Seems like this still doesn't fix the infamous "1E1" == "010", which I guess can't be fixed without breaking a lot of old code and assumptions.