While I still believe that a USB port would be valuable (and feasible) for use from the Workbench, let me throw out another solution that could solve the problem without USB.
What do you think about some kind of USB over Ethernet? Given that Ethernet is a given on some future revision of the MiniMig, this would make the 'USB' a completely software solution.
Again, my concern is getting access to modern devices on the Workbench. While having a stand alone solution would be best, I think it would be safe to say that 99.5% of all MiniMig users will also have a network at home, and a full PC with USB already implemented.
I would think that this would allow the MiniMig side of things to be very light weight. It could be a two driver system. One driver on the PC side to export the interface, and one on the MiniMig side to access the interface. This would put 100% of the USB stack on the PC side. It actually wouldn't even need to be a USB device on the PC side. I guess it would just be a 'Virtual Device' driver.
One question on behalf of the 'Classic Gaming USB' (as opposed to the 'Workbench USB') crowd would be do you think the over head of sending commands across the network would introduce more latency than would be acceptable for things like keyboards and joysticks. I ask this because wireless keyboards and mice are pretty common these days, so the CG-USB crowd could get benefit from this kind of solution as well. They could even use an old laptop with WiFI as a wireless USB adapter. It seems that it would also be possible to implement a version of the Ethernet to keyboard/mouse/joystick driver directly in the FPGA, so that it would default to those devices if you were loading directly to a classic game, instead of to the workbench.
While this solution is not perfect, it seems like it would be simpler to implement in the short run with the limited resources that the MiniMig has at it's disposal. It would also address the problems posed by both the CG-USB and WB-USB crowds. We are after all looking for a solution to a problem. Not a problem for our solution.
So, to those that can plan these things out in there head with little to no effort, does this sound like a feasible solution to the USB problem?