André: If a character in a language that's not native to my keyboard is only directly available via one of the "additional" keys, we need another method. Also it's a matter of being efficient (though it's unclear whether a deadkey scancode mechanism (see below) would actually be faster).
Tom: ANSI/ISO/JIS – thanks for giving labels to this. I'll try out the link's instructions when I next have access to my JIS keyboard, which won't be for a while. If it's easy to set up so that I can switch easily or if the JIS scancodes are a superset of the ANSI ones (I obviously can't change the physical makeup of the built-in US/ANSI MacBook keyboard, so I don't want a hassle every time I disconnect the external keyboard), that would work. All: It seems like input methods produce *characters* in encodings – but someone who knows this better please fill in details. Ideal would be a low-level mechanism that intercepts scancodes and lets me designate dead keys/scancodes that I can program in a way that they produce other * scancodes*. Stephan
