Testing with freely downloadable SW viewer is really interesting, as it seems to suffer the same bug. I'm going to check it and update bugs reports. For the Java application, it may be interesting, but i guess that it is more complicated that a simple child windows issue : if you look at this screenshot http://appdb.winehq.org/screenshots.php?iAppId=318&iVersionId=8983 , you can see that the window is divided in two parts, and the problem is that there is clearly a mismatch between the left panel, where the tree is displayed ( GDI, i guess ), and the left graphics window, which is openGL graphics, with a 2D layer on top, and maybe some compositing I guess that this type of window has to be contructed to expose the bug. I may be quite difficult in Java. For threading, the traces shows that the same thread is used for all the drawing, so I don't think that it is a threading issue.
