Layout · the PM's part 1 ruling R2-6The four entry pages share one layout, and this refusal page keeps it. Before this pass the question reference card and the cycle context card filled the whole first screen and the refusal panel started below the fold, which on this page of all pages is the wrong order: the reason a person came here is the panel telling them what to do. The facts they need without opening anything sit in the question strip under the page header, including the two that matter here, that the answer format is not supported and that the omission route is refused too. Everything else moved into the collapsed "Question and cycle details" section under the refusal area, and the state gallery below draws that section OPEN so the review can confirm every row of both former cards survived. The alternatives the PM offered were a card on the right and a card below. The right-hand card was rejected because it takes width the table answer on page 6 of this wireframe set needs and because this page already uses its right column for the "tell your administrator" block. A plain card below was rejected because it keeps reference detail permanently in the scroll. One thing to note for the review: the strip carries the SHORT form of the refusal, and the panel carries the full explanation, so the same fact appears twice by design rather than by accident.

Annotation · not proposed UIFail-closed contract (spec ii 3.9). The input vocabulary is ONE exported constant shared by the import validator, the entry validator, and the catalog reads that feed this screen. Native paths refuse an unknown answer format with 422 VOCAB_TYPE_INVALID on WRITE, READ, IMPORT, and RENDER. The renderer receives the declared format from the catalog read and, when it does not recognize it, shows this explicit error state: never a numeric fallback, never a blank control, never a text box that happens to accept anything. Test 9.2 names the UI unsupported-type state and the never-a-numeric-fallback rule as an acceptance case, so this drawing is the reviewable form of a tested requirement.

Annotation · not proposed UIWhy guessing a format is forbidden, stated as a mechanism rather than a preference. The answer payload is a CLOSED set with exactly one legal slot per format (spec ii 3.9): the legacy formats and date write value, while table and written answers write structuredValue. A guessed control picks a slot. Two outcomes follow, both bad: the write is refused anyway and the user has wasted the work with no explanation, or the value lands in a slot that means something different from what the question asked, and the question reads as answered. Since the same clause also drives later reporting, a wrong-slot value propagates into anything built on it. Refusing at render is the only outcome that leaves the record honest.

Annotation · not proposed UIHow this state is REACHED, in plain terms for the review: the catalog is versioned and published independently of the app, so a published edition can declare a format that a deployed app version predates. It is not a data fault and not a user fault, which is why the copy blames neither. The declared format string drawn here (valueDenominatorMatrix) is FICTIONAL and deliberately not one of the spec ii vocabulary members, so nothing on this page can be mistaken for a real vocabulary entry.

Annotation · not proposed UIThe omission route is closed here too, and that is now stated in the contract rather than inferred. Spec ii 3.10 as amended: on a clause whose answer format is unknown to the running version, BOTH routes fail closed, and the answer write AND the omission write refuse VOCAB_TYPE_INVALID, because an omission is still a write against a contract this version cannot validate. That is why both entry choices above are switched off rather than left to fail on save. This page's earlier draft raised the question; the spec answered it, and the drawing matches the answer.

Annotation · not proposed UIWhat identifies this answer (spec ii 6.4 and 3.8). A native answer is keyed by cycle, clause logical id, topic instance (null throughout I1), organization scope, and period. Native rows carry clauseLogicalId and do NOT populate disclosureId; they never pass through the legacy assignment and entry resolvers. The clause is resolved through the cycle's frozen edition set, so a question outside that edition set refuses with CLAUSE_NOT_IN_ASSEMBLY rather than rendering.

Annotation · not proposed UISECOND QUESTION FOR THE PM, about the copy rather than the mechanism. The panel tells the site manager to contact "your administrator", which is the wording the assignment fixed and the wording a site user understands. The actual remedy, though, is a newer app version, and in this phase deployment sits with the platform operator (Evercomm), not with the tenant's own administrator. So the tenant admin can only pass the message on. Two options: keep the current wording, which is honest about who the user should talk to first and hides the plumbing, or name the escalation explicitly ("your administrator, who will raise it with the platform operator"). Drawn with the shorter wording. The question code block below the panel exists so that whichever route is taken, the person on the receiving end gets what they need in one message.

Annotation · not proposed UIRole. Daniel Ong holds ENTRY access. Spec ii 6.2 gives the entry role read access to the two cycle catalog reads (GET /v1/reporting-cycles/:id/catalog and .../catalog/clauses/:logicalId) and 403 on every cycle mutation route; the full matrix is test 9.3. That is why no cycle action appears on this page. The application shell around the page stands in for existing NxStart chrome and is drawn identically on all twelve pages of this wireframe set so the surfaces can be compared; which menu items an entry-role user actually sees is existing product behavior this wireframe set does not restate, and it is not the mechanism that stops the write. The server is.