Dev Call Feb 26, 2018 2300 UTC

ChibiOS

  • First flight on Crazyflie2 (Sid got it flying,
  • Started work on F7 MCUs

Copter

  • Updates

Rover

  • Updates

Wiki

  • Updates

Issues


http://firmware.ardupilot.org/analog/
https://discuss.ardupilot.org/t/test-development-contribution/25894

ArduPilot Dev Call 2018-02-27

Agenda: Dev Call Feb 26, 2018 2300 UTC

Attendee count (max): 18

UTC2301 - tridge and CrazyFlie2

UTC2307 - tridge and ChibiOS

UTC2312 - Randy and Wiki

  • Randy’s been working on the Wiki a lot
  • particularly the dev wiki
    • http://ardupilot.org/dev/index.html
    • Numbers 2, 3 and 4 have been updated
    • Particularly trying to clarify the git stuff
      • Pretty pictures for forking, cloning etc
    • Chopped out redundant bits
    • Build environment on Windows 10
      • Ubuntu compat layer (Windows Subsystem for Linux)
      • Still need the .msi thing for Windows 7 etc
    • Tridge has been looking at Cygwin build
      • Path name lengths
      • Changing things to relative paths
      • Full pathname to test.cpp for waf is too long, so configure can fall over
        • Isnan / isinf
      • Need to keep AP checkout close to root directory to avoid problems
    • Windows build overhead
      • 4 different systems?
      • Vagrant is good for new users
      • A native environment is still a good idea
    • Editors?
      • Style guides etc
      • Project import guides for e.g. Eclipse?
      • JP: Jaime is working on putting something together for Atom (https://atom.io/)
        • Working on Wiki due to feedback from Unconference questionnaire
  • RealFlight
    • Runs slowly…
    • SITL under WSL makes it a lot better
    • Randy’s looking forward to improving docs on e.g. how to fly a quadplane in SITL
    • Good for dedicated AP users
      • But $200 is steep for new users
      • If you’re doing “interesting aircraft” e.g. heli or heavy quadplanes it’s VERY useful
  • Tool guys are using SimuLink and Gazebo
    • Tridge wants JSON over UDP or JSON over tcp
    • MitM interface?
      • Between AP and the model (e.g. RealFlight)
    • In Python
      • Then hook code from Simulink into the Python JSON
    • Maybe Octave instead?
    • Maybe Mathworks would like to sponsor something?
      • They did stuff with the OBC
  • Adding in more supported flight controllers
    • http://ardupilot.org/copter/docs/common-autopilots.html
    • Need multiple tiers?
      • To make it clear that if you want a stable, well-supported board for your expensive vehicle you should be getting e.g. a Cube
    • No VRBrains mentioned?!
      • The old docco was getting very out of date so it got zapped
      • They’re moving to ChibiOS which will ease the problem
      • Linking to their doc pages may be preferable
  • Wiki editing guide
    • Including Vagrant VM
  • Looking at shrinking commercial support page
  • New partners coming on boards

UTC2351 - Arming issue

  • https://github.com/ArduPilot/ardupilot/issues/7732
  • We just need to apologise to the GCS operators
    • They have to cope with the four states
    • Perhaps we could add options to avoid getting into the states
      • The GCSs would probably crap out if you did configure the vehicle to work like it currently does
  • Weekly devcall: I think we should nominate MdB to close #7732
  • Weekly devcall: (since he’s not on the call)

UTC2359 - Olivier and funding

  • Funding for RPi Compute Module carrier board
    • Sam’s team did initial design
      • James just made the ticket
    • Sam’s very busy ATM
    • James thinks there’s a call for a specially-configured companion computer that’s not hacked together
    • Hardware guy’s perspective?
      • This appears to be funding a competing product with partner’s money
      • Gumstix environment is locked down so design data wouldn’t be available
    • From other Hardware guy’s perspective
      • I can’t buy anything at the moment!
    • First dibs on satisfying a proposal goes to a partner?
  • Funding for UAVCAN DIY UC4H ESC nodes
    • Costly?
    • Details needed on implementation
    • Great to see funding committee moving on this stuff
    • JW: I’m working on a personal ESC project
    • [11:13 AM] (Channel) JW: one of the requirements is to support CAN

UTC0013 - Voting results / business interactions

UTC0020 - random videos

UTC0023 - swag shop first tagline

UTC0029 - AVSI update

  • Booth 3442
  • Many partners will be there
  • Some money will be spent
  • Divided 4 to 6 ways
  • Jeff will finish calculations on cost and update everybody
  • Craig will update Facebook stuff

UTC0037 - HandFreeHarvest

UTC0040 - Voting - very close, can we have it checked?

  • Ardupilot.devel people can see and check it
  • Tridge has checked it
  • FF: If people will continue to doubt our votes then we need a different system, Google Forms wasn’t really made for this
  • Having the votes closed on this one makes it different to others
  • (Channel) Buzz: I was the first person to use google forms for this process, and I went out of my way to make the process open and transparent. it’s just a tickbox when you setup the form/s. I believe in 100% open and transparent, and nearly all vote/s have been done the same way since then.
  • Anonymity?
  • Randy will go through how the voting works
  • Be consistent
  • Any quorum required?
    • Would any votes ever get up then?
  • (Channel) Buzz: I think this vote should have all the individual vote/s releases, as that’s the process we used till now.
  • Have to make sure we have discussion on-list before votes are created!
    • Call is not sufficient!
    • People were surprised!
  • Do we need to run this vote again?
    • Will reopen vote for a week and allow people to change their vote
    • Remember the vote is not as to whether XYZZY become a partner, it’s whether we talk to them
      • There will be a follow-up vote for them becoming a partner
  • [12:01 PM] (Channel) Jaime-mobile:
    I’m ok with reopening for a week. Not because I feel it was illegitimate, just because it is so close and we have people that did not vote. Also to Craig’s point we need to discuss why we are approaching them

UTC0105 - randy is looking forward to outcomes from Unconference!

Particularly trying to clarify the git stuff

maybe OpenTx’s attempt could provide some additional inspiration

I wouldn’t call these perfect, but I like the attempt to first describe and partially explain the overall workflow

btw: I think the current git descriptions in the docs forget about the git submodules

I didn’t see anything in the notes on what was said about this link. I want to help with autotest, but I cannot find any documentation on the test infrastructure, how I can modify, what tests are “needed” etc.