plan: OGAR sink-in + consumer-bridge removal (medcare-bridge → UnifiedBridge → delete)

.claude/board/INTEGRATION_PLANS.md

@@ -1,3 +1,7 @@
+## 2026-06-26 — ogar-sink-in + consumer-bridge removal (medcare-bridge → UnifiedBridge → delete)
+
+Plan: `.claude/plans/ogar-sink-in-and-consumer-bridge-removal-v1.md`. **Codifies the sink-in layering + sequences the final bridge deletion.** Grounded finding: the migration is **~90% shipped** — all 6 per-consumer bridges (incl. `medcare_bridge.rs`) are already `#[deprecated] type FooBridge = UnifiedBridge<FooPort>` aliases over one generic harness; `docs/CONSUMER-BRIDGE-DEPRECATION.md` (OGAR#95) is the live doctrine ("nothing removed; deletion lands after consumers migrate"). **Sink-in decision (§2):** OGAR owns the *meaning* (PortSpec / codebook / ClassView / ActionDef / ontology); `lance-graph-contract` owns the zero-dep *wire* mirror (the parity guard, never a dep); `lance-graph-ontology` owns the OGIT registry; `lance-graph-rbac` owns grants keyed on the **shared lo-u16 concept** (render hi-u16 never gates RBAC — the CLASSID-RBAC-KEYSTONE-SPEC); `lance-graph-ogar` keeps only the `UnifiedBridge<P>` mechanism + `parity`. **medcare recast (§3):** the medcare domain dissolves into **reusable patterns** (the shared `HealthcarePort::class_id` / convergent codebook surface in OGAR) **+ domain-specific Ontology Schema enrichment** (Healthcare `0x09XX` / FMA-V3 `0x1000_0A01` / CPIC-V3 `0x1000_0E00` in OGIT) — never a bridge in the spine. **Deliverables:** D-SINK-1 ratify layering → D-SINK-2 migrate consumers off `MedcareBridge` (cross-repo, Iron Rule 5) → D-SINK-3 delete the 6 aliases (scope-lock tests ported first) → D-SINK-4 reconcile the two `WoaBridge` identities → D-SINK-5 sink the medcare ontology (the cross-repo codebook-mint arc, `ISS-OGAR-GENETICS-MIRROR-PENDING`). **Gated:** no deletion before consumers migrate (OGAR#95); intersects the operator-gated classid realign (D-OVC) + Canon:Custom. Extends CONSUMER-BRIDGE-DEPRECATION + ogar-vocab-contract-codebook-migration-v1; does not supersede them. Doc-only. PROPOSED.
+
 ## 2026-06-25 — soa-value-tenant-migration v2 SEQUENCING (operator-locked; identity→V3, then V3-tenants)
 
 Plan: `.claude/plans/soa-value-tenant-migration-v2.md`. **Operator-locked ordering** for the migration the v1 BRIEF opened + the v1-HARVEST inventoried: **ONE migration, TWO ordered phases — Phase 1 identity→V3 (key-side), Phase 2 V3-shaped value tenants (value-side).** Identity-first is **forced, not chosen**: OGAR #128's envelope parser resolves `tail_variant` (key shape) UPSTREAM of `value_schema` (tenant shape), so the tenants cannot be shaped to a V3 geometry the key does not yet express (the key is the coordinate system; tenants are read *through* it; the Phase-2 `helix-place‖CAM-PQ` facet is a *reflection* of the V3 part_of/is_a key). **Phase 1** = OGAR-registry + envelope-parser wiring: a leading-`1` generation marker in the HIGH (custom) u16 → V3 `tail_variant`, canon `0xDDCC` low-u16 preserved (OSINT `0x0000_0700 → 0x1000_0700`; the earlier `0x1007` low-half form is **rejected** — it corrupts the domain byte, Codex-P1), **coexist-by-classid** (legacy zero-prefix classids keep their current tail — **V1** default or **V2** `new_v2`/`guid-v2-tail`-minted; RESERVE-DON'T-RECLAIM ⇒ zero V1/V2-corpus re-mint, layout-preserving), grade `[H]` gate **F-update**. **Phase 2** = the harvest (contained facet + 8 KEEP + 2 DEFER, F-1/F-code; the §5 body still gated on the two 5+3 panels). **Payoffs:** identity-first migrates the one shared `entity_type≡class_id` anchor to V3 *before* the A↔B reconciliation (Phase 2 reconciles in V3 coords, no redo), and dissolves the harvest's scope question (harvest = the Phase-2 input). **Watch:** Phase 1's substance IS the OGAR casing-miss gap (harvest §6.1) — the corrective `/home/user/{OGAR,MedCare-rs}` sweep gates Phase-1 start, not optional polish. Doc-only. On `claude/serene-mayer-1a09he` (rides with the harvest to main).

.claude/handovers/2026-06-26-ast-address-to-next-session-capstone.md

@@ -0,0 +1,210 @@
+# Handover Capstone — AST-as-(part_of:is_a) / OGAR-ERP-stdlib arc → next session
+
+> **From:** session on `claude/medcare-bridge-lance-graph-wmx76z` (2026-06-26)
+> **To:** next session picking up the OGAR-ERP / V3 AST-address arc
+> **Status of the arc:** CONJECTURE captured + carrier verified; ONE empirical
+> probe converts it to FINDING. No V3 *migration* was performed this session.
+> **Companion handover (READ ALONGSIDE THIS):** the parallel CPIC/genome session
+> (session `01TzqvDqbFRzyx17EkLKBoZF`) owns the V3 *identity* + genome-aligned
+> track (PRs #614/#615/#618, the `q2/cpic` NARS prototype, the CYP2D6
+> activity-score falsifier). This handover owns the *"why" doc* track (#616/#617,
+> ruff #29). Two halves of one arc — their CYP2D6 probe and my MedCare probe are
+> complementary falsifiers of the SAME genome-/AST-as-address CONJECTURE.
+
+---
+
+## 0. TL;DR (read this, then the mandatory reads)
+
+This session **did not write product code**. It (a) verified another session's
+V3 alignment of a design doc, (b) consolidated a multi-layer architectural
+thesis into one knowledge doc, and (c) scaffolded a C# harvester. The thesis:
+**a transcode source's structural AST is stored AS the `(part_of:is_a)` GUID
+address (shipped `FacetCascade`), making ERP construction in any consumer as
+cheap as importing pre-minted OGIT class primitives + classaction pointers.**
+The whole arc is CONJECTURE; the gate to FINDING is the **MedCare brick-3 probe**
+(§6). Everything else is wiring, not research.
+
+**Single most important correction to inherit:** the carrier and minter are
+already SHIPPED/built. Do not re-design them. The remaining work is *wiring*
+(mint imported OGIT → codebook; merge the minter branch; run the probe).
+
+---
+
+## 1. Mandatory reads to get up to speed (in order)
+
+1. **`.claude/knowledge/ast-as-partof-isa-address.md`** — THE doc this arc
+   produced (merged via #616, extended via #617). The full thesis + the three
+   honest boundaries. Start here.
+2. **`.claude/knowledge/guid-canon-and-prefix-routing.md`** — the V3 GUID canon.
+3. **`crates/lance-graph-contract/src/facet.rs`** — the SHIPPED carrier
+   (`FacetCascade` = `facet_classid(4) | 6×(8:8) = 16 B`, content-blind,
+   `(part_of:is_a)` is one consumer projection). Four SIMD lanes.
+4. **`crates/lance-graph-contract/src/canonical_node.rs`** — `TailVariant::V3`,
+   `mint_for`, `CLASSID_OSINT_V3 = 0x1000_0700` (the one wired V3 exemplar).
+5. **`crates/lance-graph-contract/src/class_view.rs`** — the static-shape →
+   dynamic-ClassView → askama ladder (XML/XSD/XSLT analogy), C2/N3 invariants.
+6. **`crates/lance-graph-contract/src/ogar_codebook.rs`** — the import surface
+   (`canonical_concept_id`, `render_classid_for_concept`).
+7. **`.claude/knowledge/ogar-consumer-preflight.md`** + **`core-first-transcode-doctrine.md`**
+   — the pull-don't-reconstruct iron rule + the Core-first doctrine.
+8. **OGAR repo:** `docs/OGAR-AST-CONTRACT.md` (the THINK arm `Class` / DO arm
+   `ActionDef`+`ActionInvocation`+`KausalSpec` IR) and
+   `vocab/imports/ogit/` (complete OGIT, ~1,940 TTL, IMPORTED).
+
+---
+
+## 2. What I shipped this session (all PRs)
+
+| PR | Repo | State | Content |
+|---|---|---|---|
+| **#616** | lance-graph | **MERGED** | `ast-as-partof-isa-address.md` (the *what*) + I aligned another session's V3 correction onto it. Now on `main`. |
+| **#617** | lance-graph | **OPEN** | The *why*: three-layer thesis + laws-as-patterns + OGIT-imported + DO-as-classaction-pointer + the 2 MB headline. Branch `claude/medcare-bridge-lance-graph-wmx76z`, HEAD `e34a7fa`. **Docs only**, safe to merge independently. |
+| **#29** | ruff | **OPEN** | `ruff_csharp_spo` — Roslyn (.NET) C#→SPO harvester (brick-3 input). **CI-gated, not locally verified** (no .NET SDK here; proxy blocks workspace `cargo check`). |
+
+**No code/contract surface was changed in lance-graph.** `#617`'s diff is one
+knowledge file, +186 lines.
+
+---
+
+## 3. The thesis in one screen (so you don't re-derive it)
+
+- **Layer 1 — rails-shaped semantic AST at *assembler* cost.** `facet.rs`'s four
+  SIMD lanes ARE the LSP ops (`definition`=`vpcmpeqd`+`vmovmskps`,
+  `typeHierarchy`=`lo_chain` `vpxor`+`tzcnt`, `documentSymbol`=`hi_chain`,
+  ancestry=GFNI Morton). Only the *declarative* (rails) class body flattens to
+  fixed-width tiles; an imperative syntax tree does not.
+- **Layer 2 — static OGAR / dynamic ClassView / askama.** One ClassView projects
+  the `N × 6×(part_of:is_a)` SoA two ways: row→typed **view** (THINK→askama) AND
+  row→conditional **classaction** (DO→`ActionDef` under `KausalSpec` guard).
+  C2 (presence≠semantics) + N3 (append-only positions) = a row view that evolves
+  17 years (Redmine) without breaking persisted rows.
+- **Layer 3 — OGAR as importable ERP-LAW primitives + lance as compiler.** The
+  primitives are *laws/regulations* (universal, reusable), not CRUD:
+  `ObjectView`=required fields, `ActionDef`/`KausalSpec`=rules. Content stays
+  with the consumer; only the pattern is shared. `ogar_codebook` is the import
+  surface. **DO in the consumer = a classaction POINTER** (`classid → ActionDef`
+  via `ActionInvocation` DTO) — exactly the OGAR consumer doctrine ("classid is
+  pure address; magic at the resolution target").
+- **Headline (CONJECTURE):** OpenProject + Odoo as ~2 MB of GUID-encoded
+  `(part_of:is_a)` vs ~20 MB / ~250K LOC (~10×). Dimensional check: 16 B/node ⇒
+  2 MB ≈ 131K nodes — credible for both ERPs' structural skeleton.
+
+---
+
+## 4. V3 state — what EXISTS vs what I did (IMPORTANT, do not misattribute)
+
+**I performed NO V3 migration this session.** I only *read/verified* the shipped
+V3 surfaces to ground the docs. The V3 mints shipped in OTHER sessions
+(#613/#614/#615, then **#618** which landed AFTER my #617 merged):
+
+| domain | classid | status (as of #618, 2026-06-26) |
+|---|---|---|
+| **OSINT-V3** | `0x1000_0700` | wired (first exemplar, #613-615) |
+| **FMA-V3** | `0x1000_0A01` | **wired** (#618 — `ReadMode::FMA_V3`) |
+| **CPIC-V3** | `0x1000_0E00` | **wired** (#618 — Genetics domain `0x0E` operator-allocated; `ReadMode::CPIC_V3`) |
+
+**#618 "Phase 1 V3 set complete"** — OSINT + FMA + CPIC all minted, Genetics
+domain `0x0E` allocated. So CPIC/FMA/OSINT are ALL wired now (none by me).
+
+The V3 marker lives in the **HIGH (custom) u16**; canon/domain in the **LOW u16**
+(`classid_concept_domain` routes on the low half — a deliberate Codex-P1 fix; the
+low-half `0x1007` alternative was rejected as domain `0x10`=Unassigned). **Open
+operator decision: "Canon:Custom"** — whether to flip canon to the high half
+(would require re-pointing `classid_concept_domain`). Not settled; flagged in the
+doc, not hardened.
+
+---
+
+## 5. The minter (brick 2) — BUILT and verified, on an unmerged branch
+
+- **`lance_graph_contract::facet_mint`** — `mint_facets(&[NodeDecl], facet_classid)
+  -> Result<Vec<FacetCascade>, MintError>`. Deterministic sibling ranking,
+  roundtrip-lossless, `MintError::{DuplicateId, UnknownParent, DepthOverflow(>6),
+  FanoutOverflow(>256), Cycle}`. `NodeDecl{id, part_of_parent, is_a_parent}` is
+  producer-agnostic (C++/C#/Ruby/Python harvests all fill it).
+- **Branch `claude/serene-mayer-1a09he`, commit `360fc720`. NOT merged to main.**
+- **I ran its 8 tests myself: 8 passed**, in the zero-dep `lance-graph-contract`
+  crate (no ndarray/burn needed).
+
+---
+
+## 6. THE critical path — brick 3 (the probe that converts CONJECTURE→FINDING)
+
+`ruff_csharp_spo` harvest (MedCare C#) → `facet_mint` → lance SoA → `typeHierarchy`
+/`definition` LSP query, **with `AdaWorldAPI/MedCareV2` as the parity oracle**.
+It produces the first real numbers for: (a) bytes-of-(part_of:is_a) vs source LOC
+(the 10× claim), (b) lossless round-trip of the class graph, (c) query-as-asm-op.
+
+**Its gates are wiring, not research — in order:**
+1. **Merge `serene-mayer` (`facet_mint`)** into main (verified green).
+2. **Mint imported OGIT → `ogar_codebook`** — the source is ALREADY in OGAR at
+   `vocab/imports/ogit/` (~1,940 TTL incl. `NTO/{Audit,Compliance,Legal,Security}`
+   + `SDF/Automation` = HIRO DO-arm source). The remaining step is the
+   **source→codebook wire** (`facet_mint` → publish), NOT harvesting.
+3. **Reserve one `facet_classid`** in the value-tenant migration's codec selector
+   for the AST `6×(8:8 part_of:is_a)` reading (a ClassView reading, NOT a new
+   `ValueSchema` variant — see `soa-value-tenant-migration-v1-harvest.md §5.1`).
+4. Then run the MedCare probe; needs a .NET 8 SDK to run the Roslyn harvester.
+
+---
+
+## 7. Honest caveats / corrections I made (inherit these — don't repeat them)
+
+- **`burn` is NOT a build blocker.** An earlier claim ("workspace compile blocked
+  by proxy 403 on ndarray's `burn` submodule") was **wrong** — `burn` is an
+  *excluded* workspace member (`ndarray/Cargo.toml:375`) with zero code coupling.
+  The minter's crate (`lance-graph-contract`) is zero-dep and builds standalone.
+- **The local `owl` repo is NOT W3C OWL.** It's camel-ai's "Optimized Workforce
+  Learning" multi-agent framework (arXiv:2505.23885). The operator's "OWL↔OGIT
+  bijective-lossless" result refers to W3C OWL (the ontology standard); I cited it
+  **attributed, not re-verified** (the converter is plausibly OGIT's
+  `bin/ogit-tools-jar-with-dependencies.jar`, unread).
+- **Everything in the doc is CONJECTURE.** The carrier is shipped, but no
+  end-to-end measurement exists. Keep the label until the probe runs.
+- **Sync-first discipline:** every PR this session verified `HEAD..origin/main`
+  empty + `diff --stat origin/main..HEAD` showing only intended files BEFORE
+  pushing (a prior arc orphaned commits by not doing this). Keep doing it.
+- **Workspace iron rules in play:** OGAR — no model identifier in artifacts, no
+  German PII labels, no Bash grep/sed/tail/head/awk (use Grep/Read tools).
+  Consumer side — pull via `*Port::class_id` / `canonical_concept_id`, NEVER
+  construct a `*Bridge` or copy the codebook.
+
+---
+
+## 8. Concrete next actions (pick up here)
+
+1. **Decide:** merge #617 (docs, safe) — or leave for review.
+2. **Brick-3 wiring step 1:** merge `claude/serene-mayer-1a09he` (`facet_mint`).
+3. **Brick-3 wiring step 2 (highest leverage):** mint OGIT-source→`ogar_codebook`.
+   The input (`OGAR/vocab/imports/ogit/`) is already in OGAR. **CAVEAT (per the
+   companion handover's `ISS-OGAR-GENETICS-MIRROR-PENDING` +
+   `E-CODEBOOK-MINT-IS-A-CROSS-REPO-ARC`): this is a CROSS-REPO arc, not a local
+   wire** — minting a concept moves OGAR `ogar-vocab` + the contract `CODEBOOK`
+   mirror + lance-graph-ogar `parity::domains_agree` *together*, in one coordinated
+   change. Don't push a half-mint. It still produces the first real byte-size
+   numbers for the 10× claim — just plan it as the cross-repo arc.
+4. **Decide the open gate:** the "Canon:Custom" classid half-order. **Now
+   UNBLOCKED** (the full V3 set is minted, #618) but **must be done ONCE,
+   atomically, over the COMPLETE V3 set** (OSINT+FMA+CPIC) — a half-flipped corpus
+   mis-routes (`I-LEGACY-API-FEATURE-GATED`). Operator call; reuse-clustering
+   argues canon-high, but it collides with the shipped low-half domain routing.
+5. **Board-hygiene debt (mine — owed):** #616/#617 shipped the knowledge doc +
+   findings WITHOUT the same-commit board updates the rule requires. Owed:
+   `EPIPHANIES.md` entry for the three-layer thesis + `PR_ARC_INVENTORY.md` rows
+   for #616/#617 + a `LATEST_STATE.md` Contract-Inventory line. **Fold into the
+   next PR** (the companion session also has #614/#615/#618 PR_ARC rows
+   outstanding — batch them) — do NOT push a standalone board commit to main.
+6. **Optional:** verify ruff #29 in CI (needs network + .NET SDK).
+
+---
+
+## 9. Coordinates
+
+- lance-graph branch: `claude/medcare-bridge-lance-graph-wmx76z` (HEAD `e34a7fa`,
+  synced FF to `main` which includes #616 at `b7eb02c`).
+- minter: `claude/serene-mayer-1a09he` @ `360fc720` (unmerged).
+- ruff branch: `claude/medcare-bridge-lance-graph-wmx76z` (PR #29, scaffold @ `0f463f2`).
+- OGIT-in-OGAR: `OGAR/vocab/imports/ogit/` (~1,940 TTL).
+- Operator note (not code): MedCare Railway deploy still binds `:3000` because a
+  Railway **PORT service variable** overrides `$PORT` — the operator must remove
+  it; the app correctly prefers `$PORT`.