20201209 Dev Meeting

Agenda

General:

• Update from relevant previous calls
• Tagged: issues, PR.

Notes

• Meeting on 23 Dec is still on.

  • Hamish will not attend Jan 6 meeting.
  • Lorenz to discuss getting host rights from Ramon by then (or possibly moving to new meeting system)

• MAVlink Governance Policy - Lorenz and JamesP had productive discussion.

  • James working on docs and in internal discussion.
  • Some of the material in governance doc needs to move into MAVLink spec.

• add messages for events interface #1531

  • James has concerns around use of component information metadata that is not yet "approved" and which there are competing implementations.
    • Events interface does not require the component information infrastructure, though "some system needed to share the metadata describing events"
    • There is no competing implementation. There is an architectural need for which component information is a proposed solution (PX4 are creating a reference implementation for this proposal). ArduPilot are encouraged to engage on this - everyone has this problem.
  • James - if these events are shared, then must be managed. Ditto schema. General agreement that just merging into tree without review not an acceptable process.
  • James - has performance been evaluated?
    • Events much more memory efficient in terms of bandwidth than statustext. Just a number for an event rather than string and occasional broadcast of current sequence number to allow syncing if events missed.
    • Libraries to support them not large, and have been provided as independent modules. Beat wonders if these should move into MAVLink repo?
  • James - impact on tooling - e.g. for flight analysis tools and logs been evaluated?
    • Beat - yes. Evaluation done with Flight Review tool. Obviously this will have costs for system.
  • James to get ArduPilot to review the work and reference implementation for both Component Information and Events.

• Travis-ci.org shuts down 31 December. ArduPilot is transferring some of our relevant CI to github actions ... [is this on track/any other actions for us as team?]

  • James - ArduPilot already using these for last week. Will push them upstream
  • Lorenz will add additional hooks for generating the C libraries.

• Autotune - what happens if dialect doesn't want us to move message to common.

  • James - this is instance of being overcautious on possible future change. Merging in this case makes sense and he will follow up.

• Adding STorM32 dialect - what's the model here, can we support it?

  • Lorenz
    • dialects made sense when we were still evolving common standard, not so much now.
    • examples given of what people want to do "independently" actually should be done within standard
    • We don't want to support dialects within mavlink unless they have lots of momentum and are evolving towards a standard, but we don't want to prevent people from interworking with the standard. Upshot, we can work with reserving ranges etc.
  • Call got broken early. Hoping Ollie can attend next time to discuss further.

• Mission checksum message

  • Hamish to update to use CRC32. Julian to help with wording.

• Fix CRC - James to follow up getting docs update to CRC merged in generator. i.e. this is not change to code, but to make clear what the CRC we use actually is.

Attendees: HamishW, JamesP, JulianO, Beat, Lorenz, Oliver Waldmann