Dev Call Jan 7 2019 2300 UTC

Issues & Pull Requests


  • microSD changes
  • serial pass thru
  • CAN peripheral
    *Nuttx removed


  • quad autotune



*Release update

Attendee count (max): 26


  • Follow-mode, copter chasing Plane


  • Video shows issue with yaw vs COG, PR created for option

  • Chase distance is on slider on the transmitter, used in video

  • UWB is next

  • … and landing quad on Plane

  • Uc4h software

    • AP_Periph code is moving along nicely

    • Tridge is having to add dynamic i2c bus allocation so we can do remote i2c over CAN

UTC1102 -

  • Is it open season on Nuttx?

  • Lots of code that could go

  • Easy cleanup to do

  • Do we do it now?

  • MdB asked for this to be delayed

    • Parameter speed issue

    • But would affect both NuttX and ChibiOS, so not relevant

  • Septentrino configuration failure happens will need to be looked at, but MdB will look at it

  • MdB is OK with things moving forward

  • Wormed its way into many places, so this should be a huge cleanup

  • Open Season on Nuttx has been declared!

UTC1106 -

  • Normally we don’t code stuff for the future

    • The bool would be simpler
  • May fix Rob’s 874 issue

UTC1113 -

  • Some comments on the assert_storage_size would be good

  • Tridge tested it on hardware

  • MdB to read through it

  • Radius_of_earth - we need an inverse of this

  • This PR highlights some divisions where they should me mults

    • Space-slash-space searching showed a huge number of places where we could do better
  • Change in initialiser format


UTC1128 - frsky/sbus

UTC1129 -

  • Disarm aircraft should force safety switch on (optionally)

  • We disabled toggling safety switch when armed

  • Disarming in flight is generally a problem, so forcing safety on isn’t a huge deal, especially as an option

  • Location of parameter is an open question

  • Why not under safety options?

  • [10:32 AM] (Channel) MdB: safety option

  • [10:32 AM] To Weekly devcall: I’d guess BRD_

  • [10:32 AM] To Weekly devcall: We didn’t but “rotate log on disarm” under arming!

  • [10:32 AM] (Channel) al: +1 BRD

  • [10:32 AM] (Channel) MdB: yeah lol

  • [10:32 AM] (Channel) MdB: thats why I put it in arming but it’s inconsistent :smiley:

  • [10:32 AM] (Channel) AT: BRD_SAFETYOPTION

  • [10:32 AM] (Channel) Naterater.desktop: I like it being near BRD_SAFETYENABLE

UTC1133 -

  • Reset everything so stale data isn’t used

  • You will lose your mavlink2 signing keys

    • Inconvenient

    • Some RC receiver stuff in there too?

UTC1134 -

  • Got self-scope-creeped a bit

  • Should vtol-takeoff and vtol-land be treated as takeoff and land?

  • Works on all vehicles

  • merged!

UTC1139 -

  • Uavcan compass appearing as compass2

  • Forum user had a fly-away because of this

    • Several issues came out of the same log
  • Flying near some big power tower

  • Messed with the compass

    • And the GPSs somehow?!
  • Uavcan compasses appear last, not first

  • Compass_primary to 2 would force it

  • why?

    • Because of the way compass devids work

    • If we made the primary compass a uavcan it would break people’s existing compasses

    • We don’t yet have the code to automatically reorder compasses based on the devids

      • We could add it
    • From a user’s point of view this is confusing

      • They don’t necessarily track which protocol is being used

        • I2c? Can? What?
    • Also an issue is number of compasses and forcing users to use the typemask

    • Same problem with i2c compasses appearing on CANBUS and again the ordering will change depending on when they’re identified

    • Automatically reordering compasses based on the devids

      • Change offsets etc etc around

      • Make canbus ones come up earlier

      • No re-calibration required

      • CE: And there will be more CAN users

UTC1147 -

  • Failure timer

    • Complete within a time limit

      • Or go into mode-land

        • As opposed to the “just go” option added for the OBC when height not reached (not in master)
      • Otherwise we just hang around waiting to finish the transition

  • Airspeed estimate

    • Why airspeed and transition counter?
  • 16-bit parameter for seconds?

  • 8 bits would seem to be reasonable…

    • Tridge has seen 30 or 40 seconds, so minutes is not inconceivable
  • merged!

UTC1152 -

  • Needs Randy-review

    • Bill’s PR is big and being looked at
  • Has Leonard approved the latest changes?

    • Yes

UTC1154 -

  • No Pierre

  • Needs Randy to look at it

  • And possibly Leonard

UTC1156 - need docs on mavlink signing

UTC1157 -

  • Larger binary with this compiler

  • Want to move to new compiler in build server

    • No mechanism in build script for it
  • Why not go to 7?

    • Tridge’s testing on 7 showed debugging didn’t work

    • Gdb not working correctly on stm32

    • To move forward to a really, really new compiler:

      • Stm32 gdb debugging has to work

      • Must be well flight tested

      • The Ubuntu 18.04 compiler doesn’t actually run for very long?

  • Should we do a Plane release before changing it?

    • Otherwise NuttX won’t fit…

    • Does this affect travis? Other CI servers?

    • Another PR to change

  • Also we need to work out what to do on the autotest server

  • Won’t change further things until next Plane goes out

UTC0008 - microsd update

  • Sporadic reports of stable and master releases

  • Whole bunch of changes have gone in

  • Not confident it’s 100% addressed

  • Can insert SD card after boot

    • Won’t work with log_disarmed
  • Card remounting might happen at some stage

  • Luis found it more reliable on his boards

  • Backportable?

    • Pending PR to do so
  • Still don’t fully understand issue

  • Lots of cross-product testing

    • Found as many failures under NuttX as on ChibiOS

    • Downright weird

    • Not consistent

    • Adding logic analyser made cards more reliable

  • Retries every 3 seconds to mount the card

  • Optimal storage block size?

    • Probably doesn’t matter

    • MdB is testing a little

    • Size limits when using smaller blocks

UTC0013 - serial pass-through

  • Pass-thorugh to GPS over USB

  • Previous two mechanisms

  • Nsh mavlink bridge

  • Tcp routing for running ucenter

  • Now can direct-attach against flight controller and flash the GPS

    • Added for Robotis servos
  • Can’t change baud rate at the moment

  • No notification of client changing baud rate request change on USB

  • Can’t flash SiK radios unless you run the radio at serial rate of 115200 as that’s the rate used for flashing

  • Can run Septentrino configuration tool without disconnecting from the flight controller

UTC0017 - CAN peripheral topic

  • AP_Peripheral is ongoing with PhillipK and Jani and tridge

  • Next week there’ll be something for people to try

  • There’s an open PR for it

UTC0018 - NuttX removal

  • Some found issues

    • Build issues
  • Tooling issues

    • Version in MP is coming from PX4 not ChibiOS

UTC0020 - Plane and tridge

  • QAUTOTUNE is available in master

    • People seem to really like it
  • Tom’s Onion stuff is coming along nicely

    • Tridge is reasonably happy with it

    • Needs testing of each of the modes

  • Test flights in RealFlight….

  • Need to test every flight mode

  • [11:21 AM] To Weekly devcall: /me would like to see autotests written :stuck_out_tongue:

  • [11:22 AM] (Channel) RM: it will greatly increase the ability for other developers to get involved with plane

  • If all tests pass then we can merge it

  • Takeoff mode after the onion!

  • Optimisation PR - TP and MdB disagree on opimizations, third opinion needed

  • Nate is expecting to test later this week

  • Flightmode constants from mavlink?

    • Nobody’s really fussed

UTC0023 -

  • 3.6.5 beta testing in next week or so

  • PixHawk4 fix

  • Some users getting v2 firmware?

    • All firmwares currently fit, so why are we limiting features?

      • Quadplane might not actually fit…
  • Blheli-passthrough not working?

  • Pilot takeoff-dz?

  • F405 uart issue?

    • Uart2 and 3 the same?

    • Some narrowing-down would be ideal

  • Fly-away issue again:

    • Horrible thing is that the EKF failsafe didn’t trigger

    • 40 seconds for the EKF failsafe to kick in and land the vehicle


    • Perhaps Lane-switching is delaying ekf failsafe triggering?

    • Compasses are a mess

    • But actually using mag3

    • Scariest bug we’ve seen in a long time

UTC0048 - Rover

UTC0051 - thanks to Nate for taking on the issues list

  • Nate apologies in advance if he’s getting too keen

  • Should he be applying labels?

UTC0053 - Wiki

UTC0106 - dev conference

  • Book it via email to get discounts and

  • Survey coming out in early Feb

  • Book your rooms!