ((copyop.getCopyFlags() & osg::CopyOp::DEEP_COPY_ARRAYS))
{
if (_useVertexBufferObjects)
{
// copying of arrays doesn't set up buffer objects so we'll need to force
// Geometry to assign these, we'll do this by switching off VBO's then renabling them.
setUseVertexBufferObjects(false);
setUseVertexBufferObjects(true);
}
}
Toggling the vertex buffer objects off then on again actually touches not only the arrays controlled by DEEP_COPY_ARRAYS, but also the PrimitiveSets which are controlled by DEEP_COPY_PRIMITIVES. This means if the user has copyflags of only DEEP_COPY_ARRAYS, we are modifying arrays that belong to the original const Geometry& we are copying from. I believe this shouldn't be allowed to happen because we are using a const& specifier for the original Geometry.
In my case the osgUtil::IncrementalCompileOperation was trying to compile the geometry, while in the main thread a clone operation toggled the VBO's off and on, a crash ensues.
In the attached patch, you will find a more efficient handling of VBO's in the osg::Geometry copy constructor, so that only the Arrays that were actually deep copied have their VBO assigned, and no changes are made to Arrays that already had a valid VBO assigned. In addition, the DEEP_COPY_PRIMITIVES flag is now honored so that VBO's are set up correctly should a user copy a Geometry with only that flag.
"
git-svn-id: http://svn.openscenegraph.org/osg/OpenSceneGraph/branches/OpenSceneGraph-3.4@15128 16af8721-9629-0410-8352-f15c8da7e697
setVertexAttribArrayList(array) with array containing NULL vertexAttrib.
I added a test in order to avoid it
Code:
void Geometry::setVertexAttribArrayList(const ArrayList& arrayList)
{
_vertexAttribList = arrayList;
dirtyDisplayList();
if (_useVertexBufferObjects)
{
for(ArrayList::iterator itr = _vertexAttribList.begin();
itr != _vertexAttribList.end();
++itr)
{
if(itr->get())//ADDED
addVertexBufferObjectIfRequired(itr->get());
}
}
}
"
and
"The bug i ran into is a crash reading osgt Geometry with null vertexattribs.
The only thing i added is a not nul check on array passed to setVertexAttribArrayList."
git-svn-id: http://svn.openscenegraph.org/osg/OpenSceneGraph/branches/OpenSceneGraph-3.4@15122 16af8721-9629-0410-8352-f15c8da7e697
I used osgSetGLExtensionsFuncPtr to remove the symbols. I don't know how to test this path, but it did remove the symbols from libosgViewer.so. I have also not been able yet to see if that was sufficient for our customer.
I did this by looking at other cases, and I tried to follow some of the same practices in PixelBufferX11, like using _useSGIX in a similar way to the previous _useGLX1_3."
git-svn-id: http://svn.openscenegraph.org/osg/OpenSceneGraph/branches/OpenSceneGraph-3.4@15042 16af8721-9629-0410-8352-f15c8da7e697
On a Intel HD graphics Linux system with Mesa 10.1.3, I found that osg's Extensions::isTextureCompressionS3TCSupported() returned false, even though S3TC compressed textures *are* in fact working. I tested this by loading and rendering various DXT1, DXT3 and DXT5 compressed textures in the OSG.
"glxinfo | grep s3tc" gives:
GL_S3_s3tc
Note, if I install the package "libtxc-dxtn-s2tc0", I get in addition:
glxinfo | grep s3tc
GL_EXT_texture_compression_s3tc
GL_S3_s3tc
However, S3TC compressed textures worked correctly within the OSG even without libtxc-dxtn-s2tc0 installed.
I'm not sure what the differences between these extensions are, but based on the description at https://www.opengl.org/registry/specs/S3/s3tc.txt I would assume that both will work for OSG's purposes. The attached patch changes isTextureCompressionS3TCSupported() to accept either extension."
git-svn-id: http://svn.openscenegraph.org/osg/OpenSceneGraph/branches/OpenSceneGraph-3.4@15034 16af8721-9629-0410-8352-f15c8da7e697
