On Apr 28, Amaury Jacquot wrote: > There's still something when the transformation matrix is in use. Could you please send a test case, (or just a simple code snippet), that is sufficient for reproducing the memory leak. -Carl