Stephan Assmus wrote: >> For B_KEY_DOWN, BWindow will first handle this message in >>DispatchMessage to intercept TAB, SHIFT+TAB, OPTION+TAB, etc >>keystrokes >>to that it would advance to the next focus item/group. It is after >>that >>check that it calls fFocus->KeyDown(); > > > Is this confirmed? IIRC, when I implement KeyDown() of a BView and > handle B_TAB_KEY without calling the inherited version, tab cycling is > broken. Don't know for sure, have to test it. But I think you're right, especially because you have programmed BeOS a lot more than I did. That still does not justify the change Michael did. I specificaly put B_NULL_TOKEN so that those messages would not automatically go to a BView's handler. BTW, I don't really agree with you changing RootLayer's priority. This high priority thread can help us find deadlock, loop and timing situations early in development. You know you have to change that back at some point; I think that moment is now. Bye, Adi.