Hi! When using mrb_closure_new, I encountered a case wherep -> upperis not necessarily initialized. When it is not initialized, I ignore that value, but I can not understand well whether this patch...