[GeeTwo]

Some of the questions appear to indicate a poor understanding of the valid vs invalid uses of MXP, so I'm just going to throw out a calibration here:

Assuming the rules for 2016 match the rules for 2015 (usual caveats, but I'd put the odds at 75%+% for this point this year), MXP boards that only interface to sensors, data feeds, computing resources, or decorative lighting are good to go, without certification. Likewise, any MXP board that is entirely passive (has no electronics or computer function, but consist of conductors designed to faithfully replicate signals sent from the 'RIO) is probably good to go. Anything that actuates or similarly interfaces with a motor controller or other actuator that is not equivalent to a bundle of wires needs to be vetted and approved through FIRST to be approved for use. Last year, I was aware of no a device which interfaced "intelligently" with sensors while interfacing "passively" with actuators, but I expect that such a hybrid device would have been put through the active device approval process.

There is a serious potential weirdness in the use of MXP to host an ethernet switch. A network camera is obviously a sensor. A network raspberry pi that does vision processing but touches no manipulators is also demonstrably acceptable. Unfortunately, there are a great number of network devices that perform communications illegal for FRC, or actuate real-world devices. Exactly where the GDC will come down on this is completely up for grabs. Based on past performance, I'd put the likelihood that the GDC will allow network-through-MXP communication with other CPUs, even if it just consists of data processing, at a bit under 50%. That is, they seem to prefer to err on the side of being too cautious rather than being too permissive. That's just the sort of stuff happens when you put a bunch of engineers on a committee. (Not a complaint, but an observation.)