OSRA TGC meeting minutes for March, 2025

General
1

The meeting minutes for the regular meeting of the Technical Governance Committee held on the 27th of March, 2025, were approved in the April meeting and are now available in the official minutes repository. You can find the complete minutes here, as well as a Markdown version below. The PDF version is the official version.

The next TGC meeting will take place in late May, 2025. The minutes of the April meeting should be approved in that meeting and posted publicly shortly thereafter.

TECHNICAL GOVERNANCE COMMITTEE

MEETING MINUTES

27 MAR 2025 / 22:00-23:30 UTC / VIRTUAL VIA ZOOM

ATTENDEES

OSRF:

  • Geoff Biggs (TGC Chair)
  • Kat Scott (Developer Advocate)
  • Vanessa Yamzon Orsi (Secretary)
  • Vicky Brasseur (OSRA Governance Advisor)

OSRA Delegates/TGC Reps:

  • Bence Magyar (ros-controls PMC Lead)
  • Carlos Aguero, (Gazebo Lead Proxy)
  • Clara Berendesen (Infra PMC Representative)
  • Christian Henkel (Gold Rep, Bosch)
  • Daniel Gordon (Gold Rep, Huawei)
  • David Lu!! (Individual Rep)
  • Hemal Shah (Platinum Rep, NVIDIA)
  • Maria Vergo (Silver Rep, ROS-I)
  • Michael Carroll (ROS PMC Interim Lead)
  • Michael Grey (Open-RMF)
  • Michel Hidalgo (Silver Rep, Ekumen)
  • Steve Peters (Gazebo)
  • Steven! Ragnarok (Infrastructure PMC Lead)
  • Tully Foote (Platinum Rep, Intrinsic)
  • Zhaoyuan Cheng (Platinum Rep, Qualcomm)

OSRA Observers

  • Carmelo Di Franco (Aitronik)
  • Henning Kayser (PickNik)
  • Nick Hehr (Viam)
  • Tom Panzarella (Seegrid)
  • Todd Sutton (Zachry)

AGENDA

Geoff Biggs, TGC Chair called the meeting to order at 22:06 on 27 Mar 2025 UTC

Existing Business

  1. Confirmation of minutes of meeting TGC-20250227-Regular

    • Approved by consensus

  2. Project reports

    • Project update: ROS
      • Presented by Michael Carroll, PMC Lead
        • Committer updates:
          • Christophe Bedard from Apex has proposed to mentor Emerson Knapp from Polymath Robotics to move up from Committer to PMC member
          • Janosch Machowinski Is now a full Committer for the ROS project
        • Official votes coming up:
          • Change needed in REP-2005 to rename FastRTPS
          • Remove ROS Geometry 2 from core-maintained repositories
          • Proposed schedule for ROS 1 Noetic EOL - tentatively scheduled for May 31, 2025
        • ROS Developments:
          • Completed the pixi installation for Windows CI and user installations; great community feedback received for ease of use
          • 1.5 weeks away from Kilted feature freeze, working on getting rmw-zenoh to Tier 1 status beforehand
          • Also working on FastDDS updates to 3.2.x, action introspection, rclpy executor implementation
          • Moving python launch files to default the community to xml and yaml
          • Common simulation interfaces pull request completed
        • GSoC season underway
        • docs.ros.org rebrand has launched
        • Key Kilted release dates
        • PMC will start L-Turtle roadmap, will seek TGC input
    • Project update: Gazebo
      • Presented by Carlos Aguero, PMC Member on behalf of the Project Leader
        • Launched effort to collect data on CI jobs in order to understand trends and usage to improve efficiency on allocating resources
        • Testing migration from BC package to Conda to reproduce cloud environment for testing and debugging
        • Prototype for zenoh on gz-transport considered complete, includes basic pubsub and basic services. Design document circulated to the community for feedback.
        • Efforts to migrate from qt5 to 6 in the summer
        • Exploring possibility of removing immersion numbers from package names to reduce overhead for releases; proposal will remove ability to run side-by-side versions of Gazebo, considering implications vs more efficiency for releases
        • Reducing discrepancy between optical frames between ROS and Gazebo via the rosbridge
        • Deprecating gz launch in favor of other more used tools, tutorials now available for other methods
        • Efforts to increase contributions - curated list of issues from maintainers, focused on making them more scoped and prioritized per repo then made public to the community. Reception has been great and issues are being crossed off.
        • GSoC season in progress
    • Project update: Open-RMF
      • Presented by Michael Grey, PMC Lead
        • Runtime generation of workflows now supports buffers useful for synchronizing multiple threads that are running simultaneously.
        • We will soon support a concept of sections for runtime generated workflows
        • Adding new capability called dynamic events to allow external support for more complex behaviors as requested by the community
    • Project update: ROS Controls
      • Presented by Bence Magyar, PMC Lead
        • Started adding organization-wide templates for issue report, feature request and useful websites following the ROS project
        • Created a ros2 control cmake package and to help move users from Humble, also created gz ros2 control in addition to ignition ros2 control for seamless migration
      • Excited about the structured parameter handling project
      • Participating in GSoC with two projects
      • Reiterated call for input on adding joint limit support
    • Project update: Infrastructure
      • Presented by Steven! Ragnarok, PMC Lead
        • ROS Build farm now runs Ubuntu 24.0.4
        • Support for EmPy 4
        • Support for non-GPU Windows agents on the Gazebo Build farm
        • Build farmers working on a scaling schedule to try and optimize resources and improve throughout on the Gazebo CI
        • Assessed security risks from a broad software supply chain compromise in a popular GitHub action
          • Conclusion: no impact on OSRF projects, tracked down instances and prevented any chance of compromise
        • RedHat’s Fedora Robotics SIG is exploring the possibility of getting Fedora built binaries on the official ROS Build farm
        • Upcoming: Getting ROS migration distro script ready to be used for Kilted branching, moving snapshot and bootstrap servers, getting Windows 11 compatible server running on the ROS 2 CI farm

  3. Existing Business

    • Status of SIGs/TCs

      • Technical Committee on Enhancement Proposal Processes

        • Goal: Upgrade and improvements over REP
        • Chair: Geoff Biggs
        • Members: PMC Leads
        • Update:

          • Geoff presented the current draft of the Enhancement Proposal Process steps, from drafting to ratification. The high-level planning is now complete.

          • Next steps:

            • Documentation of detailed process description, including implementation, tools, steps, discussion forums, document archiving, example/s.

            • Feedback from PMCs and TGC

          • Vote requested: Approve extension of TC to June 2025 and update members (removing 2 who are now unavailable and adding 1 new member)

            • Approved by consensus

      • SIG on Interoperability

        • Chair: Michael Grey
        • Grey presented a preview of how async programming will work on rclrs, observed uptick in activity of community users on rclrs GitHub

    • Results of vote to recommend adoption of GAI policy

      • Following the failed motion prior to the previous TGC meeting, this item needs to be revisited and a second attempt made at explicitly accepting or rejecting the proposed Generative AI Policy and Guidance. No changes have been made since the previous vote.

        • Approved by consensus

    • Discussion on communications and processes in the TGC

      • An in-depth discussion on the communications methods and processes used in the TGC was held.

      • Data from the survey sent to all TGC Members was discussed, including pros and cons of various platforms, what features are requested, effort to standardize across projects and groups if possible to allow for easier oversight and cross-collaboration.

      • Suggestions for improving interactions during TGC meetings.

      • Explored ideas for asynchronous communication and collaboration.

      • Discussed processes around decision-making, with the goal of understanding how we can ensure all TGC Members have enough time to participate fully. Issues discussed include: time to discuss with their constituents (other Gold and Silver OSRA members, their Project’s PMC, or their organisation, depending on the TGC Member); how votes are carried out within and outside of meetings; inbuilt process delays between raising a topic and making the decision; enabling dissent; open proposals of agenda items; and increasing between-meeting discussion amongst TGC Members.

      • Location of pertinent documents and reference materials for items discussed during meetings.

        • Action Item: The TGC Chair will write up the items discussed and circulate to the TGC for further review and iteration.

New Business

  • None

Other Items

  • None

Adjournment - Geoff Biggs adjourned the meeting at 22:56 UTC.

ACTION ITEMS

  1. The meeting minutes will be circulated to the TGC for review and approval before the next meeting by the TGC Chair.
  2. The TGC Chair will write up the items discussed regarding communication methods and procedures within the TGC/OSRA and circulate to the TGC for further review and iteration.
1 Like
2

Could anyone provide some context for this please?

3 Likes
3

Yep! I think in this case, the notes are actually incorrect.

The vote was to remove ros/geometry, which is the ROS 1 only version of the package. ros2/geometry2, the ROS 2 port of the geometry package remains maintained and under the PMC.

This was mostly a bookkeeping note since Noetic will be winding down at the end of next month, and pure ROS 1 repos don’t make sense under PMC authority.

The repo will remain, though I believe the plan of record is to put it into an archived, read-only state.

1 Like
4

Thank you! That seems clearly more reasonable and no problem. Of course some people will still not be happy because especially the Python-API of tf remains much more intuitive to use than in tf2_ros and family, but that is a ROS2 design decision and unrelated.

Shout out to @DLu for providing tf.transformations in ROS2 still. :sports_medal:

1 Like
5

Of course some people will still not be happy because especially the Python-API of tf remains much more intuitive to use than in tf2_ros and family,

Do you have an issue that I can use to get more context on this? I don’t know that I’ve seen this feedback.

6

Thanks for asking, and sorry to clutter yet another thread with progress reports on all fronts by discussing tiny details and idealism.

There are no issues I’m aware of because the concerns are - by now old - design decisions.

  • My personal qualms are with the very powerful modular transform method, which does not behave pythonic: Importing syntactically unused packages is required for correct behavior. It can be quite annoying trying to find actually used implementations, also in C++ code bases.
  • Others expressed their preference for a single listener structure in the past, and tf2 famously splits the separate buffer and ROS updater concerns into two objects.

Beyond these two things, tf example use and tf2 example use are naturally quite similar.

  • The third aspect comparing packages though is obviously the tf.transformations library which was dropped (in theory rightly so) in favor of external libraries. I believe @Dlu’s fork and README speaks for itself though and the library has further relevance.
2 Likes
7

FWIW I personally think ‘tiny details’ presented as you have and properly clustered are what build to big improvements. It’s easy for people to make high level abstract statements about issues, but those high level statements are nearly unusable without further context.

Conversely, specific comments like yours, @peci1 's recent-ish ‘pain.md’ document, and other comments of the ilk (Offhand I recall @MoffKalast has done some similar lists) are far easier for people (whether ‘officially’ PMC/TGC/WG/SIG members or not) to parse, categorize, and prioritize.

3 Likes
8

Ah thanks for letting me know there’s a proper list, I’ll see if I can add my collection of nitpicks to it so we have it all in one place.

Even if it doesn’t lead to any fixes in the short term given that a lot of them are heavily ingrained design decisions as v4hn mentions, it’ll be very useful so we can link it the next time Katie decides to go on another Noetic bashing spree :winking_face_with_tongue:

2 Likes