Tier 5, the dual demo: model your own Recommendation

A curriculum chapter of Learn OIML SMART.

The artifact: a kernel-validated Primmel model of OIML R 7 (1979), Clinical thermometers, built with Primmel Studio, plus the live proof that it simulates, maps to its source document, and previews as a certificate.

Prerequisites: Tier 4, authoring. The r60 classroom helps but is not required.

Why a second instrument

Every earlier tier used OIML R 60 (load cells), deliberately: one instrument, learned deeply. This tier exists to prove the methodology TRANSFERS. If you can model a mercury-in-glass clinical thermometer with the same seven-step loop you used for load cells, the method is yours; if you can’t, the earlier tiers were memorization. That is the whole point of the dual demo: r60 taught you the method, R 7 makes you apply it.

The seven-step loop

The full walkthrough lives with the model , demo/r7-clinical-thermometer/TUTORIAL.md in the Primmel Studio repo. The loop it teaches:

1. Read the document like a modeller, instrument / truths / checks / verdict (the IS-HAS-DOES read).
2. Open the model in the Studio, the OIML program layer activates through the plugin, never a kernel branch.
3. The subject anatomy, IS identity + design parameters (the MPE table), HAS attributes + characteristics, DOES behaviors, and the certificate preview as the first secondary view.
4. Requirements with provenance, each a constraint on the subject’s IS/HAS/DOES, each carrying its clause URN (source: { doc: urn:oiml:pub:r:7:1979, clause: "…" }).
5. Conformance tests and forms, tests interrogate the subject to produce evidence; forms are the evidence skeleton.
6. The workflow, simulated, the clause-8 ambient gate and the MPE gate, walked live with registers (conform / invalid ambient / out-of-MPE).
7. The doc map + review + save, the model mapped back onto the document’s statements; the kernel’s model-diff as the review-before-commit.

The assessment gate

Reproduce the gate, as with every tier:

cd ~/src/primmel/editor
npx vitest run src/lib/__tests__/r7-tutorial.test.ts   # 6 tests
npx tsx e2e/r7-smoke.ts                                 # the live loop

• the package loads strict and validates clean;
• every doc-map target resolves against the real R 7 document (the anchor sentences pin to the right texts);
• the workflow simulates all three paths.

Then, the honest exercise: pick a third Recommendation (R 49 water meters, R 76 non-automatic weighing) and run the same seven steps. If the loop holds for you on a document you have never seen, the curriculum is done.

The dual demo, side by side