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.
Meeting Link
Agenda
Finalize maintainer role description
Identify maintainer opportunities
Open mic for maintainers
Meeting Notes
HITL Issues
There are a lot of inconsistencies with the HITL implementation, we should mark this feature as experimental in the documentation unless we can find a maintainer to make it stable. We need to add HITL to our CI infrastructure. TODO: Contributors Needed.
The Gazebo classic HITL implementation was broken when the actuator changes came in, the HITL plane model is reported to be working well by @Jaeyoung-Lim
Simulation build targets
With the introduction of the new Gazebo simulator as the default simulator solution for PX4, we were forced to rename our previous gazebo implementation as Gazebo Classic. This forced us to adopt new build targets independent for each, which created confusion for upgrading users not being able to build the simulation targets they were expecting. Users having problems should be directed to the list of gazebo vehicles for the new implementation or to the classic gazebo list for the older implementation. Please note that the vehicle models are not shared between implementations. The models are specific to a simulator.
PX4 v1.14 Release Discussion
We are happy with the state of PX4 v1.14-rc2 and are moving forward with a silent, stable release. We are going to need to coordinate don’t the following steps:
Release notes: we have a version 99% ready to go, and @rroche will be sending a PR to the docs and will also be in charge of updating the GH release tag.
Docs versioning:@hamishwillee, we know this takes time, can you please advise on the best timeline for you to kick off the versioning? We are ready whenever you are. (finally!)
Release announcement: we are likely going to hold off an announcement until the PX4 Developer Summit 2023 and try to get the news in front of as many as possible. @rroche is also taking the lead here
There are a lot of inconsistencies with the HITL implementation, we should mark this feature as experimental in the documentation unless we can find a maintainer to make it stable.
The Gazebo classic HITL implementation was broken when the actuator changes came in, the HITL plane model is reported to be working well by @Jaeyoung-Lim
So to confirm:
“All” HITL is experimental in PX4 v1.14 and current main?
What does it mean to say Gazebo Classic was broken but Plane works well? Does this mean it was broken but is now fixed in v1.14? Or does it mean that it is broken generally, but you can still use Plane if you like?
Does HITL work with Gazebo (Gz)?
Does HITL work with ROS2?
Generally “what is supported and what is not”.
I should fix this before taking a docs cut.
W.r.t. Gazebo confusion, this is inevitable from the renaming. As a result of this though I have added a warning up the top of the Gazebo (new) page pointing to gazebo classic vehicles and noting that the vehicle models are not interchangeable.
@rroche How was the SET_ACTUATOR mavlink issue resolved? Did someone confirm this does actually work? If this wasn’t fixed there needs to be a docs fix.
Release notes: we have a version 99% ready to go, and @rroche will be sending a PR to the docs and will also be in charge of updating the GH release tag.
@rroche If you can do your release note PR today I can merge it and do a docs cut tomorrow for PX4 v1.14.
Docs versioning:@hamishwillee, we know this takes time, can you please advise on the best timeline for you to kick off the versioning? We are ready whenever you are. (finally!)
@rroche As soon as we have clarified what is experimental/supported and you have done release note I can take a docs cut. That would be best done tomorrow because that gives me a week of tidy up before heading off to DevDay the following week.
I could do it next week, but better to de-risk it by doing it earlier. Is getting the docs PR done today/tomorrow possible?
@mwbb@rroche@dagar We’re not going to maintain jMAVSim long term, so the designation should be deprecated not experimental (one is theoretically “on the way up” and you’d expect it to be fixed, while the other is “on the way down” and perhaps you wouldn’t).
The important there here is to set expectation appropriately and enable a smooth transition for anyone that needs it. Would it be OK to add a note in the v1.14 release note indicating that we expect to drop support in the next release (say)?
I plan to look at removing jMAVSim as the default “:try this first simulator” in the main build guides this week (and in the near future refocus simulation around “Gazebo first” - as discussed with @dagar at dev conference).
I would be happy to have everyone else’s inputs on this.
I might be a bit biased about removing jMAVSim since that had always been the solution to go for me on macOS so far. I have never had any successful experience with Gazebo on macOS!
I might be a bit biased about removing jMAVSim since that had always been the solution to go for me on macOS so far. I have never had any successful experience with Gazebo on macOS!
JMAVSim has been my consistent go-to for the last 6 years because most of the time it works, and those times it hasn’t, fixing the problems has mostly been about finding a good version of Java to install. It’s fast to load and it works on all the platforms. Gazebo Classic by contrast has sucked. Gazebo sucks, but with less vehicle types supported.
@dagar says these are fixable problems. If so that’s a good way forward. I am concerned that no one seems to have any idea about macOS. Perhaps we should drop support for that too?
But there are still many using MacOS and also some are swithing to cause of arm64 support including me! How much effort it needs to be maintained?
@mwbb I don’t know, and we can’t ask because it has no maintainer. What we need first is for the setup scripts to be part of the PX4 tree and run in CI so that it can be maintained. Right now now one seems to know anything about it.
October 03, 2023
Meeting Link
Agenda
Meeting Notes
How much effort it needs to be maintained?