ME.15:4.4 - Minimal Constructed Variant Replay

Continue the temperature-sensor review chain. M-Unit-Review-1 is admitted with reusable semantics: identify every source unit, normalize values, compare with declared tolerance, and stop on an unresolved mismatch. ME.14 retained it for reports whose sources can use different unit systems.

A team proposes an approximation rule for missing conversion metadata: infer a likely scale from sensor range, mark the inference, and continue only when the tolerance decision is unchanged under both plausible scales. This is not merely a new checklist. It changes the reusable operation and stop conditions, so create candidate account C-M-Unit-Review-Approx-1 derived from M-Unit-Review-1. Preserve identification, explicit normalization when metadata exists, tolerance comparison, and unresolved-mismatch stop; add the bounded inference operation and a two-scale invariance stop.

Do not admit the candidate from parent status. Link candidate-account content CA-M-Unit-Review-Approx-Content-1, proposed representation C-REP-Approx-1 of that candidate account, and the same conversion-table support result only as candidate-serving material; none is a U.MethodDescription for the candidate whole. Keep performance on missing-metadata reports unresolved. Retain that limit, the reviewer capability conditions and the comparison against stopping or metadata recovery where they matter to the entry’s use. To maintain the catalogue, record the proposed approximation rule, its origin in M-Unit-Review-1, its two-scale stopping condition and its untested candidate status. This is enough to retain the proposal in the catalogue. Choose a trial only when its useful attainable answer warrants the work and delay. Until independent admission and evidence exist, the repertoire contains admitted M-Unit-Review-1 plus candidate lineage C-M-Unit-Review-Approx-1; no Work is said to enact the candidate whole.

If the team only updates the diagram to show the already-required normalization step, first identify its exact action. For one action, maintain the complete C.37 representation-use result and the diagram under its direct governor; ME.9 is not invoked. When that result belongs to a complete ME.9 profile for several unlike actions, also maintain only the cross-use correspondences, conflicting omissions, edition relations, decisions to keep representations separate, and conditions for reconsidering several selections affected by the diagram change. Neither case creates a variant. If the team replaces the manual table with a verified tool while operations and stops remain unchanged, maintain the ME.10 support result and no variant.