ROS 2 Package and document, with evidence for every claim
By this phase the lab should have a proven cross-worker path to describe. Now the job is to explain it truthfully: a clear README, benchmarks anyone can rerun, tutorials that stay correct as versions change, and a first release that comes only after validation.
Where this phase starts
The roadmap's audit (10 to 11 October 2026) found gaps that this phase exists to close. These are starting points, not results.
No proof-of-function demo yet
The root README has no central demo animation and no validated modern install profile. Its main example starts a CPU stress test (stress-ng), not a ROS 2 talker and listener.
No tagged release
At the audit the repository had no GitHub release. The plan holds the first one back until a clean install has been validated.
Old measurements exist
The DataAnalysis/ folder holds sample monitoring files from a single recording on 2024-04-20 on the historical Vagrant lab cluster. They stay labelled historical and are never quoted as modern results.
Every claim carries its evidence
This phase follows one rule: a statement the lab publishes must point to something that backs it. Step through the four artifacts and watch each claim get its evidence. The last row shows a claim with nothing behind it, which stays out.
No evidence, no shipping
Illustration · the rule, not progressThe amber row at the bottom is an invented example of a claim with no evidence. The picture shows the rule the plan will follow. It does not report that any of this evidence exists yet.
The four issues and what finishes each
The completion gates below come straight from the roadmap. They describe what must be true before an issue counts as done. None is met yet.
| Issue | Required outcome | Completion gate | Status |
|---|---|---|---|
| #17 ↗ | Visual README and accurate quickstart | Every featured command and diagram matches checked-in, tested code. | Planned |
| #18 ↗ | Reproducible DDS, autoscaling and monitoring benchmarks | A runnable protocol, raw data, regenerated plots and stated limitations. | Planned |
| #19 ↗ | Searchable, versioned tutorials and troubleshooting | A built and published docs site, verified links and an honest support matrix. | Planned |
| #20 ↗ | First verified release and a contributor-friendly workflow | A versioned release with integration proof, and contribution guides. | Planned |
Three rules this phase keeps
Historical stays historical
New benchmarks come from controlled new runs. Old sample files are kept and clearly labelled as old. They never appear as results for the modern profile.
Release only after validation
The first tagged release follows the validation of a clean install. If that takes longer, the release waits. The plan holds a milestone rather than skip a gate.
Legacy steps are separated
The old Foxy steps stay available but are marked as historical, and the phase adds a migration note for people still on them.
Read the full list of checks in the release gates.