Singapore ROS 2 peer learning: compare publisher and subscriber QoS

The problem

A robotics builder has matching ROS 2 topic and message-type labels but has not checked whether the publisher’s offered QoS policies satisfy the subscription’s requested policies; a partial policy match can be mistaken for established communication.

Check the communication pair before choosing a robotics session

Matching a topic name and message type leaves a specific ROS 2 question open: can this publisher’s offered communication policies satisfy this subscription’s requested policies? Quality of Service, or QoS, is a collection of policies forming a profile. Prepare that pair for discussion before asking a peer to explain an absent sensor reading. Identify the interface observation, accepted material and participant task, then read this specific communication contract before drawing a deployment conclusion. An absent reading alone does not identify a cause, and a handwritten compatibility comparison is separate from an observed connection or delivered message.

• Prepare the interface observation and permitted material →

Read offers and requests in the correct direction

The official Jazzy documentation uses a request-versus-offer model and requires every policy affecting compatibility to be compatible. Its reliability table allows a reliable publisher with a best-effort subscription; a best-effort publisher with a reliable subscription is incompatible. These directions are therefore different. Its durability table also marks a volatile publisher with a transient-local subscription incompatible. Passing a reliability row cannot override a failing durability row. The source identifies Jazzy as an older supported distribution at the recorded check. Bind your learning case to its actual distribution and relevant documentation rather than assuming another version or middleware uses all the same settings or defaults.

• ROS 2 Jazzy documentation: Quality of Service settings ↗ · Source checked 21 October 2026

Build one comparison for each publisher and subscription

Use seven entries: named distribution and exercise context; publisher identity; subscription identity; each policy offered; corresponding policy requested; documented comparison result; and connection or delivery evidence actually supplied. Add an unknown row for any compatibility condition the material omits, including deadline, liveliness or lease duration where unresolved. Keep a profile label beside its established values instead of filling missing fields from memory. The source notes that middleware defaults can differ. Compare each endpoint pair on its own: another subscription’s compatible row does not supply this one’s missing policy. Preserve the input version and whether it is an authorised extract or a wholly invented case. The card is a discussion aid, with no live inspection or configuration action implied.

Complete a fictional two-subscription comparison

All endpoints, settings and results in this example are stipulated for a handwritten Jazzy learning case. Publisher P offers best-effort reliability and volatile durability. Subscription A requests best effort and volatile: both examined rows are compatible. Its other compatibility policies and actual connection remain unknown. Subscription B requests reliable and transient local: the reliability row is incompatible, and the durability row is also incompatible. Under those stated policy inputs, B has two established compatibility failures; A has two passing rows and an incomplete overall answer. The case supplies no network observation, sensor reading or operating robot. It also supplies no basis for treating B’s failed comparison as a failure of A’s independent pair.

Keep a one-row improvement at its actual scope

Now consider another invented paper comparison in which P offers the reliable policy while remaining volatile. B still requests reliable and transient local. The reliability row now passes, but the durability row still fails. The pair therefore remains incompatible on durability, despite the passing reliability row. This comparison is not an instruction to change either endpoint. Which policies an actual task needs remains a separate requirement question. Likewise, A’s earlier two passing rows cannot establish every other policy or prove that messages arrived. Retain three entries separately: the documented row result; overall compatibility marked compatible, incompatible or incomplete; and any observed connection or delivery. Use an incompatible row to formulate a precise question, while preserving any other possible explanation for a real symptom.

Choose an exercise that lets you examine the policy relationship

Compare the actual participant task with this missing answer. In an invented shortlist, Offer C demonstrates a robot and accepts general questions, without giving participants endpoint settings. Offer D explicitly supplies a publisher/subscription policy worksheet and lets participants explain which rows match, fail or remain unknown. D provides the stated comparison task; its prerequisites, material access and admission still need confirmation. C can provide context without supplying that exercise. You could ask, “I want to compare publisher offers with subscription requests for this permitted ROS 2 case, including a reliability match with a durability mismatch. Does the session examine those pairs, which distribution is used and what preparation is required?” Use the organiser’s actual contribution route; a topic match does not allocate individual troubleshooting.

Carry the remaining question into a focused search

Retain the endpoint pair, supplied policy values, source version, row results and exact missing condition after a discussion. In the fictional B comparison, explaining the remaining durability mismatch completes a reasoning task; it records no changed setting or successful communication. Seek the relevant current documentation or responsible technical route when the actual case needs further evidence. Use Boardzip’s free ROS 2 event search for candidates, then check whether they offer this compatibility exercise, the starting knowledge, accepted material and participant role. Search is free; admission and practice conditions keep their own terms. If no verified listing supplies the required task, keep that gap visible. A general robotics result or a prepared demonstration does not establish a current QoS clinic, working connection or prototype repair.