If the implementation is supposed to be on the native side, USB host support is indeed much more difficult. Even to the degree of being impossible for a non-accelerated machine. While PS/2 is doable, and was already implemented by me for three QL-style machines.bwinkel67 wrote:Oh, I thought USB was more difficult to support that's why I was asking about PS2.
Some people are considering to outsource USB host support to a CPU external to the QL, in effect using a second machine that is magnitudes faster and more complex than the QL. I always found this a matter of philosphy. E.g. a Raspberry or the likes could provide USB host support for the QL - but it could as well emulate the whole machine, improving more than just mouse support. Where does "outsourcing" start and where does it stop?
My own design decisions always stayed fully on the native side, rather dealing with the challenges and shortcomings than "outsourcing" or "emulating". It is a matter of taste. For a purely pragmatic view, "outsourcing" and "emulating" is the easier way to go. But it tends to move the focus of our small QL development manpower away from the native side to other, non-68K platforms.