Dev Call Nov 19, 2018 2300 UTC

Issues & Pull Requests



ChibiOS

Plane

  • Release Update

Copter

  • Release Update

Rover

  • Release Update

Attendee count (max): 24

UTC2305 - https://github.com/ArduPilot/ardupilot/pull/9631

  • Is this a good change?

  • Wrapper?

  • Makes code larger!

  • Having to go look at the wrapper sucks

  • On heliplane the indirection is actually useful

  • [10:07 AM] (Channel) MdB: If you are going to have to overload it fine, but I think the PR is alot more readable…

  • Randy thinks indirection for one line is not useful

  • The wrapper functions would make things slower but more compact

UTC2309 - https://github.com/ArduPilot/ardupilot/pull/8409

  • MdB has commented on the issue now

  • Tridge will go back through MdB’s comments now!

  • Randy’s keen for this one!

  • Parameters are one concern

    • Crash bug

    • Will constrain

  • [10:13 AM] (Channel) Naterater: Didn’t Tom say there is a unit that can do 100Hz GPS? Not worth supporting?

  • [10:14 AM] (Channel) MdB: Nate: Not that we support at this time, and we don’t poll that fast anyways

UTC2310 - https://github.com/ArduPilot/ardupilot/pull/9358

  • Thanks to Tom in absentia for fixing this up

  • Should get original bug submitter to come back and test this

UTC2318 - ChibiOS

  • Removing the NuttX builds

  • Progress update

  • Aiming to get rid of the NuttX build

  • Issue numbers are dropping!

  • Solved a nunber of issues

  • DSMX spectrum issue

  • Safety mask issue

  • Solo gimbal twitch puzzling and hard to track down

    • Will probably remove NuttX anyway on this one

    • Need to MitM the Solo gimbal cable to nut this one out

    • Does Philip have the relevant parts or the part numbers

    • Tridge tried some buffer-size changes etc etc and it didn’t help

    • Still a puzzle!

    • OreoLED drivers still not working

    • Tridge and Peter might get together this week with all the parts to nut this out

  • I2c fix

    • Thanks to yakk54colin

    • Probably worth backporting to the stable releases

    • Timeout i2c transactions

      • Never under NuttX

        • If a device locks up you never get your bus back

        • OTOH, a very slow device will still work

        • Yakk54colin has a device which is sometimes slow

          • 10s of milliseconds
      • Chibios will time out the transaction

        • We need to cycle the clock line to destick devices whose transaction has timed out

        • Device will hold sda line low forever if it doesn’t get clocked

      • LED on same bus as airspeed sensor possibly interfering by causing noise?

  • NuttX support to be removed next week?

    • Major churn
  • Waf on eclipse is working now!

    • Don’t need the px4 toolchain stuff any more

    • Use stuff straight from ARM consortium and eclipse project

      • Much newer version of eclipse!

      • Cygwin package is still painful

        • Having something prepackaged for this would be good
    • Can now target all boards

    • Can we build the 3.5 tree with this new setup?

  • Should we remove NuttX entirely from the tree or just make the compiles fail?

    • “px4-v3 has been removed, see <webpage> for details”?

    • Should we provide a get-out clause?

    • How hard of a switch do we want to make?

    • Maybe had a #error for a week or two?

    • How about Leonard and Paul and the like - are they ready for waf etc etc?

      • Leonard is, tridge helped him

      • Paul is, tridge helped him…

      • Randy might be the last hold-out?

        • Randy doesn’t think so

        • Jacob and Rob still use Make

    • Call For Testers - Final ChibiOS testing before NuttX is removed

  • Tridge wants half a page of warnings explaining what’s going on if someone tries a make / NuttX build

    • So a #error may not be sufficient
  • Randy wants to give more warning that things are going to die

  • Tridge was going to PR in a week

    • And discussion then

    • First PR would remove build options and submodules

    • Not remove the #ifdef px4 through vehicle code

    • Second stage to remove the _PX4 drivers etc etc and the #ifdef’d code

    • Need to do the submodule remove very, very carefully

      • It is a bit of an art
  • [10:43 AM] (Channel) LV: Just get a Banner on ALL ArduPilot sites and FB announcing the break from make and PX4 and linking to a wiki article

  • Randy will create a PR to #error

  • More advertising required!

    • Need a blog post and post that to facebook
  • LV thinks a banner on every page

    • Randy will patch the dev wiki
  • Should we remove the px4-v3 etc etc builds from the Wiki?

    • Tridge thinks yes

    • People using Copter 3.4.6?

      • Randy thinks deprecating

      • “Building with Make not recommended” already there

  • While we’re at it - we should kill the APM stuff from the Wiki

    • Long term

    • [10:49 AM] (Channel) Jani: we have still huge userbase of APM firmwares

    • [10:50 AM] (Channel) Jani: tens of thousands of downloads every month

  • Will discuss again in a week’s time

  • Craig thinks more visibility would be good

UTC2338 - documenting the supported boards

UTC0007 - Plane update

  • No new releases this week

  • Probably in next week or two

    • Benewake

    • Tridge is flogging Copter bits and Randy is flogging tridge’s bits

  • Will get differential spoiler stuff in in the next 3.9 point release

UTC0008 - Randy and Copter

UTC0011 - Randy and Rover

  • Rc2 will come out in next few weeks

  • https://www.youtube.com/watch?v=RRjOyaj2hNk

    • Using Cartographer
  • Next thing to focus on is the timing between companion computer and flight controller

  • Most bits are there

    • AP_RTC

    • Need to align the clocks!

    • ROS stuff needs to know when our IMU data was generated

    • ROS should be aligning clocks on the same way we do for input

    • [11:15 AM] To Weekly devcall: We don’t drift the clocks IIRC. tridge’s stuff doesn’t need it.

    • [11:16 AM] To Weekly devcall: "determining age"

    • [11:16 AM] To Weekly devcall: Rather than "aligning clocks"

  • We assume they are timestamping correctly

  • We assume uart timing jitter

  • We assume some sort of processing lag

  • We need the “glass time”, not message sending time!

  • ROS is spitting out error because we’re not responding quickly enough on timesync messages

UTC0020 - back to NuttX builds

UTC0029 - OlliW’s stuff is now being made by Jani

  • Will send some to tridge

  • Firmware is closed-source

    • No reason given

    • Bare-metal?

  • Schematics are available

  • Compiled binaries

  • New BetaCopter release

    • Jani would like to see patches come back
  • Tridge would be happier with firmware available

    • Not useful for tinkering as is

    • Need the source or new firmware

  • Stm32f103

    • Could probably run ChibiOS on it

    • E.g. io controller

    • Has a bootloader

    • Can update over uavcan

    • Could be a really, really useful tool with fiddleable firmware!

  • Still great to have without the firmware source

    • But better with

UTC0038 - https://github.com/ArduPilot/ardupilot/pull/8397

  • Tridge will rebase and have another look

UTC0040 - KDE CAN

  • FF will look again on Thursday