church_exp_2_2 measurementYour third verifier-boundary find fixed, and the measurement you asked for before any film work. No implementation on the runtime frontier in this round.
Gate: grid v1.40.0 (87 entries / 375 citations) · lowering.mjs 0.7.2 · negative battery 286/286 · lowering 23/23 · derive 45/45 · realm 24/24 · bridge 48/48 · film 16/16 · pack 24/24, 0 skipped. cert_id a08ee15d… byte-identical — thirty-fifth consecutive round.
verifyEmissionReceipt(T.church(2),
emissionReceipt(closedTemplateSemId(T.church(2)), "deadbeef"),
() => "deadbeef") → { ok: true }Your reading is exactly right: that is what a parametric verifier means, and the defect was the spelling. Renamed to verifyEmissionReceiptAgainst / verifyEmissionReceiptOwnedAgainst, and makeEmissionVerifier({ canonicaliseTarget }) binds the trusted oracle at a composition root. The bound verifier has arity 2 — no parameter in which to nominate a judge. Binding without one throws.
I kept no alias for the old name: an alias is a second path to one relation with the weaker spelling still reachable, which is the defect B2 removed from lower().
The forgery that makes emissionReceipt refuse caused grid_check to crash with a stack trace instead of reporting a diagnostic — the battery saw a nonzero exit and the wrong reason. Climbing to the behavioural rung means running adversary-influenced code; every probe on that rung is wrapped now.
That is two rounds, two costs: first silent weakening (arity invisible to typeof), now silent crashing. Both found by the battery, neither by review. Your rule is recorded verbatim where the hierarchy is stated.
measure_exp22.mjs, declared as a tool, asserts nothingFRAMES: 21 terminal: NORMAL_FORM nf: λa.(S (S (S (S a)))) — matches the corpus
FIRE: APP-LAM 6 · DUP-SUP= 4 · DUP-LAM 3 · DUP-VAR 3 · APP-SUP 2 · DUP-APP 2 · DUP-SUP! 1
NEVER: APP-ERA · DUP-ERA
LOCI: d: 13 · t: 4 · v: 4Full frame table with pre→post ids and the enabled-count per step is in the tool's output; it is in the pack and reruns in a second.
Every rule the round was aiming at fires, and all three locus families appear. Three things worth having before building:
TWO rules remain unexercised, not one. DUP-ERA was expected. `APP-ERA` was not — the
dedicated witness has to cover both, or say why not.
The final signature is 50 characters, under the §5 compaction bound, so the decoder does not
refuse this normal form. The fixture is decodable end to end.
The 21 is a coincidence until proven otherwise. The corpus records ref_interactions: 21 and
the measured film is 21 frames — different quantities, an AST reference interaction count and a float-plane frame count. Your warning is in the ledger verbatim: the theorem stays given the frozen film strategy, C and the independent kernel agree frame by frame on the same transitions and the same terminal, and a different legitimate count gets investigated rather than forced.
The historical wrinkle is in this tree's own record, and worse than "worth remembering": l_prog_history.round_4_diagnosis names church_exp_2_2 at step 15 as the FALSE QUIESCENCE witness that falsified law:sched.free.ast-term@1. The AST relation is retracted and the float-plane enumeration reaches a normal form cleanly — but the C emitter will have to be explicit about its transition strategy, what counts as ENABLED, what TERMINAL means, and whether BUDGET_EXHAUSTED is distinguishable from NORMAL_FORM. That fixture has fooled a scheduler once.
My reason for deferring EMISSION_CONFORMANCE-v1 was wrong. I said church_exp_2_2 would force the template grammar to grow, so the fixtures would be rebuilt. It will not — church_exp_2_2 is a raw runtime corpus term, independent of Template := church | add | port, and the work on it is native-film transitions rather than compiler grammar. Your reason replaces mine: the marginal value of seven more closed-template fixtures is lower than the first independently replayed native film containing real DUP/SUP transitions. Sequence recorded as church_exp_2_2 → dedicated ERA witness → EMISSION_CONFORMANCE-v1.
Build only what the measurement requires: get ic32_film through the seven rules that actually fire and the d:/v: loci, with the C emitter's strategy, enabledness and terminal conditions stated explicitly rather than inherited. Then the ERA witness covering both unexercised rules. I will not touch compiler apparatus.
One question: the measurement drives the JS kernel. The theorem needs C and the kernel to agree frame by frame — so should the next round's first artifact be the C-side measurement printed in the same shape, compared against this one before any assertion is written? That is what I would do, but it front-loads a second measurement round before any gate moves.
governance/lowering.mjs 0.7.2 — …Against naming, makeEmissionVerifier
governance/lowering_check.mjs 23/23 — emission-verdict-names-its-oracle
governance/measure_exp22.mjs — new, declared in artifacts.json as a tool; asserts nothing
governance/grid_check.mjs — B2.1.2 assertions; behavioural probes wrapped; the corollary recorded
governance/negative_battery.sh 286/286 — 5 new
governance/round-11-ledger.md — items 266–274