…er green

The re-scoped C-FIRST gate is built in ruff: reassemble() (generator stage 1,
inverse of expand) + the round-trip falsifier that replaced the council-
rejected self-golden. CPP-REASSEMBLE-RT runs green on real Tesseract ccutil
(67 classes; class-set preservation + idempotence). The falsifier found a real
bug class — 19/67 const-overload IRI collisions (GAP-CONST-OVERLOAD, queued).
truth-architect's PREMATURE flag resolved: real measurement on real data.

Co-Authored-By: Claude <noreply@anthropic.com>
https://claude.ai/code/session_016b33swuXE23hKtqxsHu9p1

Records the step-2 emitter council. Major reframe (3/5 agents independently):
the plan's premise was false — ClassView is a field/render vocabulary with NO
method-resolution surface (has_function does not appear in lance-graph-contract).

5-agent consolidation: manifest-first cut (not stubs/bodies), additive
(container ADDITIVE-CONFIRMED), placement = a new ruff_cpp_codegen crate that
emits TEXT only (no ruff->lance-graph compile edge), D (const-overload fix)
before the emitter, PARITY markers + Frankenstein deny-list.

Unresolved fork handed to the 3-brutal panel: (A) mint a minimal MethodSig POD
+ classid_methods LazyLock registry (method-axis sibling of classid_read_mode,
EXTEND-CORE but additive) vs (B) ride the SHIPPED codegen_spine::TripletProjection
+ roundtrip_eq (no new type; warns MethodSig re-implements CppMethod and that
re-emitting the harvest is a tautology). codegen_spine.rs verified real (632 lines,
TripletProjection trait + roundtrip_eq fn + Genericity enum).

Co-Authored-By: Claude <noreply@anthropic.com>
https://claude.ai/code/session_016b33swuXE23hKtqxsHu9p1

…are IRI)

The brutal panel resolved the A-vs-B fork as a false binary (honest-tester):
B's codegen_spine::roundtrip_eq pattern is the build-time GATE, A's MethodSig
shape is the emitted-text target, A's runtime registry is a deferred additive
EXTEND-CORE. baton-auditor's decisive correction (confirmed by honest-tester):
the const-overload merge is UPSTREAM in expand's (s,p,o) dedup, so the round-trip
cannot observe it — it is a fixed point. Therefore D (cv-aware method IRI) must
run FIRST, unanimously (behavior-monitor's emitter-first rationale refuted).

Final order: D (cv-aware IRI, autonomous correctness fix — adds no predicate,
falsifier CPP-REASSEMBLE-RT 48/67 -> 67/67) -> emitter scaffold (ruff_cpp_codegen,
emit-text-only, ruff-side round-trip gate, no classid mint) -> MethodSig
EXTEND-CORE in lance-graph (additive) -> wire/byte-parity (operator).

Co-Authored-By: Claude <noreply@anthropic.com>
https://claude.ai/code/session_016b33swuXE23hKtqxsHu9p1

…nst assumption

Records D's outcome: the cv-aware method IRI + two round-trip-metric fixes
(cpp_projection dedup; is_const in the methods sort key) closed the entire
collision tail (48/67 -> 67/67, now a hard gate). The falsifier overturned the
council's "19/67 = const overloads" inference: only 3 were const; 13 were benign
duplicate template_instantiates, 2 duplicate-harvested methods, 1 a sort-order
artifact. GAP-CONST-OVERLOAD resolved. Next: emitter scaffold (ruff_cpp_codegen)
then the MethodSig EXTEND-CORE.

Co-Authored-By: Claude <noreply@anthropic.com>
https://claude.ai/code/session_016b33swuXE23hKtqxsHu9p1

…real corpus

The C-FIRST step-2 emitter shipped in ruff: project(ModelGraph)->MethodSig
manifest + render to lance-graph-naming Rust text (emit-text-only, no
lance-graph edge), gated by a decompile==expand signature-plane round-trip with
teeth (dropped-method test proves it fails). CPP-CODEGEN-RT on real ccutil:
67 classes, 857 methods -> 124 KB MethodSig manifest, round-trip holds, PARITY
markers + Frankenstein deny-list present. Next (operator-gated, additive canon
growth): the MethodSig EXTEND-CORE in lance-graph-contract.

Co-Authored-By: Claude <noreply@anthropic.com>
https://claude.ai/code/session_016b33swuXE23hKtqxsHu9p1

…ompile target

The MethodSig EXTEND-CORE (C-FIRST step 2's deferred-runtime-registry piece).
A new codegen_manifest module:

- MethodSig: the dispatch-relevant C++ method signature in a const-constructible
  shape (all fields &'static: name, params: &'static [&'static str], ret,
  is_const, is_static, overrides). It is the exact literal ruff_cpp_codegen::render
  emits, so the generated text now has a real compile target. The &'static shape
  is load-bearing: class_view::FieldRef is String-backed and cannot appear in a
  const; MethodSig is the method-axis sibling that can.
- ClassMethods{classid, methods} + methods_for(registry, classid): the
  registry-entry type + pure zero-fallback lookup (unregistered classid -> empty
  slice). classid is bound OGAR-side, never minted here; the runtime
  classid->methods registry DATA is generated downstream (consumer repo), NOT
  stored here (honest-tester's "defer the runtime registry").

Additive (container-architect ADDITIVE-CONFIRMED): a sibling module, zero
NodeRow/ValueTenant/ValueSchema/stride/ENVELOPE_LAYOUT_VERSION impact. Body-shaping
flags (pure-virtual/constexpr/noexcept/operator/requires) are out of scope.
Board hygiene: LATEST_STATE Contract Inventory updated same commit
(D-CPP-CODEGEN-1). +2 tests (const-constructibility proof + zero-fallback);
640 contract lib green; clippy -D warnings clean.

C-FIRST: D + emitter scaffold + MethodSig EXTEND-CORE all landed; the in-env arc
is complete. Remaining is operator-gated (tesseract-rs wiring + byte-parity).

Co-Authored-By: Claude <noreply@anthropic.com>
https://claude.ai/code/session_016b33swuXE23hKtqxsHu9p1

…e's Rust side

The deferred Option A content-store tier, built (operator's "keep building here"
+ the leptonica-is-an-install-not-a-transcode epiphany). New unicharset module:
UniCharSet (deepnsm::Vocabulary-shaped: reverse id->unichar + lookup
unichar->id), load_from_str/load_from_file parsing the .unicharset text format
(line 1 = count; first whitespace token per line = unichar; id = position;
property columns ignored — the old_style_included_ plain-table scope),
id_to_unichar/unichar_to_id (the two adapter leaves), and dump() rendering the
<id>\t<unichar> table matching the C++ oracle.

This is the Rust side of PROBE-OGAR-ADAPTER-UNICHARSET. The unicharset path is
pure text parsing — ZERO leptonica (never touches Pix) — so it builds and
unit-tests in-env with no C deps. leptonica is only an *install* (a link dep of
the C++ oracle harness), never a transcode and never in the Rust path. Byte
parity is now one `diff`: combine_tessdata to get eng.unicharset, a ~10-line
libtesseract harness dumps id_to_unichar, `cargo run --example unicharset_dump`
dumps the Rust side, diff. Byte-identical => CONJECTURE -> FINDING.

Additive (sibling content-store module, zero NodeRow/tenant impact). Board:
LATEST_STATE Contract Inventory (D-UNICHARSET-1). +4 tests + the unicharset_dump
example; 644 contract lib green; clippy -D warnings + fmt clean. The
classid->&UniCharSet LazyLock resolver (OGAR wiring) remains the follow-up.

Co-Authored-By: Claude <noreply@anthropic.com>
https://claude.ai/code/session_016b33swuXE23hKtqxsHu9p1

…ICHARSET FINDING

leptonica installed in-env (apt-get — an install, not a transcode), so the
byte-parity probe RAN and passed. UniCharSet dump vs a C++ UNICHARSET FFI oracle
on the real eng.lstm-unicharset: 112/112 byte-identical.

The falsifier did its job: the documented-format parser matched 111/112; the
oracle named the one real convention it missed — the NULL file-token IS the space
unichar (unicharset.cpp:882 remaps "NULL" -> " "). One-line fix
(load_from_str maps "NULL" -> " "), re-diff, 0 differences. NOT a Core gap.

CONJECTURE -> FINDING for the unicharset adapter: the variable-length bijection
rides the content-store tier with no Core gap and is byte-exact with libtesseract.
Doctrine flipped (core-first-transcode-doctrine.md falsifier RESULT); EPIPHANIES
E-CPP-PARITY-1; plan BYTE-PARITY ACHIEVED. The classid->ClassView->UnifiedStep
dispatch wiring is mechanical remainder; the lookups themselves are now proven.
+1 test (null_token_maps_to_space); contract lib green; clippy + fmt clean.

Co-Authored-By: Claude <noreply@anthropic.com>
https://claude.ai/code/session_016b33swuXE23hKtqxsHu9p1