PX4 Maintainers Call: July 04, 2023

:calendar: July 04, 2023

The maintainers meeting is a meeting for the developer team to coordinate on pressing issues and to plan the development of the PX4 Autopilot project, the community is welcome to join and listen, but won’t be able to speak unless specific access is granted ahead of time.

:selfie: Meeting Link

:notebook_with_decorative_cover: Agenda

  • v1.14 Release Candidate
  • Remote ID development
  • Release procedure review
  • The Next release brief discussion

:memo: Meeting Notes

v1.14 Release Candidate

The beta2 was from March 22nd, and we haven’t had any beta/release candidate tags yet.

  • This resulted in 3+months discrepancy in current development vs user experience
  • Getting the release candidate out is therefore very needed & if not, even a beta should be cut.

We have fixes in for Optical Flow preflight check (needs to be backported):

v5 target having flash overflow

PWM configuration PR is still hanging: https://github.com/PX4/PX4-Autopilot/pull/21513

  • What is left to do? Can’t we include these changes and cut a release candidate?

Update: Matthias is porting it to 1.14 now, it is important to alert people so that the ESCs don’t start spinning randomly when upgrading to 1.14.

Backport PR that needs to be merged:

Note: The PR to a main branch hasn’t been merged, and currently it’s uncertain if the fix is actually correct.

CP_DIST parameter doesn’t exist QGC error → Will create issue in QGC to resolve.

Summary: We will merge the PRs above, but will have to wait until we resolve all the remaining PR / Issues, and we can’t conclude on the exact release date

v5 target overflowing

px4_fmu-v5_test target overflowing by 600+ bytes: (Mission) FeasibilityChecker: remove duplicate line · PX4/PX4-Autopilot@dccfcb8 · GitHub

We could bump the compiler version, Ramon will check on this.

Remote ID development

A lot of things are unclear about the exact implementation of the Remote ID feature in PX4. Such as:

  • What happens when the remote ID module loses signal during flight? What should PX4 do?
  • Which messages should PX4 actually send/receive from the Remote ID module?
  • How can we make the UAV ID tamper-proof (as required by regulation)? A simple parameter won’t cut it.

Therefore, it seems reasonable to say that Remote ID shouldn’t be officially supported in v1.14, and should be the focus of the next release.

  • What are your opinions on this?
  • Anyone with suggestions on who can lead the Remote ID discussion in the working group?

Current development coordination issue: Remote ID Development Coordination · Issue #21777 · PX4/PX4-Autopilot · GitHub

Summary: Maintainers aren’t currently very familliar with Remote ID, so we will have to form a work group to organize this.

Release procedure review

  • It was quite easy to have a big delay in beta release (e.g. beta2 for v1.14 was cut more than 3 months ago), should we cut the beta every week instead?

Discussion: Beta release can be frequent, wouldn’t affect safety concerns. But also, we may not have some issues resolved, so we don’t want to cut it too frequently.

  • If we have a release blocker that isn’t getting resolved for more than a week or two, what should we do?

Feedback: To have a stable release, we should have the issues resolved, and we often have focus getting scattered, so we don’t always achieve timely release due to that.

Core components list: PX4 Release procedure by junwoo091400 · Pull Request #2331 · PX4/PX4-user_guide · GitHub

  • We had the progress staled for a while, but it is relevant, and we will discuss next week’s call.

Further comments on backporting process here: https://github.com/PX4/PX4-user_guide/pull/2331#issuecomment-1619210692

  • Individual backports are time consuming. Shouldn’t we have one person cherry pick all the PRs to backport at once & merge that beta PR & cut the beta each time?

Feedback: Right now, most delay / problem is merging to main itself, instead of backporting things (which is easier). We don’t need to have one person backport everything (it is also more prone to test / featuer failures, since PR is too big).

The next release after v1.14

Note: We will discuss this in the next week’s meeting.

(next release question)
I fixed all Jenkins errors/checks on:

Would you mind taking a look at it and, maybe, assigning a team member to follow up in the future?
What is the plan? Timeline?

Feedback: We will need to update the flight review for broken compatibility. Changes in QGC should be fine. Beat will give a review on this

According to the MAVLink spec it should behave in the same way as is speced for loss of the remoteID heartbeat - which is to post remoteID malfunction in the LOCATION message status field.

There might be other requirements not in the MAVLink spec I guess.

1 Like