Dev Call Nov 12, 2018 2300 UTC

Issues & Pull Requests


  • Release Update


  • Release Update


  • Release Update


  • Update


Can we add
to the call please

Attendee count (max): 24


  • Tridge and fr-sky pass-through code

    • How well does it work?

    • Massive lag?

      • 30 seconds to a minute?

      • Luis sees 1 second-ish

    • “Shouldn’t be working”

      • But does work for people…
    • Maybe the receiver?

    • Maybe the binding?

UTC2305 -

  • MdB not present

  • FF is going to review it quickly

  • Revisit later in the call

  • Looks good to FF

  • merged!

UTC2305 -

  • FrSky

  • Distance to waypoints and things

  • Need to make stuff generically available

  • Stop getting stuff from vehicle directories

  • Randy doesn’t want Notify to be used as an information bus

  • Tridge wants a structure somewhere where we just add fields to which the vehicle code writes to

  • Peter suggests it should go into GCS

  • How do singletons affect this?

  • How about AP_Vehicle?

    • Make it a singleton?

    • Put the structure in there?

    • Think of it like a “blackboard”

  • No inherited vehicle class at this point!

  • Put this PR off until we have an AP_Vehicle singleton and converted OSD over to it

  • Tridge to work with Alex on existing frsky code

UTC2324 -


  • Randy would like this merged as he’s looking at supporting

  • Semaphore issues

UTC2328 - crow breaks

UTC2329 -

  • Waf fixes

  • FF has been using it all the time

  • Merged!

  • Anything that did work should work

    • There’s more stuff that should be documented
  • Windows CI….

    • Tridge is thinking of doing CI on AppVeyor

      • The build artifacts are useful!
    • Azure instead?

      • Scripting via?
    • Ideally a

UTC2339 -

  • Merged!

UTC2341 - random Plane PRs

  • Several merged but several more to look at

UTC2341 -

  • Memsetting?

  • Need do read/write

  • ABI guarantees no field reordering

  • Some computers don’t do 32-bit integers!

  • Locking of home

  • Mission Planner and setting of home….

    • Make mission at home, go to field

    • MissionPlanner asks if you want to set home

    • What does this actually do?

    • Mission Planner has own “home”

    • MP only sets vehicle home if you right-click and say set-home

    • Locking barometric vertical home is the issue

      • Might think it has already taken off

UTC0000 -

  • Needs someone to look at it

UTC0000 -

  • Missing MdB and Nate to discuss this one

  • Sid is going to try to get to the bottom of these issues

  • 10Hz

  • Some related PRs

UTC0008 - autotuning yaw

UTC0013 -

  • another MdB thing!

  • Recent feature: When doing mission based landing do a fixed wing landing until facing into wind at appropriate altitude

    • Which is a superset of what David has asked for here
  • Automatically using q-trans should be the default

  • No MdB to discuss with, so moving on

UTC0016 -

  • Adding support with with-wip

  • Lots of work being done in upstream mavlink

  • Lots of stuff marked as WIP

  • This PR adds supports so that WIP messages are stripped from documentation and generationed code

  • Reduces distance between forks

  • Can bring WIP messages in from upstream and they won’t appear

  • We already have wip messages…

    • WIP in fields make no sense
  • Pymavlink releases won’t get any WIP messages

  • Won’t be able to mavlogdump logs…

    • Or graph them
  • What’s the point of WIP?

    • There shouldn’t be WIP messages in there at all
  • WIP markup is fine

    • “Might change”
  • Would be nice if you used a WIP message you got a warning when compiling

    • Use attribute((deprecated))
  • What does this gain us?

  • Say you’re adding a new feature, a new WIP camera message

    • Would like users to be able to try new message

    • Users using pre-built mavexplorer or mavlogdump or qgc or whatever

    • They want to be able to see those messages?

      • They can’t unless they rebuild the tool
  • What’s the downside of always generating the WIP messages?

    • Not doing so will cause pain
  • This means stable releases are stable

  • how about, instead:

    • In Python

      • Generate wip messages but first time you use it it warns you are using a WIP message

        • Or require an environment variable before it will understand a message
      • C headers

        • Could have a define saying no-wip-messages

        • Maybe just a warning during build?

  • Trying to prevent a WIP message getting into a stable release?

    • Is that the intent?
  • Choose something from ?

UTC0038 - Plane 3.9 point release

  • Fixed bug in RC channels

  • Nothing major

  • 3.10 series will start soon based on master

UCT0039 - Copter update

  • Several enhancements into the next point release (3.6.1)

    • Garmin driver

    • Relative vs absolute devop

    • A few new boards supported

    • Battery failsafe voltage parameter conversion fix

  • Several issues still to address


    • Logging

      • Breakout cable for microsd would be really handy

      • Tridge’s mindpx doesn’t log at all



      • Tridge is thinking really low level problem here

        • Length of traces and the like

        • Drain resistor presence and size

      • Scope of problem?

        • Which boards are really problematic?
      • Reminds people of an old PX4v1 bug where the debug cable would cause the SD card to not work

      • NuttX instead of ChibiOS on boards where it is a problem?

        • Possibly, and certainly vice-versa
      • SD card implementation guides are massive

        • Probably lots of things that aren’t conformant
      • EMI suppression register?

        • Basically oversampling
      • Probably the most serious issue we have outstanding

    • 12m offset in some LIDARs

    • Benewake tf-mini issues

      • Reporting unhealthy when out of range

      • Already have a fix for this

    • D-shot issues

      • Just a documentation issue, user needs to fix parameters

      • D-shot on iomcu should be possible…

[in Skype]

  • James, 11:46
    My devcall connection is rubbish, so I’m hanging up. Finance update: vote for Bugmaster is out. EOI for financial assistance to attend Conference will go out in a few hours.
    Apart from Maintainers, intent is to make people sing for their supper and link conference support to a wiki refresh push. This is so we can release more money (outcome of last conference was to hire in some wiki help, but if we do the work ourselves there’s more benefit).
    Also will put out a vote for “Contribution of the Year”, for a community member to attend.

UTC0059 - 3.5.0rc1 has gone out for Rover

UTC0101 - marketing update

  • Last Wed team had a relatively long call to figure out some marketing matters

  • Pushing social media stuff

    • Who does what, that sort of thing
  • Discussed tradeshow plans

  • Discussed regulatory interfacing

  • Content issues on Wiki

    • Outdated information and pointing to outdated information
  • Talking with augmented reality people that overlays street names and the like over live video streams

    • Was looking for a one or two link thing to pass onto them

    • They currently do DJI API

      • Maybe we could implement that interface?

        • Tridge is kind of interested

          • Significant lump of work

            • 2 months

              • Subset of API

              • Lots of bits that are probably unused

      • Jeff will follow up

UTC0110 - CI

  • Plagued with failing CI tests

  • Fails in CI but not locally?

  • Tridge has pushed a feature into master which allows us to get logs from failed travis tests

  • Should be able to narrow the causes down

  • Lots of logs due to rebooting

  • Peter is looking at renaming the dataflash log files to be more useful

  • 80% packet loss on Rover?

UTC0116 - high packet loss

  • Could be sequencing

    • Use mavlogdump to show the sequence numbers

    • Also

  • Could be mavlink system IDs

  • Pixhawk connected to companion computer via serial link

    • WiFi or wired it doesn’t matter

    • Mavproxy doesn’t show packet loss

    • Qgc doesn’t see it

Qgc is 252

UTC01120 - vote on bugmaster thing!