> Okay. I got a nice trigger from this. I've added some extra > debugging info after the backtrace. Thanks. Nice but not amusing. Things like that never can happen. So presumably *Server* is the buffer you want to get rid off. I can only add another diff based on that assumption. Please put a breakpoint on that non-sensical best_window = Qt; line 3050 in window.c. When it triggers please use "s" to painfully step through the entire set_window_buffer and other_buffer_safely below and tell me whether you find something fishy: Is other_buffer_safely called at all and what is the name of the "buf" it returns? Does set_window_buffer for some reason refuse to wset_buffer to the one returned by other_buffer_safely? martin