Subject matter
Subject : **Reproducible labs offline**. Domain : « Academy, community and contribution ». Form : « Overview ». Intention : define the perimeter, actors, inputs and exits. Purpose of the topic : allow learning without implicit download. This reference applies to HexaOS 3.0.0-rc1 and must be adapted to explicitly studied publishing and artefact.
Reference state
The snapshot of 20 August 2026 has eight out of thirteen qualified editions and maintains a global verdict NO-GO. This meter is a campaign photograph, not a lasting property of the document. Any republication after evolution of RC23 must regenerate this passage from the source of truth.
Prerequisites
1. Remove product version, edition, build ID, full name and print of the artifact.
2. Describe machine or hypervisor, firmware, storage, network and relevant resources.
3. Prepare a reproducible initial state and a reverse method.
4. Define the expected result, the prohibited result and the return code.
5. Use only test data and predict the removal of traces.
Reference procedure
Initiation of proceedings : define the perimeter, actors, inputs and exits. Control of the subject : inventory data, dependencies, duration, initial condition and cleaning. Perform actions one by one, without rewriting newspapers or replacing a failure with an interpretation. When the scenario depends on the runtime state, restart under the expected conditions and then reproduce the observation. Keep the time, order, redacted output and final code.
Evidence expected
•- minimal preconditions and configuration; - reproducible actions, including expected refusals; - observed result compared to the criterion defined before the test; - redacted traces, prints and verdict limited to the scenario; - new execution after any relevant modification.
Limitations and sequels
a third-party resource requires compatible licensing and allocation. Sort out the result PASS, FAIL Or PENDING According to the observation, without converting a lack of evidence into success. Validations GPU And HPC material remains open until their physical scenarios are completed.
Supplement R2 — Maintenance and non-regression
Purpose of this treatment : define the controls to be replayed after a change in source, configuration, or artifact. Evidence expected : configuration diff, new hash, gate history, comparison result and explicit decision. Limit of conclusion : any relevant modification invalidates the previous evidence until it is re-enacted. All information remains linked to HexaOS 3.0.0-rc1, Snapshot RC23 of 20 August 2026, eight editions qualified out of thirteen and to the overall verdict NO-GO.