ROS PMC Minutes for August 18th, 2026

<p>The <a href="https://osralliance.org/staging/wp-content/uploads/2024/03/ros_project_charter.pdf">ROS PMC</a> meeting for this week was on Tuesday, August 18, 2026. The notes from the meeting are available at <a href="https://docs.google.com/document/d/1TGXYp33N00ZJtilsRzfClFXcSyMa4QKVNmA03Lx9D7w/edit?tab=t.0">ROS PMC weekly meeting agenda</a>.</p> <h2><a class="anchor" href="https://discourse.openrobotics.org#p-117269-ros-pmc-business-1" name="p-117269-ros-pmc-business-1"></a>ROS PMC Business</h2> <ul> <li><strong>ROSCon is one month away.</strong> <ul> <li><strong>Presentations are due Monday, August 25, 2026.</strong></li> <li><a class="mention" href="https://discourse.openrobotics.org/u/mjcarroll">@mjcarroll</a> is happy to review slides or run through dry runs for anyone who wants feedback. Reach out in the private PMC chat if you would rather not spoil your talk.</li> <li>The recent speaker session recording is now up on the Open Robotics Vimeo account.</li> </ul> </li> <li><strong>Membership changes proposed.</strong> Nothing is decided yet. These will go out as an asynchronous vote, one item at a time, with approve, reject, or abstain options. The aim is to have this settled before ROSCon. <ul> <li>Proposed for PMC membership: <ul> <li><a class="mention" href="https://discourse.openrobotics.org/u/skye.galaxy">@skye.galaxy</a>, sponsored by <a class="mention" href="https://discourse.openrobotics.org/u/emersonknapp">@emersonknapp</a></li> <li><a class="mention" href="https://discourse.openrobotics.org/u/jmachowinski">@jmachowinski</a>, sponsored by <a class="mention" href="https://discourse.openrobotics.org/u/sloretz">@sloretz</a></li> <li><a class="mention" href="https://discourse.openrobotics.org/u/andrew_symington">@andrew_symington</a>, who put his name forward in the meeting. <strong>Still needs a sponsor.</strong></li> </ul> </li> <li>Proposed for emeritus status: <a class="mention" href="https://discourse.openrobotics.org/u/nuclearsandwich">@nuclearsandwich</a> and <a class="mention" href="https://discourse.openrobotics.org/u/audrow">@audrow</a>, both of whom have stepped back from active participation. This is housekeeping that has been outstanding for a while rather than any change in circumstance.</li> <li>Net effect would be roughly 17 voting members.</li> <li><a class="mention" href="https://discourse.openrobotics.org/u/emersonknapp">@emersonknapp</a> asked whether there is an upper limit on voting members. There is no target number. Quorum was a real constraint earlier this year, but moving to asynchronous votes with a week-long window has largely relieved that pressure. If the PMC eventually grows past what is workable, splitting it is the conversation, as discussed in the project handbook.</li> <li>On sponsorship: a sponsor is required by process, though for candidates who are already long-standing committers there is not much mentoring left to do. <a class="mention" href="https://discourse.openrobotics.org/u/mjcarroll">@mjcarroll</a> will re-read the governance documents to confirm whether a committer can sponsor another committer, or whether that is restricted to PMC members.</li> </ul> </li> <li><strong>New committers proposed.</strong> This came out of an audit of who is actually contributing versus who currently holds committer status, and these were the obvious gaps. <ul> <li><a class="mention" href="https://discourse.openrobotics.org/u/kimmcg">@KimMcG</a>, currently a committer in Infrastructure, who contributes substantially to ROS 2 and would benefit from write and CI access on the ROS side, particularly for CI and Windows work.</li> <li>Nadav Elkabets, who has done significant work on the asyncio executor in <code>rclpy</code> and is presenting on it at ROSCon. Write and CI access in <code>rclpy</code> at minimum.</li> <li>Timple, who appears across many ROS repositories and has expressed interest</li> <li>Martin Pecka.</li> <li><strong>Scoped access going forward.</strong> Rather than granting broad access by default, new committers should start with a defined scope. This cuts two ways: it is what you have write access to and are trusted to merge and release in, but it also gives Waffle triage a name to route incoming work toward.</li> <li><strong>Volunteers wanted</strong> to sponsor or mentor these candidates. Reach out to <a class="mention" href="https://discourse.openrobotics.org/u/mjcarroll">@mjcarroll</a> asynchronously. Bringing people on as a batch should also help us shore up the documentation and process around it, which is a muscle the project has not flexed enough. Committers are low cost to add and add a great deal of value.</li> </ul> </li> <li><strong>Bazel Working Group.</strong> <a class="mention" href="https://discourse.openrobotics.org/u/andrew_symington">@andrew_symington</a> will draft a charter and lead the group, following on from the elections. Targeting pre-ROSCon so it can anchor build system conversations there. <a class="mention" href="https://discourse.openrobotics.org/u/emersonknapp">@emersonknapp</a> encouraged getting a first draft down without setting too high a bar, since developing the tooling is exactly what the working group is for.</li> <li><strong>Rolling sync automation.</strong> <a class="mention" href="https://discourse.openrobotics.org/u/cottsay">@cottsay</a> is back from vacation, and <a class="mention" href="https://discourse.openrobotics.org/u/yadunund">@yadunund</a> will set up time with him this week to move the implementation forward.</li> </ul> <h2><a class="anchor" href="https://discourse.openrobotics.org#p-117269-regular-business-2" name="p-117269-regular-business-2"></a>Regular Business</h2> <h3><a class="anchor" href="https://discourse.openrobotics.org#p-117269-buildfarm-update-3" name="p-117269-buildfarm-update-3"></a>Buildfarm update</h3> <p>The buildfarm builds and tests ROS packages across all supported distributions and platforms. This section covers new issues, current priorities, and infrastructure changes that affect maintainers.</p> <ul> <li><img alt=":new_button:" class="emoji" height="20" src="https://emoji.discourse-cdn.com/noto/new_button.png?v=15" title=":new_button:" width="20" /> Issues <ul> <li>No new issues opened this week.</li> </ul> </li> <li><img alt=":warning:" class="emoji" height="20" src="https://emoji.discourse-cdn.com/noto/warning.png?v=15" title=":warning:" width="20" /> High priority items already open <ul> <li><strong>RMW build regression.</strong> The first PR merged but the backports did not, so the dashboard is still red. <a class="mention" href="https://discourse.openrobotics.org/u/mjcarroll">@mjcarroll</a> will merge the remaining three, which are ready to go.</li> <li><strong><code>test_launch</code> failures.</strong> One PR merged with others pending, and a Windows job failed again today. The behavior has changed from a segfault to a test timeout, so the situation improved without fully resolving. <a class="mention" href="https://discourse.openrobotics.org/u/kimmcg">@KimMcG</a> raised whether the ROS 2 repos point at a released Fast DDS version rather than master; the relevant branches turned out to be synced, so something else is in play. <a class="mention" href="https://discourse.openrobotics.org/u/mjcarroll">@mjcarroll</a> will work through it with <a class="mention" href="https://discourse.openrobotics.org/u/miguelcompany">@MiguelCompany</a>.</li> <li><strong>Performance package regressions</strong>, possibly a Cyclone problem. The natural owner has very little availability for Cyclone work at present. <a class="mention" href="https://discourse.openrobotics.org/u/wjwwood">@wjwwood</a> volunteered to pick it up.</li> <li><strong><code>ros2topic</code> failure.</strong> The suggested Cyclone fix did not resolve it and it is still reproducing consistently. <a class="mention" href="https://discourse.openrobotics.org/u/miguelgonrod">@miguelgonrod</a> will note that on the issue.</li> </ul> </li> <li><img alt=":locked:" class="emoji" height="20" src="https://emoji.discourse-cdn.com/noto/locked.png?v=15" title=":locked:" width="20" /> <strong>Infrastructure notice from <a class="mention" href="https://discourse.openrobotics.org/u/claraberendsen">@claraberendsen</a>.</strong> <a href="http://build.ros.org">build.ros.org</a> was hit hard by bots last week. A web application firewall deployed on Sunday has stabilized things. <ul> <li><strong>You will now encounter the WAF when fetching test reports.</strong> If this blocks any of your workloads, say so and the buildfarmers will accommodate you. Scope will keep tightening as bots find their way back in.</li> </ul> </li> </ul> <h3><a class="anchor" href="https://discourse.openrobotics.org#p-117269-waffle-note-taker-4" name="p-117269-waffle-note-taker-4"></a>Waffle note-taker</h3> <p>Waffle is a weekly meeting where maintainers triage incoming issues and pull requests that haven’t been otherwise acted upon. One person per week is assigned to be the “waffle note-taker” to both take notes in the meeting to keep the meeting as efficient as possible.</p> <ul> <li>Aug 20, 2026 <ul> <li><a class="mention" href="https://discourse.openrobotics.org/u/andrew_symington">@andrew_symington</a></li> <li>Note that <a class="mention" href="https://discourse.openrobotics.org/u/andrew_symington">@andrew_symington</a> is away from after this week until mid-September, so the next three weeks need cover.</li> </ul> </li> </ul> <p>Last week went well, with a number of new participants and work assigned to people outside the usual group, which is the point. Thanks to <a class="mention" href="https://discourse.openrobotics.org/u/skye.galaxy">@skye.galaxy</a> for running it.</p> <p><a class="mention" href="https://discourse.openrobotics.org/u/katherine_scott">@Katherine_Scott</a> raised the question of closing the loop: knowing whether assigned items actually get worked on between meetings. <a class="mention" href="https://discourse.openrobotics.org/u/andrew_symington">@andrew_symington</a> is willing to build something if we can define what “done” means, whether that is merged, responded to, or reviewed, but flagged that the meeting is already time-compressed. The consensus was that follow-up does not need to happen in the meeting; a list to check against afterward would be enough.</p> <h3><a class="anchor" href="https://discourse.openrobotics.org#p-117269-rosdistro-assignmentshttpsgithubcomrosrosdistro-5" name="p-117269-rosdistro-assignmentshttpsgithubcomrosrosdistro-5"></a><a href="https://github.com/ros/rosdistro">Rosdistro assignments</a></h3> <p>Rosdistro is a weekly assignment to monitor incoming issues and pull requests in <a href="https://github.com/ros/rosdistro">GitHub - ros/rosdistro</a>. Two people per week are assigned to be the rosdistro maintainers.</p> <ul> <li>Aug 18, 2026 - Aug 25, 2026 <ul> <li><a class="mention" href="https://discourse.openrobotics.org/u/tomoyafujita">@tomoyafujita</a></li> <li><a class="mention" href="https://discourse.openrobotics.org/u/cottsay">@cottsay</a></li> </ul> </li> <li>Outgoing rosdistro <ul> <li><a class="mention" href="https://discourse.openrobotics.org/u/sloretz">@sloretz</a></li> <li><a class="mention" href="https://discourse.openrobotics.org/u/emersonknapp">@emersonknapp</a></li> <li>The Creative Commons licensing question remains open with no progress made.</li> <li><a class="mention" href="https://discourse.openrobotics.org/u/emersonknapp">@emersonknapp</a> added a <strong><code>reviewer discussion</code> label</strong> to mark issues where we said we would come back with an answer and have not yet.</li> <li><strong>Gurum DDS</strong>: already went through the import process and has OSRF approval, so this just needs the version bump. <a class="mention" href="https://discourse.openrobotics.org/u/cottsay">@cottsay</a> will follow up.</li> <li><strong>Swift DDS (Greenstone)</strong>: blocked. Not an open source license, so importing the binaries requires a specific agreement, and the repository is a folder of libraries. <a class="mention" href="https://discourse.openrobotics.org/u/tfoote">@tfoote</a> suggested the requester contact <a href="mailto:info@openrobotics.org">info@openrobotics.org</a>.</li> </ul> </li> </ul> <h3><a class="anchor" href="https://discourse.openrobotics.org#p-117269-release-managementhttpsosrfzulipchatcomnarrowchannel526029-ros-pmctopicrelease20management20thread-6" name="p-117269-release-managementhttpsosrfzulipchatcomnarrowchannel526029-ros-pmctopicrelease20management20thread-6"></a><a href="https://osrf.zulipchat.com/#narrow/channel/526029-ROS-PMC/topic/Release.20Management.20Thread">Release Management</a></h3> <p>Release Management tracks the status of upcoming releases, patches, and syncs across the active ROS distributions.</p> <ul> <li><a class="mention" href="https://discourse.openrobotics.org/u/christophebedard">@christophebedard</a> (Humble) <ul> <li>Synced recently.</li> </ul> </li> <li><a class="mention" href="https://discourse.openrobotics.org/u/yadunund">@yadunund</a> (Rolling) <ul> <li>Automation work restarting this week with <a class="mention" href="https://discourse.openrobotics.org/u/cottsay">@cottsay</a>.</li> </ul> </li> <li><a class="mention" href="https://discourse.openrobotics.org/u/marcogg">@marcogg</a> (Jazzy) <ul> <li><img alt=":warning:" class="emoji" height="20" src="https://emoji.discourse-cdn.com/noto/warning.png?v=15" title=":warning:" width="20" /> Roughly two months since the last sync. <a class="mention" href="https://discourse.openrobotics.org/u/mjcarroll">@mjcarroll</a> will reach out.</li> </ul> </li> <li><a class="mention" href="https://discourse.openrobotics.org/u/cottsay">@cottsay</a> (Kilted) <ul> <li><img alt=":warning:" class="emoji" height="20" src="https://emoji.discourse-cdn.com/noto/warning.png?v=15" title=":warning:" width="20" /> Roughly two months since the last sync, with one expected soon.</li> </ul> </li> <li><a class="mention" href="https://discourse.openrobotics.org/u/sloretz">@sloretz</a> (Lyrical) <ul> <li><strong>Sync tomorrow, August 19, 2026</strong>, a normal sync rather than a patch release, with around 250 updates.</li> <li>Target cadence is every two weeks. Provisionally: <strong>freeze August 31 with a sync September 2, then freeze September 14 with a sync September 16</strong>, regressions permitting. That gives us potentially two more Lyrical syncs before ROSCon.</li> <li><strong>Make your releases at the end of this week</strong> if you want maximum time in the testing repository ahead of the next sync.</li> <li><span class="mention">@nvcyc</span> noted bugs found in the CUDA buffer backend that are important to land before ROSCon so people are exercising the fixed version.</li> </ul> </li> <li>Patch syncs are also being sought by <a class="mention" href="https://discourse.openrobotics.org/u/emersonknapp">@emersonknapp</a> and <a class="mention" href="https://discourse.openrobotics.org/u/skye.galaxy">@skye.galaxy</a>.</li> </ul> <h2><a class="anchor" href="https://discourse.openrobotics.org#p-117269-working-group-updates-7" name="p-117269-working-group-updates-7"></a>Working Group Updates</h2> <ul> <li>Client Library Working Group (<a class="mention" href="https://discourse.openrobotics.org/u/alsora">@alsora</a>) <ul> <li>Meeting this week. <a class="mention" href="https://discourse.openrobotics.org/u/skye.galaxy">@skye.galaxy</a> has added items on executors and pybind.</li> <li>Also on the agenda: the shared memory <code>rosidl</code> buffer work recently posted to Discourse, which is interesting because it could potentially remove the intra-process manager from C++ entirely. Early numbers look flatter and more stable, though intra-process still shows somewhat lower latency, within the same order of magnitude.</li> </ul> </li> <li>Accelerated Memory Transport Working Group (<a class="mention" href="https://discourse.openrobotics.org/u/ahcorde">@ahcorde</a>) <ul> <li>Meeting this week.</li> </ul> </li> <li>ROSGraph Working Group (<a class="mention" href="https://discourse.openrobotics.org/u/emersonknapp">@emersonknapp</a>) <ul> <li><strong>This week’s meeting moved from Tuesday to Thursday</strong> due to the Actuate conference. Back to the regular schedule in two weeks.</li> <li>Nodal is making steady progress, and work has started on the conference presentation.</li> <li>Graph monitor upgrades in progress, with the goal of producing nodal descriptions of an entire running graph automatically by observing it, rather than running one-off tooling per package or per node.</li> <li><strong>Topic statistics prototype</strong> using trace tools instead of a dynamic library shim to intercept messages, since the RMW discussion did not favor introducing the shim as a core feature. The monitor side is unchanged because the message contract stays the same. Not ready for review yet. <a class="mention" href="https://discourse.openrobotics.org/u/jmachowinski">@jmachowinski</a> has a PR removing the existing topic statistics implementation outright, which <a class="mention" href="https://discourse.openrobotics.org/u/emersonknapp">@emersonknapp</a> agrees is the right move given it does not do anything useful outside toy cases. Open questions: whether a deprecation cycle is needed, and platform coverage, since the trace tools approach may not work on Windows. The design intent is a swappable ingest point so different backends can be used per platform.</li> <li><strong>Prototype GitHub Actions wrapping the <code>ros_buildfarm</code> scripts</strong>, so anyone can run buildfarm-like CI on their own runners before releasing, and catch the build issues that typically only surface after bloom. A secondary benefit would be producing debs for internal or side-channel mini-distros. <strong>If you have already built something like this, or want to help, please get in touch.</strong></li> <li><a class="mention" href="https://discourse.openrobotics.org/u/yadunund">@yadunund</a> asked whether a shared graph cache across nodes on the same host is in scope. Every participant currently maintains its own. There may be efficiency gains in sharing one, and Zenoh’s router already holds something similar, though any solution should be RMW-agnostic. <a class="mention" href="https://discourse.openrobotics.org/u/yadunund">@yadunund</a> and <a class="mention" href="https://discourse.openrobotics.org/u/emersonknapp">@emersonknapp</a> will discuss separately.</li> </ul> </li> </ul> <h2><a class="anchor" href="https://discourse.openrobotics.org#p-117269-agenda-8" name="p-117269-agenda-8"></a>Agenda</h2> <ul> <li>[<a class="mention" href="https://discourse.openrobotics.org/u/katherine_scott">@Katherine_Scott</a>] <strong>ROSCon hackathon: Friday, September 25, 2026</strong> <ul> <li>Confirmed and going ahead at the Google Toronto office, next door to the ROSCon venue. Space for just over 100 people.</li> <li>Full day, starting around 9:00 or 9:30 with an introduction to reviewing and submitting pull requests across the OSRA projects, then open time for people to review PRs and fix issues.</li> <li><strong>Everyone: start collecting issues, pull requests, and good first issues</strong> that can be handed out to participants on the day. This is the main thing needed from the PMC right now.</li> <li>Informal prizes are planned, likely a one-off sticker or swag run.</li> <li>The Luma invite is out and everyone is welcome to register. <strong>Important: your Luma registration name must match your official ID</strong>, or Google office security will not let you in. Message <a class="mention" href="https://discourse.openrobotics.org/u/katherine_scott">@Katherine_Scott</a> if there is a mismatch.</li> <li>If you have not booked ROSCon travel, plan to be there Friday. If you have, see whether you can extend.</li> <li>Separately, the SIG AI announcement is going out on LinkedIn shortly. <strong>Please help spread the word.</strong> Details will also go to Zulip.</li> </ul> </li