ROS PMC Minutes for April 22, 2025

[mjcarroll]

The ROS PMC meeting for this week was on Tuesday. The notes from the meeting are available at ROS 2 PMC weekly meeting agenda - April 22, 2025

PMC Business

• Formal Decision: @ahcorde is the sole nominee for ROS representative to TGC
• Formal Decision: Add rosidl_rust to REP 2005 (ROS Project Charter - article 4.9.ii)
  • Motion: @wjwwood, Second: @ahcorde, passed by consensus
• There is a TGC Meeting this week, @mjcarroll to circulate update slides

Ongoing Tasks

• Buildfarm:
  • PTAL: Buildfarm Issues Triage Sheet
  • Increase Rci_release timeout 👨‍🌾 Increase timeout for Rci__nightly-release_ubuntu_noble_amd64 by Blast545 · Pull Request #323 · ros2/ros_buildfarm_config · GitHub
    • Increases build times from 300 to 360 minutes on the buildfarm as we have recently been timing out. Likely due to the addition of rmw_zenoh to the test matrix.
    • @cottsay reports that test parallelization should help better help resource utilization once all isolation patches have landed.
  • Test to be disabled + added to kilted board:
    • 🧑‍🌾 `multi_nested_message_instrospection_test` failing consistently in nightly_linux_coverage · Issue #812 · ros2/rosidl · GitHub
  • RHEL rmw_cyclonedds problem
    • 🧑‍🌾 rmw_cyclonedds_cpp: `ddsi_xt_type_init_impl with invalid type object` error in RHEL 9 · Issue #535 · ros2/rmw_cyclonedds · GitHub
  • Reported issues with jetson and ARM compilation:
    • ARM Binaries Not Functioning on Jetson Platforms (Nav2/MPPI) · Issue #1092 · ros-infrastructure/ros_buildfarm · GitHub
    • Addition of new server instances in AWS have more architecture-specific flags than Jetson or RPi boards have.
    • Short term fix is to remove the new AWS instances temporarily, long term fix is to ensure that builds only use a specific subset of arch-flags when building generic binaries.
• Waffle assignments:
  • April 24 + May 1: @MichaelOrlov
• Rosdistro Assignments
  • April 22 - April 29: @cottsay + @mjcarroll
  • Outgoing rosdistro issues: None
• ROS Bosses:
  • Audrow (Humble)
  • Shane (Noetic)
    • Noetic Syncthis week, freeze today
    • @sloretz to close feature PRs on ROS 1 repos
  • Marco (Jazzy)
  • Yadu (Rolling)

Kilted Development

• Next milestone will be branching for kilted and bumps for package versions in rolling.
  • Scheduled for yesterday, didn’t happen
  • Two critical fixes in rmw_desert and rmw_gurumdds to eliminate cyclic dependencies.
  • Connextdds now works in arm64, rebuild started this morning, scaled to meet demand
  • Plan to run migration today once critical mass of arm64 builds
• Next steps:
  • Beta announcement on Monday.
  • We have regressions that need to be addressed

Agenda Items

• @tomoyafujita - ros2 service info -v support
  • What information should be presented when running ros2 service info -v depending on the middleware implementation?
  • add: get clients, servers info by leeminju531 · Pull Request #371 · ros2/rmw · GitHub
  • Conclusion: by default, the user should really on see things that are potentially useful when instancing things in ROS. If there are implementation-specific details, those are still potentially useful, but should be hidden behind a higher verbosity level flag.
• @mjcarroll ROS 2 Migration Pain Points
  • @peci1 has put together a good list of pain points they have run into with ROS 2 migration. Things on this list that aren’t currently represented in issue trackers should probably be turned into issues.
  • This is worth reading over to see if there are easy documentation gaps to address
• @claraberendsen Windows 2025 on ci.ros2.org.
  • We now have a CI machine configured for Windows 2025 (Windows 11)
  • We will run a few nightlies to find regressions between Windows 10 + Windows 11
• @mjcarroll Informal ROSCon Talk Proposal brainstorming

As a reminder, ROS PMC meetings are open to the public, though only committers and members may speak without being called on. If you have topics that you would like discussed, feel free to respond here.

[JRTG]

@mjcarroll

• @peci1 has put together a good list of pain points they have run into with ROS 2 migration.
  […]
• This is worth reading

Where can we find this list?

[christophebedard]

Here: ROS 2 migration pain points · GitHub

[mjcarroll]

Apologies, I must not have copied it over when I was condensing the notes.

For future reference, all of the meeting notes can be found in the agenda/minutes document.

Maybe in the future I will be on par with an AI at generating notes

[peci1]

Thanks for noticing my list I haven’t known I’ve published it yet, but let it be Maybe I leaked the link in some discussion…

I was just writing down notes when I really started to rewrite parts of our codebase. What might be interesting is that although I’ve been closely following many ROS 2 issues for a long time, I haven’t really started working with it until Jazzy. So you might take it as a view of a seasoned ROS 1 dev who is novice to ROS 2.

I definitely wanted to turn them into actual issues, but if someone is faster, I don’t mind =)

I really appreciate the PMCs proactively looking into it. Kudos!

[mjcarroll]

I think I found it linked in a discussion somewhere, but cannot find where at this point. I put it in my list of things to review this week and thought it was interesting enough to surface at the PMC level. Sorry if you weren’t quite ready to circulate publicly, it was just too valuable to pass up.

Many of these are things that we are aware of (and have tracking issues for), but having a concise list from someone I know has a large amount of ROS experience is very helpful in being able to prioritize going forward, especially as we start to think about plans for the next release cycle.

Thanks for putting it together!

[RFRIEDM]

Do we think it might be useful to start linking the “gotchas” in the ROS 1 migration guide for transparency? We got bit by this on our last migration finding unsolved DDS issues and it blew up our timelines/cost for the migration by 2x.

[peci1]

Do you mean docs issues or something else?

[RFRIEDM]

A bit of docs on how to get the middleware functional for our common use case (VIO).