gnomonlearn
Docs/Formal proof scope comment correction

Formal proof scope comment correction

Versioned draft snapshot

RFCs, reference catalogs, and evidence reflect Spec 3b2e3e7a114fb826b1445d39a48756b8fff629cd. Counts and pending/status notes belong to that snapshot. Read the dated project update for later progress.

Source repositories are private during staging. Links marked “repository access required” need access to the linked repository.

This is a trace and gap record from the pinned draft specification. It does not establish proof or implementation conformance.

This independently sealed editorial follow-up preserves the native compilation capsule in spec e6248540ca0f3d4d7496de5ff8c27f4e97500eff. The held candidate is 0e5470251e3c9dc307ebe9fe3ca5790e927f9273, directly above the compiled complete consumer candidate bcf32923406ea4515b9fe9d0e46a38e7da8ca1ab, in the separate .work/formal-consumer-comments-20261013 checkout. The compiled source checkouts, native logs, dependency closure, and outputs remain unchanged.

The 13 area headers and umbrella now explain that imported assumptions and explicit premises bound the authored propositions. Historical “100% coverage,” “Closes deriv_*,” and blanket proof-completion labels were corrected to describe intended citations and candidate implications. One nearby field-equality note now states its lack of runtime write-transaction evidence. Normative quoted RFC headers, theorem identities, types, bodies, imports, strings, and all noncomment bytes remain exactly equal to the compiled candidate.

check_scope.py removes only Lean comments, retaining strings, quoted identifiers, and every byte outside comments. Nested-comment controls pass; code/string mutations and malformed lexical inputs are rejected. It compares all 14 theorem/umbrella files to the immutable parent, then verifies that the only other changed file is the source-bound review ledger. All 722 theorem-file hash bindings were refreshed. Both exact ledger text comparison and parsed protobuf comparison establish that every other review field, verdict, finding, statement/narrative hash, and identity remains unchanged.

The unchanged counts are 775 authored declarations, 12 authored deferrals, zero unassigned sorry tokens, 721 generated entries, 718 generated deferrals, 722 reviewed citations, and 715 mismatches. Authored and generated deferrals describe separate source streams; they must not be summed as unique claims. Strict release remains blocked. This lexical and source-binding control is not a new native compilation or proof-acceptance receipt. No native Lean rerun was performed for comments alone; the previous capsule records compilation of the exact parent source. Comments can affect documentation/scanner output, which is the intended editorial change.

The delta bundle and mail patch require the compiled parent candidate. Typed ValidationEvidence binds the new source and actual focused terminal logs to the exact previous evidence record. seal.py --verify checks these bindings, including the immutable prior record and its 14 compiled source-file hashes, without rebuilding the proof graph. There is no upstream model change, external write, merge, release tag, registry publication, dependency pin adoption, site update, runtime deployment, or new consistency certification in this follow-up. The local candidate is prepared for review and later authorized consumer adoption.