the trigger by wp distance, trigger with throttle off and trigger a
transistor all don't work, and are not structured correctly, plus are
dangerous.
The existing relay support can (correctly!) handle the resistor case
without hard-wiring a pin. The "turn off throttle to trigger" idea is
badly broken, it would crash a copter if it worked. We can make it
work properly on planes if there is demand. As it was it didn't work
anyway.
The triggger by wp distance method was broken, and had no way to
initiate a trigger anyway
The dead_zone was being used inconsistently, used in PWM space in one
place, and in control output space in another.
The fix required us to move the index number of the RC channel eeprom
value for RCn_DZ, as users will have a throttle deadzone of 3 set in
their eeprom due to a bug that Randy just fixed that caused the value
to always be saved to eeprom. In plane we then need to fix the
deadzone for the throttle to be 30
this fixes issue #303
Thanks to Soren Kuula for spotting this!
these sends did nothing, as MAVLINK_COMM_3 is not connected to
anything. If it was connected they would be dangerous, as a blocking
serial write could cause the stabilization of a copter not to run. If
the serial port was non-blocking it would corrupt the packet.
If we needed something like this it would have to use the MAVLink
packet queueing logic we use elsewhere
* added comment
* replaced "~0" with the appropriate macro
* resolved -Wconversion
* saved some space by modifying AP_Param::add_vector_suffix (writing
behind buffer could happen, if buffer_size is almost uint16_max but this
can surely be ignored)
further reading:
http://stackoverflow.com/questions/809227/is-it-safe-to-use-1-to-set-all-bits-to-true
the function:
virtual size_t AP_HAL::Print::write(const uint8_t *buffer, size_t size);
was hidden in all derived classes by their
virtual size_t write(uint8_t) = 0; implementations.
To solve this, a non-virtual write(const uint8_t *, size_t) that calls a
virtual write_implementation was added.
This isn't necessary atm, because the derived classes don't call
write(const uint8_t *, size_t), BUT this decreases the apm2-quad Program
size by 40 bytes :D and removes warnings.
