i ve only done  a handful of test flights with the spark from an open field but will be using it mostly from a boat and (obviously) over water. I have to say that all the discussions about compass errors have me a little worried.  I have flown both a P2 and a Mavic from the boat without any issue. am i right to assume the spark seems to be more sensitive to compass calibration problems?

Turning it on for diagnostics shows no calibration error but all launches will be within 30 to 40' of an engine room with over 15k lbs of engines and generators... if people report ceiling fans causing interference I have to ASS-ume a 27KW generator may be a problem

Also, while I understand the need to rely on a compass to determine the drone heading, if the compass fails while the drone is moving why does the software not fall back on a GPS derived heading to eliminate risks of flyaways? Woudl be better than a failed compass

I understand  that for every crash and flyaway there are thousands of successful flights... I'm just trying to figure out my odds.

The event of compass error mid-flight is rare. Unless you fly into highly electromagnetic interference source like high-voltage transmission line, power grid, huge structure of ferro-magnetic, radio/mobile station that will be the case of compass error, otherwise the electronic compass itself is malfunction mid-flight which is rare. Launching from non-magnetic surface boat should not be problem. Interference from generator installed should not be problem also because the chassis itsself enclose the electro-magnetic inside or the ship navigation system will fail.  You will know the compass error at the initialization. If encountered, try move the aircraft higher or away. When launched in the air, the interference is no longer effect.