[AdaWorldAPI]

⊘ Correction (2026-10-23, commit f9db51966). The diagram drew layer [7], "ONNX mathematical conformance adapters", inside the vertical chain between egglog [6] and Tract [8]. [7] is an external leaf facade: an ONNX name maps straight onto an existing native function, and no native layer depends on it. lance-graph-onnx (#1754, draft) implements it. Its only normal dependency is ndarray, it does not import R2IL's router [1], and it has no dispatch.

Branch state (commit 0f88b75eb): main moved; #1744 (the D-ONNX-RELALG-CAP S0 draft) was merged by the operator. This PR merged main back in. The conflicts were in board files only: both STATUS_BOARD sections were kept and the entries index was regenerated. The S0 discussion in §6 is unchanged here and is still the owners' call.

⊘ Correction (2026-10-23, commit ace69d64c). This description and the first version of the diagram called layer [1] "the R2IL polyfill — lance-graph-r2il::procs + FoldDialect". That conflated R2IL with one of its strategies.

R2IL is an opcode table (ogar-r2il) with two readings:

• machine (0xC400): r2sleigh's r2conc / r2sym;
• fold (0xC401): FoldDialect.

Layer [1] is the router. It resolves a route once per ClassId (DIRECT / COMPOSE onto ndarray::simd, LOWER to r2sleigh only for machine semantics, or REFUSE) and stores the callable in the binding. The hard invariant is ClassId → LazyLock → prepared dispatch → direct kernel, with no per-call routing.

Measured in #1752 (D-MPF-4): the prebound route costs the same as direct ndarray::simd; the frame/focus/scope guard adds 27–28 ns. The production ClassId-keyed table does not exist yet (seam S17a, LAB).

Docs only: two plans, the board rows, and nothing else. Merges nothing, adds no ABI, changes no execution contract. The corrections to #1746–#1751 and #1750 were approved by the operator on 2026-10-23.

The two P0 invariants

1. ONNX is an external mathematical compatibility specification. The existing R2IL router (lab(mpf): D-MPF-1 — the Stage 1 math polyfill already exists; one expression, oracle, refusals, cost split #1752; originally written "polyfill", see the correction above) is the native starting point. No ONNX graph, Tract runtime or Optimum dependency is in the internal execution path.
2. Quack binding + DR64³ + Palette256³ is the intended native representation. DR64³ owns the proposed local relational and focus geometry. Palette256³ owns the proposed numerical representation. Existing R2IL, mask-risc, BGZ, Helix and ndarray primitives are reused.

d-native-repr-contract-v1.md: which seams exist (read in source at f0b046b9f)

status	seams
SHIPPED	Quack binds names to numeric IDs once and reuses the result (bind.rs); ordered-lane version witness; bgz17 Palette256³ (PaletteEdge = one u8 per S/P/O plane, three codebooks); R2IL / mask-risc / ndarray execution
LAB	dr-math-binding-probe: frame seal with stale refusal, focus and scope guards, typed method axes, 64×64 DR register. dir-rmap-probe: 256-slot global → local focus. egglog-proved Boolean composition. The R2IL route per ClassId (S17a, D-MPF-4)
PROPOSED	the DR64³ third axis (no type; the register is DR64²); a production table registrar
MISSING	a binding store in Quack; codebook runtime identity (no id, generation or digest on Palette); the binding from DR coordinates to palette values

Smallest missing seam: a Quack-bound query does not know its frame (D-NFR-1). Section 3 of the contract separates reusable validated bindings from frame, focus and scope state. Those checks are re-run on every use and are never cached away.

d-native-first-dependency-correction-v1.md: the canonical correction

• One dependency diagram that every affected PR references.
• The 218 harvested passages reduce to seven root assumptions (A1–A7). Each is corrected once, at the PR that owns it.
• Native fallback and exact / approximate rules are made explicit.
• The lab(mpf): D-MPF-1 — the Stage 1 math polyfill already exists; one expression, oracle, refusals, cost split #1752 baseline is recorded:
  • the 50 ns one-word target is not met through the mask-risc and FoldDialect paths;
  • the fair i64 loop supersedes the 1.6× claim;
  • the D-MPF-4 route rows are added.
• An owner-facing S0 proposal is included. It decides nothing about D-ONNX-RELALG-CAP S0: ONNX as an import membrane over the existing substrate; board closeout #1726–#1741 #1744 / docs(onnx) S0: convergence amendment (HOLD, competing capstones #1744/#1750) #1745 / docs(board): D-ONNX-RELALG-CAP S0 — ONNX intake onto existing surfaces + board reconciliation #1750.

Board

• STATUS_BOARD.md: new D-NFC-0 and D-NFR-0..5 rows.
• INTEGRATION_PLANS.md: new entry.
• entries/2026-10-23-d-nfr-0-native-repr-contract.md: new entry file.
• Entries index regenerated.

---

Generated by Claude Code

🤖 Generated with Claude Code

https://claude.ai/code/session_01G22yT6htkcdyXsihxxXdrg

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 9f96fb7e-5e80-476e-aa85-dde62f034460

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Warning

Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption.

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-23T22:15:07.441193Z	ace69d6	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ace69d64cb

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Mark the ClassId-prebound router as LAB rather than SHIPPED

The canonical diagram labels the ClassId-prebound router as SHIPPED, but the production lance-graph-r2il/src/lib.rs implements the FoldDialect strategy, not ClassId-based routing, and S17a in the companion contract explicitly says that no production ClassId-keyed table exists. Searching the reviewed tree also finds neither the cited math-polyfill-probe/examples/classid_route.rs nor its D-MPF plan. Because this diagram is designated as the dependency reference for subsequent work, its current label incorrectly makes the proposed routing integration look available to production consumers. Label the router LAB and distinguish it from the shipped execution strategies.

Useful? React with 👍 / 👎.

[AdaWorldAPI]

Agreed and fixed in 7a72923d4. The router row now reads LAB: it exists only as math-polyfill-probe/examples/classid_route.rs (#1752), and no production ClassId table exists (S17a). A separate SHIPPED line lists the strategies beneath it: ndarray::simd, mask-risc programs, and r2il's FoldDialect and procedures. The old label is kept as a marked supersession note.

The example and the D-MPF plan were missing from this branch only because they live on #1752.

---

Generated by Claude Code