Stefan Gehrer wrote: > Ok, so the last patch in this thread (attached again for clarity) seems > to be the right thing. I will apply soon as I think it is better than > leaving that pointer (which depends on input data) unchecked. Done. Stefan