On 18.10.2013 18:39, Robert Osfield wrote:
One thing I don't understand yet is why you call reset() then set the
_state pointer to null. Is it that there is another reference to the State
object that _state points to prevent the _state = 0 from deleting the
object?
Yes I think so, but I'm not sure. The reset was there even before my
changes and I can only assume that the reason was to derefernce _state.
The whole point of this patch is to ensure the State is destroyed before
the call to osg::deleteAllGLObjects(contextID):
if ( !sharedContextExists )
{
OSG_INFO<<"Doing delete of GL objects"<<std::endl;
_state->reset();
_state = 0;
osg::deleteAllGLObjects(contextID);
OSG_INFO<<"Done delete of GL objects"<<std::endl;
}
However, there is still a possibility that the user code is holding
count to the _state and _state will not be deleted. That being said, it
might be a good idea to issue a warning, something like:
if ( !sharedContextExists )
{
OSG_INFO<<"Doing delete of GL objects"<<std::endl;
_state->reset();
if (_state->referenceCount() > 1)
{
OSG_WARN<<"GraphicsContext::close(): Some of the GL
objects might not be deleted."<<std::endl;
}
_state = 0;
osg::deleteAllGLObjects(contextID);
OSG_INFO<<"Done delete of GL objects"<<std::endl;
}
BTW, it is usually not a problem when GL objects with an not-in-use id
gets deleted at the wrong context, but the object id might as well be a
valid one and the object (the display list, program ...) gets
unexpectedly wiped out of the existence.
Best Regards
Robert Milharcic
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
