VT420 experiment needed

I'm writing a software module that turns a DEC LK201 keyboard into an IFCTF keyboard (one that generates ASCII/ISO codes and sequences) and among other features supported on LK201 in various DEC products I want to implement hex compose on the numeric keypad like on VT420. The problem is, some fine details of how it should work are unclear and I don't have a VT420 to experiment on, only VT320 and VT340 which don't have hex compose. I'm wondering if someone on this list has a VT420 and could test something for me. First, what happens if you press keypad Enter or keypad '.' in a compose sequence? I can think of 3 reasonable ways the firmware can handle this: * Do what it does for Tab, Return, arrow and functions keys in a compose sequence, i.e., cancel the compose sequence and let the pressed key do its normal non-compose action. * Ignore the key (suppress the keyclick) and remain in the compose sequence in the same state. * Store the invalid key in the compose sequence accumulator and proceed to the next character of the sequence. Then at the end of the compose sequence complain with a warning bell about the invalid sequence. Can someone with a VT420 please try this and see which of the above (or something else altogether) does the firmware do? Also what happens if you mix a main keypad key and a numeric keypad key in a compose sequence? MS P.S. Reverse-engineering state machines is fun, isn't it?