Repository navigation
feat(std): T-Numeric-Construction Slice 3 — Int = AbelianGroup<GroupCompletion<Nat>> - #1466
Merged
Merged
Conversation
# Conflicts: # src/v3/compiler/src/bootstrap_generated.rs # src/v3/compiler/src/bootstrap_generated_without_parse_surface.rs
# Conflicts: # src/v3/compiler/src/bootstrap_generated.rs # src/v3/compiler/src/bootstrap_generated_without_parse_surface.rs
# Conflicts: # src/v3/compiler/src/bootstrap_generated.rs # src/v3/compiler/src/bootstrap_generated_without_parse_surface.rs
briansrls
added a commit
that referenced
this pull request
May 6, 2026
…gration Per proud-lynx-311 dispatch ack at gunbc#1739 #issuecomment-... 2026-05-06: re-grep at HEAD revealed substrate state diverges materially from S9 brief's "Option A vs B decide" framing. Reality check at HEAD: - `dsl/std/integer.dag:55` already carries `type Int = AbelianGroup<GroupCompletion<Nat>>` per Slice 3 PR #1466 / commit `4ba0a6d04`. Single-authority Q6 audit doc at `docs/audit/t-numeric-construction-group-completion-6q.md` ratified this shape and explicitly rejected the compact form. - `dsl/std/integer.dag:56` still on legacy `type UInt = UInt64` alias; UInt migration is the actual Phase-1 worker scope. Brief Phase-1 step 1 reframed: Int landing acknowledged; worker confirms HEAD match. Phase-1 step 2 reframed: UInt migration target named (`type UInt = CommutativeMonoid<Nat>`); fixed-width rows preserved per existing legacy-rows policy. Historical Option A vs B framing kept as note (superseded). Confirmed S9 dispatch ack from worker; UInt-only Phase-1 PR is in-scope. Emission entries deferred per S3 (valiant-ant-72 pre-landing); Phase-2 deferred per S8 (quiet-boar-160 pre-landing).
briansrls
added a commit
that referenced
this pull request
May 6, 2026
…id<Nat> Migrates the default `UInt` alias from the legacy fixed-width `UInt = UInt64` to the abstract algebra carrier `UInt = CommutativeMonoid<Nat>`, consistent with the landed Slice 3 form `Int = AbelianGroup<GroupCompletion<Nat>>` (PR #1466, docs/audit/t-numeric-construction-group-completion-6q.md Q6 single-authority resolution). Per S9 worker brief (R3 design schedule §1 S9, ratified 2026-05-06): Nat plays as the commutative-monoid carrier under addition (closure / associativity / identity 0 / commutativity; no inverses, since Nat lacks additive inverses by construction). UInt as `CommutativeMonoid<Nat>` is the additive structure on that carrier — the honest unsigned-integer algebra without inverses. Fixed-width rows `UInt8..UInt128` remain on `Semiring<Word*>` per existing legacy-rows policy in `dsl/std/integer.dag` (only the default alias migrates; storage rows stay shape-conflated until v3 supports trait/witness syntax separation). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
5 tasks done
briansrls
added a commit
that referenced
this pull request
May 6, 2026
…t Compose<witness, ...> Codex BLOCKING at sha 8099ace on PR #1817 caught a real type-level error: Compose<AbelianGroup, MachineWidth<64>> composes the *algebra witness shape* rather than the integer carrier. Per dsl/std/algebra.dag:148-150: "T = GroupCompletion<M> is the carrier; AbelianGroup<T> carries op/identity/inverse over that carrier" AbelianGroup<T> is a witness shape generic over carrier T, not a carrier constructor. The bare AbelianGroup (without args) is just the kind/ constructor — putting it in Compose<...> slot-1 doesn't yield an integer carrier. Correct form: first slot of Compose<...> holds the **algebraic concept** (the fully-applied carrier+witness composite that IS the Int / UInt / Real / Nat type), second slot holds the machine-constraint: - Int<64> = Compose<Int, MachineWidth<64>> (Int = AbelianGroup<GroupCompletion<Nat>> per #1466) - UInt<64> = Compose<UInt, MachineWidth<64>> (UInt = CommutativeMonoid<Nat> per #1818) - Real<64> = Compose<Real, MachineWidth<64>> (Real = ApproximateField<Rational>) - Nat<8> = Compose<Nat, MachineWidth<8>> Updates sub-decision 3 (type-level spelling) + sub-decision 4 (approximate- algebra layering) in both r3-design-schedule-2026-05-06.md §S3 and r3-program-plan.md §10.3 Q-MachineConstraint-Carrier row. Wrong-form references retained as anti-references only ("prior phrasing was wrong"; critical correction note + dsl/std/algebra.dag citation). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
4 of 5 tasks
briansrls
added a commit
that referenced
this pull request
May 6, 2026
* WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * fix(briefs): retire INVARIANTS.md line-number cites in r3-426 brief Per codex review on PR #1782 (sha c2512cd): the new briefs introduced fresh `INVARIANTS.md:<line>` citations — the same stale-line-anchor shape the :425 sweep brief itself names as needing dissolution. Replaces four line-number references in r3-426-sfp-subdoc-rename-and-author-worker.md (lines 16, 28, 58, 85) with prose-anchored locators citing the section anchor + bullet text. The other PB-lane briefs codex flagged (r3-pb-runtime-equivalence- corpus-seed-audit.md, r3-pb-t-fixedpoint-worker.md) belong to neat-bear-351's lane and are not in this PR's scope; will route separately. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): correct L6 type count + r3-435 dual-path retirement receipt Per openai-pro review on PR #1782 (sha c2512cd): 1. L6 brief said acceptance ratchet asserts "five types" exist but declares six (ShapeATarget, FormAxis, BehaviorAxis, MethodTemplateContractKey, EmissionCell, EmissionPathProjection). Updated both occurrences to enumerate all six explicitly. 2. r3-435 brief authorized both Path A (typed-AST reader, deletes IntegrationRsScan) and Path B (scanner widening with Char state), but retirement receipt language only covered Path B. Updated ROADMAP/ledger retirement and authority-audit step 5 to cover both authorized dissolution paths. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): L6 second 'five types' site + T-E-P lookup-gate authority Per cursor review on PR #1782 (sha 2072b98): 1. L6 brief acceptance bullet at :131 still said "five types" while the parenthetical and the Slice step at :117 said six. The prior fix only caught one of two sites; corrected the second. 2. T-E-P brief `e_p_call_pattern_lookup_authoritative` acceptance named only `lower_call_pattern` as "the only route from call site to LoweringTarget" — but Phase 2 explicitly requires lenses route through `per_call_pattern_at` (the L-7 single- authority lookup), NOT `lower_call_pattern` (compiler-internal). Acceptance text rewritten to name `per_call_pattern_at` as the consumer-facing gate and clarify `lower_call_pattern`'s internal-only scope. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): strengthen L6 ratchet from count-parity to per-row bijection Per openai-pro REQUEST_CHANGES on PR #1782 (sha 2072b98): the L6 acceptance ratchet asserted only row-count parity between emission_path_projections and the union of MethodTemplateContract rows. Count parity permits duplicate projection keys and missing source rows — the bridge from list-non-empty proxy to per-row projection would not be mechanically proven. The whole point of introducing MethodTemplateContractKey { target, dag_method } is to provide 1:1 row identity. Rewrites the ratchet to assert per-row bijection: every MethodTemplateContract source row has exactly one matching EmissionPathProjection row by key; every projection key resolves to exactly one source row; duplicate projection keys fail closed. (The other openai-pro BLOCKING finding — T-E-P e_p_call_pattern_lookup_authoritative naming the wrong authority — was already fixed at 6282a07; the review citation was against the pre-fix sha 2072b98.) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): split L6 ratchet into slice-active vs deferred-activation gates Per BLOCKING inline review on PR #1782 at d924594: the prior bijection ratchet was unsatisfiable as authored. The *_method_template_contracts lists at HEAD (rust/python/go) are non-empty; shipping emission_path_projections: [] alongside a strict bijection requirement means the bijection cannot pass when the slice lands — source rows would be uncovered. Prior text incorrectly said "vacuously true on the empty source set" — the SOURCE set isn't empty, only the projection set is. Splits the ratchet into two gate categories: - Slice-active gates: types/data shape exist + empty-state predicate (emission_path_projections == []). - Deferred-activation gate: per-row key bijection, authored as a test scaffold by this slice but #[ignore]'d (or feature-gated on len > 0) until Grounding's follow-up populates rows. On activation, asserts the bijection over the populated set. This honors P1 doc faithfulness: the slice's acceptance contract is satisfiable as authored, and the bijection becomes the load-bearing gate the row-population PR satisfies. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): require L6 bootstrap enrollment to honor P2 facts-flow-forward Per BLOCKING inline review on PR #1782 at L6:34: the slice authored src/v3/std/cross_target_coverage.dag but never required enrolling it in the bootstrap/load authority. Without enrollment the file is dead-letter on disk and EmissionPathProjection can be absent from the downstream Dag — silent P2 facts-flow-forward violation. Adds a new Slice step 2 requiring a cross_target_coverage field on BootstrapFixtures in src/v3/std/extdeps_bootstrap_fixtures.dag mirroring the *_method_template_contracts enrollment pattern, plus a loader-visibility ratchet. Renumbers subsequent Slice steps. Adds matching Acceptance bullet. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: R3 Substrate * WIP: R3 Substrate * docs(briefs): normalize Practice 4 citation in S3 brief per codex review Codex review on PR #1782 (sha 2614217) flagged stale citation "Practice 4 (Step 4 of the type-introduction checklist)" in S3 brief — no such checklist step exists in the pinned rubric. Live rule per docs/modeling-discipline.md:131-134 is the "What to check" checkpoint-comment requirement under Practice 4 (coproduct dissolution). Updated the S3 reference to cite that line directly and drop the stale "Step 4 of type-introduction checklist" wording. S2/S8/S9 already use bare "Practice 4" without the stale steps wording. S1 (committed earlier today) does not contain the stale phrase. S6 (pre-existing brief flagged by reviewer as carrying the same pattern) is out of this branch's scope; flag for separate paydown. * docs(briefs): sweep stale Practice 4 citations corpus-wide Codex review on PR #1782 flagged the stale "Step 4 of the type-introduction checklist" wording — no such checklist exists in docs/modeling-discipline.md. The live rule is the "What to check" checkpoint-comment requirement under Practice 4 (coproduct dissolution) at docs/modeling-discipline.md:131. Sweep across docs/ found 2 additional briefs carrying the stale phrasing (S3 fixed previously at 9db003d): - docs/briefs/r3-l6-emission-path-projection-substrate-worker.md - docs/briefs/r3-x1b-s1-transform-dispatch-substrate-worker.md Both updated to cite the live rule directly. Verified zero remaining "type-introduction checklist" or "Step 4 of the" references in docs/ tree. * docs(briefs): add Practice 4 citation discipline note to authoring checklist Per parent inbox status flag at gunbc#828 #issuecomment-4385346221: add a Citation discipline section to brief-authoring-checklist.md naming the correct shape — Practice 4 (coproduct dissolution) + "What to check" rule at docs/modeling-discipline.md:131 — and calling out the stale "Step 4 of the type-introduction checklist" phrasing as a failure mode. References the corpus sweep at 2dc92f3. Prevents future briefs that copy old templates from re-introducing the stale citation pattern. * docs(r3): cascade work post-Director ratification — S1 worker, R4 routing, Unit/Quantity canvas Per Director ratification at gunbc#828 inbox response 2026-05-06 (zesty-bear-812): 1. S1 Q1+Q2 RATIFIED → author S1 gap-test representative worker brief (`r3-substrate-s1-gap-test-representative-worker.md`). Dispatchable on T-E-P Phase 1 + E6-G0d landing; closes ledger row #61. 2. S2 Q2 RATIFIED (option-(b) T-LBP narrowing) + S3 + S8 deferrals → R4 carve-out routing ledger (`docs/r4-carve-out-routing.md`). Captures C1 (parallelism lens), C2 (effect_enum lens), C3 (register narrowing), C4 (deferred MachineConstraint axes), C5 (rounding-mode product extension), C6 (Unit/Quantity). Cascade messages to Verification / Grounding / Debt-Paydown Mgrs. 3. S9 Phase-3 dimensional refinement boundary → Unit/Quantity carrier canvas (`r3-substrate-s9-unit-quantity-carrier-canvas.md`). Surfaces Q-Unit-1..5 for Director ratification: carrier name, 2-axis vs 3-axis Phase-1, Duration/Seconds collapse, Refined composition order, Practice 4 classification. Updates §1.8 ledger receipts for carved items; R3 closure receipts treat carved items as explicitly-deferred (not drift). * docs(briefs): citation-discipline fix-forward — anchors/quotes, not line numbers Director redirect at gunbc#828 #issuecomment-... noted the citation paydown's replacement form (`docs/modeling-discipline.md:131`) is the same drift class as the original fabricated-step error. Bare line numbers shift silently on any doc edit above the cited line. Generalized fix: - brief-authoring-checklist.md "Citation discipline" section rewritten to prohibit bare line-numbers across ALL cross-doc references (modeling-discipline.md, r3-program-plan.md, r3-structure.md, INVARIANTS.md, MODELING.md, etc.). Use section anchors (#section-name) or rule-text quotes. - 3 briefs that swapped in :131 (S3, S6, X1B) re-converted: - docs/modeling-discipline.md#4-coproduct-dissolution anchor - "What to check" rule quoted inline (self-contained) Brief-authoring-checklist failure-mode section names both vectors (fabricated step + bare line numbers) and references both sweep commits in audit trail. Broader cross-doc citation sweep across briefs I authored on this branch (frontmatter authority-docs lists, body prose) is recommended per Director note — flagging as separate paydown task; this commit closes the immediate recurrence vector. * docs(briefs): fix unfaithful Int=AbelianGroup<Nat> + sweep stale 'Step 4' refs Per openai-pro review on PR #1782 sha 0ecd481 (gunbc#1782 #...): BLOCKING fix (P1 modeling faithfulness): - S9 + S3 directed workers to model `Int = AbelianGroup<Nat>`, but Nat is not an Abelian-group carrier (additive inverses not representable). Reframed: - `Int` is the additive Abelian group on integers (carrier Z); `Nat` is a commutative monoid under addition. - Algebraic relationship is **explicit group-completion**: `Int ≡ GroupCompletion<CommutativeMonoid<Nat>>` — NOT `AbelianGroup<Nat>` parameterization. - Worker either models `Int` directly as `AbelianGroup` (terminal Abelian-group instance) OR introduces explicit `GroupCompletion<M>` substrate carrier; decision lands in Phase-1 PR per P1 procedure. - S3 cross-reference + Phase-4 algebra-axis preservation note reframed accordingly. NON-BLOCKING in-PR cleanup (Practice 4 citation discipline): - 6 briefs still carrying stale 'Practice 4 Step 4' wording (r3-coproduct-1/2/3, r3-l6, r3-t-e-p, r3-x1b) updated to use the section anchor (`docs/modeling-discipline.md#4-coproduct-dissolution`) + inline rule-text quote ("What to check") form per the citation discipline established in `b4cae5299`. Closes both findings from openai-pro review. * docs(briefs): Q-Unit-1..5 RATIFIED — apply Scale.Unit→One rename + ratification banner Director ratified Q-Unit-1..5 at gunbc#828 #issuecomment-... 2026-05-06 (zesty-bear-812). All 5 questions ratified; one rename required: - Scale enum's `Unit` value collides with outer `Unit<Q, S>` carrier name (`Unit<Time, Unit>` is shadowing-confusing). Rename to `One`. - Duration ≡ Unit<Time, One> (collapses with Seconds per Q-Unit-3). - Milliseconds ≡ Unit<Time, Milli>. Canvas updated: - Top-level RATIFICATION banner records all 5 ratifications + rename - Scale enum: `Unit` → `One` (with inline rename receipt) - Sketch product expansions use `Unit<Time, One>` / `Unit<Time, Milli>` - Phase-3 reframe table products updated - Q-Unit-3 prose collapse reference updated Subsequent worker brief authoring (Unit/Quantity carrier landing post-prerequisite) inherits the renamed shape. * docs(briefs): Unit/Quantity carrier worker brief — Q-Unit-1..5 RATIFIED Worker brief consumes the ratified Unit/Quantity canvas (`5f22fd06e` ratification banner). Phase-1 lands `Unit<Q, S>` + `Quantity` + `Scale` declarations at the ratified shape (Scale's `One` value per Q-Unit-1 rename). Phase-2 reframes S9 Phase-3 dimensional refinements (Duration / Seconds / Milliseconds) to outer-Refined / inner-Unit composition per Q-Unit-4 RATIFIED. EpochMs deferred to R4 C6 (Aspect-axis follow-up). Practice 4 marks pre-determined per Q-Unit-5: `Unit<Q, S>` 🟢 PRIMITIVE; `Quantity` + `Scale` 🟡 SCAFFOLD with "consumed by ≥1 emission rule" dissolution trigger. Cross-program handoff to Grounding Mgr (#1745) for dimensional emission consumption documented. Worker pin TBD — dispatchable when idle pool refreshes. * docs(briefs): S9 Phase-1 reframed — Int landed at #1466; UInt-only migration Per proud-lynx-311 dispatch ack at gunbc#1739 #issuecomment-... 2026-05-06: re-grep at HEAD revealed substrate state diverges materially from S9 brief's "Option A vs B decide" framing. Reality check at HEAD: - `dsl/std/integer.dag:55` already carries `type Int = AbelianGroup<GroupCompletion<Nat>>` per Slice 3 PR #1466 / commit `4ba0a6d04`. Single-authority Q6 audit doc at `docs/audit/t-numeric-construction-group-completion-6q.md` ratified this shape and explicitly rejected the compact form. - `dsl/std/integer.dag:56` still on legacy `type UInt = UInt64` alias; UInt migration is the actual Phase-1 worker scope. Brief Phase-1 step 1 reframed: Int landing acknowledged; worker confirms HEAD match. Phase-1 step 2 reframed: UInt migration target named (`type UInt = CommutativeMonoid<Nat>`); fixed-width rows preserved per existing legacy-rows policy. Historical Option A vs B framing kept as note (superseded). Confirmed S9 dispatch ack from worker; UInt-only Phase-1 PR is in-scope. Emission entries deferred per S3 (valiant-ant-72 pre-landing); Phase-2 deferred per S8 (quiet-boar-160 pre-landing). * docs(briefs): fix two BLOCKING citation drift items per codex review Codex review on PR #1782 sha 5f22fd0 flagged two authority-receipt gaps: 1. brief-authoring-checklist.md:118 — `<this-commit>` placeholder in the failure-mode section was unverifiable post-merge. Replaced with `b4cae5299` (the actual anchor/quote conversion commit). 2. r3-378-http-path-helper-fail-open-worker.md:84 — `INVARIANTS.md C-8 fail-closed` cited without anchor or rule-text quote, in violation of the citation discipline added in this PR. Converted to `INVARIANTS.md#p3-fail-closed` (P3 heading anchor) + inline rule-text quote naming P3 + C-8. PR's own discipline now consistently applied across the diff. * docs(briefs): sweep bare .md:NNN citations in Substrate-Mgr-authored briefs Per Director recommendation in citation-discipline thread (gunbc#828 follow-on): bare line-number references drift silently. Swept 19 remaining instances across 5 briefs I authored on this branch: - docs/briefs/r3-substrate-s1-q-class-2-chain-break-gap-test-canvas.md - docs/briefs/r3-substrate-s1-gap-test-representative-worker.md - docs/briefs/r3-substrate-s2-t-lbp-scope-calibration-canvas.md - docs/briefs/r3-substrate-s3-machine-constraint-carrier-worker.md Conversion: bare `file.md:NNN` → `file.md §"Section name"` form. Stable; survives doc edits; human-readable. Section names per HEAD heading audit. Verification: `grep -nE '.md:[0-9]+' docs/briefs/r3-substrate-*.md docs/r4-carve-out-routing.md` returns zero results in Substrate-Mgr-authored scope. Broader corpus scope (~232 references across non-Substrate briefs) remains as separate paydown task. * docs(briefs): fix BLOCKING + NON-BLOCKING items per openai-pro review on a462f00 Three findings addressed; 2 BLOCKING + 1 NON-BLOCKING: BLOCKING — R3/R4 dual authority for Unit/Quantity carrier: R4 carve-out C6 narrowed from "Unit/Quantity carrier landing" (deferred all of it) to **"Aspect-axis (PointKind) follow-on for EpochMs and instant/rate-shaped refinements only"**. The 2-axis Measure<Q,S> carrier landing IS in R3 per Q-Unit-1..5 RATIFIED; only the Aspect axis (PointKind = Magnitude | Instant | Rate) carves to R4. Cascade-message C6 reference + R4 program- plan input list updated to match. Cross-Mgr message clarifies Grounding consumes 2-axis dimensional refinements in R3. BLOCKING — Locked design partially applied (Scale.Unit→One rename in canvas alternate-sketch sections): - Canvas line 105-108 alternate sketch ("Open question: Instant vs Duration axis") still used pre-rename `Unit` Scale value; added pre-ratification disclaimer marking the section as deferred to R4 C6 + noting `Unit` is pre-rename - Canvas line 163 Q-Unit-3 collapse note: `Scale = Unit` → `Scale = One` (BLOCKING — bare line-numbers in S1 canvas frontmatter): NOT APPLICABLE — already fixed at sweep commit 127287a prior to review; reviewer was on sha a462f00 pre-sweep. NON-BLOCKING — placeholder receipts: - brief-authoring-checklist.md `#issuecomment-...` → concrete ID 4385393971 (Director's anchor-discipline reply) - r3-substrate-unit-quantity-carrier-worker.md (3 occurrences) + canvas RATIFICATION banner: `#issuecomment-...` → concrete ID 4385412256 (Q-Unit ratification reply) - `<this-commit>` placeholder in checklist already fixed at 5be1c99 Verification: zero `#issuecomment-...` or `<this-commit>` placeholders in docs/briefs/ + docs/r4-carve-out-routing.md. * docs(briefs): codex BLOCKING review fixes — Practice 4 source-level + policy scope Three BLOCKING findings on PR #1782 (codex sha b913e2f): 1. **Citation-discipline policy scope**: addressed by scoping the rule prospectively to briefs landing after the policy commit (`b4cae5299`). Pre-existing corpus references are flagged as paydown but NOT individually blocking; coordinated single-PR sweep is the right paydown shape. Substrate-Mgr scope already swept at `127287a66`. 2. **R4 C6 reconciliation**: already fixed at `6e5a5481a` (R4 C6 narrowed to Aspect-axis follow-on for EpochMs only; 2-axis carrier landing IS in R3). Codex sha b913e2f was pre-fix. 3. **Practice 4 source-level vs PR-body conflation**: corrected. Acceptance bullet for Unit/Quantity worker brief now requires inline checkpoint comments on the LIVE declarations in `dsl/std/units.dag` (load-bearing artifact per `docs/modeling-discipline.md#4-coproduct-dissolution` "What to check"). PR-body summary is supplementary, not substitute. Closes the review's reconciliation gates. * docs(briefs): absorb Q-MachineConstraint + Q-Unit-1-Recanvas ratifications Two major Director ratifications absorbed: 1. Q-MachineConstraint-Carrier RATIFIED at gunbc#828 #issuecomment-4385530115 (Brian directive 2026-05-06: "universal substrate, ratify defaults"; PR #1817 queued). 6 sub-decisions: - MachineWidth<bits> only for R3; defer RegisterClass / EndianMode / Alignment / signedness post-R3 - Parametric Compose<Algebra, MachineConstraint> type-level interaction shape (lookup-maps REJECTED — AlgebraMachineProduct record-keyed table form retracted) - Int<64> ≡ Compose<AbelianGroup, MachineWidth<64>> (algebra-side options A and B preserve interaction semantics) - Real<64> ≡ Compose<ApproximateField<Rational>, MachineWidth<64>> (algebra approx + machine approx as independent composing axes) - ≥3 pairs is minimum, not target - UNIVERSAL substrate: every target carries machine-constraint facts; targets without native machine-width semantics handle omission at Grounding-level discharge (target-conditioned lowering NOT target-conditioned substrate) Reframed: S3 brief Phase-1, S3 Phase-3/4 emission entries, S8 Phase-1/2/3 (ApproximateField<Real> → ApproximateField<Rational>; AlgebraMachineProduct → Compose<...>), S9 emission entries. 2. Q-Unit-1-Recanvas RATIFIED Measure<Q, S> at gunbc#828 #issuecomment-4385539791. valiant-ibex-312's substrate-state grep caught Unit<Q, S> collision with kernel Unit at dsl/std/types.dag:176; Director ratified Measure<Q, S> as outer carrier name (conventional in scientific computing; no dsl/std/ collision). Reframed: Unit/Quantity worker brief throughout (Unit<Q,S> → Measure<Q,S>; dsl/std/units.dag → dsl/std/measure.dag); canvas spec sections + sketch products (Refined<Measure<Time, ...>...>); R4 carve-out C6 with informal/formal name disambiguation note. Both rename + reshape are now consistent across S3 / S8 / S9 / Unit-Q briefs + canvas + R4 routing. Workers (valiant-ant-72 on S3, quiet-boar-160 on S8, proud-lynx-311 on S9, valiant-ibex-312 on Unit-Q) all dispatch against the ratified shape. * docs(briefs): S11 Slice C worker brief — ROADMAP :425 retirement post-#1801 Cascade-clearance trigger fired post-#1801 (Slice B) merge 2026-05-06. S11 brief authored per B1 RATIFIED prose+regen bundling discipline. Phase 1: per-file edits across the post-A+B residual list (~10 files per smart-ram-167's Slice B post-rebase enumeration); anchor + rule-text-quote pattern per current citation discipline; bootstrap regen in same PR (no split). Phase 2: ROADMAP :425 row PARTIAL → Retired with 3-slice receipt; Q-Slice-C-Retirement-Receipt resolution. Worker pin smart-ram-167 (#1759) — Slice B precedent owner. Dispatch packet sent separately to #1759. --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 6, 2026
* WIP: proud-lynx-311 * WIP: proud-lynx-311 * WIP: proud-lynx-311 * WIP: proud-lynx-311 * chore(v3): regen bootstrap snapshot for substrate.dag :83 anchor swap The :425 Slice A anchor swap (\§P1` → `#p1-modeling-faithfulness`) on src/v3/std/substrate.dag:83 shifted SourceSpan byte offsets for every node parsed below that line, drifting bootstrap_generated.rs. Regenerated via `cargo run -p v3-compiler --features bootstrap-regen-fresh --bin regen_bootstrap`. No semantic change — pure span-offset refresh. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(:425): retire trailing `§P5` shorthand on r2-manager-brief-authority-matrix.md:205 Cursor review on a402dca flagged residual `\`§P5\` (dispatch-discipline)` shorthand inside the same sentence as a P2 anchor. Slice A's mechanical sweep matched only `INVARIANTS.md §P5` prefixed forms; the bare shorthand survived. Swap to `INVARIANTS.md#p5-progress-is-dissolution` matches adjacent P2 form. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test(v3): refresh parse_corpus_manifest after substrate.dag :83 anchor swap The :425 Slice A anchor swap shifted byte offsets in src/v3/std/substrate.dag, drifting SG-2 parse snapshot hashes (same root cause as the bootstrap_generated regen in a402dca). Regenerated via: cargo test -p v3-compiler --test integration -- --ignored \ refresh_handwritten_parse_snapshot_manifest Verified: handwritten_parse_snapshot_matches_manifest passes locally. Pure offset/hash refresh — no semantic change. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * ci: re-trigger v3 after t_demo_fixture transient flake * fix(:425): align link URL with anchor label on r2-manager-brief-authority-matrix.md:222 Codex review on 9e67e1f noted the Markdown label was swapped to the #p5-progress-is-dissolution anchor but the URL still pointed at the INVARIANTS.md root. Aligned URL to match label. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * feat(std): T-Numeric-Construction S9 Phase-1 — UInt = CommutativeMonoid<Nat> Migrates the default `UInt` alias from the legacy fixed-width `UInt = UInt64` to the abstract algebra carrier `UInt = CommutativeMonoid<Nat>`, consistent with the landed Slice 3 form `Int = AbelianGroup<GroupCompletion<Nat>>` (PR #1466, docs/audit/t-numeric-construction-group-completion-6q.md Q6 single-authority resolution). Per S9 worker brief (R3 design schedule §1 S9, ratified 2026-05-06): Nat plays as the commutative-monoid carrier under addition (closure / associativity / identity 0 / commutativity; no inverses, since Nat lacks additive inverses by construction). UInt as `CommutativeMonoid<Nat>` is the additive structure on that carrier — the honest unsigned-integer algebra without inverses. Fixed-width rows `UInt8..UInt128` remain on `Semiring<Word*>` per existing legacy-rows policy in `dsl/std/integer.dag` (only the default alias migrates; storage rows stay shape-conflated until v3 supports trait/witness syntax separation). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(std): UInt = Nat (preserve CommutativeSemiring surface) Codex review at PR #1818 / Substrate Mgr inline review at dsl/std/integer.dag:102 surface a P2 facts-flow-forward violation in the prior `type UInt = CommutativeMonoid<Nat>` pivot: wrapping Nat (already `CommutativeSemiring<Magnitude>`) in `CommutativeMonoid<Nat>` projects out the multiplicative monoid + identity 1, dropping substrate facts already established at the Nat layer. The substrate cannot structurally name "the additive monoid projection of Nat" without losing Nat's other monoid. Pivot to the design-doc-named direct alias: `type UInt = Nat`. Per the existing comment header in this file, the design doc states `UInt IS Nat`; Nat already carries the full unsigned-integer algebra surface (additive monoid + multiplicative monoid + distribution + commutativity on both). No new operations are needed — the alias IS the algebra. Asymmetry with `Int`: `Int` requires additive inverses Nat lacks, so `GroupCompletion<Nat>` is the honest carrier introduction. `UInt` requires no new operations, so no carrier introduction is needed. Updates the PR's deferred Phase-1 Step 3 receipt to the `Compose<CommutativeSemiring, MachineWidth<N>>` parametric form (post-S3 landing). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test(v3): refresh parse-snapshot manifest for integer.dag UInt change Mechanical refresh after PR #1818 fix commit. integer.dag debug-bytes shifted 12659→12662, fnv1a64 1eef7086b17908de→104a413082264303 with the type swap from `UInt = CommutativeMonoid<Nat>` to `UInt = Nat` plus surrounding doc-comment rewrite. Generated by: cargo test -p v3-compiler refresh_handwritten_parse_snapshot_manifest -- --ignored Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 6, 2026
…s unification (post-#1817) (#1821) * docs(r3): Q-MachineConstraint-Carrier RATIFIED — universal substrate + 5 PM defaults (Brian directive) Brian directive 2026-05-06: "universal substrate, ratify defaults". Resolves Q-MachineConstraint-Carrier (was OPEN per r3-program-plan.md §10.3) with 6 sub-decisions ratified: 1. Axes in R3 scope: MachineWidth<bits> only. RegisterClass<R> / EndianMode<E> / alignment / signedness-as-axis deferred post-R3. 2. Interaction substrate shape: parametric Compose<Algebra, MachineConstraint> type-level construction. Closed-enumeration lookup-maps rejected (bridges). 3. Type-level spelling: Int<64> = Compose<AbelianGroup, MachineWidth<64>> parametrically. Equivalent under both algebra-side options per PR #1815. 4. Approximate-algebra layering: algebra approx + machine approx are independent composing axes. Real<64> = Compose<ApproximateField<Rational>, MachineWidth<64>> carries both layers per S8 discipline. 5. Class 1 demonstration breadth: ≥3 algebra×constraint pairs is minimum, not target. Broader coverage follows automatically once parser handles generic interaction syntax. 6. Target-specificity: MachineConstraint<C> is UNIVERSAL substrate (Brian directive). Every target carries machine-constraint facts as substrate; targets lacking native machine-width semantics (Python int/float) handle omission at Grounding-level discharge — target-conditioned lowering, NOT target-conditioned substrate. Substrate Mgr S3 dispatch unblocked on machine-constraint side; independent of T-Numeric-Construction algebra-side Option A vs B selection per PR #1815 (interaction semantics carries through under either). Updates: - docs/r3-program-plan.md §10.3 Q-MachineConstraint-Carrier row: OPEN → RATIFIED 2026-05-06 with all 6 sub-decisions named. - docs/r3-design-schedule-2026-05-06.md §S3: drops "open scope" framing; explicit ratified-scope block citing the 6 decisions; S3 dispatch noted as unblocked on machine-constraint side. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): Q-MachineConstraint sub-decisions 1/2/5/6 elaborated per Brian clarifying inline Brian clarifying directive 2026-05-06 on the 6 sub-decisions: 1. AXES — accept MachineWidth<bits> only; add separation-of-facts/concerns discipline note: algebra/machine-constraint/target-lowering are distinct axes. Integer modeling is the canonical exercise. 2. INTERACTION SHAPE (revised from "parametric Compose<...>" framing): "these models interact a la rest apis interacting — scalable / no dual representations (.dag generated)". The interaction substrate is .dag-DECLARED + .dag-GENERATED, never hand-coded. Hard constraint: no dual representations — any Rust counterpart must be generated, never hand-maintained (feedback_isomorphism_or_generation_for_mirrors + feedback_no_generated_code_on_disk). 3. SPELLING — Compose<AbelianGroup, MachineWidth<64>> ratified ("ok fine"). Note added: the interaction substrate elaborating Compose<...> is .dag-generated per sub-decision 2. 4. LAYERING — independent composing axes ratified ("ok fine"). Real<64> = Compose<ApproximateField<Rational>, MachineWidth<64>> stays. 5. DEMONSTRATION BREADTH — strengthened: "3 is not a target, we shouldn't be 'targeting' modeling, we are just doing our best job to faithfully represent the concepts". Closure criterion is "concept is faithfully modeled", not "≥3 pairs land". 3-pair demonstration is minimum existence proof, not closure target. 6. TARGET-SPECIFICITY — universal substrate stays. Python lowering uses faithful target-available primitives (numpy.uint8 / ctypes.c_uint8 / explicit literal-bits), NOT omission. "if python lacks — we would have to actually represent the struct using something that python faithfully provides — i.e. either literal bits or otherwise". Faithful representation > target-conditioned omission. Updates docs/r3-program-plan.md §10.3 Q-MachineConstraint-Carrier row + docs/r3-design-schedule-2026-05-06.md §S3 ratified-scope block to reflect clarifying inline. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): closure predicate aligned to faithful-modeling criterion (openai-pro BLOCKING fix) openai-pro REQUEST_CHANGES at sha 8f5483b on PR #1817 caught a real internal inconsistency: ratification text said "Faithful representation of the concept is the closure criterion; pair-counting is not" but the actual closure predicates across r3-structure.md / r3-program-plan.md / r3-design-schedule-2026-05-06.md still phrased Pass as "≥3 pairs emit to target primitives". This violated single-authority + facts-flow-forward. Fix: align all 5 Pass-condition / closure-predicate locations to the faithful-modeling criterion ratified in Q-MachineConstraint sub-decision 5: - docs/r3-program-plan.md:74 (gate #60 Pass condition in §1.4 ledger): "Gate Pass = concept faithfully modeled (per Brian directive 2026-05-06) — once substrate carries algebra + machine-constraint as independent composing axes with .dag-generated interaction substrate, all valid pairs work by construction. Minimum existence-proof evidence: 3 pairs lower without v2-fallback (evidence, not target)." - docs/r3-program-plan.md:257 (§1.8 canonical ledger row): "concept faithfully modeled... min existence-proof = Int<64>/Real<64>/Nat<8> lower without v2-fallback (evidence, not target)" - docs/r3-program-plan.md:430 (§1.4 Class 1 representative gap-test): "Representative gap-test (minimum existence-proof, NOT target) — substrate faithfully modeling algebra + machine-constraint separation IS the closure criterion; the 3 pairs below are the minimum existence-proof" - docs/r3-structure.md:173 (canonical gate description): "Pass condition: concept faithfully modeled... Minimum existence-proof evidence (not closure target): 3 pairs lower without v2-fallback" - docs/r3-design-schedule-2026-05-06.md:60 (S3 closure predicate cite): "Pass = concept faithfully modeled... 3-pair set is minimum existence-proof evidence, NOT the closure target" Sub-decision 5 explanatory text retains "≥3 pairs land" as anti-reference to the prior phrasing for traceability of the change. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): apply 2 claude APPROVE-with-observations precisions on Q-MachineConstraint Claude review at sha e1ea7cf (post-BLOCKING-fix) APPROVE with two exploratory observations. Both worth applying: 1. Sub-decision 3: clarify Compose<...> elaboration vs interaction substrate. Sub-decision 2 says the *interaction substrate* (composition/projection/ coercion machinery) is .dag-generated, not the Compose<...> type itself. Avoid an S3 worker reading "the Compose type is generated." 2. Closure predicate: explicit note that §1.4 conjunctive-closure rule still binds. Sub-decision 5 narrows what counts as Pass (concept faithfully modeled, not pair-counting), but does NOT relax the conjunctive form (representative gap-test executes AND class-bridge enumeration = 0). Both are tightening clarifications; no disposition change. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): codex BLOCKING fix — Compose<concept, MachineWidth<...>> not Compose<witness, ...> Codex BLOCKING at sha 8099ace on PR #1817 caught a real type-level error: Compose<AbelianGroup, MachineWidth<64>> composes the *algebra witness shape* rather than the integer carrier. Per dsl/std/algebra.dag:148-150: "T = GroupCompletion<M> is the carrier; AbelianGroup<T> carries op/identity/inverse over that carrier" AbelianGroup<T> is a witness shape generic over carrier T, not a carrier constructor. The bare AbelianGroup (without args) is just the kind/ constructor — putting it in Compose<...> slot-1 doesn't yield an integer carrier. Correct form: first slot of Compose<...> holds the **algebraic concept** (the fully-applied carrier+witness composite that IS the Int / UInt / Real / Nat type), second slot holds the machine-constraint: - Int<64> = Compose<Int, MachineWidth<64>> (Int = AbelianGroup<GroupCompletion<Nat>> per #1466) - UInt<64> = Compose<UInt, MachineWidth<64>> (UInt = CommutativeMonoid<Nat> per #1818) - Real<64> = Compose<Real, MachineWidth<64>> (Real = ApproximateField<Rational>) - Nat<8> = Compose<Nat, MachineWidth<8>> Updates sub-decision 3 (type-level spelling) + sub-decision 4 (approximate- algebra layering) in both r3-design-schedule-2026-05-06.md §S3 and r3-program-plan.md §10.3 Q-MachineConstraint-Carrier row. Wrong-form references retained as anti-references only ("prior phrasing was wrong"; critical correction note + dsl/std/algebra.dag citation). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): flag Python width-lowering divergence with design-numeric-construction.md (codex non-blocking) Codex non-blocking improvement at sha 83be121: sub-decision 6's Python faithful-lowering directive (numpy.uint8 / ctypes.c_uint8 / literal-bits) competes with the existing default at docs/design-numeric-construction.md: 194-198 which lowers Nat<N> for any N to plain Python int with size-hint metadata (width-omission default). Brian's directive ("represent the struct using something that python faithfully provides") shifts the standing default to faithful concrete representation, but the existing design doc isn't updated yet. Fix: explicit "Tracked divergence" callout in sub-decision 6 (both r3-design- schedule.md §S3 and r3-program-plan.md §10.3 row). Routes reconciliation to T-Numeric-Construction / Grounding lane paydown — update design-numeric-construction.md Python row when the Grounding-side per-target lowering authority lands. Not bundled in this PR per scope; the divergence is tracked-not-silent. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): sub-decision 6 unified with cost lens — universal faithful representation, cost as discriminator (Brian directive) Brian directive 2026-05-06 elaborated 2x: "basically this would be represented by cost right - in all cases we should be able to represent the structure with literal bits/objects or whatever - the cost would be enormous - in those cases, we would look for alternatives like better modeling/libraries/int & 0xFF" This unifies sub-decision 6 with T-CostLens-Composition: faithful representation is universal (every target CAN represent every concept, literal-bits is the always-available floor); the cost lens is the discriminator that orders the alternatives. Reframing: - Universality of faithful representation: every target carries every concept faithfully (literal-bits as the floor) - Cost lens (T-CostLens-Composition gates cost_lens_reads_target_realization + coercion_cost_equals_complexity_by_construction) reads per-primitive realization cost from target language spec - Grounding selects the lowest-cost faithful representation by reading cost-lens output - For Python u8: cost-tier 1 = numpy.uint8 (low cost), cost-tier 2 = ctypes.c_uint8 (medium), cost-tier 3 = int & 0xFF discipline (universal floor, high cost). All three faithful; cost lens orders. Implication: existing docs/design-numeric-construction.md:194-198 Python row (Nat<N> → int with size-hint metadata) is NON-FAITHFUL — drops width structure to opaque metadata, cost lens cannot read what isn't there. Reframe routes to T-Numeric-Construction / Grounding lane paydown: replace with cost-tiered faithful-representation table. This is the structurally-elegant unification: faithful-representation universality + cost-lens-as-discriminator. Sub-decision 6 IS an instance of "coercion cost = complexity by construction" thesis (T-CostLens-Composition load-bearing claim). Updates both r3-design-schedule-2026-05-06.md §S3 and r3-program-plan.md §10.3 Q-MachineConstraint-Carrier row sub-decision 6. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 6, 2026
…-06 (#1824) * docs(r3): Q-MachineConstraint-Carrier RATIFIED — universal substrate + 5 PM defaults (Brian directive) Brian directive 2026-05-06: "universal substrate, ratify defaults". Resolves Q-MachineConstraint-Carrier (was OPEN per r3-program-plan.md §10.3) with 6 sub-decisions ratified: 1. Axes in R3 scope: MachineWidth<bits> only. RegisterClass<R> / EndianMode<E> / alignment / signedness-as-axis deferred post-R3. 2. Interaction substrate shape: parametric Compose<Algebra, MachineConstraint> type-level construction. Closed-enumeration lookup-maps rejected (bridges). 3. Type-level spelling: Int<64> = Compose<AbelianGroup, MachineWidth<64>> parametrically. Equivalent under both algebra-side options per PR #1815. 4. Approximate-algebra layering: algebra approx + machine approx are independent composing axes. Real<64> = Compose<ApproximateField<Rational>, MachineWidth<64>> carries both layers per S8 discipline. 5. Class 1 demonstration breadth: ≥3 algebra×constraint pairs is minimum, not target. Broader coverage follows automatically once parser handles generic interaction syntax. 6. Target-specificity: MachineConstraint<C> is UNIVERSAL substrate (Brian directive). Every target carries machine-constraint facts as substrate; targets lacking native machine-width semantics (Python int/float) handle omission at Grounding-level discharge — target-conditioned lowering, NOT target-conditioned substrate. Substrate Mgr S3 dispatch unblocked on machine-constraint side; independent of T-Numeric-Construction algebra-side Option A vs B selection per PR #1815 (interaction semantics carries through under either). Updates: - docs/r3-program-plan.md §10.3 Q-MachineConstraint-Carrier row: OPEN → RATIFIED 2026-05-06 with all 6 sub-decisions named. - docs/r3-design-schedule-2026-05-06.md §S3: drops "open scope" framing; explicit ratified-scope block citing the 6 decisions; S3 dispatch noted as unblocked on machine-constraint side. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): Q-MachineConstraint sub-decisions 1/2/5/6 elaborated per Brian clarifying inline Brian clarifying directive 2026-05-06 on the 6 sub-decisions: 1. AXES — accept MachineWidth<bits> only; add separation-of-facts/concerns discipline note: algebra/machine-constraint/target-lowering are distinct axes. Integer modeling is the canonical exercise. 2. INTERACTION SHAPE (revised from "parametric Compose<...>" framing): "these models interact a la rest apis interacting — scalable / no dual representations (.dag generated)". The interaction substrate is .dag-DECLARED + .dag-GENERATED, never hand-coded. Hard constraint: no dual representations — any Rust counterpart must be generated, never hand-maintained (feedback_isomorphism_or_generation_for_mirrors + feedback_no_generated_code_on_disk). 3. SPELLING — Compose<AbelianGroup, MachineWidth<64>> ratified ("ok fine"). Note added: the interaction substrate elaborating Compose<...> is .dag-generated per sub-decision 2. 4. LAYERING — independent composing axes ratified ("ok fine"). Real<64> = Compose<ApproximateField<Rational>, MachineWidth<64>> stays. 5. DEMONSTRATION BREADTH — strengthened: "3 is not a target, we shouldn't be 'targeting' modeling, we are just doing our best job to faithfully represent the concepts". Closure criterion is "concept is faithfully modeled", not "≥3 pairs land". 3-pair demonstration is minimum existence proof, not closure target. 6. TARGET-SPECIFICITY — universal substrate stays. Python lowering uses faithful target-available primitives (numpy.uint8 / ctypes.c_uint8 / explicit literal-bits), NOT omission. "if python lacks — we would have to actually represent the struct using something that python faithfully provides — i.e. either literal bits or otherwise". Faithful representation > target-conditioned omission. Updates docs/r3-program-plan.md §10.3 Q-MachineConstraint-Carrier row + docs/r3-design-schedule-2026-05-06.md §S3 ratified-scope block to reflect clarifying inline. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): closure predicate aligned to faithful-modeling criterion (openai-pro BLOCKING fix) openai-pro REQUEST_CHANGES at sha 8f5483b on PR #1817 caught a real internal inconsistency: ratification text said "Faithful representation of the concept is the closure criterion; pair-counting is not" but the actual closure predicates across r3-structure.md / r3-program-plan.md / r3-design-schedule-2026-05-06.md still phrased Pass as "≥3 pairs emit to target primitives". This violated single-authority + facts-flow-forward. Fix: align all 5 Pass-condition / closure-predicate locations to the faithful-modeling criterion ratified in Q-MachineConstraint sub-decision 5: - docs/r3-program-plan.md:74 (gate #60 Pass condition in §1.4 ledger): "Gate Pass = concept faithfully modeled (per Brian directive 2026-05-06) — once substrate carries algebra + machine-constraint as independent composing axes with .dag-generated interaction substrate, all valid pairs work by construction. Minimum existence-proof evidence: 3 pairs lower without v2-fallback (evidence, not target)." - docs/r3-program-plan.md:257 (§1.8 canonical ledger row): "concept faithfully modeled... min existence-proof = Int<64>/Real<64>/Nat<8> lower without v2-fallback (evidence, not target)" - docs/r3-program-plan.md:430 (§1.4 Class 1 representative gap-test): "Representative gap-test (minimum existence-proof, NOT target) — substrate faithfully modeling algebra + machine-constraint separation IS the closure criterion; the 3 pairs below are the minimum existence-proof" - docs/r3-structure.md:173 (canonical gate description): "Pass condition: concept faithfully modeled... Minimum existence-proof evidence (not closure target): 3 pairs lower without v2-fallback" - docs/r3-design-schedule-2026-05-06.md:60 (S3 closure predicate cite): "Pass = concept faithfully modeled... 3-pair set is minimum existence-proof evidence, NOT the closure target" Sub-decision 5 explanatory text retains "≥3 pairs land" as anti-reference to the prior phrasing for traceability of the change. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): apply 2 claude APPROVE-with-observations precisions on Q-MachineConstraint Claude review at sha e1ea7cf (post-BLOCKING-fix) APPROVE with two exploratory observations. Both worth applying: 1. Sub-decision 3: clarify Compose<...> elaboration vs interaction substrate. Sub-decision 2 says the *interaction substrate* (composition/projection/ coercion machinery) is .dag-generated, not the Compose<...> type itself. Avoid an S3 worker reading "the Compose type is generated." 2. Closure predicate: explicit note that §1.4 conjunctive-closure rule still binds. Sub-decision 5 narrows what counts as Pass (concept faithfully modeled, not pair-counting), but does NOT relax the conjunctive form (representative gap-test executes AND class-bridge enumeration = 0). Both are tightening clarifications; no disposition change. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): codex BLOCKING fix — Compose<concept, MachineWidth<...>> not Compose<witness, ...> Codex BLOCKING at sha 8099ace on PR #1817 caught a real type-level error: Compose<AbelianGroup, MachineWidth<64>> composes the *algebra witness shape* rather than the integer carrier. Per dsl/std/algebra.dag:148-150: "T = GroupCompletion<M> is the carrier; AbelianGroup<T> carries op/identity/inverse over that carrier" AbelianGroup<T> is a witness shape generic over carrier T, not a carrier constructor. The bare AbelianGroup (without args) is just the kind/ constructor — putting it in Compose<...> slot-1 doesn't yield an integer carrier. Correct form: first slot of Compose<...> holds the **algebraic concept** (the fully-applied carrier+witness composite that IS the Int / UInt / Real / Nat type), second slot holds the machine-constraint: - Int<64> = Compose<Int, MachineWidth<64>> (Int = AbelianGroup<GroupCompletion<Nat>> per #1466) - UInt<64> = Compose<UInt, MachineWidth<64>> (UInt = CommutativeMonoid<Nat> per #1818) - Real<64> = Compose<Real, MachineWidth<64>> (Real = ApproximateField<Rational>) - Nat<8> = Compose<Nat, MachineWidth<8>> Updates sub-decision 3 (type-level spelling) + sub-decision 4 (approximate- algebra layering) in both r3-design-schedule-2026-05-06.md §S3 and r3-program-plan.md §10.3 Q-MachineConstraint-Carrier row. Wrong-form references retained as anti-references only ("prior phrasing was wrong"; critical correction note + dsl/std/algebra.dag citation). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): flag Python width-lowering divergence with design-numeric-construction.md (codex non-blocking) Codex non-blocking improvement at sha 83be121: sub-decision 6's Python faithful-lowering directive (numpy.uint8 / ctypes.c_uint8 / literal-bits) competes with the existing default at docs/design-numeric-construction.md: 194-198 which lowers Nat<N> for any N to plain Python int with size-hint metadata (width-omission default). Brian's directive ("represent the struct using something that python faithfully provides") shifts the standing default to faithful concrete representation, but the existing design doc isn't updated yet. Fix: explicit "Tracked divergence" callout in sub-decision 6 (both r3-design- schedule.md §S3 and r3-program-plan.md §10.3 row). Routes reconciliation to T-Numeric-Construction / Grounding lane paydown — update design-numeric-construction.md Python row when the Grounding-side per-target lowering authority lands. Not bundled in this PR per scope; the divergence is tracked-not-silent. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): sub-decision 6 unified with cost lens — universal faithful representation, cost as discriminator (Brian directive) Brian directive 2026-05-06 elaborated 2x: "basically this would be represented by cost right - in all cases we should be able to represent the structure with literal bits/objects or whatever - the cost would be enormous - in those cases, we would look for alternatives like better modeling/libraries/int & 0xFF" This unifies sub-decision 6 with T-CostLens-Composition: faithful representation is universal (every target CAN represent every concept, literal-bits is the always-available floor); the cost lens is the discriminator that orders the alternatives. Reframing: - Universality of faithful representation: every target carries every concept faithfully (literal-bits as the floor) - Cost lens (T-CostLens-Composition gates cost_lens_reads_target_realization + coercion_cost_equals_complexity_by_construction) reads per-primitive realization cost from target language spec - Grounding selects the lowest-cost faithful representation by reading cost-lens output - For Python u8: cost-tier 1 = numpy.uint8 (low cost), cost-tier 2 = ctypes.c_uint8 (medium), cost-tier 3 = int & 0xFF discipline (universal floor, high cost). All three faithful; cost lens orders. Implication: existing docs/design-numeric-construction.md:194-198 Python row (Nat<N> → int with size-hint metadata) is NON-FAITHFUL — drops width structure to opaque metadata, cost lens cannot read what isn't there. Reframe routes to T-Numeric-Construction / Grounding lane paydown: replace with cost-tiered faithful-representation table. This is the structurally-elegant unification: faithful-representation universality + cost-lens-as-discriminator. Sub-decision 6 IS an instance of "coercion cost = complexity by construction" thesis (T-CostLens-Composition load-bearing claim). Updates both r3-design-schedule-2026-05-06.md §S3 and r3-program-plan.md §10.3 Q-MachineConstraint-Carrier row sub-decision 6. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): Q-PAFS Path A ACCEPTED — Brian Director countersignature 2026-05-06 Per Brian directive 2026-05-06: "approved path A countersign". Resolves three bundled §10.3 questions to ACCEPTED: - Q-PAFS: TC1 first slice = static representative via E6-G1.a (Path A); Path B (TC1 generic G1.b/X1.b) deferred; Path C (RustDagIsomorphism before TC1) deferred under Q-PAFS default. - Q-Pattern-A-First-Slice-Subscope: G1.a static representative ratified; TC1-generic via X1.b S1/S3 + G1.b deferred-not-blocked. - Q-EVAL-Lens-Fold-First-Slice: G1.a first-slice shape locked for Evaluator E3 sequencing. Implementation dispatch unblocked simultaneously across two lanes (same release step): - Verification V1 — Pattern-A executable cluster, TC1 first slice - Evaluator E3 — E6-G1.a static lens fold Both Mgrs (cool-owl-579 #1740, merry-gull-128 #1743) author worker briefs and dispatch in same release step. Substrate/evaluator carrier-shape changes remain STOP+PING until routed implementation PRs land — ACCEPTED closes the policy-layer scope fork (which slice first), not the substrate routing. Updates: - docs/r3-program-plan.md §10.3: Q-PAFS / Q-Pattern-A-First-Slice-Subscope / Q-EVAL-Lens-Fold-First-Slice rows from PENDING DIRECTOR COUNTERSIGNATURE to ACCEPTED 2026-05-06 (Path A / G1.a). - docs/r3-design-schedule-2026-05-06.md §V1 + §E3: dispatch trigger updated from "pending Director countersignature" to "DISPATCH UNBLOCKED 2026-05-06" with cross-references to the bundled gate. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 6, 2026
* WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * fix(briefs): retire INVARIANTS.md line-number cites in r3-426 brief Per codex review on PR #1782 (sha c2512cd): the new briefs introduced fresh `INVARIANTS.md:<line>` citations — the same stale-line-anchor shape the :425 sweep brief itself names as needing dissolution. Replaces four line-number references in r3-426-sfp-subdoc-rename-and-author-worker.md (lines 16, 28, 58, 85) with prose-anchored locators citing the section anchor + bullet text. The other PB-lane briefs codex flagged (r3-pb-runtime-equivalence- corpus-seed-audit.md, r3-pb-t-fixedpoint-worker.md) belong to neat-bear-351's lane and are not in this PR's scope; will route separately. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): correct L6 type count + r3-435 dual-path retirement receipt Per openai-pro review on PR #1782 (sha c2512cd): 1. L6 brief said acceptance ratchet asserts "five types" exist but declares six (ShapeATarget, FormAxis, BehaviorAxis, MethodTemplateContractKey, EmissionCell, EmissionPathProjection). Updated both occurrences to enumerate all six explicitly. 2. r3-435 brief authorized both Path A (typed-AST reader, deletes IntegrationRsScan) and Path B (scanner widening with Char state), but retirement receipt language only covered Path B. Updated ROADMAP/ledger retirement and authority-audit step 5 to cover both authorized dissolution paths. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): L6 second 'five types' site + T-E-P lookup-gate authority Per cursor review on PR #1782 (sha 2072b98): 1. L6 brief acceptance bullet at :131 still said "five types" while the parenthetical and the Slice step at :117 said six. The prior fix only caught one of two sites; corrected the second. 2. T-E-P brief `e_p_call_pattern_lookup_authoritative` acceptance named only `lower_call_pattern` as "the only route from call site to LoweringTarget" — but Phase 2 explicitly requires lenses route through `per_call_pattern_at` (the L-7 single- authority lookup), NOT `lower_call_pattern` (compiler-internal). Acceptance text rewritten to name `per_call_pattern_at` as the consumer-facing gate and clarify `lower_call_pattern`'s internal-only scope. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): strengthen L6 ratchet from count-parity to per-row bijection Per openai-pro REQUEST_CHANGES on PR #1782 (sha 2072b98): the L6 acceptance ratchet asserted only row-count parity between emission_path_projections and the union of MethodTemplateContract rows. Count parity permits duplicate projection keys and missing source rows — the bridge from list-non-empty proxy to per-row projection would not be mechanically proven. The whole point of introducing MethodTemplateContractKey { target, dag_method } is to provide 1:1 row identity. Rewrites the ratchet to assert per-row bijection: every MethodTemplateContract source row has exactly one matching EmissionPathProjection row by key; every projection key resolves to exactly one source row; duplicate projection keys fail closed. (The other openai-pro BLOCKING finding — T-E-P e_p_call_pattern_lookup_authoritative naming the wrong authority — was already fixed at 6282a07; the review citation was against the pre-fix sha 2072b98.) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): split L6 ratchet into slice-active vs deferred-activation gates Per BLOCKING inline review on PR #1782 at d924594: the prior bijection ratchet was unsatisfiable as authored. The *_method_template_contracts lists at HEAD (rust/python/go) are non-empty; shipping emission_path_projections: [] alongside a strict bijection requirement means the bijection cannot pass when the slice lands — source rows would be uncovered. Prior text incorrectly said "vacuously true on the empty source set" — the SOURCE set isn't empty, only the projection set is. Splits the ratchet into two gate categories: - Slice-active gates: types/data shape exist + empty-state predicate (emission_path_projections == []). - Deferred-activation gate: per-row key bijection, authored as a test scaffold by this slice but #[ignore]'d (or feature-gated on len > 0) until Grounding's follow-up populates rows. On activation, asserts the bijection over the populated set. This honors P1 doc faithfulness: the slice's acceptance contract is satisfiable as authored, and the bijection becomes the load-bearing gate the row-population PR satisfies. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): require L6 bootstrap enrollment to honor P2 facts-flow-forward Per BLOCKING inline review on PR #1782 at L6:34: the slice authored src/v3/std/cross_target_coverage.dag but never required enrolling it in the bootstrap/load authority. Without enrollment the file is dead-letter on disk and EmissionPathProjection can be absent from the downstream Dag — silent P2 facts-flow-forward violation. Adds a new Slice step 2 requiring a cross_target_coverage field on BootstrapFixtures in src/v3/std/extdeps_bootstrap_fixtures.dag mirroring the *_method_template_contracts enrollment pattern, plus a loader-visibility ratchet. Renumbers subsequent Slice steps. Adds matching Acceptance bullet. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: R3 Substrate * WIP: R3 Substrate * docs(briefs): normalize Practice 4 citation in S3 brief per codex review Codex review on PR #1782 (sha 2614217) flagged stale citation "Practice 4 (Step 4 of the type-introduction checklist)" in S3 brief — no such checklist step exists in the pinned rubric. Live rule per docs/modeling-discipline.md:131-134 is the "What to check" checkpoint-comment requirement under Practice 4 (coproduct dissolution). Updated the S3 reference to cite that line directly and drop the stale "Step 4 of type-introduction checklist" wording. S2/S8/S9 already use bare "Practice 4" without the stale steps wording. S1 (committed earlier today) does not contain the stale phrase. S6 (pre-existing brief flagged by reviewer as carrying the same pattern) is out of this branch's scope; flag for separate paydown. * docs(briefs): sweep stale Practice 4 citations corpus-wide Codex review on PR #1782 flagged the stale "Step 4 of the type-introduction checklist" wording — no such checklist exists in docs/modeling-discipline.md. The live rule is the "What to check" checkpoint-comment requirement under Practice 4 (coproduct dissolution) at docs/modeling-discipline.md:131. Sweep across docs/ found 2 additional briefs carrying the stale phrasing (S3 fixed previously at 9db003d): - docs/briefs/r3-l6-emission-path-projection-substrate-worker.md - docs/briefs/r3-x1b-s1-transform-dispatch-substrate-worker.md Both updated to cite the live rule directly. Verified zero remaining "type-introduction checklist" or "Step 4 of the" references in docs/ tree. * docs(briefs): add Practice 4 citation discipline note to authoring checklist Per parent inbox status flag at gunbc#828 #issuecomment-4385346221: add a Citation discipline section to brief-authoring-checklist.md naming the correct shape — Practice 4 (coproduct dissolution) + "What to check" rule at docs/modeling-discipline.md:131 — and calling out the stale "Step 4 of the type-introduction checklist" phrasing as a failure mode. References the corpus sweep at 2dc92f3. Prevents future briefs that copy old templates from re-introducing the stale citation pattern. * docs(r3): cascade work post-Director ratification — S1 worker, R4 routing, Unit/Quantity canvas Per Director ratification at gunbc#828 inbox response 2026-05-06 (zesty-bear-812): 1. S1 Q1+Q2 RATIFIED → author S1 gap-test representative worker brief (`r3-substrate-s1-gap-test-representative-worker.md`). Dispatchable on T-E-P Phase 1 + E6-G0d landing; closes ledger row #61. 2. S2 Q2 RATIFIED (option-(b) T-LBP narrowing) + S3 + S8 deferrals → R4 carve-out routing ledger (`docs/r4-carve-out-routing.md`). Captures C1 (parallelism lens), C2 (effect_enum lens), C3 (register narrowing), C4 (deferred MachineConstraint axes), C5 (rounding-mode product extension), C6 (Unit/Quantity). Cascade messages to Verification / Grounding / Debt-Paydown Mgrs. 3. S9 Phase-3 dimensional refinement boundary → Unit/Quantity carrier canvas (`r3-substrate-s9-unit-quantity-carrier-canvas.md`). Surfaces Q-Unit-1..5 for Director ratification: carrier name, 2-axis vs 3-axis Phase-1, Duration/Seconds collapse, Refined composition order, Practice 4 classification. Updates §1.8 ledger receipts for carved items; R3 closure receipts treat carved items as explicitly-deferred (not drift). * docs(briefs): citation-discipline fix-forward — anchors/quotes, not line numbers Director redirect at gunbc#828 #issuecomment-... noted the citation paydown's replacement form (`docs/modeling-discipline.md:131`) is the same drift class as the original fabricated-step error. Bare line numbers shift silently on any doc edit above the cited line. Generalized fix: - brief-authoring-checklist.md "Citation discipline" section rewritten to prohibit bare line-numbers across ALL cross-doc references (modeling-discipline.md, r3-program-plan.md, r3-structure.md, INVARIANTS.md, MODELING.md, etc.). Use section anchors (#section-name) or rule-text quotes. - 3 briefs that swapped in :131 (S3, S6, X1B) re-converted: - docs/modeling-discipline.md#4-coproduct-dissolution anchor - "What to check" rule quoted inline (self-contained) Brief-authoring-checklist failure-mode section names both vectors (fabricated step + bare line numbers) and references both sweep commits in audit trail. Broader cross-doc citation sweep across briefs I authored on this branch (frontmatter authority-docs lists, body prose) is recommended per Director note — flagging as separate paydown task; this commit closes the immediate recurrence vector. * docs(briefs): fix unfaithful Int=AbelianGroup<Nat> + sweep stale 'Step 4' refs Per openai-pro review on PR #1782 sha 0ecd481 (gunbc#1782 #...): BLOCKING fix (P1 modeling faithfulness): - S9 + S3 directed workers to model `Int = AbelianGroup<Nat>`, but Nat is not an Abelian-group carrier (additive inverses not representable). Reframed: - `Int` is the additive Abelian group on integers (carrier Z); `Nat` is a commutative monoid under addition. - Algebraic relationship is **explicit group-completion**: `Int ≡ GroupCompletion<CommutativeMonoid<Nat>>` — NOT `AbelianGroup<Nat>` parameterization. - Worker either models `Int` directly as `AbelianGroup` (terminal Abelian-group instance) OR introduces explicit `GroupCompletion<M>` substrate carrier; decision lands in Phase-1 PR per P1 procedure. - S3 cross-reference + Phase-4 algebra-axis preservation note reframed accordingly. NON-BLOCKING in-PR cleanup (Practice 4 citation discipline): - 6 briefs still carrying stale 'Practice 4 Step 4' wording (r3-coproduct-1/2/3, r3-l6, r3-t-e-p, r3-x1b) updated to use the section anchor (`docs/modeling-discipline.md#4-coproduct-dissolution`) + inline rule-text quote ("What to check") form per the citation discipline established in `b4cae5299`. Closes both findings from openai-pro review. * docs(briefs): Q-Unit-1..5 RATIFIED — apply Scale.Unit→One rename + ratification banner Director ratified Q-Unit-1..5 at gunbc#828 #issuecomment-... 2026-05-06 (zesty-bear-812). All 5 questions ratified; one rename required: - Scale enum's `Unit` value collides with outer `Unit<Q, S>` carrier name (`Unit<Time, Unit>` is shadowing-confusing). Rename to `One`. - Duration ≡ Unit<Time, One> (collapses with Seconds per Q-Unit-3). - Milliseconds ≡ Unit<Time, Milli>. Canvas updated: - Top-level RATIFICATION banner records all 5 ratifications + rename - Scale enum: `Unit` → `One` (with inline rename receipt) - Sketch product expansions use `Unit<Time, One>` / `Unit<Time, Milli>` - Phase-3 reframe table products updated - Q-Unit-3 prose collapse reference updated Subsequent worker brief authoring (Unit/Quantity carrier landing post-prerequisite) inherits the renamed shape. * docs(briefs): Unit/Quantity carrier worker brief — Q-Unit-1..5 RATIFIED Worker brief consumes the ratified Unit/Quantity canvas (`5f22fd06e` ratification banner). Phase-1 lands `Unit<Q, S>` + `Quantity` + `Scale` declarations at the ratified shape (Scale's `One` value per Q-Unit-1 rename). Phase-2 reframes S9 Phase-3 dimensional refinements (Duration / Seconds / Milliseconds) to outer-Refined / inner-Unit composition per Q-Unit-4 RATIFIED. EpochMs deferred to R4 C6 (Aspect-axis follow-up). Practice 4 marks pre-determined per Q-Unit-5: `Unit<Q, S>` 🟢 PRIMITIVE; `Quantity` + `Scale` 🟡 SCAFFOLD with "consumed by ≥1 emission rule" dissolution trigger. Cross-program handoff to Grounding Mgr (#1745) for dimensional emission consumption documented. Worker pin TBD — dispatchable when idle pool refreshes. * docs(briefs): S9 Phase-1 reframed — Int landed at #1466; UInt-only migration Per proud-lynx-311 dispatch ack at gunbc#1739 #issuecomment-... 2026-05-06: re-grep at HEAD revealed substrate state diverges materially from S9 brief's "Option A vs B decide" framing. Reality check at HEAD: - `dsl/std/integer.dag:55` already carries `type Int = AbelianGroup<GroupCompletion<Nat>>` per Slice 3 PR #1466 / commit `4ba0a6d04`. Single-authority Q6 audit doc at `docs/audit/t-numeric-construction-group-completion-6q.md` ratified this shape and explicitly rejected the compact form. - `dsl/std/integer.dag:56` still on legacy `type UInt = UInt64` alias; UInt migration is the actual Phase-1 worker scope. Brief Phase-1 step 1 reframed: Int landing acknowledged; worker confirms HEAD match. Phase-1 step 2 reframed: UInt migration target named (`type UInt = CommutativeMonoid<Nat>`); fixed-width rows preserved per existing legacy-rows policy. Historical Option A vs B framing kept as note (superseded). Confirmed S9 dispatch ack from worker; UInt-only Phase-1 PR is in-scope. Emission entries deferred per S3 (valiant-ant-72 pre-landing); Phase-2 deferred per S8 (quiet-boar-160 pre-landing). * docs(briefs): fix two BLOCKING citation drift items per codex review Codex review on PR #1782 sha 5f22fd0 flagged two authority-receipt gaps: 1. brief-authoring-checklist.md:118 — `<this-commit>` placeholder in the failure-mode section was unverifiable post-merge. Replaced with `b4cae5299` (the actual anchor/quote conversion commit). 2. r3-378-http-path-helper-fail-open-worker.md:84 — `INVARIANTS.md C-8 fail-closed` cited without anchor or rule-text quote, in violation of the citation discipline added in this PR. Converted to `INVARIANTS.md#p3-fail-closed` (P3 heading anchor) + inline rule-text quote naming P3 + C-8. PR's own discipline now consistently applied across the diff. * docs(briefs): sweep bare .md:NNN citations in Substrate-Mgr-authored briefs Per Director recommendation in citation-discipline thread (gunbc#828 follow-on): bare line-number references drift silently. Swept 19 remaining instances across 5 briefs I authored on this branch: - docs/briefs/r3-substrate-s1-q-class-2-chain-break-gap-test-canvas.md - docs/briefs/r3-substrate-s1-gap-test-representative-worker.md - docs/briefs/r3-substrate-s2-t-lbp-scope-calibration-canvas.md - docs/briefs/r3-substrate-s3-machine-constraint-carrier-worker.md Conversion: bare `file.md:NNN` → `file.md §"Section name"` form. Stable; survives doc edits; human-readable. Section names per HEAD heading audit. Verification: `grep -nE '.md:[0-9]+' docs/briefs/r3-substrate-*.md docs/r4-carve-out-routing.md` returns zero results in Substrate-Mgr-authored scope. Broader corpus scope (~232 references across non-Substrate briefs) remains as separate paydown task. * docs(briefs): fix BLOCKING + NON-BLOCKING items per openai-pro review on a462f00 Three findings addressed; 2 BLOCKING + 1 NON-BLOCKING: BLOCKING — R3/R4 dual authority for Unit/Quantity carrier: R4 carve-out C6 narrowed from "Unit/Quantity carrier landing" (deferred all of it) to **"Aspect-axis (PointKind) follow-on for EpochMs and instant/rate-shaped refinements only"**. The 2-axis Measure<Q,S> carrier landing IS in R3 per Q-Unit-1..5 RATIFIED; only the Aspect axis (PointKind = Magnitude | Instant | Rate) carves to R4. Cascade-message C6 reference + R4 program- plan input list updated to match. Cross-Mgr message clarifies Grounding consumes 2-axis dimensional refinements in R3. BLOCKING — Locked design partially applied (Scale.Unit→One rename in canvas alternate-sketch sections): - Canvas line 105-108 alternate sketch ("Open question: Instant vs Duration axis") still used pre-rename `Unit` Scale value; added pre-ratification disclaimer marking the section as deferred to R4 C6 + noting `Unit` is pre-rename - Canvas line 163 Q-Unit-3 collapse note: `Scale = Unit` → `Scale = One` (BLOCKING — bare line-numbers in S1 canvas frontmatter): NOT APPLICABLE — already fixed at sweep commit 127287a prior to review; reviewer was on sha a462f00 pre-sweep. NON-BLOCKING — placeholder receipts: - brief-authoring-checklist.md `#issuecomment-...` → concrete ID 4385393971 (Director's anchor-discipline reply) - r3-substrate-unit-quantity-carrier-worker.md (3 occurrences) + canvas RATIFICATION banner: `#issuecomment-...` → concrete ID 4385412256 (Q-Unit ratification reply) - `<this-commit>` placeholder in checklist already fixed at 5be1c99 Verification: zero `#issuecomment-...` or `<this-commit>` placeholders in docs/briefs/ + docs/r4-carve-out-routing.md. * docs(briefs): codex BLOCKING review fixes — Practice 4 source-level + policy scope Three BLOCKING findings on PR #1782 (codex sha b913e2f): 1. **Citation-discipline policy scope**: addressed by scoping the rule prospectively to briefs landing after the policy commit (`b4cae5299`). Pre-existing corpus references are flagged as paydown but NOT individually blocking; coordinated single-PR sweep is the right paydown shape. Substrate-Mgr scope already swept at `127287a66`. 2. **R4 C6 reconciliation**: already fixed at `6e5a5481a` (R4 C6 narrowed to Aspect-axis follow-on for EpochMs only; 2-axis carrier landing IS in R3). Codex sha b913e2f was pre-fix. 3. **Practice 4 source-level vs PR-body conflation**: corrected. Acceptance bullet for Unit/Quantity worker brief now requires inline checkpoint comments on the LIVE declarations in `dsl/std/units.dag` (load-bearing artifact per `docs/modeling-discipline.md#4-coproduct-dissolution` "What to check"). PR-body summary is supplementary, not substitute. Closes the review's reconciliation gates. * docs(briefs): absorb Q-MachineConstraint + Q-Unit-1-Recanvas ratifications Two major Director ratifications absorbed: 1. Q-MachineConstraint-Carrier RATIFIED at gunbc#828 #issuecomment-4385530115 (Brian directive 2026-05-06: "universal substrate, ratify defaults"; PR #1817 queued). 6 sub-decisions: - MachineWidth<bits> only for R3; defer RegisterClass / EndianMode / Alignment / signedness post-R3 - Parametric Compose<Algebra, MachineConstraint> type-level interaction shape (lookup-maps REJECTED — AlgebraMachineProduct record-keyed table form retracted) - Int<64> ≡ Compose<AbelianGroup, MachineWidth<64>> (algebra-side options A and B preserve interaction semantics) - Real<64> ≡ Compose<ApproximateField<Rational>, MachineWidth<64>> (algebra approx + machine approx as independent composing axes) - ≥3 pairs is minimum, not target - UNIVERSAL substrate: every target carries machine-constraint facts; targets without native machine-width semantics handle omission at Grounding-level discharge (target-conditioned lowering NOT target-conditioned substrate) Reframed: S3 brief Phase-1, S3 Phase-3/4 emission entries, S8 Phase-1/2/3 (ApproximateField<Real> → ApproximateField<Rational>; AlgebraMachineProduct → Compose<...>), S9 emission entries. 2. Q-Unit-1-Recanvas RATIFIED Measure<Q, S> at gunbc#828 #issuecomment-4385539791. valiant-ibex-312's substrate-state grep caught Unit<Q, S> collision with kernel Unit at dsl/std/types.dag:176; Director ratified Measure<Q, S> as outer carrier name (conventional in scientific computing; no dsl/std/ collision). Reframed: Unit/Quantity worker brief throughout (Unit<Q,S> → Measure<Q,S>; dsl/std/units.dag → dsl/std/measure.dag); canvas spec sections + sketch products (Refined<Measure<Time, ...>...>); R4 carve-out C6 with informal/formal name disambiguation note. Both rename + reshape are now consistent across S3 / S8 / S9 / Unit-Q briefs + canvas + R4 routing. Workers (valiant-ant-72 on S3, quiet-boar-160 on S8, proud-lynx-311 on S9, valiant-ibex-312 on Unit-Q) all dispatch against the ratified shape. * docs(briefs): S11 Slice C worker brief — ROADMAP :425 retirement post-#1801 Cascade-clearance trigger fired post-#1801 (Slice B) merge 2026-05-06. S11 brief authored per B1 RATIFIED prose+regen bundling discipline. Phase 1: per-file edits across the post-A+B residual list (~10 files per smart-ram-167's Slice B post-rebase enumeration); anchor + rule-text-quote pattern per current citation discipline; bootstrap regen in same PR (no split). Phase 2: ROADMAP :425 row PARTIAL → Retired with 3-slice receipt; Q-Slice-C-Retirement-Receipt resolution. Worker pin smart-ram-167 (#1759) — Slice B precedent owner. Dispatch packet sent separately to #1759. * docs(briefs): Q-Refined-Phantom-Composition (c) RATIFIED — defer literal Refined form Director ratified option (c) at gunbc#828 #issuecomment-... 2026-05-06: drop `non_negative` predicate at Measure (phantom) layer; rely on `Quantity = Time` tagging for non-negative-magnitude semantics. Q-Unit-4 outer-Refined / inner-Measure composition is preserved as **conceptual model**; the literal `Refined<Measure<...>, predicate>` form is **deferred to value-typed-integration follow-up** when substrate gains Refined<> as first-class type expression and dimensional refinements get value-typed integration. Updates: - Canvas RATIFICATION banner Q-Unit-4 row reframed to "conceptual model; literal form deferred" - Canvas Phase-3 reframe table: Duration / Milliseconds drop the Refined<...> wrapper; rely on Quantity tag - Worker brief Phase-2 prose + reframe table + Acceptance bullet all reflect deferred-literal-form posture valiant-ibex-312 unblocked to proceed with Phase-2 (Grounding handoff doc) per the (c)-shaped Acceptance. * docs(briefs): S12 F2 + F8 doc-sharpening worker brief — post-#1820 cascade #1820 (Slice C) merged 2026-05-06; ROADMAP :425 row Retired closes the prose-reference paydown program. S12 cascade-clearance trigger fires per B6 RATIFIED single bundled Mgr-tier doc-sharpening PR. Scope: F2 (.v3 filename-suffix grammar dispatch) + F8 (bootstrap load-order/exclusion authority) — single bundled PR per B6. Cross-program coordination with PB Mgr (#1742) per design schedule §3 P5 (comment-only OR co-author; no duplicate PR). Worker pin smart-ram-167 (#1759) — Slice C precedent owner; pattern-familiar with prose+regen bundling discipline. Dispatch packet sent separately to #1759. * docs(briefs): S9 brief — UInt landed via #1818; algebra label corrected to CommutativeSemiring Per proud-lynx-311 merge report at gunbc#1746 #issuecomment-... 2026-05-06: PR #1818 landed `type UInt = Nat` after codex BLOCKING review on initial `CommutativeMonoid<Nat>` form (P2 facts-flow-forward violation — wrap projects out Nat's multiplicative monoid + identity 1). Final landed form preserves full `CommutativeSemiring<Magnitude>` surface — Nat IS a commutative semiring (addition + multiplication + 0 + 1). Algebra-side label for UInt is CommutativeSemiring, NOT CommutativeMonoid (corrects my earlier dispatch labeling). S9 Phase-1 step 2 marked LANDED with merge ref. Phase-1 step 3 emission entries reframed: - Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>> - UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>> Worker discipline + reviewer caught a substrate-state error in my brief framing; brief now matches landed reality. * docs(briefs): fix contradictory (c) guidance in Unit/Quantity worker brief Codex BLOCKING review on PR #1825 sha 1bce041 caught a real contradiction: brief Phase-2 prose at lines 115-118 stated "predicates do NOT attach at the Measure (phantom) layer" per Q-Refined-Phantom-Composition (c) RATIFIED, but immediately following paragraph at lines 130-132 still said "Refined predicate non_negative applies at the Measure-typed level" — left over from the original Q-Unit-4 framing. Updated lines 130-137 to consistently apply (c) shape: predicates deferred to value-typed-integration follow-up; dimensional semantics ride on Quantity tag at Measure layer; worker authors 3 in-scope refinements WITHOUT predicate at this layer. P2 boundary discipline restored: brief now gives single-authority guidance for Phase-2 substrate landing. * docs(briefs): name predicate authority + concretize ratification ID — codex BLOCKING Codex BLOCKING review on PR #1825 sha 32541a5: 1. (BLOCKING) Q-Refined-Phantom-Composition (c) ratification deferred phantom-carrier predicates without naming the value-level predicate authority. Updated canvas Q-Unit-4 RATIFICATION row to explicitly name: 'predicates re-attach at value-typed Int/UInt level via where-sugar (e.g., range(min: 0) per current dsl/std/types.dag precedent)' when value-typed integration lands. 2. (BLOCKING) Worker brief option-c reframe didn't sweep the audit receipt §5 (Carrier dissolves the bridge). Updated to consistently apply (c) shape — dimensional semantics ride on Quantity tag at Measure layer; value-level predicate authority deferred; until then, range/gt_zero constraints attach at value-typed Int/UInt via where-sugar per dsl/std/types.dag convention. 3. (NON-BLOCKING) Canvas had #issuecomment-... placeholder for the (c) ratification message ID. Replaced with concrete #issuecomment-4385687075 per dashboard relay header. Brief now names a single value-level authority (Int/UInt where-sugar) for value-range facts; phantom-layer Quantity tag carries only dimensional semantics. P2 boundary discipline maintained. * docs(briefs): S12 RETIRED per Director option (2) ratification Director ratified option (2) at gunbc#828 #issuecomment-4385873... 2026-05-06 post smart-ram-167's substrate-state STOP-AND-ESCALATE: - F2 absorbed by 2026-05-04 routing + #1662 closure (`Result/DivError span.file-keyed` retired) - F8 topical concerns distributed across live rows (Class 5 Gap 3 `bootstrap.rs::std_fixtures()`; Character-level load-set / bootstrap; Filename-sentinel-bridges) with own owners - Doc-discipline goals absorbed by per-Mgr citation paydown routing at gunbc#828 #issuecomment-4385583247 Frontmatter updated to RETIRED with absorption receipt. Worker (smart-ram-167) pin freed. Director-side acknowledgment of Substrate Mgr's option (2) recommendation: "Substrate Mgr's option (2) recommendation is the right move structurally. Most Mgrs default to 'synthesize work to fill the brief' when canvas-vs-HEAD mismatch surfaces. Recommending retirement when the work is genuinely absorbed is harder discipline." * docs(briefs): S7 PR-F (BoundDeclaration consumer + ReferenceModel<T>) worker brief Per cascade-clearance trigger fired post-#1782 merge 2026-05-06. loyal-wolf-828 (#1764) acknowledged S7 dispatch with substrate-state pre-flight grep (gunbc#1764 #issuecomment-4385872...) and surfaced await-brief posture. This brief satisfies that prerequisite. Worker pre-flight inventory absorbed: - BoundDeclaration substrate landed (substrate.dag StaticBound + PlatformDependent); partial BoundDeclarationView consumer in fold.rs (ScratchIntExamples scope; doc-explicit partial) - ReferenceModel<T> NOT yet in dsl/std/ (substrate-fact-introduction required; design-emission-model.md §Q2 framing) Phase-1: BoundDeclaration consumer broadening to full design-emission-model.md surface (u128 / isize / usize / walker arms / pilot mirror). Phase-2: ReferenceModel<T> P1 substrate-fact-introduction with DFS-of-concept-DAG + Practice 4 in-source checkpoint comment + named consumer demand (Grounding G1). Phase-3: cross-program handoff to Grounding Mgr (#1745) for G1 (T-Ground-Rust Phase 1). Closure predicate: unblocks Grounding T-Ground-Rust Phase 1. Worker pin loyal-wolf-828 holds; dispatch fires immediately on this brief landing. --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This was referenced May 6, 2026
briansrls
added a commit
that referenced
this pull request
May 6, 2026
…nt = Nat (#1840) * WIP: proud-lynx-311 * WIP: proud-lynx-311 * WIP: proud-lynx-311 * WIP: proud-lynx-311 * chore(v3): regen bootstrap snapshot for substrate.dag :83 anchor swap The :425 Slice A anchor swap (\§P1` → `#p1-modeling-faithfulness`) on src/v3/std/substrate.dag:83 shifted SourceSpan byte offsets for every node parsed below that line, drifting bootstrap_generated.rs. Regenerated via `cargo run -p v3-compiler --features bootstrap-regen-fresh --bin regen_bootstrap`. No semantic change — pure span-offset refresh. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(:425): retire trailing `§P5` shorthand on r2-manager-brief-authority-matrix.md:205 Cursor review on a402dca flagged residual `\`§P5\` (dispatch-discipline)` shorthand inside the same sentence as a P2 anchor. Slice A's mechanical sweep matched only `INVARIANTS.md §P5` prefixed forms; the bare shorthand survived. Swap to `INVARIANTS.md#p5-progress-is-dissolution` matches adjacent P2 form. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test(v3): refresh parse_corpus_manifest after substrate.dag :83 anchor swap The :425 Slice A anchor swap shifted byte offsets in src/v3/std/substrate.dag, drifting SG-2 parse snapshot hashes (same root cause as the bootstrap_generated regen in a402dca). Regenerated via: cargo test -p v3-compiler --test integration -- --ignored \ refresh_handwritten_parse_snapshot_manifest Verified: handwritten_parse_snapshot_matches_manifest passes locally. Pure offset/hash refresh — no semantic change. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * ci: re-trigger v3 after t_demo_fixture transient flake * fix(:425): align link URL with anchor label on r2-manager-brief-authority-matrix.md:222 Codex review on 9e67e1f noted the Markdown label was swapped to the #p5-progress-is-dissolution anchor but the URL still pointed at the INVARIANTS.md root. Aligned URL to match label. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * feat(std): T-Numeric-Construction S9 Phase-1 — UInt = CommutativeMonoid<Nat> Migrates the default `UInt` alias from the legacy fixed-width `UInt = UInt64` to the abstract algebra carrier `UInt = CommutativeMonoid<Nat>`, consistent with the landed Slice 3 form `Int = AbelianGroup<GroupCompletion<Nat>>` (PR #1466, docs/audit/t-numeric-construction-group-completion-6q.md Q6 single-authority resolution). Per S9 worker brief (R3 design schedule §1 S9, ratified 2026-05-06): Nat plays as the commutative-monoid carrier under addition (closure / associativity / identity 0 / commutativity; no inverses, since Nat lacks additive inverses by construction). UInt as `CommutativeMonoid<Nat>` is the additive structure on that carrier — the honest unsigned-integer algebra without inverses. Fixed-width rows `UInt8..UInt128` remain on `Semiring<Word*>` per existing legacy-rows policy in `dsl/std/integer.dag` (only the default alias migrates; storage rows stay shape-conflated until v3 supports trait/witness syntax separation). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(std): UInt = Nat (preserve CommutativeSemiring surface) Codex review at PR #1818 / Substrate Mgr inline review at dsl/std/integer.dag:102 surface a P2 facts-flow-forward violation in the prior `type UInt = CommutativeMonoid<Nat>` pivot: wrapping Nat (already `CommutativeSemiring<Magnitude>`) in `CommutativeMonoid<Nat>` projects out the multiplicative monoid + identity 1, dropping substrate facts already established at the Nat layer. The substrate cannot structurally name "the additive monoid projection of Nat" without losing Nat's other monoid. Pivot to the design-doc-named direct alias: `type UInt = Nat`. Per the existing comment header in this file, the design doc states `UInt IS Nat`; Nat already carries the full unsigned-integer algebra surface (additive monoid + multiplicative monoid + distribution + commutativity on both). No new operations are needed — the alias IS the algebra. Asymmetry with `Int`: `Int` requires additive inverses Nat lacks, so `GroupCompletion<Nat>` is the honest carrier introduction. `UInt` requires no new operations, so no carrier introduction is needed. Updates the PR's deferred Phase-1 Step 3 receipt to the `Compose<CommutativeSemiring, MachineWidth<N>>` parametric form (post-S3 landing). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test(v3): refresh parse-snapshot manifest for integer.dag UInt change Mechanical refresh after PR #1818 fix commit. integer.dag debug-bytes shifted 12659→12662, fnv1a64 1eef7086b17908de→104a413082264303 with the type swap from `UInt = CommutativeMonoid<Nat>` to `UInt = Nat` plus surrounding doc-comment rewrite. Generated by: cargo test -p v3-compiler refresh_handwritten_parse_snapshot_manifest -- --ignored Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * test(v3): refresh parse-snapshot manifest for Slice 2 types.dag edit Mechanical refresh after the Slice 2 Nat-alignment migration (`NonNegativeInt` / `PositiveInt` carrier swap to `Nat`). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * fix(std): partial Slice 2 — NonNegativeInt only; PositiveInt stays Int CI failure on PR #1840 surfaced two real substrate gaps in the attempted full Slice 2 migration: 1. Adding `import std.nat { Nat }` to `dsl/std/types.dag` triggers a circular-dependency false-positive in the resolver during transitive load of test fixtures via `std.render` chains (verified via p0_std_render_repeat_string_test). 2. Shipping `type PositiveInt = Nat where range(min: 1)` triggers `Diagnostic::ResolveError` for the `range` predicate — the resolver does not currently entail `where range(min: …)` over `Nat` (only over `Int` and Int-row carriers). This matches dispatch STOP-AND-ESCALATE bullet 2 (predicate-entailment failure). Partial migration shipped: - `NonNegativeInt = Nat` declaration moved to `dsl/std/integer.dag` (which already imports `std.nat { Nat }`); structural alias, no predicate, resolves cleanly. - `PositiveInt = Int where range(min: 1)` stays in `dsl/std/types.dag` unchanged. PositiveInt's Nat-alignment migration sequences after the predicate-entailment substrate supports `range` over Nat. `src/v3/std/approximate_field.dag` keeps `import std.types { PositiveInt }` (unchanged consumer surface for PositiveInt). Bootstrap snapshots re-regenerated; parse-corpus manifest refreshed. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 6, 2026
* WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * fix(briefs): retire INVARIANTS.md line-number cites in r3-426 brief Per codex review on PR #1782 (sha c2512cd): the new briefs introduced fresh `INVARIANTS.md:<line>` citations — the same stale-line-anchor shape the :425 sweep brief itself names as needing dissolution. Replaces four line-number references in r3-426-sfp-subdoc-rename-and-author-worker.md (lines 16, 28, 58, 85) with prose-anchored locators citing the section anchor + bullet text. The other PB-lane briefs codex flagged (r3-pb-runtime-equivalence- corpus-seed-audit.md, r3-pb-t-fixedpoint-worker.md) belong to neat-bear-351's lane and are not in this PR's scope; will route separately. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): correct L6 type count + r3-435 dual-path retirement receipt Per openai-pro review on PR #1782 (sha c2512cd): 1. L6 brief said acceptance ratchet asserts "five types" exist but declares six (ShapeATarget, FormAxis, BehaviorAxis, MethodTemplateContractKey, EmissionCell, EmissionPathProjection). Updated both occurrences to enumerate all six explicitly. 2. r3-435 brief authorized both Path A (typed-AST reader, deletes IntegrationRsScan) and Path B (scanner widening with Char state), but retirement receipt language only covered Path B. Updated ROADMAP/ledger retirement and authority-audit step 5 to cover both authorized dissolution paths. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): L6 second 'five types' site + T-E-P lookup-gate authority Per cursor review on PR #1782 (sha 2072b98): 1. L6 brief acceptance bullet at :131 still said "five types" while the parenthetical and the Slice step at :117 said six. The prior fix only caught one of two sites; corrected the second. 2. T-E-P brief `e_p_call_pattern_lookup_authoritative` acceptance named only `lower_call_pattern` as "the only route from call site to LoweringTarget" — but Phase 2 explicitly requires lenses route through `per_call_pattern_at` (the L-7 single- authority lookup), NOT `lower_call_pattern` (compiler-internal). Acceptance text rewritten to name `per_call_pattern_at` as the consumer-facing gate and clarify `lower_call_pattern`'s internal-only scope. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): strengthen L6 ratchet from count-parity to per-row bijection Per openai-pro REQUEST_CHANGES on PR #1782 (sha 2072b98): the L6 acceptance ratchet asserted only row-count parity between emission_path_projections and the union of MethodTemplateContract rows. Count parity permits duplicate projection keys and missing source rows — the bridge from list-non-empty proxy to per-row projection would not be mechanically proven. The whole point of introducing MethodTemplateContractKey { target, dag_method } is to provide 1:1 row identity. Rewrites the ratchet to assert per-row bijection: every MethodTemplateContract source row has exactly one matching EmissionPathProjection row by key; every projection key resolves to exactly one source row; duplicate projection keys fail closed. (The other openai-pro BLOCKING finding — T-E-P e_p_call_pattern_lookup_authoritative naming the wrong authority — was already fixed at 6282a07; the review citation was against the pre-fix sha 2072b98.) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): split L6 ratchet into slice-active vs deferred-activation gates Per BLOCKING inline review on PR #1782 at d924594: the prior bijection ratchet was unsatisfiable as authored. The *_method_template_contracts lists at HEAD (rust/python/go) are non-empty; shipping emission_path_projections: [] alongside a strict bijection requirement means the bijection cannot pass when the slice lands — source rows would be uncovered. Prior text incorrectly said "vacuously true on the empty source set" — the SOURCE set isn't empty, only the projection set is. Splits the ratchet into two gate categories: - Slice-active gates: types/data shape exist + empty-state predicate (emission_path_projections == []). - Deferred-activation gate: per-row key bijection, authored as a test scaffold by this slice but #[ignore]'d (or feature-gated on len > 0) until Grounding's follow-up populates rows. On activation, asserts the bijection over the populated set. This honors P1 doc faithfulness: the slice's acceptance contract is satisfiable as authored, and the bijection becomes the load-bearing gate the row-population PR satisfies. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): require L6 bootstrap enrollment to honor P2 facts-flow-forward Per BLOCKING inline review on PR #1782 at L6:34: the slice authored src/v3/std/cross_target_coverage.dag but never required enrolling it in the bootstrap/load authority. Without enrollment the file is dead-letter on disk and EmissionPathProjection can be absent from the downstream Dag — silent P2 facts-flow-forward violation. Adds a new Slice step 2 requiring a cross_target_coverage field on BootstrapFixtures in src/v3/std/extdeps_bootstrap_fixtures.dag mirroring the *_method_template_contracts enrollment pattern, plus a loader-visibility ratchet. Renumbers subsequent Slice steps. Adds matching Acceptance bullet. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: R3 Substrate * WIP: R3 Substrate * docs(briefs): normalize Practice 4 citation in S3 brief per codex review Codex review on PR #1782 (sha 2614217) flagged stale citation "Practice 4 (Step 4 of the type-introduction checklist)" in S3 brief — no such checklist step exists in the pinned rubric. Live rule per docs/modeling-discipline.md:131-134 is the "What to check" checkpoint-comment requirement under Practice 4 (coproduct dissolution). Updated the S3 reference to cite that line directly and drop the stale "Step 4 of type-introduction checklist" wording. S2/S8/S9 already use bare "Practice 4" without the stale steps wording. S1 (committed earlier today) does not contain the stale phrase. S6 (pre-existing brief flagged by reviewer as carrying the same pattern) is out of this branch's scope; flag for separate paydown. * docs(briefs): sweep stale Practice 4 citations corpus-wide Codex review on PR #1782 flagged the stale "Step 4 of the type-introduction checklist" wording — no such checklist exists in docs/modeling-discipline.md. The live rule is the "What to check" checkpoint-comment requirement under Practice 4 (coproduct dissolution) at docs/modeling-discipline.md:131. Sweep across docs/ found 2 additional briefs carrying the stale phrasing (S3 fixed previously at 9db003d): - docs/briefs/r3-l6-emission-path-projection-substrate-worker.md - docs/briefs/r3-x1b-s1-transform-dispatch-substrate-worker.md Both updated to cite the live rule directly. Verified zero remaining "type-introduction checklist" or "Step 4 of the" references in docs/ tree. * docs(briefs): add Practice 4 citation discipline note to authoring checklist Per parent inbox status flag at gunbc#828 #issuecomment-4385346221: add a Citation discipline section to brief-authoring-checklist.md naming the correct shape — Practice 4 (coproduct dissolution) + "What to check" rule at docs/modeling-discipline.md:131 — and calling out the stale "Step 4 of the type-introduction checklist" phrasing as a failure mode. References the corpus sweep at 2dc92f3. Prevents future briefs that copy old templates from re-introducing the stale citation pattern. * docs(r3): cascade work post-Director ratification — S1 worker, R4 routing, Unit/Quantity canvas Per Director ratification at gunbc#828 inbox response 2026-05-06 (zesty-bear-812): 1. S1 Q1+Q2 RATIFIED → author S1 gap-test representative worker brief (`r3-substrate-s1-gap-test-representative-worker.md`). Dispatchable on T-E-P Phase 1 + E6-G0d landing; closes ledger row #61. 2. S2 Q2 RATIFIED (option-(b) T-LBP narrowing) + S3 + S8 deferrals → R4 carve-out routing ledger (`docs/r4-carve-out-routing.md`). Captures C1 (parallelism lens), C2 (effect_enum lens), C3 (register narrowing), C4 (deferred MachineConstraint axes), C5 (rounding-mode product extension), C6 (Unit/Quantity). Cascade messages to Verification / Grounding / Debt-Paydown Mgrs. 3. S9 Phase-3 dimensional refinement boundary → Unit/Quantity carrier canvas (`r3-substrate-s9-unit-quantity-carrier-canvas.md`). Surfaces Q-Unit-1..5 for Director ratification: carrier name, 2-axis vs 3-axis Phase-1, Duration/Seconds collapse, Refined composition order, Practice 4 classification. Updates §1.8 ledger receipts for carved items; R3 closure receipts treat carved items as explicitly-deferred (not drift). * docs(briefs): citation-discipline fix-forward — anchors/quotes, not line numbers Director redirect at gunbc#828 #issuecomment-... noted the citation paydown's replacement form (`docs/modeling-discipline.md:131`) is the same drift class as the original fabricated-step error. Bare line numbers shift silently on any doc edit above the cited line. Generalized fix: - brief-authoring-checklist.md "Citation discipline" section rewritten to prohibit bare line-numbers across ALL cross-doc references (modeling-discipline.md, r3-program-plan.md, r3-structure.md, INVARIANTS.md, MODELING.md, etc.). Use section anchors (#section-name) or rule-text quotes. - 3 briefs that swapped in :131 (S3, S6, X1B) re-converted: - docs/modeling-discipline.md#4-coproduct-dissolution anchor - "What to check" rule quoted inline (self-contained) Brief-authoring-checklist failure-mode section names both vectors (fabricated step + bare line numbers) and references both sweep commits in audit trail. Broader cross-doc citation sweep across briefs I authored on this branch (frontmatter authority-docs lists, body prose) is recommended per Director note — flagging as separate paydown task; this commit closes the immediate recurrence vector. * docs(briefs): fix unfaithful Int=AbelianGroup<Nat> + sweep stale 'Step 4' refs Per openai-pro review on PR #1782 sha 0ecd481 (gunbc#1782 #...): BLOCKING fix (P1 modeling faithfulness): - S9 + S3 directed workers to model `Int = AbelianGroup<Nat>`, but Nat is not an Abelian-group carrier (additive inverses not representable). Reframed: - `Int` is the additive Abelian group on integers (carrier Z); `Nat` is a commutative monoid under addition. - Algebraic relationship is **explicit group-completion**: `Int ≡ GroupCompletion<CommutativeMonoid<Nat>>` — NOT `AbelianGroup<Nat>` parameterization. - Worker either models `Int` directly as `AbelianGroup` (terminal Abelian-group instance) OR introduces explicit `GroupCompletion<M>` substrate carrier; decision lands in Phase-1 PR per P1 procedure. - S3 cross-reference + Phase-4 algebra-axis preservation note reframed accordingly. NON-BLOCKING in-PR cleanup (Practice 4 citation discipline): - 6 briefs still carrying stale 'Practice 4 Step 4' wording (r3-coproduct-1/2/3, r3-l6, r3-t-e-p, r3-x1b) updated to use the section anchor (`docs/modeling-discipline.md#4-coproduct-dissolution`) + inline rule-text quote ("What to check") form per the citation discipline established in `b4cae5299`. Closes both findings from openai-pro review. * docs(briefs): Q-Unit-1..5 RATIFIED — apply Scale.Unit→One rename + ratification banner Director ratified Q-Unit-1..5 at gunbc#828 #issuecomment-... 2026-05-06 (zesty-bear-812). All 5 questions ratified; one rename required: - Scale enum's `Unit` value collides with outer `Unit<Q, S>` carrier name (`Unit<Time, Unit>` is shadowing-confusing). Rename to `One`. - Duration ≡ Unit<Time, One> (collapses with Seconds per Q-Unit-3). - Milliseconds ≡ Unit<Time, Milli>. Canvas updated: - Top-level RATIFICATION banner records all 5 ratifications + rename - Scale enum: `Unit` → `One` (with inline rename receipt) - Sketch product expansions use `Unit<Time, One>` / `Unit<Time, Milli>` - Phase-3 reframe table products updated - Q-Unit-3 prose collapse reference updated Subsequent worker brief authoring (Unit/Quantity carrier landing post-prerequisite) inherits the renamed shape. * docs(briefs): Unit/Quantity carrier worker brief — Q-Unit-1..5 RATIFIED Worker brief consumes the ratified Unit/Quantity canvas (`5f22fd06e` ratification banner). Phase-1 lands `Unit<Q, S>` + `Quantity` + `Scale` declarations at the ratified shape (Scale's `One` value per Q-Unit-1 rename). Phase-2 reframes S9 Phase-3 dimensional refinements (Duration / Seconds / Milliseconds) to outer-Refined / inner-Unit composition per Q-Unit-4 RATIFIED. EpochMs deferred to R4 C6 (Aspect-axis follow-up). Practice 4 marks pre-determined per Q-Unit-5: `Unit<Q, S>` 🟢 PRIMITIVE; `Quantity` + `Scale` 🟡 SCAFFOLD with "consumed by ≥1 emission rule" dissolution trigger. Cross-program handoff to Grounding Mgr (#1745) for dimensional emission consumption documented. Worker pin TBD — dispatchable when idle pool refreshes. * docs(briefs): S9 Phase-1 reframed — Int landed at #1466; UInt-only migration Per proud-lynx-311 dispatch ack at gunbc#1739 #issuecomment-... 2026-05-06: re-grep at HEAD revealed substrate state diverges materially from S9 brief's "Option A vs B decide" framing. Reality check at HEAD: - `dsl/std/integer.dag:55` already carries `type Int = AbelianGroup<GroupCompletion<Nat>>` per Slice 3 PR #1466 / commit `4ba0a6d04`. Single-authority Q6 audit doc at `docs/audit/t-numeric-construction-group-completion-6q.md` ratified this shape and explicitly rejected the compact form. - `dsl/std/integer.dag:56` still on legacy `type UInt = UInt64` alias; UInt migration is the actual Phase-1 worker scope. Brief Phase-1 step 1 reframed: Int landing acknowledged; worker confirms HEAD match. Phase-1 step 2 reframed: UInt migration target named (`type UInt = CommutativeMonoid<Nat>`); fixed-width rows preserved per existing legacy-rows policy. Historical Option A vs B framing kept as note (superseded). Confirmed S9 dispatch ack from worker; UInt-only Phase-1 PR is in-scope. Emission entries deferred per S3 (valiant-ant-72 pre-landing); Phase-2 deferred per S8 (quiet-boar-160 pre-landing). * docs(briefs): fix two BLOCKING citation drift items per codex review Codex review on PR #1782 sha 5f22fd0 flagged two authority-receipt gaps: 1. brief-authoring-checklist.md:118 — `<this-commit>` placeholder in the failure-mode section was unverifiable post-merge. Replaced with `b4cae5299` (the actual anchor/quote conversion commit). 2. r3-378-http-path-helper-fail-open-worker.md:84 — `INVARIANTS.md C-8 fail-closed` cited without anchor or rule-text quote, in violation of the citation discipline added in this PR. Converted to `INVARIANTS.md#p3-fail-closed` (P3 heading anchor) + inline rule-text quote naming P3 + C-8. PR's own discipline now consistently applied across the diff. * docs(briefs): sweep bare .md:NNN citations in Substrate-Mgr-authored briefs Per Director recommendation in citation-discipline thread (gunbc#828 follow-on): bare line-number references drift silently. Swept 19 remaining instances across 5 briefs I authored on this branch: - docs/briefs/r3-substrate-s1-q-class-2-chain-break-gap-test-canvas.md - docs/briefs/r3-substrate-s1-gap-test-representative-worker.md - docs/briefs/r3-substrate-s2-t-lbp-scope-calibration-canvas.md - docs/briefs/r3-substrate-s3-machine-constraint-carrier-worker.md Conversion: bare `file.md:NNN` → `file.md §"Section name"` form. Stable; survives doc edits; human-readable. Section names per HEAD heading audit. Verification: `grep -nE '.md:[0-9]+' docs/briefs/r3-substrate-*.md docs/r4-carve-out-routing.md` returns zero results in Substrate-Mgr-authored scope. Broader corpus scope (~232 references across non-Substrate briefs) remains as separate paydown task. * docs(briefs): fix BLOCKING + NON-BLOCKING items per openai-pro review on a462f00 Three findings addressed; 2 BLOCKING + 1 NON-BLOCKING: BLOCKING — R3/R4 dual authority for Unit/Quantity carrier: R4 carve-out C6 narrowed from "Unit/Quantity carrier landing" (deferred all of it) to **"Aspect-axis (PointKind) follow-on for EpochMs and instant/rate-shaped refinements only"**. The 2-axis Measure<Q,S> carrier landing IS in R3 per Q-Unit-1..5 RATIFIED; only the Aspect axis (PointKind = Magnitude | Instant | Rate) carves to R4. Cascade-message C6 reference + R4 program- plan input list updated to match. Cross-Mgr message clarifies Grounding consumes 2-axis dimensional refinements in R3. BLOCKING — Locked design partially applied (Scale.Unit→One rename in canvas alternate-sketch sections): - Canvas line 105-108 alternate sketch ("Open question: Instant vs Duration axis") still used pre-rename `Unit` Scale value; added pre-ratification disclaimer marking the section as deferred to R4 C6 + noting `Unit` is pre-rename - Canvas line 163 Q-Unit-3 collapse note: `Scale = Unit` → `Scale = One` (BLOCKING — bare line-numbers in S1 canvas frontmatter): NOT APPLICABLE — already fixed at sweep commit 127287a prior to review; reviewer was on sha a462f00 pre-sweep. NON-BLOCKING — placeholder receipts: - brief-authoring-checklist.md `#issuecomment-...` → concrete ID 4385393971 (Director's anchor-discipline reply) - r3-substrate-unit-quantity-carrier-worker.md (3 occurrences) + canvas RATIFICATION banner: `#issuecomment-...` → concrete ID 4385412256 (Q-Unit ratification reply) - `<this-commit>` placeholder in checklist already fixed at 5be1c99 Verification: zero `#issuecomment-...` or `<this-commit>` placeholders in docs/briefs/ + docs/r4-carve-out-routing.md. * docs(briefs): codex BLOCKING review fixes — Practice 4 source-level + policy scope Three BLOCKING findings on PR #1782 (codex sha b913e2f): 1. **Citation-discipline policy scope**: addressed by scoping the rule prospectively to briefs landing after the policy commit (`b4cae5299`). Pre-existing corpus references are flagged as paydown but NOT individually blocking; coordinated single-PR sweep is the right paydown shape. Substrate-Mgr scope already swept at `127287a66`. 2. **R4 C6 reconciliation**: already fixed at `6e5a5481a` (R4 C6 narrowed to Aspect-axis follow-on for EpochMs only; 2-axis carrier landing IS in R3). Codex sha b913e2f was pre-fix. 3. **Practice 4 source-level vs PR-body conflation**: corrected. Acceptance bullet for Unit/Quantity worker brief now requires inline checkpoint comments on the LIVE declarations in `dsl/std/units.dag` (load-bearing artifact per `docs/modeling-discipline.md#4-coproduct-dissolution` "What to check"). PR-body summary is supplementary, not substitute. Closes the review's reconciliation gates. * docs(briefs): absorb Q-MachineConstraint + Q-Unit-1-Recanvas ratifications Two major Director ratifications absorbed: 1. Q-MachineConstraint-Carrier RATIFIED at gunbc#828 #issuecomment-4385530115 (Brian directive 2026-05-06: "universal substrate, ratify defaults"; PR #1817 queued). 6 sub-decisions: - MachineWidth<bits> only for R3; defer RegisterClass / EndianMode / Alignment / signedness post-R3 - Parametric Compose<Algebra, MachineConstraint> type-level interaction shape (lookup-maps REJECTED — AlgebraMachineProduct record-keyed table form retracted) - Int<64> ≡ Compose<AbelianGroup, MachineWidth<64>> (algebra-side options A and B preserve interaction semantics) - Real<64> ≡ Compose<ApproximateField<Rational>, MachineWidth<64>> (algebra approx + machine approx as independent composing axes) - ≥3 pairs is minimum, not target - UNIVERSAL substrate: every target carries machine-constraint facts; targets without native machine-width semantics handle omission at Grounding-level discharge (target-conditioned lowering NOT target-conditioned substrate) Reframed: S3 brief Phase-1, S3 Phase-3/4 emission entries, S8 Phase-1/2/3 (ApproximateField<Real> → ApproximateField<Rational>; AlgebraMachineProduct → Compose<...>), S9 emission entries. 2. Q-Unit-1-Recanvas RATIFIED Measure<Q, S> at gunbc#828 #issuecomment-4385539791. valiant-ibex-312's substrate-state grep caught Unit<Q, S> collision with kernel Unit at dsl/std/types.dag:176; Director ratified Measure<Q, S> as outer carrier name (conventional in scientific computing; no dsl/std/ collision). Reframed: Unit/Quantity worker brief throughout (Unit<Q,S> → Measure<Q,S>; dsl/std/units.dag → dsl/std/measure.dag); canvas spec sections + sketch products (Refined<Measure<Time, ...>...>); R4 carve-out C6 with informal/formal name disambiguation note. Both rename + reshape are now consistent across S3 / S8 / S9 / Unit-Q briefs + canvas + R4 routing. Workers (valiant-ant-72 on S3, quiet-boar-160 on S8, proud-lynx-311 on S9, valiant-ibex-312 on Unit-Q) all dispatch against the ratified shape. * docs(briefs): S11 Slice C worker brief — ROADMAP :425 retirement post-#1801 Cascade-clearance trigger fired post-#1801 (Slice B) merge 2026-05-06. S11 brief authored per B1 RATIFIED prose+regen bundling discipline. Phase 1: per-file edits across the post-A+B residual list (~10 files per smart-ram-167's Slice B post-rebase enumeration); anchor + rule-text-quote pattern per current citation discipline; bootstrap regen in same PR (no split). Phase 2: ROADMAP :425 row PARTIAL → Retired with 3-slice receipt; Q-Slice-C-Retirement-Receipt resolution. Worker pin smart-ram-167 (#1759) — Slice B precedent owner. Dispatch packet sent separately to #1759. * docs(briefs): Q-Refined-Phantom-Composition (c) RATIFIED — defer literal Refined form Director ratified option (c) at gunbc#828 #issuecomment-... 2026-05-06: drop `non_negative` predicate at Measure (phantom) layer; rely on `Quantity = Time` tagging for non-negative-magnitude semantics. Q-Unit-4 outer-Refined / inner-Measure composition is preserved as **conceptual model**; the literal `Refined<Measure<...>, predicate>` form is **deferred to value-typed-integration follow-up** when substrate gains Refined<> as first-class type expression and dimensional refinements get value-typed integration. Updates: - Canvas RATIFICATION banner Q-Unit-4 row reframed to "conceptual model; literal form deferred" - Canvas Phase-3 reframe table: Duration / Milliseconds drop the Refined<...> wrapper; rely on Quantity tag - Worker brief Phase-2 prose + reframe table + Acceptance bullet all reflect deferred-literal-form posture valiant-ibex-312 unblocked to proceed with Phase-2 (Grounding handoff doc) per the (c)-shaped Acceptance. * docs(briefs): S12 F2 + F8 doc-sharpening worker brief — post-#1820 cascade #1820 (Slice C) merged 2026-05-06; ROADMAP :425 row Retired closes the prose-reference paydown program. S12 cascade-clearance trigger fires per B6 RATIFIED single bundled Mgr-tier doc-sharpening PR. Scope: F2 (.v3 filename-suffix grammar dispatch) + F8 (bootstrap load-order/exclusion authority) — single bundled PR per B6. Cross-program coordination with PB Mgr (#1742) per design schedule §3 P5 (comment-only OR co-author; no duplicate PR). Worker pin smart-ram-167 (#1759) — Slice C precedent owner; pattern-familiar with prose+regen bundling discipline. Dispatch packet sent separately to #1759. * docs(briefs): S9 brief — UInt landed via #1818; algebra label corrected to CommutativeSemiring Per proud-lynx-311 merge report at gunbc#1746 #issuecomment-... 2026-05-06: PR #1818 landed `type UInt = Nat` after codex BLOCKING review on initial `CommutativeMonoid<Nat>` form (P2 facts-flow-forward violation — wrap projects out Nat's multiplicative monoid + identity 1). Final landed form preserves full `CommutativeSemiring<Magnitude>` surface — Nat IS a commutative semiring (addition + multiplication + 0 + 1). Algebra-side label for UInt is CommutativeSemiring, NOT CommutativeMonoid (corrects my earlier dispatch labeling). S9 Phase-1 step 2 marked LANDED with merge ref. Phase-1 step 3 emission entries reframed: - Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>> - UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>> Worker discipline + reviewer caught a substrate-state error in my brief framing; brief now matches landed reality. * docs(briefs): fix contradictory (c) guidance in Unit/Quantity worker brief Codex BLOCKING review on PR #1825 sha 1bce041 caught a real contradiction: brief Phase-2 prose at lines 115-118 stated "predicates do NOT attach at the Measure (phantom) layer" per Q-Refined-Phantom-Composition (c) RATIFIED, but immediately following paragraph at lines 130-132 still said "Refined predicate non_negative applies at the Measure-typed level" — left over from the original Q-Unit-4 framing. Updated lines 130-137 to consistently apply (c) shape: predicates deferred to value-typed-integration follow-up; dimensional semantics ride on Quantity tag at Measure layer; worker authors 3 in-scope refinements WITHOUT predicate at this layer. P2 boundary discipline restored: brief now gives single-authority guidance for Phase-2 substrate landing. * docs(briefs): name predicate authority + concretize ratification ID — codex BLOCKING Codex BLOCKING review on PR #1825 sha 32541a5: 1. (BLOCKING) Q-Refined-Phantom-Composition (c) ratification deferred phantom-carrier predicates without naming the value-level predicate authority. Updated canvas Q-Unit-4 RATIFICATION row to explicitly name: 'predicates re-attach at value-typed Int/UInt level via where-sugar (e.g., range(min: 0) per current dsl/std/types.dag precedent)' when value-typed integration lands. 2. (BLOCKING) Worker brief option-c reframe didn't sweep the audit receipt §5 (Carrier dissolves the bridge). Updated to consistently apply (c) shape — dimensional semantics ride on Quantity tag at Measure layer; value-level predicate authority deferred; until then, range/gt_zero constraints attach at value-typed Int/UInt via where-sugar per dsl/std/types.dag convention. 3. (NON-BLOCKING) Canvas had #issuecomment-... placeholder for the (c) ratification message ID. Replaced with concrete #issuecomment-4385687075 per dashboard relay header. Brief now names a single value-level authority (Int/UInt where-sugar) for value-range facts; phantom-layer Quantity tag carries only dimensional semantics. P2 boundary discipline maintained. * docs(briefs): S12 RETIRED per Director option (2) ratification Director ratified option (2) at gunbc#828 #issuecomment-4385873... 2026-05-06 post smart-ram-167's substrate-state STOP-AND-ESCALATE: - F2 absorbed by 2026-05-04 routing + #1662 closure (`Result/DivError span.file-keyed` retired) - F8 topical concerns distributed across live rows (Class 5 Gap 3 `bootstrap.rs::std_fixtures()`; Character-level load-set / bootstrap; Filename-sentinel-bridges) with own owners - Doc-discipline goals absorbed by per-Mgr citation paydown routing at gunbc#828 #issuecomment-4385583247 Frontmatter updated to RETIRED with absorption receipt. Worker (smart-ram-167) pin freed. Director-side acknowledgment of Substrate Mgr's option (2) recommendation: "Substrate Mgr's option (2) recommendation is the right move structurally. Most Mgrs default to 'synthesize work to fill the brief' when canvas-vs-HEAD mismatch surfaces. Recommending retirement when the work is genuinely absorbed is harder discipline." * docs(briefs): S7 PR-F (BoundDeclaration consumer + ReferenceModel<T>) worker brief Per cascade-clearance trigger fired post-#1782 merge 2026-05-06. loyal-wolf-828 (#1764) acknowledged S7 dispatch with substrate-state pre-flight grep (gunbc#1764 #issuecomment-4385872...) and surfaced await-brief posture. This brief satisfies that prerequisite. Worker pre-flight inventory absorbed: - BoundDeclaration substrate landed (substrate.dag StaticBound + PlatformDependent); partial BoundDeclarationView consumer in fold.rs (ScratchIntExamples scope; doc-explicit partial) - ReferenceModel<T> NOT yet in dsl/std/ (substrate-fact-introduction required; design-emission-model.md §Q2 framing) Phase-1: BoundDeclaration consumer broadening to full design-emission-model.md surface (u128 / isize / usize / walker arms / pilot mirror). Phase-2: ReferenceModel<T> P1 substrate-fact-introduction with DFS-of-concept-DAG + Practice 4 in-source checkpoint comment + named consumer demand (Grounding G1). Phase-3: cross-program handoff to Grounding Mgr (#1745) for G1 (T-Ground-Rust Phase 1). Closure predicate: unblocks Grounding T-Ground-Rust Phase 1. Worker pin loyal-wolf-828 holds; dispatch fires immediately on this brief landing. * docs(briefs): S9 Slice 2.5 — predicate-registry + PB-1 shim retirement (Path 3 RATIFIED) Director ratified Path 3 at gunbc#828 #issuecomment-4390333451 post proud-lynx-311's pre-flight DFS at gunbc#1746 #issuecomment-4390253616. Brief reframes Slice 2.5 from S-sized (primitive + 1 consumer) to L-sized (predicate registry + PB-1 shim retirement). 5 deliverables: 1. gt_zero primitive predicate 2. Predicate registry infrastructure (resolver-side substrate-fact-introduction) 3. PB-1 shim retirement (span.file=='dsl/std/types.dag' special case dissolves) 4. Migration of existing types.dag predicates (range, pattern, etc.) 5. PositiveInt = Refined<Nat, gt_zero> consumer update Per Director hard scope bars: no parallel predicate-resolution path; no per-call-site special-casing; PB-1 shim retirement non-negotiable in same slice. Single PR per Path 3 RATIFIED. Discipline anchors: feedback_construction_over_ratchets (PB-1 shim is bridge); feedback_dissolve_bridges (registry is structural fix); feedback_no_textual_enforcement_bridges (file-path special-case is anti-pattern shape); P1 substrate-fact-introduction procedure. Worker pin proud-lynx-311 — Director pre-cleared on pre-flight DFS context. PR #1840 (partial Slice 2 — NonNegativeInt = Nat) MERGED at 16:54Z; Slice 2.5 dispatchable immediately on this brief landing. * docs(briefs): Slice 2.5 audit receipt — Mgr independent grep parallel-verifies worker DFS Codex BLOCKING review on PR #1845 sha 2d23094 flagged: external PR receipt absorbed as live substrate state without re-grepping local std declarations at brief authoring time. Authority receipt §1 cited proud-lynx-311's pre-flight DFS as authority but didn't record Mgr-side independent verification. Updated audit receipt §1 + §4 to record Mgr's independent grep at HEAD as parallel verification: - gt_zero zero matches in dsl/std/ + src/v3/ (verified) - NonNegativeInt = Nat at dsl/std/integer.dag:133 (#1840 land) - PositiveInt = Int where range(min: 1) at dsl/std/types.dag:255 - PB-1 shim doc-block confirms bridge-text framing - Predicate registry confirmed not present Worker DFS remains the dispatch-time discipline (substrate-state grep at dispatch); Mgr's independent grep is parallel verification, not substitute. Per Director's pattern note at gunbc#828 #issuecomment-4390199218 + #issuecomment-4390333451: dispatch-packet authoring discipline includes Mgr-side substrate-state-grep, not just relay of worker findings. * docs(briefs): Slice 2.5 brief — Int extensibility scope-bound; lower.rs cites anchored openai-pro REQUEST_CHANGES on PR #1845 sha 2d23094 flagged 2 issues: BLOCKING — P1 modeling faithfulness / substrate scope leak: `gt_zero` Int-side extensibility was left to 'worker judgment if generalization cheap'. Per feedback_construction_over_ratchets: don't speculatively extend predicate-carrier compatibility ahead of named consumer demand. Reframed to **EXPLICITLY OUT-OF-SCOPE** for Slice 2.5; Nat-only registration. Future Int-side consumer surfaces as separate substrate-fact-introduction (P1 procedure) brief with own consumer-demand receipt. NON-BLOCKING — citation discipline self-violation: 3 instances of bare `lower.rs:821-835` line-range citations contradict the brief's own 'no bare :NNN' requirement at line 211. Converted to function-name anchored form: `src/v3/compiler/src/lower.rs` (`lower_type_alias_refinements_phase` doc-block). Both findings absorbed; no parallel scope ambiguity remains; brief now self-consistent on citation discipline. * WIP: R3 Substrate * docs(briefs): Slice 2.5 brief — audit-receipt cites converted to declaration-name anchors cursor APPROVE_WITH_COMMENTS on PR #1845 sha ebf6ac1 noted citation self-inconsistency: audit-receipt section §1 + §4 used bare `dsl/std/integer.dag:133` and `dsl/std/types.dag:255` line-number pins despite the brief's own no-bare-:NNN rule. Converted to declaration-name-anchored form: - 'declaration named NonNegativeInt' (with reproducible grep command) - 'declaration named PositiveInt' (with reproducible grep command) - 'lower_type_alias_refinements_phase doc-block' for PB-1 location Brief now fully self-consistent on its own citation discipline. Verified zero remaining .dag:NNN / .rs:NNN / .md:NNN bare cites. * docs(briefs): Slice 2.5 brief — Path (a) RATIFIED carrier+argument contract verification Director Path (a) RATIFIED at gunbc#828 #issuecomment-4390760353 (triage) + #issuecomment-4390794121 (convergence ack): predicate-body lowering with carrier+argument contract verification at lower-time IN SAME SLICE. Placeholder semantics REMOVED from registry path. Codex REQUEST_CHANGES surfaced that registered-predicate-without- contract = silent acceptance per feedback_fail_closed_discipline. Brief reframe absorbs Path (a): Hard scope bar #5 added: - Same-slice carrier+argument-contract verification - Mismatched carrier OR malformed argument → Diagnostic::ResolveError - No 'named but not checked' path remains Scope expansion section added (L → XL): - Predicate-body lowering for all 7 registered predicates - Per-predicate arg-shape validation - Multi-session work acknowledged - Carrier-check at 788d6ac retained; predicate-body builds on top Acceptance bullets updated: gt_zero with arg-shape validation + all 7 predicates with carrier+argument contract verification. Director-pinned brief-authoring discipline: dissolution trigger named in a brief authored as substrate-fact-introduction must specify same-slice acceptance, not deferred follow-up. Otherwise the brief encodes 'land bridge + defer dissolution' anti-pattern, the same shape feedback_construction_over_ratchets warns against at implementation layer applied at brief authoring layer. This is the 11th miss this session — but the pattern surfacing it (direct sweep + Codex REQUEST_CHANGES + Director ratification) produces structural fixes that prevent the next miss class. Brief revision is the structural recovery. --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 6, 2026
* WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * WIP: R3 Substrate * fix(briefs): retire INVARIANTS.md line-number cites in r3-426 brief Per codex review on PR #1782 (sha c2512cd): the new briefs introduced fresh `INVARIANTS.md:<line>` citations — the same stale-line-anchor shape the :425 sweep brief itself names as needing dissolution. Replaces four line-number references in r3-426-sfp-subdoc-rename-and-author-worker.md (lines 16, 28, 58, 85) with prose-anchored locators citing the section anchor + bullet text. The other PB-lane briefs codex flagged (r3-pb-runtime-equivalence- corpus-seed-audit.md, r3-pb-t-fixedpoint-worker.md) belong to neat-bear-351's lane and are not in this PR's scope; will route separately. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): correct L6 type count + r3-435 dual-path retirement receipt Per openai-pro review on PR #1782 (sha c2512cd): 1. L6 brief said acceptance ratchet asserts "five types" exist but declares six (ShapeATarget, FormAxis, BehaviorAxis, MethodTemplateContractKey, EmissionCell, EmissionPathProjection). Updated both occurrences to enumerate all six explicitly. 2. r3-435 brief authorized both Path A (typed-AST reader, deletes IntegrationRsScan) and Path B (scanner widening with Char state), but retirement receipt language only covered Path B. Updated ROADMAP/ledger retirement and authority-audit step 5 to cover both authorized dissolution paths. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): L6 second 'five types' site + T-E-P lookup-gate authority Per cursor review on PR #1782 (sha 2072b98): 1. L6 brief acceptance bullet at :131 still said "five types" while the parenthetical and the Slice step at :117 said six. The prior fix only caught one of two sites; corrected the second. 2. T-E-P brief `e_p_call_pattern_lookup_authoritative` acceptance named only `lower_call_pattern` as "the only route from call site to LoweringTarget" — but Phase 2 explicitly requires lenses route through `per_call_pattern_at` (the L-7 single- authority lookup), NOT `lower_call_pattern` (compiler-internal). Acceptance text rewritten to name `per_call_pattern_at` as the consumer-facing gate and clarify `lower_call_pattern`'s internal-only scope. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): strengthen L6 ratchet from count-parity to per-row bijection Per openai-pro REQUEST_CHANGES on PR #1782 (sha 2072b98): the L6 acceptance ratchet asserted only row-count parity between emission_path_projections and the union of MethodTemplateContract rows. Count parity permits duplicate projection keys and missing source rows — the bridge from list-non-empty proxy to per-row projection would not be mechanically proven. The whole point of introducing MethodTemplateContractKey { target, dag_method } is to provide 1:1 row identity. Rewrites the ratchet to assert per-row bijection: every MethodTemplateContract source row has exactly one matching EmissionPathProjection row by key; every projection key resolves to exactly one source row; duplicate projection keys fail closed. (The other openai-pro BLOCKING finding — T-E-P e_p_call_pattern_lookup_authoritative naming the wrong authority — was already fixed at 6282a07; the review citation was against the pre-fix sha 2072b98.) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): split L6 ratchet into slice-active vs deferred-activation gates Per BLOCKING inline review on PR #1782 at d924594: the prior bijection ratchet was unsatisfiable as authored. The *_method_template_contracts lists at HEAD (rust/python/go) are non-empty; shipping emission_path_projections: [] alongside a strict bijection requirement means the bijection cannot pass when the slice lands — source rows would be uncovered. Prior text incorrectly said "vacuously true on the empty source set" — the SOURCE set isn't empty, only the projection set is. Splits the ratchet into two gate categories: - Slice-active gates: types/data shape exist + empty-state predicate (emission_path_projections == []). - Deferred-activation gate: per-row key bijection, authored as a test scaffold by this slice but #[ignore]'d (or feature-gated on len > 0) until Grounding's follow-up populates rows. On activation, asserts the bijection over the populated set. This honors P1 doc faithfulness: the slice's acceptance contract is satisfiable as authored, and the bijection becomes the load-bearing gate the row-population PR satisfies. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(briefs): require L6 bootstrap enrollment to honor P2 facts-flow-forward Per BLOCKING inline review on PR #1782 at L6:34: the slice authored src/v3/std/cross_target_coverage.dag but never required enrolling it in the bootstrap/load authority. Without enrollment the file is dead-letter on disk and EmissionPathProjection can be absent from the downstream Dag — silent P2 facts-flow-forward violation. Adds a new Slice step 2 requiring a cross_target_coverage field on BootstrapFixtures in src/v3/std/extdeps_bootstrap_fixtures.dag mirroring the *_method_template_contracts enrollment pattern, plus a loader-visibility ratchet. Renumbers subsequent Slice steps. Adds matching Acceptance bullet. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: R3 Substrate * WIP: R3 Substrate * docs(briefs): normalize Practice 4 citation in S3 brief per codex review Codex review on PR #1782 (sha 2614217) flagged stale citation "Practice 4 (Step 4 of the type-introduction checklist)" in S3 brief — no such checklist step exists in the pinned rubric. Live rule per docs/modeling-discipline.md:131-134 is the "What to check" checkpoint-comment requirement under Practice 4 (coproduct dissolution). Updated the S3 reference to cite that line directly and drop the stale "Step 4 of type-introduction checklist" wording. S2/S8/S9 already use bare "Practice 4" without the stale steps wording. S1 (committed earlier today) does not contain the stale phrase. S6 (pre-existing brief flagged by reviewer as carrying the same pattern) is out of this branch's scope; flag for separate paydown. * docs(briefs): sweep stale Practice 4 citations corpus-wide Codex review on PR #1782 flagged the stale "Step 4 of the type-introduction checklist" wording — no such checklist exists in docs/modeling-discipline.md. The live rule is the "What to check" checkpoint-comment requirement under Practice 4 (coproduct dissolution) at docs/modeling-discipline.md:131. Sweep across docs/ found 2 additional briefs carrying the stale phrasing (S3 fixed previously at 9db003d): - docs/briefs/r3-l6-emission-path-projection-substrate-worker.md - docs/briefs/r3-x1b-s1-transform-dispatch-substrate-worker.md Both updated to cite the live rule directly. Verified zero remaining "type-introduction checklist" or "Step 4 of the" references in docs/ tree. * docs(briefs): add Practice 4 citation discipline note to authoring checklist Per parent inbox status flag at gunbc#828 #issuecomment-4385346221: add a Citation discipline section to brief-authoring-checklist.md naming the correct shape — Practice 4 (coproduct dissolution) + "What to check" rule at docs/modeling-discipline.md:131 — and calling out the stale "Step 4 of the type-introduction checklist" phrasing as a failure mode. References the corpus sweep at 2dc92f3. Prevents future briefs that copy old templates from re-introducing the stale citation pattern. * docs(r3): cascade work post-Director ratification — S1 worker, R4 routing, Unit/Quantity canvas Per Director ratification at gunbc#828 inbox response 2026-05-06 (zesty-bear-812): 1. S1 Q1+Q2 RATIFIED → author S1 gap-test representative worker brief (`r3-substrate-s1-gap-test-representative-worker.md`). Dispatchable on T-E-P Phase 1 + E6-G0d landing; closes ledger row #61. 2. S2 Q2 RATIFIED (option-(b) T-LBP narrowing) + S3 + S8 deferrals → R4 carve-out routing ledger (`docs/r4-carve-out-routing.md`). Captures C1 (parallelism lens), C2 (effect_enum lens), C3 (register narrowing), C4 (deferred MachineConstraint axes), C5 (rounding-mode product extension), C6 (Unit/Quantity). Cascade messages to Verification / Grounding / Debt-Paydown Mgrs. 3. S9 Phase-3 dimensional refinement boundary → Unit/Quantity carrier canvas (`r3-substrate-s9-unit-quantity-carrier-canvas.md`). Surfaces Q-Unit-1..5 for Director ratification: carrier name, 2-axis vs 3-axis Phase-1, Duration/Seconds collapse, Refined composition order, Practice 4 classification. Updates §1.8 ledger receipts for carved items; R3 closure receipts treat carved items as explicitly-deferred (not drift). * docs(briefs): citation-discipline fix-forward — anchors/quotes, not line numbers Director redirect at gunbc#828 #issuecomment-... noted the citation paydown's replacement form (`docs/modeling-discipline.md:131`) is the same drift class as the original fabricated-step error. Bare line numbers shift silently on any doc edit above the cited line. Generalized fix: - brief-authoring-checklist.md "Citation discipline" section rewritten to prohibit bare line-numbers across ALL cross-doc references (modeling-discipline.md, r3-program-plan.md, r3-structure.md, INVARIANTS.md, MODELING.md, etc.). Use section anchors (#section-name) or rule-text quotes. - 3 briefs that swapped in :131 (S3, S6, X1B) re-converted: - docs/modeling-discipline.md#4-coproduct-dissolution anchor - "What to check" rule quoted inline (self-contained) Brief-authoring-checklist failure-mode section names both vectors (fabricated step + bare line numbers) and references both sweep commits in audit trail. Broader cross-doc citation sweep across briefs I authored on this branch (frontmatter authority-docs lists, body prose) is recommended per Director note — flagging as separate paydown task; this commit closes the immediate recurrence vector. * docs(briefs): fix unfaithful Int=AbelianGroup<Nat> + sweep stale 'Step 4' refs Per openai-pro review on PR #1782 sha 0ecd481 (gunbc#1782 #...): BLOCKING fix (P1 modeling faithfulness): - S9 + S3 directed workers to model `Int = AbelianGroup<Nat>`, but Nat is not an Abelian-group carrier (additive inverses not representable). Reframed: - `Int` is the additive Abelian group on integers (carrier Z); `Nat` is a commutative monoid under addition. - Algebraic relationship is **explicit group-completion**: `Int ≡ GroupCompletion<CommutativeMonoid<Nat>>` — NOT `AbelianGroup<Nat>` parameterization. - Worker either models `Int` directly as `AbelianGroup` (terminal Abelian-group instance) OR introduces explicit `GroupCompletion<M>` substrate carrier; decision lands in Phase-1 PR per P1 procedure. - S3 cross-reference + Phase-4 algebra-axis preservation note reframed accordingly. NON-BLOCKING in-PR cleanup (Practice 4 citation discipline): - 6 briefs still carrying stale 'Practice 4 Step 4' wording (r3-coproduct-1/2/3, r3-l6, r3-t-e-p, r3-x1b) updated to use the section anchor (`docs/modeling-discipline.md#4-coproduct-dissolution`) + inline rule-text quote ("What to check") form per the citation discipline established in `b4cae5299`. Closes both findings from openai-pro review. * docs(briefs): Q-Unit-1..5 RATIFIED — apply Scale.Unit→One rename + ratification banner Director ratified Q-Unit-1..5 at gunbc#828 #issuecomment-... 2026-05-06 (zesty-bear-812). All 5 questions ratified; one rename required: - Scale enum's `Unit` value collides with outer `Unit<Q, S>` carrier name (`Unit<Time, Unit>` is shadowing-confusing). Rename to `One`. - Duration ≡ Unit<Time, One> (collapses with Seconds per Q-Unit-3). - Milliseconds ≡ Unit<Time, Milli>. Canvas updated: - Top-level RATIFICATION banner records all 5 ratifications + rename - Scale enum: `Unit` → `One` (with inline rename receipt) - Sketch product expansions use `Unit<Time, One>` / `Unit<Time, Milli>` - Phase-3 reframe table products updated - Q-Unit-3 prose collapse reference updated Subsequent worker brief authoring (Unit/Quantity carrier landing post-prerequisite) inherits the renamed shape. * docs(briefs): Unit/Quantity carrier worker brief — Q-Unit-1..5 RATIFIED Worker brief consumes the ratified Unit/Quantity canvas (`5f22fd06e` ratification banner). Phase-1 lands `Unit<Q, S>` + `Quantity` + `Scale` declarations at the ratified shape (Scale's `One` value per Q-Unit-1 rename). Phase-2 reframes S9 Phase-3 dimensional refinements (Duration / Seconds / Milliseconds) to outer-Refined / inner-Unit composition per Q-Unit-4 RATIFIED. EpochMs deferred to R4 C6 (Aspect-axis follow-up). Practice 4 marks pre-determined per Q-Unit-5: `Unit<Q, S>` 🟢 PRIMITIVE; `Quantity` + `Scale` 🟡 SCAFFOLD with "consumed by ≥1 emission rule" dissolution trigger. Cross-program handoff to Grounding Mgr (#1745) for dimensional emission consumption documented. Worker pin TBD — dispatchable when idle pool refreshes. * docs(briefs): S9 Phase-1 reframed — Int landed at #1466; UInt-only migration Per proud-lynx-311 dispatch ack at gunbc#1739 #issuecomment-... 2026-05-06: re-grep at HEAD revealed substrate state diverges materially from S9 brief's "Option A vs B decide" framing. Reality check at HEAD: - `dsl/std/integer.dag:55` already carries `type Int = AbelianGroup<GroupCompletion<Nat>>` per Slice 3 PR #1466 / commit `4ba0a6d04`. Single-authority Q6 audit doc at `docs/audit/t-numeric-construction-group-completion-6q.md` ratified this shape and explicitly rejected the compact form. - `dsl/std/integer.dag:56` still on legacy `type UInt = UInt64` alias; UInt migration is the actual Phase-1 worker scope. Brief Phase-1 step 1 reframed: Int landing acknowledged; worker confirms HEAD match. Phase-1 step 2 reframed: UInt migration target named (`type UInt = CommutativeMonoid<Nat>`); fixed-width rows preserved per existing legacy-rows policy. Historical Option A vs B framing kept as note (superseded). Confirmed S9 dispatch ack from worker; UInt-only Phase-1 PR is in-scope. Emission entries deferred per S3 (valiant-ant-72 pre-landing); Phase-2 deferred per S8 (quiet-boar-160 pre-landing). * docs(briefs): fix two BLOCKING citation drift items per codex review Codex review on PR #1782 sha 5f22fd0 flagged two authority-receipt gaps: 1. brief-authoring-checklist.md:118 — `<this-commit>` placeholder in the failure-mode section was unverifiable post-merge. Replaced with `b4cae5299` (the actual anchor/quote conversion commit). 2. r3-378-http-path-helper-fail-open-worker.md:84 — `INVARIANTS.md C-8 fail-closed` cited without anchor or rule-text quote, in violation of the citation discipline added in this PR. Converted to `INVARIANTS.md#p3-fail-closed` (P3 heading anchor) + inline rule-text quote naming P3 + C-8. PR's own discipline now consistently applied across the diff. * docs(briefs): sweep bare .md:NNN citations in Substrate-Mgr-authored briefs Per Director recommendation in citation-discipline thread (gunbc#828 follow-on): bare line-number references drift silently. Swept 19 remaining instances across 5 briefs I authored on this branch: - docs/briefs/r3-substrate-s1-q-class-2-chain-break-gap-test-canvas.md - docs/briefs/r3-substrate-s1-gap-test-representative-worker.md - docs/briefs/r3-substrate-s2-t-lbp-scope-calibration-canvas.md - docs/briefs/r3-substrate-s3-machine-constraint-carrier-worker.md Conversion: bare `file.md:NNN` → `file.md §"Section name"` form. Stable; survives doc edits; human-readable. Section names per HEAD heading audit. Verification: `grep -nE '.md:[0-9]+' docs/briefs/r3-substrate-*.md docs/r4-carve-out-routing.md` returns zero results in Substrate-Mgr-authored scope. Broader corpus scope (~232 references across non-Substrate briefs) remains as separate paydown task. * docs(briefs): fix BLOCKING + NON-BLOCKING items per openai-pro review on a462f00 Three findings addressed; 2 BLOCKING + 1 NON-BLOCKING: BLOCKING — R3/R4 dual authority for Unit/Quantity carrier: R4 carve-out C6 narrowed from "Unit/Quantity carrier landing" (deferred all of it) to **"Aspect-axis (PointKind) follow-on for EpochMs and instant/rate-shaped refinements only"**. The 2-axis Measure<Q,S> carrier landing IS in R3 per Q-Unit-1..5 RATIFIED; only the Aspect axis (PointKind = Magnitude | Instant | Rate) carves to R4. Cascade-message C6 reference + R4 program- plan input list updated to match. Cross-Mgr message clarifies Grounding consumes 2-axis dimensional refinements in R3. BLOCKING — Locked design partially applied (Scale.Unit→One rename in canvas alternate-sketch sections): - Canvas line 105-108 alternate sketch ("Open question: Instant vs Duration axis") still used pre-rename `Unit` Scale value; added pre-ratification disclaimer marking the section as deferred to R4 C6 + noting `Unit` is pre-rename - Canvas line 163 Q-Unit-3 collapse note: `Scale = Unit` → `Scale = One` (BLOCKING — bare line-numbers in S1 canvas frontmatter): NOT APPLICABLE — already fixed at sweep commit 127287a prior to review; reviewer was on sha a462f00 pre-sweep. NON-BLOCKING — placeholder receipts: - brief-authoring-checklist.md `#issuecomment-...` → concrete ID 4385393971 (Director's anchor-discipline reply) - r3-substrate-unit-quantity-carrier-worker.md (3 occurrences) + canvas RATIFICATION banner: `#issuecomment-...` → concrete ID 4385412256 (Q-Unit ratification reply) - `<this-commit>` placeholder in checklist already fixed at 5be1c99 Verification: zero `#issuecomment-...` or `<this-commit>` placeholders in docs/briefs/ + docs/r4-carve-out-routing.md. * docs(briefs): codex BLOCKING review fixes — Practice 4 source-level + policy scope Three BLOCKING findings on PR #1782 (codex sha b913e2f): 1. **Citation-discipline policy scope**: addressed by scoping the rule prospectively to briefs landing after the policy commit (`b4cae5299`). Pre-existing corpus references are flagged as paydown but NOT individually blocking; coordinated single-PR sweep is the right paydown shape. Substrate-Mgr scope already swept at `127287a66`. 2. **R4 C6 reconciliation**: already fixed at `6e5a5481a` (R4 C6 narrowed to Aspect-axis follow-on for EpochMs only; 2-axis carrier landing IS in R3). Codex sha b913e2f was pre-fix. 3. **Practice 4 source-level vs PR-body conflation**: corrected. Acceptance bullet for Unit/Quantity worker brief now requires inline checkpoint comments on the LIVE declarations in `dsl/std/units.dag` (load-bearing artifact per `docs/modeling-discipline.md#4-coproduct-dissolution` "What to check"). PR-body summary is supplementary, not substitute. Closes the review's reconciliation gates. * docs(briefs): absorb Q-MachineConstraint + Q-Unit-1-Recanvas ratifications Two major Director ratifications absorbed: 1. Q-MachineConstraint-Carrier RATIFIED at gunbc#828 #issuecomment-4385530115 (Brian directive 2026-05-06: "universal substrate, ratify defaults"; PR #1817 queued). 6 sub-decisions: - MachineWidth<bits> only for R3; defer RegisterClass / EndianMode / Alignment / signedness post-R3 - Parametric Compose<Algebra, MachineConstraint> type-level interaction shape (lookup-maps REJECTED — AlgebraMachineProduct record-keyed table form retracted) - Int<64> ≡ Compose<AbelianGroup, MachineWidth<64>> (algebra-side options A and B preserve interaction semantics) - Real<64> ≡ Compose<ApproximateField<Rational>, MachineWidth<64>> (algebra approx + machine approx as independent composing axes) - ≥3 pairs is minimum, not target - UNIVERSAL substrate: every target carries machine-constraint facts; targets without native machine-width semantics handle omission at Grounding-level discharge (target-conditioned lowering NOT target-conditioned substrate) Reframed: S3 brief Phase-1, S3 Phase-3/4 emission entries, S8 Phase-1/2/3 (ApproximateField<Real> → ApproximateField<Rational>; AlgebraMachineProduct → Compose<...>), S9 emission entries. 2. Q-Unit-1-Recanvas RATIFIED Measure<Q, S> at gunbc#828 #issuecomment-4385539791. valiant-ibex-312's substrate-state grep caught Unit<Q, S> collision with kernel Unit at dsl/std/types.dag:176; Director ratified Measure<Q, S> as outer carrier name (conventional in scientific computing; no dsl/std/ collision). Reframed: Unit/Quantity worker brief throughout (Unit<Q,S> → Measure<Q,S>; dsl/std/units.dag → dsl/std/measure.dag); canvas spec sections + sketch products (Refined<Measure<Time, ...>...>); R4 carve-out C6 with informal/formal name disambiguation note. Both rename + reshape are now consistent across S3 / S8 / S9 / Unit-Q briefs + canvas + R4 routing. Workers (valiant-ant-72 on S3, quiet-boar-160 on S8, proud-lynx-311 on S9, valiant-ibex-312 on Unit-Q) all dispatch against the ratified shape. * docs(briefs): S11 Slice C worker brief — ROADMAP :425 retirement post-#1801 Cascade-clearance trigger fired post-#1801 (Slice B) merge 2026-05-06. S11 brief authored per B1 RATIFIED prose+regen bundling discipline. Phase 1: per-file edits across the post-A+B residual list (~10 files per smart-ram-167's Slice B post-rebase enumeration); anchor + rule-text-quote pattern per current citation discipline; bootstrap regen in same PR (no split). Phase 2: ROADMAP :425 row PARTIAL → Retired with 3-slice receipt; Q-Slice-C-Retirement-Receipt resolution. Worker pin smart-ram-167 (#1759) — Slice B precedent owner. Dispatch packet sent separately to #1759. * docs(briefs): Q-Refined-Phantom-Composition (c) RATIFIED — defer literal Refined form Director ratified option (c) at gunbc#828 #issuecomment-... 2026-05-06: drop `non_negative` predicate at Measure (phantom) layer; rely on `Quantity = Time` tagging for non-negative-magnitude semantics. Q-Unit-4 outer-Refined / inner-Measure composition is preserved as **conceptual model**; the literal `Refined<Measure<...>, predicate>` form is **deferred to value-typed-integration follow-up** when substrate gains Refined<> as first-class type expression and dimensional refinements get value-typed integration. Updates: - Canvas RATIFICATION banner Q-Unit-4 row reframed to "conceptual model; literal form deferred" - Canvas Phase-3 reframe table: Duration / Milliseconds drop the Refined<...> wrapper; rely on Quantity tag - Worker brief Phase-2 prose + reframe table + Acceptance bullet all reflect deferred-literal-form posture valiant-ibex-312 unblocked to proceed with Phase-2 (Grounding handoff doc) per the (c)-shaped Acceptance. * docs(briefs): S12 F2 + F8 doc-sharpening worker brief — post-#1820 cascade #1820 (Slice C) merged 2026-05-06; ROADMAP :425 row Retired closes the prose-reference paydown program. S12 cascade-clearance trigger fires per B6 RATIFIED single bundled Mgr-tier doc-sharpening PR. Scope: F2 (.v3 filename-suffix grammar dispatch) + F8 (bootstrap load-order/exclusion authority) — single bundled PR per B6. Cross-program coordination with PB Mgr (#1742) per design schedule §3 P5 (comment-only OR co-author; no duplicate PR). Worker pin smart-ram-167 (#1759) — Slice C precedent owner; pattern-familiar with prose+regen bundling discipline. Dispatch packet sent separately to #1759. * docs(briefs): S9 brief — UInt landed via #1818; algebra label corrected to CommutativeSemiring Per proud-lynx-311 merge report at gunbc#1746 #issuecomment-... 2026-05-06: PR #1818 landed `type UInt = Nat` after codex BLOCKING review on initial `CommutativeMonoid<Nat>` form (P2 facts-flow-forward violation — wrap projects out Nat's multiplicative monoid + identity 1). Final landed form preserves full `CommutativeSemiring<Magnitude>` surface — Nat IS a commutative semiring (addition + multiplication + 0 + 1). Algebra-side label for UInt is CommutativeSemiring, NOT CommutativeMonoid (corrects my earlier dispatch labeling). S9 Phase-1 step 2 marked LANDED with merge ref. Phase-1 step 3 emission entries reframed: - Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>> - UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>> Worker discipline + reviewer caught a substrate-state error in my brief framing; brief now matches landed reality. * docs(briefs): fix contradictory (c) guidance in Unit/Quantity worker brief Codex BLOCKING review on PR #1825 sha 1bce041 caught a real contradiction: brief Phase-2 prose at lines 115-118 stated "predicates do NOT attach at the Measure (phantom) layer" per Q-Refined-Phantom-Composition (c) RATIFIED, but immediately following paragraph at lines 130-132 still said "Refined predicate non_negative applies at the Measure-typed level" — left over from the original Q-Unit-4 framing. Updated lines 130-137 to consistently apply (c) shape: predicates deferred to value-typed-integration follow-up; dimensional semantics ride on Quantity tag at Measure layer; worker authors 3 in-scope refinements WITHOUT predicate at this layer. P2 boundary discipline restored: brief now gives single-authority guidance for Phase-2 substrate landing. * docs(briefs): name predicate authority + concretize ratification ID — codex BLOCKING Codex BLOCKING review on PR #1825 sha 32541a5: 1. (BLOCKING) Q-Refined-Phantom-Composition (c) ratification deferred phantom-carrier predicates without naming the value-level predicate authority. Updated canvas Q-Unit-4 RATIFICATION row to explicitly name: 'predicates re-attach at value-typed Int/UInt level via where-sugar (e.g., range(min: 0) per current dsl/std/types.dag precedent)' when value-typed integration lands. 2. (BLOCKING) Worker brief option-c reframe didn't sweep the audit receipt §5 (Carrier dissolves the bridge). Updated to consistently apply (c) shape — dimensional semantics ride on Quantity tag at Measure layer; value-level predicate authority deferred; until then, range/gt_zero constraints attach at value-typed Int/UInt via where-sugar per dsl/std/types.dag convention. 3. (NON-BLOCKING) Canvas had #issuecomment-... placeholder for the (c) ratification message ID. Replaced with concrete #issuecomment-4385687075 per dashboard relay header. Brief now names a single value-level authority (Int/UInt where-sugar) for value-range facts; phantom-layer Quantity tag carries only dimensional semantics. P2 boundary discipline maintained. * docs(briefs): S12 RETIRED per Director option (2) ratification Director ratified option (2) at gunbc#828 #issuecomment-4385873... 2026-05-06 post smart-ram-167's substrate-state STOP-AND-ESCALATE: - F2 absorbed by 2026-05-04 routing + #1662 closure (`Result/DivError span.file-keyed` retired) - F8 topical concerns distributed across live rows (Class 5 Gap 3 `bootstrap.rs::std_fixtures()`; Character-level load-set / bootstrap; Filename-sentinel-bridges) with own owners - Doc-discipline goals absorbed by per-Mgr citation paydown routing at gunbc#828 #issuecomment-4385583247 Frontmatter updated to RETIRED with absorption receipt. Worker (smart-ram-167) pin freed. Director-side acknowledgment of Substrate Mgr's option (2) recommendation: "Substrate Mgr's option (2) recommendation is the right move structurally. Most Mgrs default to 'synthesize work to fill the brief' when canvas-vs-HEAD mismatch surfaces. Recommending retirement when the work is genuinely absorbed is harder discipline." * docs(briefs): S7 PR-F (BoundDeclaration consumer + ReferenceModel<T>) worker brief Per cascade-clearance trigger fired post-#1782 merge 2026-05-06. loyal-wolf-828 (#1764) acknowledged S7 dispatch with substrate-state pre-flight grep (gunbc#1764 #issuecomment-4385872...) and surfaced await-brief posture. This brief satisfies that prerequisite. Worker pre-flight inventory absorbed: - BoundDeclaration substrate landed (substrate.dag StaticBound + PlatformDependent); partial BoundDeclarationView consumer in fold.rs (ScratchIntExamples scope; doc-explicit partial) - ReferenceModel<T> NOT yet in dsl/std/ (substrate-fact-introduction required; design-emission-model.md §Q2 framing) Phase-1: BoundDeclaration consumer broadening to full design-emission-model.md surface (u128 / isize / usize / walker arms / pilot mirror). Phase-2: ReferenceModel<T> P1 substrate-fact-introduction with DFS-of-concept-DAG + Practice 4 in-source checkpoint comment + named consumer demand (Grounding G1). Phase-3: cross-program handoff to Grounding Mgr (#1745) for G1 (T-Ground-Rust Phase 1). Closure predicate: unblocks Grounding T-Ground-Rust Phase 1. Worker pin loyal-wolf-828 holds; dispatch fires immediately on this brief landing. * docs(briefs): S9 Slice 2.5 — predicate-registry + PB-1 shim retirement (Path 3 RATIFIED) Director ratified Path 3 at gunbc#828 #issuecomment-4390333451 post proud-lynx-311's pre-flight DFS at gunbc#1746 #issuecomment-4390253616. Brief reframes Slice 2.5 from S-sized (primitive + 1 consumer) to L-sized (predicate registry + PB-1 shim retirement). 5 deliverables: 1. gt_zero primitive predicate 2. Predicate registry infrastructure (resolver-side substrate-fact-introduction) 3. PB-1 shim retirement (span.file=='dsl/std/types.dag' special case dissolves) 4. Migration of existing types.dag predicates (range, pattern, etc.) 5. PositiveInt = Refined<Nat, gt_zero> consumer update Per Director hard scope bars: no parallel predicate-resolution path; no per-call-site special-casing; PB-1 shim retirement non-negotiable in same slice. Single PR per Path 3 RATIFIED. Discipline anchors: feedback_construction_over_ratchets (PB-1 shim is bridge); feedback_dissolve_bridges (registry is structural fix); feedback_no_textual_enforcement_bridges (file-path special-case is anti-pattern shape); P1 substrate-fact-introduction procedure. Worker pin proud-lynx-311 — Director pre-cleared on pre-flight DFS context. PR #1840 (partial Slice 2 — NonNegativeInt = Nat) MERGED at 16:54Z; Slice 2.5 dispatchable immediately on this brief landing. * docs(briefs): Slice 2.5 audit receipt — Mgr independent grep parallel-verifies worker DFS Codex BLOCKING review on PR #1845 sha 2d23094 flagged: external PR receipt absorbed as live substrate state without re-grepping local std declarations at brief authoring time. Authority receipt §1 cited proud-lynx-311's pre-flight DFS as authority but didn't record Mgr-side independent verification. Updated audit receipt §1 + §4 to record Mgr's independent grep at HEAD as parallel verification: - gt_zero zero matches in dsl/std/ + src/v3/ (verified) - NonNegativeInt = Nat at dsl/std/integer.dag:133 (#1840 land) - PositiveInt = Int where range(min: 1) at dsl/std/types.dag:255 - PB-1 shim doc-block confirms bridge-text framing - Predicate registry confirmed not present Worker DFS remains the dispatch-time discipline (substrate-state grep at dispatch); Mgr's independent grep is parallel verification, not substitute. Per Director's pattern note at gunbc#828 #issuecomment-4390199218 + #issuecomment-4390333451: dispatch-packet authoring discipline includes Mgr-side substrate-state-grep, not just relay of worker findings. * docs(briefs): Slice 2.5 brief — Int extensibility scope-bound; lower.rs cites anchored openai-pro REQUEST_CHANGES on PR #1845 sha 2d23094 flagged 2 issues: BLOCKING — P1 modeling faithfulness / substrate scope leak: `gt_zero` Int-side extensibility was left to 'worker judgment if generalization cheap'. Per feedback_construction_over_ratchets: don't speculatively extend predicate-carrier compatibility ahead of named consumer demand. Reframed to **EXPLICITLY OUT-OF-SCOPE** for Slice 2.5; Nat-only registration. Future Int-side consumer surfaces as separate substrate-fact-introduction (P1 procedure) brief with own consumer-demand receipt. NON-BLOCKING — citation discipline self-violation: 3 instances of bare `lower.rs:821-835` line-range citations contradict the brief's own 'no bare :NNN' requirement at line 211. Converted to function-name anchored form: `src/v3/compiler/src/lower.rs` (`lower_type_alias_refinements_phase` doc-block). Both findings absorbed; no parallel scope ambiguity remains; brief now self-consistent on citation discipline. * WIP: R3 Substrate * docs(briefs): Slice 2.5 brief — audit-receipt cites converted to declaration-name anchors cursor APPROVE_WITH_COMMENTS on PR #1845 sha ebf6ac1 noted citation self-inconsistency: audit-receipt section §1 + §4 used bare `dsl/std/integer.dag:133` and `dsl/std/types.dag:255` line-number pins despite the brief's own no-bare-:NNN rule. Converted to declaration-name-anchored form: - 'declaration named NonNegativeInt' (with reproducible grep command) - 'declaration named PositiveInt' (with reproducible grep command) - 'lower_type_alias_refinements_phase doc-block' for PB-1 location Brief now fully self-consistent on its own citation discipline. Verified zero remaining .dag:NNN / .rs:NNN / .md:NNN bare cites. * docs(briefs): Slice 2.5 brief — Path (a) RATIFIED carrier+argument contract verification Director Path (a) RATIFIED at gunbc#828 #issuecomment-4390760353 (triage) + #issuecomment-4390794121 (convergence ack): predicate-body lowering with carrier+argument contract verification at lower-time IN SAME SLICE. Placeholder semantics REMOVED from registry path. Codex REQUEST_CHANGES surfaced that registered-predicate-without- contract = silent acceptance per feedback_fail_closed_discipline. Brief reframe absorbs Path (a): Hard scope bar #5 added: - Same-slice carrier+argument-contract verification - Mismatched carrier OR malformed argument → Diagnostic::ResolveError - No 'named but not checked' path remains Scope expansion section added (L → XL): - Predicate-body lowering for all 7 registered predicates - Per-predicate arg-shape validation - Multi-session work acknowledged - Carrier-check at 788d6ac retained; predicate-body builds on top Acceptance bullets updated: gt_zero with arg-shape validation + all 7 predicates with carrier+argument contract verification. Director-pinned brief-authoring discipline: dissolution trigger named in a brief authored as substrate-fact-introduction must specify same-slice acceptance, not deferred follow-up. Otherwise the brief encodes 'land bridge + defer dissolution' anti-pattern, the same shape feedback_construction_over_ratchets warns against at implementation layer applied at brief authoring layer. This is the 11th miss this session — but the pattern surfacing it (direct sweep + Codex REQUEST_CHANGES + Director ratification) produces structural fixes that prevent the next miss class. Brief revision is the structural recovery. * docs(briefs): Slice 2.5 brief — Path 2 RATIFIED (Gap 1 discharge + Q-Regex carve-out) Director Path 2 RATIFIED at gunbc#828 #issuecomment-4391985613 post Rung 3 STOP at gunbc#1746 #issuecomment-4391946213. Path 2 narrowing of Path (a): - gt_zero + range body synthesis already landed (5f3c40c) - Gap 1 discharge mechanism: extends scalar-literal predicate evaluation at lower-time. L-class; lands SAME SLICE per same-slice-dissolution discipline - non_empty + brand body synthesis: lands via Gap 1 discharge - pattern / format / content: fail-closed-with-named-dep at user-code authoring layer (Diagnostic::ResolveError naming Q-Regex-Primitive). NOT placeholder; structural rejection - PB-1 shim FULLY RETIRED per hard scope bar #3 - Q-Regex-Primitive carved as follow-on substrate-fact-introduction (XL substrate; not pre-authored per feedback_construction_over_ratchets) Acceptance updated: - 4 of 7 body-synthesized in same slice - 3 of 7 fail-closed-with-named-dep (no placeholder semantics) - Gap 1 discharge mechanism landed (no partial implementation) Worker disposition: proud-lynx-311 continues Rung 3 with Gap 1 implementation + non_empty/brand body synthesis + pattern/format/ content fail-closed-with-named-dep diagnostic. * docs(briefs): S9 Phase-1 Step 3 emission entries — Tier-1 brief authoring (1/5) First Tier-1 brief landing post Director auto-nudge on assignment #1858. Authored against Q-MachineConstraint sub-decisions ratified shape now landed at #1856: - Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>> → Rust i32/i64/i128 - UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>> → Rust u32/u64/u128 - 6 concrete instantiations (≥3 minimum per sub-decision 5) - Cross-program handoff to Grounding Mgr (#1745) for G2 emission consumption - numeric_construction_demonstration Acceptance bullet (Int<32> round-trip); Real<64> half deferred to S9 Phase 2 / S8 Float cascade Brief authored with substrate-state-grep discipline + same-slice- dissolution discipline (no deferred trigger; demonstration in same slice or explicitly carved). 5-question authority audit completed in brief body. Worker pin proud-lynx-311 (S9 holds); brief queues post-Slice-2.5 completion as natural follow-on. Tier-1 brief queue progress: - 1/5 authored (this brief) - 4 remaining: S9 Phase-2 Float coordination; T-LBP complexity cementing; T-LBP cost cementing; S3 Phase-2 parser-grammar * docs(briefs): S3 Phase-2 parser-grammar surface — Tier-1 brief 2/5 Authored per Director freed-pool pressure at gunbc#828 #issuecomment-4392095857 + Tier-1 brief-queue commitment. Brief covers parser-grammar surface for Compose<Algebra, MachineConstraint> interaction syntax post-#1856 (Phase-1 carrier slice landed): - Phase 2.1: Parser surface per ratified Q-MachineConstraint-Grammar-Shape - Phase 2.2: Bootstrap demonstrator with >=3 algebra*machine-axis pairs - Phase 2.3: Class 1 5-criteria Pass receipt (criteria 1-4; criterion 5 v2-oracle parity may carve) - Phase 2.4: numeric_construction_demonstration co-receipt with S9 Phase-1 Step 3 emission entries brief Director ratification HOLDS on grammar bikeshed: - Candidate A: Int @ Width<32> (annotation form) - Candidate B: Int with MachineWidth<32> (with-form) - Candidate C: Int<32> (positional generic — fits existing convention) - Candidate D: Compose<AbelianGroup, MachineWidth<32>> (no parser change; type aliases provide convenience names) Worker pin valiant-ant-72 holds (S3 Phase-1 precedent owner); freed-pool until Director ratifies grammar shape. Tier-1 brief queue progress: 2/5 authored (this brief + S9 Phase-1 Step 3). Remaining: S9 Phase-2 Float coordination; T-LBP complexity cementing; T-LBP cost cementing. * docs(briefs): S3 Phase-2 brief — fix internal contradiction on Class 1 closure scope Codex BLOCKING review on PR #1889 sha ec609b4 flagged real contradiction: - Provenance section claimed '5-criteria Pass closed' (line 229) - Phase 2.3 said 'criterion 5 may queue separately' (line 154) - STOP-AND-ESCALATE bullet 1 ratified 4-of-5 carve-out (line 181) Updated Provenance to consistently state criteria 1-4 close in this slice; criterion 5 (v2-oracle parity) cementing test may carve to separate slice. Single 4-of-5 framing throughout. INVARIANTS.md 'Documentation Describes Live State' satisfied. Per feedback_construction_over_ratchets: don't claim closure that isn't structurally complete. 4-of-5 + named carve is the honest shape; 5/5 only on cementing-test land. * briefs: cite Q-MachineConstraint sub-decision 6 supersession of T-V2-Retirement gate Address Codex BLOCKING #1 on PR #1889: S3 Phase-2 + S9 Phase-1 Step 3 worker briefs consumed Q-MachineConstraint ratification without reconciling the historical T-V2-Retirement-landing-first cascade gate. Add an explicit precondition-framing section to each brief citing Director ratification at gunbc#828 #issuecomment-4385530115 (Q-MachineConstraint sub-decision 6 UNIVERSAL substrate posture) as the implicit supersession; PR #1856 landing without T-V2-Retirement first confirms supersession in practice. STOP-AND-ESCALATE if supersession contested at dispatch. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * slice-2.5 brief: mark pattern/format/content Path 2 acceptance pending Director ratification Address Codex BLOCKING #2 on PR #1889: Path 2 deliverable + acceptance language asserted fail-closed-with-named-dep semantics for pattern/format/content, but Director rejected that shape (Option 4) at gunbc#828 #issuecomment-4392.... Options A (consumer migration), B (Q-Regex bundled into slice), C (typed fail-closed exception with named trigger) await ratification per migration-cost catalog at #issuecomment-4392116215. Brief now flags those three predicates + PB-1 shim retirement coupling as PENDING DIRECTOR RATIFICATION; worker STOP-AND-ESCALATE if dispatched pre-ratification. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * slice-2.5 brief: reconcile PB-1 retirement shape with Options A/B/C gating Address Codex BLOCKING #4 on PR #1889: prior text claimed "PB-1 FULLY RETIRED" while preceding bullet said retirement gates on chosen path — internal contradiction. Reframe to gate the pattern/format/content half of shim retirement on Options A/B/C ratification while preserving Director hard scope bar #3 binding on the gt_zero/range/non_empty/brand half (cleanly retired via Gap 1). No path leaves PB-1 alive as parallel-authority shim on slice landing. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * briefs: tier-1 3/5 — S9 Phase-2 Float emission coordination worker brief Tier-1 brief queue progress (3/5): S9 Phase-2 Float<N> emission entries (Float<32>/Float<64> via Compose<ApproximateField<Real>, MachineWidth<N>>) + Real<64> round-trip demonstration completing §1.8 #67 Acceptance bullet. Coordination/synthesis brief — gates on both S8 (ApproximateField<F> + Real base carrier) and S9 Phase-1 Step 3 (Int<N>/UInt<N> emission-entry pattern) landing before dispatch. Inherits T-V2-Retirement supersession framing from Phase-1 Step 3. STOP-AND-ESCALATE on missing preconditions or unratified Q-ApproximateField-Axiom-Set. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * briefs: ledger advancement honesty (PRODUCER_LANDED not CONSUMER_LANDED) + restore exact issuecomment IDs Address openai-pro BLOCKING + NON-BLOCKING findings on PR #1889: BLOCKING (boundary discipline / same-PR consumer): S9 Phase-1 Step 3 + Phase-2 Float briefs claimed DECLARED → CONSUMER_LANDED upon merge while only requiring cross-program handoff receipt to Grounding G2. For target-primitive substrate entries, handoff is not landed-consumer proof. Reframed both briefs to advance rows DECLARED → PRODUCER_LANDED; Grounding G2 follow-on PR (per-pair lowering rules + emitted-Rust- primitive verification) advances to CONSUMER_LANDED. Default split-PR producer-then-consumer per bundled-scope discipline ratified at gunbc#1739 #issuecomment-4392225548 (parallel infrastructure DISALLOWED in same PR). NON-BLOCKING (citation hygiene): replaced ellipsized "#issuecomment-4392..." stubs in slice-2.5 brief with the exact Director Option-4-rejection comment ID #issuecomment-4392081719 so the ratification claim is mechanically checkable. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * slice-2.5 brief: absorb Director Option A revised RATIFICATION (final shape) Director ratified Option A revised at gunbc#1739 #issuecomment-4392329709 (2026-05-06). Final brief shape: - 4 predicates body-synthesize via Gap 1 (gt_zero / range / non_empty / brand) - pattern / format / content STRUCTURALLY ABSENT from registry (not pending-with-placeholder); user-code where-clause produces Diagnostic::ResolveError naming Q-Regex-Primitive as separate substrate-fact-introduction follow-on - ~12 types.dag refined-type declarations migrate to drop pattern/format/content where-clauses with inline-comment receipts pointing to Q-Regex restoration path; 1 extdep (sts_endpoint) adapts as needed - PB-1 shim FULLY RETIRED — no exception, no preserved branch - Q-Regex-Primitive as separate brief / worker pin / Director ratification cycle when concretely needed (Option B XXL bundling REJECTED) Phase ordering rewritten to absorb migration as explicit phase between predicate enrollment and shim retirement. proud-lynx-311 re-dispatches against revised brief; PR #1846 absorbs additional scope per Director directive. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * s3-phase-2 brief: absorb Q-MC sub-decision 3 ratification (grammar shape was already answered) Director observation at gunbc#1739 #issuecomment-4392382517: Q-MC sub-decision 3 (Brian directive 2026-05-06 at gunbc#828 #issuecomment-4385530115) already ratified the grammar shape — `Int<N>` user surface desugars to `Compose<Int, MachineWidth<N>>` substrate parametrically. Earlier brief HOLD framing was redundant. Brief revised: - Replace bikeshed section with Q-MC sub-decision 3 quotation + surface/substrate framing (Candidate C user-facing; Candidate D substrate elaboration; A/B rejected per Director assessment) - Phase 2.1 parser surface scoped to numeric-literal-position recognition + parametric desugar - Front-matter status: dispatchable per ratification (not HOLD) - Provenance: ratification-state-grep discipline pin folded into standing dispatch-checklist valiant-ant-72 dispatch fires immediately per Director directive. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * briefs: remove truncated issuecomment-4392... stub from S9 Phase-1 Step 3 provenance Address cursor NON-BLOCKING finding on PR #1889 sha 94292e4: line 144 truncated issuecomment ID. The originally-intended Director auto-nudge comment doesn't map cleanly to a single ID, so replace with the standing assignment reference (gunbc#1858) which is unambiguous. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * briefs: tier-1 4/5 + 5/5 — T-LBP complexity + cost lens cementing-test worker briefs Tier-1 brief queue completes (4/5 + 5/5): 4/5 — T-LBP complexity-lens cementing test (§1.8 rows #79 + #87 first entry). Consumer-side cementing test against frozen v2-oracle snapshot; producer-side Lens<Complexity> instance is upstream precondition. Per Q-Lens-Behavioral-Parity-R3-Closeability option (b) RATIFIED at gunbc#828 #issuecomment-4385329180 — T-LBP narrowed to complexity + cost lenses only. 5/5 — T-LBP cost-lens cementing test + cost_lens_demonstration (§1.8 rows #80 + #70 + #87 second entry). Absorbs row #70 demo naturally — same source corpus + lens run; demonstration adds observable-cost-bound assertions on top of parity check. ≥2 algebra-instances composed + ≥1 recursive call required per row #70. Both briefs gate on: - T-LBP per-lens substrate-producer brief landing (Lens<C> instance) - Frozen v2-oracle snapshot capture (separate brief, pre-v2-retirement) - DO NOT capture-against-live-oracle (violates v2_oracle_no_remaining_test_consumers) - DO NOT bundle producer/snapshot/algebra-instance authoring per Director bundled-scope ratification at gunbc#1739 #issuecomment-4392225548 Tier-1 5/5 complete. Worker pins assigned at dispatch time once preconditions land. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * canvas: emission-provenance substrate shape disposition (supersedes PR #1902 PROPOSAL) Substrate Mgr canvas authored post-codex BLOCKING + PM disposition handoff at gunbc#846 #issuecomment-4392510255. Two real substrate-shape questions stacked: Q1 (category mismatch): Lens<C>.read is per-Behavior; emission provenance is per-emitted-line. 4 options surfaced — (a) per-Behavior Lens-compatible / (b) per-line instrumentation NOT a lens / (c) withdraw + re-canvas / (d) Mgr-surface both substrates distinct. Mgr recommends (d) if visualization is load-bearing; (a) suffices if per-Behavior aggregate works for the slide. Brian-channel sub-question surfaced. Q2 (fold-rule enumerability): T-Rule-Enumeration substrate-fact- introduction is prerequisite for every Q1 path (rust_target.rs uses inline &str template names; rules not enumerable). Mgr recommends parallel-dispatch with Q1 deliberation — satisfies Brian's ASAP framing while shape ratifies. Q3 (§1.8 ledger retarget): #89 already taken by T-LAS; new cluster location follows Q1 disposition. Director ratification ask + Brian-channel sub-question surfaced. PR #1902 close-superseded post-ratification. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * briefs: reconcile Compose<...> slot-1 spellings to algebraic-concept names per Q-MC sub-decision 3 Address cursor NON-BLOCKING on PR #1889 sha 2b49b30: S3 Phase-2 brief mixed Compose<AbelianGroup, ...> with Compose<Int, ...>; S9 Phase-1 Step 3 used Compose<AbelianGroup, ...> + Compose<CommutativeSemiring, ...>; S9 Phase-2 used Compose<ApproximateField<Real>, ...>. All wrong per Q-MC sub-decision 3 critical correction at gunbc#828 #issuecomment-4385530115: "prior phrasing Compose<AbelianGroup, MachineWidth<64>> was wrong — AbelianGroup<T> is a witness shape over carrier T, not a carrier constructor; bare AbelianGroup in slot-1 composes witness rather than concept." Slot-1 is the fully-applied algebraic-concept name: - Int<N> = Compose<Int, MachineWidth<N>> (Int = AbelianGroup<GroupCompletion<Nat>>) - UInt<N> = Compose<UInt, MachineWidth<N>> (UInt = CommutativeSemiring<Nat>) - Real<N> = Compose<Real, MachineWidth<N>> (Real = ApproximateField<Rational>) Phase-2 Float brief also retargets user-facing surface from Float<N> to Real<N> (Float is target-language name for the Rust primitive; Real is the algebraic-concept name per Q-MC sub-decision 3 example). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * s9-p1-s3 brief: scope Deliverable 4 to producer-side verification (defer demonstration to Grounding G2) Address Codex BLOCKING (line 82 inline + bundled review): brief claimed end-to-end Int<32> → Rust i32 round-trip in Acceptance while Deliverable 3 said lowering rules belong to Grounding G2 follow-on. Internal contradiction. Reframe per Director bundled-scope discipline (gunbc#1739 #issuecomment-4392225548 — parallel infrastructure DISALLOWED same-PR): - Deliverable 4: producer-side substrate-axis verification only (type-check + carrier acceptance + bootstrap snapshot/manifest hold). NO end-to-end Rust emission in this PR. - §1.8 #67 numeric_construction_demonstration receipts split: producer half here (entries exist), consumer half on Grounding G2 PR (full Int<32> → Rust i32 round-trip). - Same-PR cross-program co-author available if Substrate Mgr coordinates with Grounding Mgr at dispatch; default is split. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 7, 2026
…PB-1 shim retirement (#1846) * WIP: proud-lynx-311 * WIP: proud-lynx-311 * WIP: proud-lynx-311 * WIP: proud-lynx-311 * chore(v3): regen bootstrap snapshot for substrate.dag :83 anchor swap The :425 Slice A anchor swap (\§P1` → `#p1-modeling-faithfulness`) on src/v3/std/substrate.dag:83 shifted SourceSpan byte offsets for every node parsed below that line, drifting bootstrap_generated.rs. Regenerated via `cargo run -p v3-compiler --features bootstrap-regen-fresh --bin regen_bootstrap`. No semantic change — pure span-offset refresh. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(:425): retire trailing `§P5` shorthand on r2-manager-brief-authority-matrix.md:205 Cursor review on a402dca flagged residual `\`§P5\` (dispatch-discipline)` shorthand inside the same sentence as a P2 anchor. Slice A's mechanical sweep matched only `INVARIANTS.md §P5` prefixed forms; the bare shorthand survived. Swap to `INVARIANTS.md#p5-progress-is-dissolution` matches adjacent P2 form. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test(v3): refresh parse_corpus_manifest after substrate.dag :83 anchor swap The :425 Slice A anchor swap shifted byte offsets in src/v3/std/substrate.dag, drifting SG-2 parse snapshot hashes (same root cause as the bootstrap_generated regen in a402dca). Regenerated via: cargo test -p v3-compiler --test integration -- --ignored \ refresh_handwritten_parse_snapshot_manifest Verified: handwritten_parse_snapshot_matches_manifest passes locally. Pure offset/hash refresh — no semantic change. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * ci: re-trigger v3 after t_demo_fixture transient flake * fix(:425): align link URL with anchor label on r2-manager-brief-authority-matrix.md:222 Codex review on 9e67e1f noted the Markdown label was swapped to the #p5-progress-is-dissolution anchor but the URL still pointed at the INVARIANTS.md root. Aligned URL to match label. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * feat(std): T-Numeric-Construction S9 Phase-1 — UInt = CommutativeMonoid<Nat> Migrates the default `UInt` alias from the legacy fixed-width `UInt = UInt64` to the abstract algebra carrier `UInt = CommutativeMonoid<Nat>`, consistent with the landed Slice 3 form `Int = AbelianGroup<GroupCompletion<Nat>>` (PR #1466, docs/audit/t-numeric-construction-group-completion-6q.md Q6 single-authority resolution). Per S9 worker brief (R3 design schedule §1 S9, ratified 2026-05-06): Nat plays as the commutative-monoid carrier under addition (closure / associativity / identity 0 / commutativity; no inverses, since Nat lacks additive inverses by construction). UInt as `CommutativeMonoid<Nat>` is the additive structure on that carrier — the honest unsigned-integer algebra without inverses. Fixed-width rows `UInt8..UInt128` remain on `Semiring<Word*>` per existing legacy-rows policy in `dsl/std/integer.dag` (only the default alias migrates; storage rows stay shape-conflated until v3 supports trait/witness syntax separation). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(std): UInt = Nat (preserve CommutativeSemiring surface) Codex review at PR #1818 / Substrate Mgr inline review at dsl/std/integer.dag:102 surface a P2 facts-flow-forward violation in the prior `type UInt = CommutativeMonoid<Nat>` pivot: wrapping Nat (already `CommutativeSemiring<Magnitude>`) in `CommutativeMonoid<Nat>` projects out the multiplicative monoid + identity 1, dropping substrate facts already established at the Nat layer. The substrate cannot structurally name "the additive monoid projection of Nat" without losing Nat's other monoid. Pivot to the design-doc-named direct alias: `type UInt = Nat`. Per the existing comment header in this file, the design doc states `UInt IS Nat`; Nat already carries the full unsigned-integer algebra surface (additive monoid + multiplicative monoid + distribution + commutativity on both). No new operations are needed — the alias IS the algebra. Asymmetry with `Int`: `Int` requires additive inverses Nat lacks, so `GroupCompletion<Nat>` is the honest carrier introduction. `UInt` requires no new operations, so no carrier introduction is needed. Updates the PR's deferred Phase-1 Step 3 receipt to the `Compose<CommutativeSemiring, MachineWidth<N>>` parametric form (post-S3 landing). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test(v3): refresh parse-snapshot manifest for integer.dag UInt change Mechanical refresh after PR #1818 fix commit. integer.dag debug-bytes shifted 12659→12662, fnv1a64 1eef7086b17908de→104a413082264303 with the type swap from `UInt = CommutativeMonoid<Nat>` to `UInt = Nat` plus surrounding doc-comment rewrite. Generated by: cargo test -p v3-compiler refresh_handwritten_parse_snapshot_manifest -- --ignored Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * test(v3): refresh parse-snapshot manifest for Slice 2 types.dag edit Mechanical refresh after the Slice 2 Nat-alignment migration (`NonNegativeInt` / `PositiveInt` carrier swap to `Nat`). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * fix(std): partial Slice 2 — NonNegativeInt only; PositiveInt stays Int CI failure on PR #1840 surfaced two real substrate gaps in the attempted full Slice 2 migration: 1. Adding `import std.nat { Nat }` to `dsl/std/types.dag` triggers a circular-dependency false-positive in the resolver during transitive load of test fixtures via `std.render` chains (verified via p0_std_render_repeat_string_test). 2. Shipping `type PositiveInt = Nat where range(min: 1)` triggers `Diagnostic::ResolveError` for the `range` predicate — the resolver does not currently entail `where range(min: …)` over `Nat` (only over `Int` and Int-row carriers). This matches dispatch STOP-AND-ESCALATE bullet 2 (predicate-entailment failure). Partial migration shipped: - `NonNegativeInt = Nat` declaration moved to `dsl/std/integer.dag` (which already imports `std.nat { Nat }`); structural alias, no predicate, resolves cleanly. - `PositiveInt = Int where range(min: 1)` stays in `dsl/std/types.dag` unchanged. PositiveInt's Nat-alignment migration sequences after the predicate-entailment substrate supports `range` over Nat. `src/v3/std/approximate_field.dag` keeps `import std.types { PositiveInt }` (unchanged consumer surface for PositiveInt). Bootstrap snapshots re-regenerated; parse-corpus manifest refreshed. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * feat(v3): T-Numeric-Construction S9 Slice 2.5 — predicate registry + PB-1 shim retirement Path 3 RATIFIED at gunbc#828 issuecomment-4390333451 (Director, post proud-lynx-311 pre-flight DFS at gunbc#1746 issuecomment-4390253616). Five deliverables, single PR per Director ratification: 1. `gt_zero` primitive predicate enrolled in registry (Nat-defined per Director Option 2 ratification at gunbc#828 issuecomment-4390199218). 2. `KNOWN_PREDICATE_REGISTRY` substrate-fact-introduction in src/v3/compiler/src/lower.rs — authoritative set of predicate identifiers recognized by `where`-sugar refinements (range, pattern, non_empty, brand, format, content, gt_zero). 3. PB-1 shim retired: prior file-path special case (`span.file == "dsl/std/types.dag"`) in `lower_type_alias_refinements_phase` replaced with registry-name dispatch via `all_predicate_names_registered`. Per feedback_no_textual_enforcement_bridges, gating structural behavior on textual file identifier was the anti-pattern; registry-name dispatch is the substrate-faithful authority. 4. Existing types.dag predicates (range, pattern, non_empty, brand, format, content) all migrate cleanly through the registry; bootstrap snapshot + parse-corpus manifest hold (no semantic drift). 5. PositiveInt = Nat where gt_zero migrated to dsl/std/integer.dag (companion to NonNegativeInt = Nat from #1840). approximate_field.dag consumer updated to import from std.integer. `alloc_deferred_types_dag_refinement_placeholder` renamed to `alloc_registered_refinement_placeholder` (file-name semantic dropped from placeholder name string; `is_deferred_refinement_placeholder` matcher updated to track the new placeholder-name prefix "<registered predicate, body not lowered: ...>"). Bootstrap snapshots regenerated; parse-corpus manifest refreshed. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * fix(v3): apply cargo fmt to placeholder format! macro Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(v3): Slice 2.5 — predicate registry carries allowed-carrier check Codex BLOCKING review at PR #1846 (sha 92758dd): the registry as shipped accepted any registered predicate name as a placeholder regardless of base-type carrier. That made the registry a broad diagnostic escape hatch — e.g. `String where gt_zero` would silently get a placeholder despite gt_zero being undefined over String. Extends the registry from `&[&str]` to `&[PredicateSpec]` where each spec pairs the predicate name with its **allowed carriers** list. The registered-placeholder branch now requires both: 1. predicate name in registry, AND 2. base-type alias chain intersects the predicate's allowed carriers. `carrier_chain_names()` walks `Atom::ResolvedByStructure` / `Atom::ResolvedByName` / `Instantiation` chains to collect every named declaration in the chain — so e.g. `brand` over `NonEmptyStr` matches because `NonEmptyStr` itself is in `brand`'s allowed list (without forcing the chain to bottom out at `String`). Allowed-carrier lists derived from existing dsl/std/types.dag surface usage: - range: Int, Nat - pattern: String, NonEmptyStr - non_empty: String, Secret, List - brand: NonEmptyStr, String, FilePath, Unit, Int, Nat, Secret - format: String - content: FilePath - gt_zero: Nat, Int Bootstrap snapshot regenerates cleanly; 0 ResolveErrors emitted. Failing test (test_types_dag_alias_refinement_is_deferred_placeholder_not_dropped) still passes (the test fixture uses `range(min: 1)` over Int — a registered+allowed pair). p0_std_render_repeat_string_test still passes. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * feat(v3): Slice 2.5 Path (a) — predicate registry arg-shape contract + tests Director Path (a) RATIFIED at gunbc#828 issuecomment-4390760353 (post Codex/openai-pro REQUEST_CHANGES on PR #1846): registered predicates must validate carrier AND argument shape; mismatches fail-closed via Diagnostic::ResolveError rather than silently take a placeholder. Ladder rungs 1+2+6 land here: Rung 1 — `PredicateSpec` arg-shape contract: - New `PredicateArgShape` enum: Bare / SingleStringLiteral / SingleIdentOrStringLiteral / SingleVariantIdent{allowed} / NumericRecord{accepted_fields, require_at_least_one}. - `PredicateSpec` extended with `arg_shape` field. - Per-predicate arg shape filled from `dsl/std/types.dag` surface usage: - gt_zero / non_empty: Bare - pattern / brand: SingleStringLiteral - format: SingleIdentOrStringLiteral - content: SingleVariantIdent{Text|Binary} - range: NumericRecord{min, max; require ≥1} Rung 2 — Arg-shape validation at lower-time: - `validate_predicate_clause` returns `PredicateValidation::{Registered, RegisteredButMalformed{reason, span}, NotRegistered}`. - `validate_where_clause` recurses into `Operator{And, …}` chains; any malformed leaf collapses the whole clause to malformed. - `arg_shape_mismatch` validates per-shape: bare predicates must parse as Var or zero-arg Call; single-arg predicates check literal kind + variant membership; record-shape predicates validate field names + integer-literal field values. - Replaces `all_predicate_names_registered` (carrier-only) with the new validator at the call site in `lower_type_alias_refinements_phase`. Malformed registered predicates emit `Diagnostic::ResolveError` directly via `report_declaration_error` and skip refinement allocation; valid registered predicates still take a placeholder (rung 3 body lowering is the named follow-on). Rung 6 — Test coverage (per openai-pro testing finding): - `test_registered_predicate_placeholder_works_outside_types_dag` cements the PB-1 file-path retirement (registered predicate in `dsl/std/integer.dag` still takes a placeholder). - `test_registered_predicate_on_disallowed_carrier_fails_closed` cements the carrier check (`String where gt_zero` surfaces diagnostic, no placeholder). - `test_registered_predicate_with_malformed_args_fails_closed` cements the arg-shape check (`Int where range(bogus: 1)` surfaces diagnostic, no placeholder). Naming-drift cleanup (per openai-pro non-blocking observation): doc-comment reference to `KNOWN_PREDICATE_REGISTRY` updated to the actual constant name `KNOWN_PREDICATES`. Bootstrap snapshots regenerated; parse-corpus manifest refreshed (declaration counts unchanged; placeholder shape preserved for valid registered predicates). Rung 3 (predicate-body lowering) NOT in this commit — registry placeholder still allocated for valid registered predicates pending synthesis of per-predicate Bool-returning body expressions (gt_zero → x > 0; range(min: m) → x >= m; etc.). That's the next ladder rung; this commit lands the fail-closed enforcement that addresses Codex/openai-pro REQUEST_CHANGES on the malformed-call path. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(v3): Slice 2.5 — sweep KNOWN_PREDICATE_REGISTRY naming drift; fix doc placement Cursor APPROVE_WITH_COMMENTS at PR #1846 (sha 75e3572): - `lower.rs:1191` rustdoc cite `[KNOWN_PREDICATE_REGISTRY]` should be `[KNOWN_PREDICATES]` — done (now reads as the registry's actual name). - `integer.dag:134` text reference fixed. - The long "predicate registry / substrate-fact-introduction" doc-block was attached to `PredicateArgShape`; moved to immediately above `KNOWN_PREDICATES` where it belongs. `PredicateArgShape` keeps a short doc that points readers at the registry doc. - `RegisteredButMalformed` doc fix: was "falls through to live resolution; diagnostic is the caller's responsibility"; behavior is actually "caller emits Diagnostic::ResolveError directly via report_declaration_error and skips refinement allocation (fail-closed; no placeholder, no live-resolution fallback)". Bootstrap snapshots regenerated. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(v3): Slice 2.5 Path (a) Rung 3 — gt_zero body synthesis lowers real predicate Director Path (a) requirement: "actual refinement body lowering required" for valid registered predicates (no placeholder semantics). This commit lands rung 3 for `gt_zero` (the new primitive) — the simplest tractable case. Body synthesis pattern: - `gt_zero` over Nat or Int → `subject > 0` (Comparison(Gt) operator applied to Var{bind_name} and Literal{Int(0)}). Equivalent for both carriers — Nat's structural lower bound is 0; gt_zero is "above structural bound." Implementation: - `synthesize_predicate_body(spec, predicate, bind_name) -> Option<SurfaceExpr>` — per-predicate body synthesizer; returns None if substrate not yet capable (e.g. pattern needs regex primitives that haven't landed). - `registered_predicate_synthesized_body` walks Operator{And,...} chains; returns Some(combined) only if every leaf is synthesizable. - Call site in `lower_type_alias_refinements_phase` Registered branch: if synthesizable, call `build_refinement_predicate_declaration` with the synthesized expr (real Arrow refinement body); else fall through to placeholder (current behavior for non-synthesized predicates). Pending substrate (still placeholder for now): - `range`: numeric comparison over min/max args — tractable next; body is `subject >= min && subject <= max`. Synthesis pending. - `non_empty`: requires substrate-uniform `length`/`size` over String / Secret / List carriers; tractable but multi-carrier shape work. - `brand`: structural-only (true for any subject); body is `Bool::True` — synthesizable trivially but semantically a no-op refinement. - `format`: format-specific (uuid → matches uuid regex, etc.) — needs regex primitives. - `content`: content-specific — needs content-typing primitives. - `pattern`: needs regex-matching primitive — STOP-AND-ESCALATE candidate per S9 Slice 2.5 brief STOP bullet 1. Rung 3 sequencing: `gt_zero` first (this commit), then `range` next (numeric primitives in scope), then surface STOP for the regex-needing predicates. Test coverage: - `test_gt_zero_lowers_to_real_refinement_body_not_placeholder` — `type Pos = Nat where gt_zero` produces a real `Declaration::refinement` (no `<registered predicate not lowered: …>` label). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(v3): Slice 2.5 — reject duplicate record-shape fields in registered predicates Codex BLOCKING inline at PR #1846 src/v3/compiler/src/lower.rs:894: NumericRecord arg-shape validation accepted duplicate named fields, so `range(min: 1, min: 2)` would still take a placeholder rather than fail-closed. Tracks seen field names during validation and emits `Diagnostic::ResolveError` for duplicates. Cementing test added: `test_registered_predicate_with_duplicate_record_field_fails_closed`. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * feat(v3): Slice 2.5 Path (a) Rung 3 — range body synthesis lowers real predicate Director continuation at gunbc#828 #issuecomment-4391621707: bundled PR per Path 3 RATIFIED; continue Rung 3 ladder; STOP for regex-needing predicates when concretely needed. Adds `range` body synthesis. Lowering rules: - range(min: m) → subject >= m - range(max: M) → subject <= M - range(min: m, max: M) → subject >= m && subject <= M Implementation: extracts integer-literal field values from the arg-shape-validated SurfaceExpr::Record arg, builds Comparison(Ge) and/or Comparison(Le) clauses, combines via Logical(And) when both bounds present. Test coverage: - `test_range_lowers_to_real_refinement_body_not_placeholder` — exercises all 3 forms (`min:`, `max:`, both); each must produce a real `Declaration::refinement` with no placeholder name marker. - Updated `test_types_dag_alias_refinement_is_deferred_placeholder_not_dropped` and `test_registered_predicate_placeholder_works_outside_types_dag` to use `non_empty` (still pending body synthesis) so the placeholder branch is exercised independently of Rung 3 progress; otherwise these would regress as more predicates pick up real bodies. Remaining Rung 3 ladder: - non_empty / brand: small substrate work next. - pattern / format / content: STOP-AND-ESCALATE candidates per S9 Slice 2.5 brief; require regex-primitive substrate-fact-introduction (Q-Regex-Primitive) that has not landed. Will surface concrete STOP when reached. Bootstrap snapshots regenerated; parse-corpus manifest refreshed. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(v3): Slice 2.5 — Bare predicate rejects zero-arg-call form Codex BLOCKING inline at PR #1846 src/v3/compiler/src/lower.rs:895: Bare predicate validation conflated "no arguments" with "no argument list" — `gt_zero()` (zero-arg Call) was accepted as Bare alongside the canonical bare-ident `gt_zero` (Var) form. Tighten Bare arg-shape contract to accept ONLY `SurfaceExpr::Var`. Explicit empty call list `gt_zero()` now produces fail-closed `Diagnostic::ResolveError` with message naming the canonical form. Cementing test added: `test_bare_predicate_with_empty_call_args_fails_closed`. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * chore(v3): Slice 2.5 — defer non_empty/brand body synthesis (discharge cascade) Attempted non_empty + brand body synthesis exposed a discharge cascade in scalar_literal_requires_refinement_discharge: any non-placeholder refinement is treated as needing static discharge, but the existing infrastructure cannot evaluate `subject != ""` (non_empty) or `true` (brand) against scalar data literals. Result: bootstrap regen emits 18+ spurious "scalar literal does not satisfy `where` refinement" diagnostics for data declarations whose fields are typed as branded/non-empty carriers (e.g. `LensRegistryEntry.name: NonEmptyStr` matches `name: "lower_helpers"` literal, which DOES satisfy the predicate but the lowerer can't evaluate that statically yet). Reverts the non_empty + brand body synthesis from this commit; both predicates fall through to placeholder until the discharge mechanism gets the ability to evaluate trivial-true and string-non-empty predicates against scalar literals at lower-time. Threads `_carrier_chain` parameter through `synthesize_predicate_body` for forward compatibility — the parameter is unused now but will be needed when non_empty / brand discharge lands. Doc-comment updated to surface the cascade as STOP-AND-ESCALATE candidate alongside the regex-primitive gap. Will compose a unified STOP packet to Substrate Mgr / Director surfacing both blockers together as the boundary for further Rung 3 progress. Tests: all 8 cementing tests pass. Bootstrap snapshots regenerated (0 ResolveErrors emitted). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * fix(v3): Slice 2.5 — disable range body synthesis (int-routing cascade) CI v3 failure on `5f3c40c3a` exposed: `range` body synthesis breaks `int_literal_cardinality_test::let_annotated_uint8_*` tests. With synthesis enabled, `let x: UInt8 = 5` resolves to `Int` instead of `UInt8` — the synthesized refinement bodies for the existing types.dag Int-where-range declarations (RetryCount, HttpStatus, Port, EpochMs, Duration, Milliseconds, Seconds) cascade through the `integer_routing_witness_walk` path used by int-literal narrowing. Bisected: `gt_zero` synthesis alone passes all tests; adding `range` synthesis breaks UInt8 narrowing. Disables `range` body synthesis (literal `if false` guard) pending investigation of the cascade path. `gt_zero` body synthesis stands. The cementing test for range body lowering is replaced with a placeholder-cementing test (range still validates carrier + arg shape; body synthesis is just deferred). Surfacing as STOP-AND-ESCALATE substrate-extension question per S9 Slice 2.5 brief. Integer-routing needs to be insensitive to refinement-body declarations on Int-derived aliases before `range` body synthesis can re-enable. All 11 cementing + UInt8-narrowing tests pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(v3): Slice 2.5 Option A revised — pattern/format/content removed from registry Director Option A revised RATIFIED at gunbc#828 issuecomment-4392245968 (post Substrate Mgr 4-sub-reading framing at gunbc#1739 issuecomment-4392050708 + concrete sts_endpoint regression case). Pattern / format / content predicates are STRUCTURALLY ABSENT from the predicate registry — not pending-with-placeholder, not named-but-not-checked, not deferred-completion-bridge. User-code `where pattern(...)` / `where format(...)` / `where content(...)` falls through to live resolution and surfaces `Diagnostic::ResolveError` naming Q-Regex-Primitive as the unblock substrate-fact-introduction follow-on. Per `feedback_construction_over_ratchets`: Q-Regex-Primitive needs its own substrate-fact-introduction brief + worker pin + Director ratification cycle, not bolted onto Slice 2.5. Per `feedback_modeling_inversion_and_paydown_flow`: types.dag's existing pattern/format/content refinements were placeholder-only (PB-1 shim absorbed them; no actual content check fired). Removing the where-clauses is the honest disposition — admitting the type ratchet wasn't structurally enforcing anything anyway. ### Changes `src/v3/compiler/src/lower.rs`: - `KNOWN_PREDICATES` registry trimmed: pattern/format/content removed. Remaining: range / non_empty / brand / gt_zero (4 predicates). - `PredicateArgShape` enum: `SingleIdentOrStringLiteral` and `SingleVariantIdent` variants removed (only used by format/content). `dsl/std/types.dag` — 11 declarations migrated: - `CommitSha`, `Sha256`, `Email`, `Url`, `SemVer`, `Timestamp`, `ServiceAccountEmail`, `MimeType`: dropped `where pattern(...)`. - `GistId`: dropped `where format(uuid)`. - `TextFilePath`, `BinaryFilePath`: dropped `where content(...)`. - `GcpProjectId`: dropped pattern half of combined predicate; kept non_empty (registry-tracked, fail-closed for malformed args). Each migrated declaration carries an inline comment receipt with the original where-clause text for restoration when Q-Regex-Primitive lands. `sts_endpoint: Url = "https://..."` (extdeps/cloud/gcp/sts.dag:53): no separate migration needed — Url's where-clause drop means String is the underlying carrier; scalar literal compiles. PB-1 shim FULLY RETIRED (Director hard scope bar #3 holds). No file-path-special-case branch. Bootstrap snapshots regenerated (0 ResolveErrors). Parse-corpus manifest refreshed. fmt clean; clippy clean. 13 cementing tests pass (incl. the previously-broken UInt8 narrowing tests). Remaining work (separate from this commit, still in slice scope): - Gap 1 discharge mechanism for non_empty / brand body synthesis - Gap 3 int-routing cascade fix for range body synthesis - non_empty / brand / range bodies synthesize once gaps land Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test(v3): Slice 2.5 — cement pattern/format/content registry-absence fail-closed Per Substrate Mgr Phase 9 (gunbc#1739 issuecomment-4392408663): add a behavior test that locks the user-code fail-closed semantics for pattern/format/content where-clauses. After Option A revised (Director RATIFIED at gunbc#828 issuecomment-4392245968), these predicates are STRUCTURALLY ABSENT from KNOWN_PREDICATES; any where-clause using them must surface a Diagnostic when authored. Test exercises 3 user-code scenarios: - `type P = Int where pattern("^x")` — registered name absent - `type F = Int where format(uuid)` — registered name absent - `type C = Int where content(Text)` — registered name absent Each must produce at least one diagnostic (cached_compile_any for diagnostic readback). Note on test fixture choice: the v3 parser accepts `where` after some bare-RHS-aliasable carriers (`Int`) but rejects it after others (`String`, `FilePath`) at parse time, independent of predicate-registry semantics. Test uses `Int` as the base so the parse path doesn't pre-empt the resolve-time registry check that this test is locking. 12 tests pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(v3): Slice 2.5 — fix stale lower_type_alias_refinements_phase rustdoc Cursor APPROVE_WITH_COMMENTS at PR #1846 (`a7a2e3ca`): the rustdoc on `lower_type_alias_refinements_phase` still listed `pattern` and `format` as registered names that lower to placeholders, and over-generalized "placeholder" for ALL registered names (even `gt_zero` which now synthesizes a real body). Rewrites the doc to match landed behavior post Option A revised: - Three-way dispatch (Registered+synthesizable / Registered+placeholder / Registered-but-malformed / Unregistered) - Pattern/format/content explicitly STRUCTURALLY ABSENT, not registered - gt_zero gets real body via synthesis; range/non_empty/brand still placeholder pending Gap 1 / Gap 3 - Dissolution scope updated to name Gap 1 (discharge) + Gap 3 (routing) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(v3): Slice 2.5 — update pre-Option-A test to current registry surface CI v3 failure on `3cd5eb47e`: existing test `test_db11_type_alias_where_accepts_types_dag_constraint_spellings` exercised `type S = String where non_empty, pattern("x")` — the combined `non_empty, pattern(...)` spelling no longer compiles cleanly because `pattern` is structurally absent from the registry post Option A revised. Reduces the test corpus to the alias-friendly `Int where range(min: 1)` shape (the parse-coverage shape this test was originally locking; String-base where-clauses hit a parser quirk independent of registry semantics, and the predicate-specific cement tests below cover the full surface). Comment explains the migration receipt. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * Merge remote-tracking branch 'origin/main' into session/proud-lynx-311 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(v3): Slice 2.5 Path (a) Rung 3 — brand body synthesis lowers real predicate Director Path 2 ratification at gunbc#828 issuecomment-4391985613 lists `brand` body synthesis among the 4 predicates that must body-synthesize in this slice. Lowering rule: brand("X") → subject == subject (reflexive Comparison(Eq)) Reflexive equality is semantically `true` for every carrier value (the brand predicate imposes no value-level constraint — branded types are nominal-only). The body REFERENCES the bind subject so the `unused_parameters` lens self-analysis test doesn't flag the synthesized refinement param as unused. Plain `Literal(Bool(true))` would discharge correctly but would surface as 14 unused-param findings (one per branded type in dsl/std/types.dag). Substrate-state grep confirmed no scalar `data` declarations use branded carriers (PathSegment / IntentId / IssueId / etc. are field types in record types, never scalar literal bases) — Gap 1 discharge cascade does not fire for branded types in the existing corpus. Test coverage: - `test_brand_lowers_to_real_refinement_body_not_placeholder` — `type B = NonEmptyStr where brand("B")` produces a real `Declaration::refinement` (no `<registered predicate not lowered: …>` label). Result: 2 of 4 in-scope predicates body-synthesized (gt_zero + brand). non_empty / range still pending (Gap 1 discharge mechanism + Gap 3 int-routing cascade). All 835 v3 integration tests pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * WIP: proud-lynx-311 * fix(v3): brand cement test uses Int carrier (parser quirk on String where) The parser rejects `where` after String/NonEmptyStr/FilePath bare-RHS aliases independent of registry semantics. Test uses `Int where brand("B")` instead — Int is in brand's allowed-carriers list and parses cleanly. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(v3): Slice 2.5 — mixed-clause fail-closed; revert brand body synthesis openai-pro REQUEST_CHANGES at PR #1846 (`79c4da09`): a `where`-clause that combines a body-synthesized predicate (e.g. `gt_zero`) with a placeholder-only predicate (e.g. `range`) routed both into a single generic placeholder, silently weakening the synthesized fact. P3 fail-closed / facts-flow-forward violation. Restructures `registered_predicate_synthesized_body` to return a `BodySynthOutcome` (AllSynthesized / AllPlaceholder / Mixed) per clause. The call site in `lower_type_alias_refinements_phase` matches on the outcome: - AllSynthesized: live predicate declaration via build_refinement_predicate_declaration. - AllPlaceholder: alloc_registered_refinement_placeholder (status quo). - Mixed: emit Diagnostic::ResolveError naming the predicates so the user knows which combination needs separation; the synthesized fact is NOT silently merged into a placeholder. Cementing test: `test_mixed_synthesizable_and_placeholder_predicates_fails_closed` — `type M = Nat where gt_zero, range(max: 10)` surfaces a diagnostic. ### Brand body synthesis reverted Brand body synthesis (landed at `c999493cb` as `subject == subject`) is reverted to placeholder for this slice because it caused the existing types.dag declarations: Milliseconds = Int where range(min: 0), brand("Milliseconds") Seconds = Int where range(min: 0), brand("Seconds") to mix synthesizable brand with placeholder-only range, tripping the new Mixed-fail-closed path. Brand re-enables when (a) per-leaf refinement representation lands so mixed clauses preserve each fact, OR (b) range body synthesis lands and the And combination is uniformly synthesizable. Updated test: `test_brand_takes_placeholder_pending_mixed_clause_fix` (replaces `test_brand_lowers_to_real_refinement_body_not_placeholder`). Bootstrap snapshots regenerated. 17 cementing tests pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(v3): Slice 2.5 — Practice 4 enum classification + gt_zero body-shape assertion openai-pro REQUEST_CHANGES at PR #1846 (`e1ec3d1a`): 1. Adds 🟢 TERMINAL classification per `docs/modeling-discipline.md` Practice 4 to the three new lowerer-internal coproducts: - `PredicateArgShape` (parsed-arg shape contract per registered predicate) - `PredicateValidation` (registered-validation outcome) - `BodySynthOutcome` (per-clause body-synthesis outcome) Each is anchored as an implementation-layer terminal (not a scaffold to dissolve). Variant set grows when a new predicate enrolls or when the registry's structural shape extends, but no variant exists in "scaffold-pending-dissolution" status. 2. Strengthens `test_gt_zero_lowers_to_real_refinement_body_not_placeholder` to assert the actual published body shape, not just "non-placeholder": - Refinement is `Arrow`-shaped with `ArrowBody::UserDefined(bind_id)`. - Bind's value port is produced by a `Transform` with target `Operator(Comparison(Gt))`. - One of the Gt operator's input ports is fed by a `Value` node with `LiteralBits::Int(0)`. A regression that lowered `gt_zero` to `subject == subject`, `Bool(true)`, or any other real-but-wrong body would have satisfied the prior test; this one fails-closed on body-shape divergence — `PositiveInt`'s substrate contract requires exactly `> 0`. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(v3): Slice 2.5 — range body synthesis re-enabled (Gap 3 resolved) Gap 3 root cause was test fixture brittleness, not a substrate issue. `int_literal_cardinality_test::let_annotated_uint8_*` used `find_map(Value(LiteralBits::Int(N)))` to find user literals — once range body synthesis lit up, types.dag declarations like `RetryCount = Int where range(min: 1, max: 5)` produced additional `Value(Int(5))` nodes that the test matched first. Resolution: filter the literal lookup by `span.file` to find the user's literal. Also migrates `Milliseconds` / `Seconds` in `dsl/std/types.dag` to drop the `brand("…")` half of their combined `range(min: 0), brand(…)` clause. Brand body synthesis is held at placeholder pending per-leaf refinement representation (separate substrate work); leaving the combined clause active triggers the openai-pro mixed-fail-closed diagnostic. Inline restoration receipt names the per-leaf-refinement follow-on as the trigger to restore brand half. Mixed-clause cementing test (`test_mixed_synthesizable_and_placeholder_predicates_fails_closed`) updated to use `gt_zero, brand(...)` — both registered, both allowed over Nat, brand still placeholder — so the Mixed path is exercised with a still-realistic combination. range cementing test renamed `test_range_takes_placeholder_pending_routing_fix` → `test_range_lowers_to_real_refinement_body_not_placeholder`. Test shape unchanged, asserts non-placeholder for all 3 range arities (min-only, max-only, both). `assert_int_value_port_resolves_to_uint8_in_file` helper added with optional `span_file` filter; existing call sites preserve prior behavior; the in-range narrowing test passes the filter. Result: 3 of 4 in-scope predicates body-synthesize (gt_zero / range + new) — brand stays at placeholder until per-leaf rep lands. 837 v3 integration tests pass; clippy clean. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(v3): Slice 2.5 — nested-And aggregation collects per-leaf names recursively openai-pro REQUEST_CHANGES at PR #1846 (`4a42cb1f`): `registered_predicate_synthesized_body` aggregated per-clause leaf names via `leaf_predicate_name(arg)` directly — but that returns `None` for an `Operator(And)` arg (the subtree is not itself a leaf). Net effect: a clause like `gt_zero, gt_zero, brand("X")` parsing as `(gt_zero, gt_zero), brand("X")` would aggregate only `brand` as an unsynthesized leaf name; `synthesized_names` stayed empty even though the inner subtree returned `AllSynthesized`. Outer aggregation then classified the whole clause as `AllPlaceholder` instead of `Mixed`, silently collapsing the nested gt_zero-derived facts into a single placeholder. P3 fail-closed / facts-flow-forward violation. Fix: `collect_leaf_predicate_names(expr)` recurses through `And` subtrees so every leaf name reaches the outer aggregator. Replaces the call to `leaf_predicate_name(arg)` at the AllSynthesized / AllPlaceholder branches. Cementing test: `test_nested_and_with_placeholder_leaf_fails_closed` — `type N = Nat where gt_zero, gt_zero, brand("X")` surfaces a Mixed diagnostic and does not collapse into a placeholder. 10 cementing tests pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 7, 2026
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* fix(briefs): retire INVARIANTS.md line-number cites in r3-426 brief
Per codex review on PR #1782 (sha c2512cde): the new briefs
introduced fresh `INVARIANTS.md:<line>` citations — the same
stale-line-anchor shape the :425 sweep brief itself names as needing
dissolution. Replaces four line-number references in
r3-426-sfp-subdoc-rename-and-author-worker.md (lines 16, 28, 58, 85)
with prose-anchored locators citing the section anchor + bullet text.
The other PB-lane briefs codex flagged (r3-pb-runtime-equivalence-
corpus-seed-audit.md, r3-pb-t-fixedpoint-worker.md) belong to
neat-bear-351's lane and are not in this PR's scope; will route
separately.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): correct L6 type count + r3-435 dual-path retirement receipt
Per openai-pro review on PR #1782 (sha c2512cde):
1. L6 brief said acceptance ratchet asserts "five types" exist but
declares six (ShapeATarget, FormAxis, BehaviorAxis,
MethodTemplateContractKey, EmissionCell, EmissionPathProjection).
Updated both occurrences to enumerate all six explicitly.
2. r3-435 brief authorized both Path A (typed-AST reader, deletes
IntegrationRsScan) and Path B (scanner widening with Char state),
but retirement receipt language only covered Path B. Updated
ROADMAP/ledger retirement and authority-audit step 5 to cover
both authorized dissolution paths.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): L6 second 'five types' site + T-E-P lookup-gate authority
Per cursor review on PR #1782 (sha 2072b986):
1. L6 brief acceptance bullet at :131 still said "five types" while
the parenthetical and the Slice step at :117 said six. The prior
fix only caught one of two sites; corrected the second.
2. T-E-P brief `e_p_call_pattern_lookup_authoritative` acceptance
named only `lower_call_pattern` as "the only route from call
site to LoweringTarget" — but Phase 2 explicitly requires
lenses route through `per_call_pattern_at` (the L-7 single-
authority lookup), NOT `lower_call_pattern` (compiler-internal).
Acceptance text rewritten to name `per_call_pattern_at` as the
consumer-facing gate and clarify `lower_call_pattern`'s
internal-only scope.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): strengthen L6 ratchet from count-parity to per-row bijection
Per openai-pro REQUEST_CHANGES on PR #1782 (sha 2072b986): the L6
acceptance ratchet asserted only row-count parity between
emission_path_projections and the union of MethodTemplateContract
rows. Count parity permits duplicate projection keys and missing
source rows — the bridge from list-non-empty proxy to per-row
projection would not be mechanically proven. The whole point of
introducing MethodTemplateContractKey { target, dag_method } is to
provide 1:1 row identity.
Rewrites the ratchet to assert per-row bijection: every
MethodTemplateContract source row has exactly one matching
EmissionPathProjection row by key; every projection key resolves
to exactly one source row; duplicate projection keys fail closed.
(The other openai-pro BLOCKING finding — T-E-P
e_p_call_pattern_lookup_authoritative naming the wrong authority
— was already fixed at 6282a0704; the review citation was against
the pre-fix sha 2072b986.)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): split L6 ratchet into slice-active vs deferred-activation gates
Per BLOCKING inline review on PR #1782 at d92459423: the prior
bijection ratchet was unsatisfiable as authored. The
*_method_template_contracts lists at HEAD (rust/python/go) are
non-empty; shipping emission_path_projections: [] alongside a
strict bijection requirement means the bijection cannot pass when
the slice lands — source rows would be uncovered. Prior text
incorrectly said "vacuously true on the empty source set" — the
SOURCE set isn't empty, only the projection set is.
Splits the ratchet into two gate categories:
- Slice-active gates: types/data shape exist + empty-state
predicate (emission_path_projections == []).
- Deferred-activation gate: per-row key bijection, authored as a
test scaffold by this slice but #[ignore]'d (or feature-gated
on len > 0) until Grounding's follow-up populates rows. On
activation, asserts the bijection over the populated set.
This honors P1 doc faithfulness: the slice's acceptance contract
is satisfiable as authored, and the bijection becomes the
load-bearing gate the row-population PR satisfies.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): require L6 bootstrap enrollment to honor P2 facts-flow-forward
Per BLOCKING inline review on PR #1782 at L6:34: the slice authored
src/v3/std/cross_target_coverage.dag but never required enrolling
it in the bootstrap/load authority. Without enrollment the file
is dead-letter on disk and EmissionPathProjection can be absent
from the downstream Dag — silent P2 facts-flow-forward violation.
Adds a new Slice step 2 requiring a cross_target_coverage field
on BootstrapFixtures in src/v3/std/extdeps_bootstrap_fixtures.dag
mirroring the *_method_template_contracts enrollment pattern, plus
a loader-visibility ratchet. Renumbers subsequent Slice steps.
Adds matching Acceptance bullet.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* WIP: R3 Substrate
* WIP: R3 Substrate
* docs(briefs): normalize Practice 4 citation in S3 brief per codex review
Codex review on PR #1782 (sha 2614217e) flagged stale citation
"Practice 4 (Step 4 of the type-introduction checklist)" in
S3 brief — no such checklist step exists in the pinned rubric.
Live rule per docs/modeling-discipline.md:131-134 is the "What to
check" checkpoint-comment requirement under Practice 4 (coproduct
dissolution). Updated the S3 reference to cite that line directly
and drop the stale "Step 4 of type-introduction checklist" wording.
S2/S8/S9 already use bare "Practice 4" without the stale steps wording.
S1 (committed earlier today) does not contain the stale phrase.
S6 (pre-existing brief flagged by reviewer as carrying the same
pattern) is out of this branch's scope; flag for separate paydown.
* docs(briefs): sweep stale Practice 4 citations corpus-wide
Codex review on PR #1782 flagged the stale "Step 4 of the
type-introduction checklist" wording — no such checklist exists in
docs/modeling-discipline.md. The live rule is the "What to check"
checkpoint-comment requirement under Practice 4 (coproduct
dissolution) at docs/modeling-discipline.md:131.
Sweep across docs/ found 2 additional briefs carrying the stale
phrasing (S3 fixed previously at 9db003d40):
- docs/briefs/r3-l6-emission-path-projection-substrate-worker.md
- docs/briefs/r3-x1b-s1-transform-dispatch-substrate-worker.md
Both updated to cite the live rule directly. Verified zero
remaining "type-introduction checklist" or "Step 4 of the"
references in docs/ tree.
* docs(briefs): add Practice 4 citation discipline note to authoring checklist
Per parent inbox status flag at gunbc#828 #issuecomment-4385346221:
add a Citation discipline section to brief-authoring-checklist.md
naming the correct shape — Practice 4 (coproduct dissolution) +
"What to check" rule at docs/modeling-discipline.md:131 — and
calling out the stale "Step 4 of the type-introduction checklist"
phrasing as a failure mode. References the corpus sweep at
2dc92f34f.
Prevents future briefs that copy old templates from re-introducing
the stale citation pattern.
* docs(r3): cascade work post-Director ratification — S1 worker, R4 routing, Unit/Quantity canvas
Per Director ratification at gunbc#828 inbox response 2026-05-06
(zesty-bear-812):
1. S1 Q1+Q2 RATIFIED → author S1 gap-test representative worker brief
(`r3-substrate-s1-gap-test-representative-worker.md`). Dispatchable
on T-E-P Phase 1 + E6-G0d landing; closes ledger row #61.
2. S2 Q2 RATIFIED (option-(b) T-LBP narrowing) + S3 + S8 deferrals →
R4 carve-out routing ledger (`docs/r4-carve-out-routing.md`).
Captures C1 (parallelism lens), C2 (effect_enum lens), C3
(register narrowing), C4 (deferred MachineConstraint axes),
C5 (rounding-mode product extension), C6 (Unit/Quantity).
Cascade messages to Verification / Grounding / Debt-Paydown Mgrs.
3. S9 Phase-3 dimensional refinement boundary → Unit/Quantity carrier
canvas (`r3-substrate-s9-unit-quantity-carrier-canvas.md`).
Surfaces Q-Unit-1..5 for Director ratification: carrier name,
2-axis vs 3-axis Phase-1, Duration/Seconds collapse, Refined
composition order, Practice 4 classification.
Updates §1.8 ledger receipts for carved items; R3 closure receipts
treat carved items as explicitly-deferred (not drift).
* docs(briefs): citation-discipline fix-forward — anchors/quotes, not line numbers
Director redirect at gunbc#828 #issuecomment-... noted the citation
paydown's replacement form (`docs/modeling-discipline.md:131`) is the
same drift class as the original fabricated-step error. Bare line
numbers shift silently on any doc edit above the cited line.
Generalized fix:
- brief-authoring-checklist.md "Citation discipline" section
rewritten to prohibit bare line-numbers across ALL cross-doc
references (modeling-discipline.md, r3-program-plan.md,
r3-structure.md, INVARIANTS.md, MODELING.md, etc.). Use section
anchors (#section-name) or rule-text quotes.
- 3 briefs that swapped in :131 (S3, S6, X1B) re-converted:
- docs/modeling-discipline.md#4-coproduct-dissolution anchor
- "What to check" rule quoted inline (self-contained)
Brief-authoring-checklist failure-mode section names both vectors
(fabricated step + bare line numbers) and references both sweep
commits in audit trail.
Broader cross-doc citation sweep across briefs I authored on this
branch (frontmatter authority-docs lists, body prose) is recommended
per Director note — flagging as separate paydown task; this commit
closes the immediate recurrence vector.
* docs(briefs): fix unfaithful Int=AbelianGroup<Nat> + sweep stale 'Step 4' refs
Per openai-pro review on PR #1782 sha 0ecd4818 (gunbc#1782 #...):
BLOCKING fix (P1 modeling faithfulness):
- S9 + S3 directed workers to model `Int = AbelianGroup<Nat>`,
but Nat is not an Abelian-group carrier (additive inverses
not representable). Reframed:
- `Int` is the additive Abelian group on integers (carrier Z);
`Nat` is a commutative monoid under addition.
- Algebraic relationship is **explicit group-completion**:
`Int ≡ GroupCompletion<CommutativeMonoid<Nat>>` — NOT
`AbelianGroup<Nat>` parameterization.
- Worker either models `Int` directly as `AbelianGroup`
(terminal Abelian-group instance) OR introduces explicit
`GroupCompletion<M>` substrate carrier; decision lands in
Phase-1 PR per P1 procedure.
- S3 cross-reference + Phase-4 algebra-axis preservation note
reframed accordingly.
NON-BLOCKING in-PR cleanup (Practice 4 citation discipline):
- 6 briefs still carrying stale 'Practice 4 Step 4' wording
(r3-coproduct-1/2/3, r3-l6, r3-t-e-p, r3-x1b) updated to use
the section anchor (`docs/modeling-discipline.md#4-coproduct-dissolution`)
+ inline rule-text quote ("What to check") form per the
citation discipline established in `b4cae5299`.
Closes both findings from openai-pro review.
* docs(briefs): Q-Unit-1..5 RATIFIED — apply Scale.Unit→One rename + ratification banner
Director ratified Q-Unit-1..5 at gunbc#828 #issuecomment-... 2026-05-06
(zesty-bear-812). All 5 questions ratified; one rename required:
- Scale enum's `Unit` value collides with outer `Unit<Q, S>` carrier
name (`Unit<Time, Unit>` is shadowing-confusing). Rename to `One`.
- Duration ≡ Unit<Time, One> (collapses with Seconds per Q-Unit-3).
- Milliseconds ≡ Unit<Time, Milli>.
Canvas updated:
- Top-level RATIFICATION banner records all 5 ratifications + rename
- Scale enum: `Unit` → `One` (with inline rename receipt)
- Sketch product expansions use `Unit<Time, One>` / `Unit<Time, Milli>`
- Phase-3 reframe table products updated
- Q-Unit-3 prose collapse reference updated
Subsequent worker brief authoring (Unit/Quantity carrier landing
post-prerequisite) inherits the renamed shape.
* docs(briefs): Unit/Quantity carrier worker brief — Q-Unit-1..5 RATIFIED
Worker brief consumes the ratified Unit/Quantity canvas
(`5f22fd06e` ratification banner). Phase-1 lands `Unit<Q, S>` +
`Quantity` + `Scale` declarations at the ratified shape (Scale's
`One` value per Q-Unit-1 rename). Phase-2 reframes S9 Phase-3
dimensional refinements (Duration / Seconds / Milliseconds) to
outer-Refined / inner-Unit composition per Q-Unit-4 RATIFIED.
EpochMs deferred to R4 C6 (Aspect-axis follow-up).
Practice 4 marks pre-determined per Q-Unit-5: `Unit<Q, S>` 🟢
PRIMITIVE; `Quantity` + `Scale` 🟡 SCAFFOLD with "consumed by ≥1
emission rule" dissolution trigger.
Cross-program handoff to Grounding Mgr (#1745) for dimensional
emission consumption documented.
Worker pin TBD — dispatchable when idle pool refreshes.
* docs(briefs): S9 Phase-1 reframed — Int landed at #1466; UInt-only migration
Per proud-lynx-311 dispatch ack at gunbc#1739 #issuecomment-...
2026-05-06: re-grep at HEAD revealed substrate state diverges
materially from S9 brief's "Option A vs B decide" framing.
Reality check at HEAD:
- `dsl/std/integer.dag:55` already carries
`type Int = AbelianGroup<GroupCompletion<Nat>>` per Slice 3 PR
#1466 / commit `4ba0a6d04`. Single-authority Q6 audit doc at
`docs/audit/t-numeric-construction-group-completion-6q.md`
ratified this shape and explicitly rejected the compact form.
- `dsl/std/integer.dag:56` still on legacy `type UInt = UInt64`
alias; UInt migration is the actual Phase-1 worker scope.
Brief Phase-1 step 1 reframed: Int landing acknowledged; worker
confirms HEAD match. Phase-1 step 2 reframed: UInt migration
target named (`type UInt = CommutativeMonoid<Nat>`); fixed-width
rows preserved per existing legacy-rows policy.
Historical Option A vs B framing kept as note (superseded).
Confirmed S9 dispatch ack from worker; UInt-only Phase-1 PR is
in-scope. Emission entries deferred per S3 (valiant-ant-72
pre-landing); Phase-2 deferred per S8 (quiet-boar-160 pre-landing).
* docs(briefs): fix two BLOCKING citation drift items per codex review
Codex review on PR #1782 sha 5f22fd06 flagged two authority-receipt
gaps:
1. brief-authoring-checklist.md:118 — `<this-commit>` placeholder in
the failure-mode section was unverifiable post-merge. Replaced
with `b4cae5299` (the actual anchor/quote conversion commit).
2. r3-378-http-path-helper-fail-open-worker.md:84 — `INVARIANTS.md
C-8 fail-closed` cited without anchor or rule-text quote, in
violation of the citation discipline added in this PR.
Converted to `INVARIANTS.md#p3-fail-closed` (P3 heading anchor)
+ inline rule-text quote naming P3 + C-8.
PR's own discipline now consistently applied across the diff.
* docs(briefs): sweep bare .md:NNN citations in Substrate-Mgr-authored briefs
Per Director recommendation in citation-discipline thread (gunbc#828
follow-on): bare line-number references drift silently. Swept 19
remaining instances across 5 briefs I authored on this branch:
- docs/briefs/r3-substrate-s1-q-class-2-chain-break-gap-test-canvas.md
- docs/briefs/r3-substrate-s1-gap-test-representative-worker.md
- docs/briefs/r3-substrate-s2-t-lbp-scope-calibration-canvas.md
- docs/briefs/r3-substrate-s3-machine-constraint-carrier-worker.md
Conversion: bare `file.md:NNN` → `file.md §"Section name"` form.
Stable; survives doc edits; human-readable. Section names per
HEAD heading audit.
Verification: `grep -nE '.md:[0-9]+' docs/briefs/r3-substrate-*.md
docs/r4-carve-out-routing.md` returns zero results in
Substrate-Mgr-authored scope.
Broader corpus scope (~232 references across non-Substrate briefs)
remains as separate paydown task.
* docs(briefs): fix BLOCKING + NON-BLOCKING items per openai-pro review on a462f007
Three findings addressed; 2 BLOCKING + 1 NON-BLOCKING:
BLOCKING — R3/R4 dual authority for Unit/Quantity carrier:
R4 carve-out C6 narrowed from "Unit/Quantity carrier landing"
(deferred all of it) to **"Aspect-axis (PointKind) follow-on
for EpochMs and instant/rate-shaped refinements only"**. The
2-axis Measure<Q,S> carrier landing IS in R3 per Q-Unit-1..5
RATIFIED; only the Aspect axis (PointKind = Magnitude | Instant
| Rate) carves to R4. Cascade-message C6 reference + R4 program-
plan input list updated to match. Cross-Mgr message clarifies
Grounding consumes 2-axis dimensional refinements in R3.
BLOCKING — Locked design partially applied (Scale.Unit→One
rename in canvas alternate-sketch sections):
- Canvas line 105-108 alternate sketch ("Open question: Instant
vs Duration axis") still used pre-rename `Unit` Scale value;
added pre-ratification disclaimer marking the section as
deferred to R4 C6 + noting `Unit` is pre-rename
- Canvas line 163 Q-Unit-3 collapse note: `Scale = Unit` → `Scale = One`
(BLOCKING — bare line-numbers in S1 canvas frontmatter): NOT
APPLICABLE — already fixed at sweep commit 127287a66 prior to
review; reviewer was on sha a462f007 pre-sweep.
NON-BLOCKING — placeholder receipts:
- brief-authoring-checklist.md `#issuecomment-...` → concrete ID
4385393971 (Director's anchor-discipline reply)
- r3-substrate-unit-quantity-carrier-worker.md (3 occurrences)
+ canvas RATIFICATION banner: `#issuecomment-...` → concrete ID
4385412256 (Q-Unit ratification reply)
- `<this-commit>` placeholder in checklist already fixed at 5be1c99f2
Verification: zero `#issuecomment-...` or `<this-commit>` placeholders
in docs/briefs/ + docs/r4-carve-out-routing.md.
* docs(briefs): codex BLOCKING review fixes — Practice 4 source-level + policy scope
Three BLOCKING findings on PR #1782 (codex sha b913e2f4):
1. **Citation-discipline policy scope**: addressed by scoping
the rule prospectively to briefs landing after the policy
commit (`b4cae5299`). Pre-existing corpus references are
flagged as paydown but NOT individually blocking; coordinated
single-PR sweep is the right paydown shape. Substrate-Mgr scope
already swept at `127287a66`.
2. **R4 C6 reconciliation**: already fixed at `6e5a5481a` (R4 C6
narrowed to Aspect-axis follow-on for EpochMs only; 2-axis
carrier landing IS in R3). Codex sha b913e2f4 was pre-fix.
3. **Practice 4 source-level vs PR-body conflation**: corrected.
Acceptance bullet for Unit/Quantity worker brief now requires
inline checkpoint comments on the LIVE declarations in
`dsl/std/units.dag` (load-bearing artifact per
`docs/modeling-discipline.md#4-coproduct-dissolution` "What to
check"). PR-body summary is supplementary, not substitute.
Closes the review's reconciliation gates.
* docs(briefs): absorb Q-MachineConstraint + Q-Unit-1-Recanvas ratifications
Two major Director ratifications absorbed:
1. Q-MachineConstraint-Carrier RATIFIED at gunbc#828
#issuecomment-4385530115 (Brian directive 2026-05-06: "universal
substrate, ratify defaults"; PR #1817 queued). 6 sub-decisions:
- MachineWidth<bits> only for R3; defer RegisterClass / EndianMode
/ Alignment / signedness post-R3
- Parametric Compose<Algebra, MachineConstraint> type-level
interaction shape (lookup-maps REJECTED — AlgebraMachineProduct
record-keyed table form retracted)
- Int<64> ≡ Compose<AbelianGroup, MachineWidth<64>> (algebra-side
options A and B preserve interaction semantics)
- Real<64> ≡ Compose<ApproximateField<Rational>, MachineWidth<64>>
(algebra approx + machine approx as independent composing axes)
- ≥3 pairs is minimum, not target
- UNIVERSAL substrate: every target carries machine-constraint
facts; targets without native machine-width semantics handle
omission at Grounding-level discharge (target-conditioned
lowering NOT target-conditioned substrate)
Reframed: S3 brief Phase-1, S3 Phase-3/4 emission entries,
S8 Phase-1/2/3 (ApproximateField<Real> → ApproximateField<Rational>;
AlgebraMachineProduct → Compose<...>), S9 emission entries.
2. Q-Unit-1-Recanvas RATIFIED Measure<Q, S> at gunbc#828
#issuecomment-4385539791. valiant-ibex-312's substrate-state
grep caught Unit<Q, S> collision with kernel Unit at
dsl/std/types.dag:176; Director ratified Measure<Q, S> as
outer carrier name (conventional in scientific computing;
no dsl/std/ collision).
Reframed: Unit/Quantity worker brief throughout (Unit<Q,S> →
Measure<Q,S>; dsl/std/units.dag → dsl/std/measure.dag); canvas
spec sections + sketch products (Refined<Measure<Time, ...>...>);
R4 carve-out C6 with informal/formal name disambiguation note.
Both rename + reshape are now consistent across S3 / S8 / S9 /
Unit-Q briefs + canvas + R4 routing. Workers (valiant-ant-72 on S3,
quiet-boar-160 on S8, proud-lynx-311 on S9, valiant-ibex-312 on
Unit-Q) all dispatch against the ratified shape.
* docs(briefs): S11 Slice C worker brief — ROADMAP :425 retirement post-#1801
Cascade-clearance trigger fired post-#1801 (Slice B) merge
2026-05-06. S11 brief authored per B1 RATIFIED prose+regen
bundling discipline.
Phase 1: per-file edits across the post-A+B residual list (~10
files per smart-ram-167's Slice B post-rebase enumeration);
anchor + rule-text-quote pattern per current citation discipline;
bootstrap regen in same PR (no split).
Phase 2: ROADMAP :425 row PARTIAL → Retired with 3-slice receipt;
Q-Slice-C-Retirement-Receipt resolution.
Worker pin smart-ram-167 (#1759) — Slice B precedent owner.
Dispatch packet sent separately to #1759.
* docs(briefs): Q-Refined-Phantom-Composition (c) RATIFIED — defer literal Refined form
Director ratified option (c) at gunbc#828 #issuecomment-... 2026-05-06:
drop `non_negative` predicate at Measure (phantom) layer; rely on
`Quantity = Time` tagging for non-negative-magnitude semantics.
Q-Unit-4 outer-Refined / inner-Measure composition is preserved as
**conceptual model**; the literal `Refined<Measure<...>, predicate>`
form is **deferred to value-typed-integration follow-up** when
substrate gains Refined<> as first-class type expression and
dimensional refinements get value-typed integration.
Updates:
- Canvas RATIFICATION banner Q-Unit-4 row reframed to "conceptual
model; literal form deferred"
- Canvas Phase-3 reframe table: Duration / Milliseconds drop the
Refined<...> wrapper; rely on Quantity tag
- Worker brief Phase-2 prose + reframe table + Acceptance bullet
all reflect deferred-literal-form posture
valiant-ibex-312 unblocked to proceed with Phase-2 (Grounding
handoff doc) per the (c)-shaped Acceptance.
* docs(briefs): S12 F2 + F8 doc-sharpening worker brief — post-#1820 cascade
#1820 (Slice C) merged 2026-05-06; ROADMAP :425 row Retired closes
the prose-reference paydown program. S12 cascade-clearance trigger
fires per B6 RATIFIED single bundled Mgr-tier doc-sharpening PR.
Scope: F2 (.v3 filename-suffix grammar dispatch) + F8 (bootstrap
load-order/exclusion authority) — single bundled PR per B6.
Cross-program coordination with PB Mgr (#1742) per design schedule
§3 P5 (comment-only OR co-author; no duplicate PR).
Worker pin smart-ram-167 (#1759) — Slice C precedent owner;
pattern-familiar with prose+regen bundling discipline. Dispatch
packet sent separately to #1759.
* docs(briefs): S9 brief — UInt landed via #1818; algebra label corrected to CommutativeSemiring
Per proud-lynx-311 merge report at gunbc#1746 #issuecomment-...
2026-05-06: PR #1818 landed `type UInt = Nat` after codex BLOCKING
review on initial `CommutativeMonoid<Nat>` form (P2 facts-flow-forward
violation — wrap projects out Nat's multiplicative monoid + identity 1).
Final landed form preserves full `CommutativeSemiring<Magnitude>`
surface — Nat IS a commutative semiring (addition + multiplication
+ 0 + 1). Algebra-side label for UInt is CommutativeSemiring, NOT
CommutativeMonoid (corrects my earlier dispatch labeling).
S9 Phase-1 step 2 marked LANDED with merge ref. Phase-1 step 3
emission entries reframed:
- Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>>
- UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>>
Worker discipline + reviewer caught a substrate-state error in my
brief framing; brief now matches landed reality.
* docs(briefs): fix contradictory (c) guidance in Unit/Quantity worker brief
Codex BLOCKING review on PR #1825 sha 1bce0418 caught a real
contradiction: brief Phase-2 prose at lines 115-118 stated
"predicates do NOT attach at the Measure (phantom) layer" per
Q-Refined-Phantom-Composition (c) RATIFIED, but immediately
following paragraph at lines 130-132 still said "Refined predicate
non_negative applies at the Measure-typed level" — left over from
the original Q-Unit-4 framing.
Updated lines 130-137 to consistently apply (c) shape: predicates
deferred to value-typed-integration follow-up; dimensional semantics
ride on Quantity tag at Measure layer; worker authors 3 in-scope
refinements WITHOUT predicate at this layer.
P2 boundary discipline restored: brief now gives single-authority
guidance for Phase-2 substrate landing.
* docs(briefs): name predicate authority + concretize ratification ID — codex BLOCKING
Codex BLOCKING review on PR #1825 sha 32541a56:
1. (BLOCKING) Q-Refined-Phantom-Composition (c) ratification deferred
phantom-carrier predicates without naming the value-level predicate
authority. Updated canvas Q-Unit-4 RATIFICATION row to explicitly
name: 'predicates re-attach at value-typed Int/UInt level via
where-sugar (e.g., range(min: 0) per current dsl/std/types.dag
precedent)' when value-typed integration lands.
2. (BLOCKING) Worker brief option-c reframe didn't sweep the audit
receipt §5 (Carrier dissolves the bridge). Updated to consistently
apply (c) shape — dimensional semantics ride on Quantity tag at
Measure layer; value-level predicate authority deferred; until
then, range/gt_zero constraints attach at value-typed Int/UInt
via where-sugar per dsl/std/types.dag convention.
3. (NON-BLOCKING) Canvas had #issuecomment-... placeholder for the
(c) ratification message ID. Replaced with concrete
#issuecomment-4385687075 per dashboard relay header.
Brief now names a single value-level authority (Int/UInt where-sugar)
for value-range facts; phantom-layer Quantity tag carries only
dimensional semantics. P2 boundary discipline maintained.
* docs(briefs): S12 RETIRED per Director option (2) ratification
Director ratified option (2) at gunbc#828 #issuecomment-4385873...
2026-05-06 post smart-ram-167's substrate-state STOP-AND-ESCALATE:
- F2 absorbed by 2026-05-04 routing + #1662 closure
(`Result/DivError span.file-keyed` retired)
- F8 topical concerns distributed across live rows
(Class 5 Gap 3 `bootstrap.rs::std_fixtures()`; Character-level
load-set / bootstrap; Filename-sentinel-bridges) with own owners
- Doc-discipline goals absorbed by per-Mgr citation paydown
routing at gunbc#828 #issuecomment-4385583247
Frontmatter updated to RETIRED with absorption receipt. Worker
(smart-ram-167) pin freed.
Director-side acknowledgment of Substrate Mgr's option (2) recommendation:
"Substrate Mgr's option (2) recommendation is the right move
structurally. Most Mgrs default to 'synthesize work to fill the brief'
when canvas-vs-HEAD mismatch surfaces. Recommending retirement when
the work is genuinely absorbed is harder discipline."
* docs(briefs): S7 PR-F (BoundDeclaration consumer + ReferenceModel<T>) worker brief
Per cascade-clearance trigger fired post-#1782 merge 2026-05-06.
loyal-wolf-828 (#1764) acknowledged S7 dispatch with substrate-state
pre-flight grep (gunbc#1764 #issuecomment-4385872...) and surfaced
await-brief posture. This brief satisfies that prerequisite.
Worker pre-flight inventory absorbed:
- BoundDeclaration substrate landed (substrate.dag StaticBound +
PlatformDependent); partial BoundDeclarationView consumer in
fold.rs (ScratchIntExamples scope; doc-explicit partial)
- ReferenceModel<T> NOT yet in dsl/std/ (substrate-fact-introduction
required; design-emission-model.md §Q2 framing)
Phase-1: BoundDeclaration consumer broadening to full
design-emission-model.md surface (u128 / isize / usize / walker
arms / pilot mirror).
Phase-2: ReferenceModel<T> P1 substrate-fact-introduction with
DFS-of-concept-DAG + Practice 4 in-source checkpoint comment +
named consumer demand (Grounding G1).
Phase-3: cross-program handoff to Grounding Mgr (#1745) for
G1 (T-Ground-Rust Phase 1).
Closure predicate: unblocks Grounding T-Ground-Rust Phase 1.
Worker pin loyal-wolf-828 holds; dispatch fires immediately
on this brief landing.
* docs(briefs): S9 Slice 2.5 — predicate-registry + PB-1 shim retirement (Path 3 RATIFIED)
Director ratified Path 3 at gunbc#828 #issuecomment-4390333451
post proud-lynx-311's pre-flight DFS at gunbc#1746 #issuecomment-4390253616.
Brief reframes Slice 2.5 from S-sized (primitive + 1 consumer) to
L-sized (predicate registry + PB-1 shim retirement). 5 deliverables:
1. gt_zero primitive predicate
2. Predicate registry infrastructure (resolver-side substrate-fact-introduction)
3. PB-1 shim retirement (span.file=='dsl/std/types.dag' special case dissolves)
4. Migration of existing types.dag predicates (range, pattern, etc.)
5. PositiveInt = Refined<Nat, gt_zero> consumer update
Per Director hard scope bars: no parallel predicate-resolution path;
no per-call-site special-casing; PB-1 shim retirement non-negotiable
in same slice. Single PR per Path 3 RATIFIED.
Discipline anchors: feedback_construction_over_ratchets (PB-1 shim
is bridge); feedback_dissolve_bridges (registry is structural fix);
feedback_no_textual_enforcement_bridges (file-path special-case is
anti-pattern shape); P1 substrate-fact-introduction procedure.
Worker pin proud-lynx-311 — Director pre-cleared on pre-flight DFS
context. PR #1840 (partial Slice 2 — NonNegativeInt = Nat) MERGED
at 16:54Z; Slice 2.5 dispatchable immediately on this brief landing.
* docs(briefs): Slice 2.5 audit receipt — Mgr independent grep parallel-verifies worker DFS
Codex BLOCKING review on PR #1845 sha 2d230949 flagged: external PR
receipt absorbed as live substrate state without re-grepping local
std declarations at brief authoring time. Authority receipt §1
cited proud-lynx-311's pre-flight DFS as authority but didn't
record Mgr-side independent verification.
Updated audit receipt §1 + §4 to record Mgr's independent grep at
HEAD as parallel verification:
- gt_zero zero matches in dsl/std/ + src/v3/ (verified)
- NonNegativeInt = Nat at dsl/std/integer.dag:133 (#1840 land)
- PositiveInt = Int where range(min: 1) at dsl/std/types.dag:255
- PB-1 shim doc-block confirms bridge-text framing
- Predicate registry confirmed not present
Worker DFS remains the dispatch-time discipline (substrate-state
grep at dispatch); Mgr's independent grep is parallel verification,
not substitute. Per Director's pattern note at gunbc#828
#issuecomment-4390199218 + #issuecomment-4390333451: dispatch-packet
authoring discipline includes Mgr-side substrate-state-grep, not
just relay of worker findings.
* docs(briefs): Slice 2.5 brief — Int extensibility scope-bound; lower.rs cites anchored
openai-pro REQUEST_CHANGES on PR #1845 sha 2d230949 flagged 2 issues:
BLOCKING — P1 modeling faithfulness / substrate scope leak:
`gt_zero` Int-side extensibility was left to 'worker judgment if
generalization cheap'. Per feedback_construction_over_ratchets:
don't speculatively extend predicate-carrier compatibility ahead
of named consumer demand. Reframed to **EXPLICITLY OUT-OF-SCOPE**
for Slice 2.5; Nat-only registration. Future Int-side consumer
surfaces as separate substrate-fact-introduction (P1 procedure)
brief with own consumer-demand receipt.
NON-BLOCKING — citation discipline self-violation:
3 instances of bare `lower.rs:821-835` line-range citations
contradict the brief's own 'no bare :NNN' requirement at line 211.
Converted to function-name anchored form:
`src/v3/compiler/src/lower.rs` (`lower_type_alias_refinements_phase` doc-block).
Both findings absorbed; no parallel scope ambiguity remains;
brief now self-consistent on citation discipline.
* WIP: R3 Substrate
* docs(briefs): Slice 2.5 brief — audit-receipt cites converted to declaration-name anchors
cursor APPROVE_WITH_COMMENTS on PR #1845 sha ebf6ac12 noted citation
self-inconsistency: audit-receipt section §1 + §4 used bare
`dsl/std/integer.dag:133` and `dsl/std/types.dag:255` line-number
pins despite the brief's own no-bare-:NNN rule.
Converted to declaration-name-anchored form:
- 'declaration named NonNegativeInt' (with reproducible grep command)
- 'declaration named PositiveInt' (with reproducible grep command)
- 'lower_type_alias_refinements_phase doc-block' for PB-1 location
Brief now fully self-consistent on its own citation discipline.
Verified zero remaining .dag:NNN / .rs:NNN / .md:NNN bare cites.
* docs(briefs): Slice 2.5 brief — Path (a) RATIFIED carrier+argument contract verification
Director Path (a) RATIFIED at gunbc#828 #issuecomment-4390760353
(triage) + #issuecomment-4390794121 (convergence ack):
predicate-body lowering with carrier+argument contract
verification at lower-time IN SAME SLICE. Placeholder semantics
REMOVED from registry path.
Codex REQUEST_CHANGES surfaced that registered-predicate-without-
contract = silent acceptance per feedback_fail_closed_discipline.
Brief reframe absorbs Path (a):
Hard scope bar #5 added:
- Same-slice carrier+argument-contract verification
- Mismatched carrier OR malformed argument → Diagnostic::ResolveError
- No 'named but not checked' path remains
Scope expansion section added (L → XL):
- Predicate-body lowering for all 7 registered predicates
- Per-predicate arg-shape validation
- Multi-session work acknowledged
- Carrier-check at 788d6acb4 retained; predicate-body builds on top
Acceptance bullets updated: gt_zero with arg-shape validation +
all 7 predicates with carrier+argument contract verification.
Director-pinned brief-authoring discipline: dissolution trigger
named in a brief authored as substrate-fact-introduction must
specify same-slice acceptance, not deferred follow-up. Otherwise
the brief encodes 'land bridge + defer dissolution' anti-pattern,
the same shape feedback_construction_over_ratchets warns against
at implementation layer applied at brief authoring layer.
This is the 11th miss this session — but the pattern surfacing it
(direct sweep + Codex REQUEST_CHANGES + Director ratification)
produces structural fixes that prevent the next miss class. Brief
revision is the structural recovery.
* docs(briefs): Slice 2.5 brief — Path 2 RATIFIED (Gap 1 discharge + Q-Regex carve-out)
Director Path 2 RATIFIED at gunbc#828 #issuecomment-4391985613
post Rung 3 STOP at gunbc#1746 #issuecomment-4391946213.
Path 2 narrowing of Path (a):
- gt_zero + range body synthesis already landed (5f3c40c3a)
- Gap 1 discharge mechanism: extends scalar-literal predicate
evaluation at lower-time. L-class; lands SAME SLICE per
same-slice-dissolution discipline
- non_empty + brand body synthesis: lands via Gap 1 discharge
- pattern / format / content: fail-closed-with-named-dep at
user-code authoring layer (Diagnostic::ResolveError naming
Q-Regex-Primitive). NOT placeholder; structural rejection
- PB-1 shim FULLY RETIRED per hard scope bar #3
- Q-Regex-Primitive carved as follow-on substrate-fact-introduction
(XL substrate; not pre-authored per feedback_construction_over_ratchets)
Acceptance updated:
- 4 of 7 body-synthesized in same slice
- 3 of 7 fail-closed-with-named-dep (no placeholder semantics)
- Gap 1 discharge mechanism landed (no partial implementation)
Worker disposition: proud-lynx-311 continues Rung 3 with Gap 1
implementation + non_empty/brand body synthesis + pattern/format/
content fail-closed-with-named-dep diagnostic.
* docs(briefs): S9 Phase-1 Step 3 emission entries — Tier-1 brief authoring (1/5)
First Tier-1 brief landing post Director auto-nudge on assignment
#1858. Authored against Q-MachineConstraint sub-decisions ratified
shape now landed at #1856:
- Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>> → Rust i32/i64/i128
- UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>> → Rust u32/u64/u128
- 6 concrete instantiations (≥3 minimum per sub-decision 5)
- Cross-program handoff to Grounding Mgr (#1745) for G2 emission consumption
- numeric_construction_demonstration Acceptance bullet (Int<32> round-trip);
Real<64> half deferred to S9 Phase 2 / S8 Float cascade
Brief authored with substrate-state-grep discipline + same-slice-
dissolution discipline (no deferred trigger; demonstration in same
slice or explicitly carved). 5-question authority audit completed
in brief body.
Worker pin proud-lynx-311 (S9 holds); brief queues post-Slice-2.5
completion as natural follow-on.
Tier-1 brief queue progress:
- 1/5 authored (this brief)
- 4 remaining: S9 Phase-2 Float coordination; T-LBP complexity
cementing; T-LBP cost cementing; S3 Phase-2 parser-grammar
* docs(briefs): S3 Phase-2 parser-grammar surface — Tier-1 brief 2/5
Authored per Director freed-pool pressure at gunbc#828
#issuecomment-4392095857 + Tier-1 brief-queue commitment.
Brief covers parser-grammar surface for Compose<Algebra, MachineConstraint>
interaction syntax post-#1856 (Phase-1 carrier slice landed):
- Phase 2.1: Parser surface per ratified Q-MachineConstraint-Grammar-Shape
- Phase 2.2: Bootstrap demonstrator with >=3 algebra*machine-axis pairs
- Phase 2.3: Class 1 5-criteria Pass receipt (criteria 1-4; criterion 5
v2-oracle parity may carve)
- Phase 2.4: numeric_construction_demonstration co-receipt with S9 Phase-1
Step 3 emission entries brief
Director ratification HOLDS on grammar bikeshed:
- Candidate A: Int @ Width<32> (annotation form)
- Candidate B: Int with MachineWidth<32> (with-form)
- Candidate C: Int<32> (positional generic — fits existing convention)
- Candidate D: Compose<AbelianGroup, MachineWidth<32>> (no parser change;
type aliases provide convenience names)
Worker pin valiant-ant-72 holds (S3 Phase-1 precedent owner); freed-pool
until Director ratifies grammar shape.
Tier-1 brief queue progress: 2/5 authored (this brief + S9 Phase-1 Step 3).
Remaining: S9 Phase-2 Float coordination; T-LBP complexity cementing;
T-LBP cost cementing.
* docs(briefs): S3 Phase-2 brief — fix internal contradiction on Class 1 closure scope
Codex BLOCKING review on PR #1889 sha ec609b4a flagged real
contradiction:
- Provenance section claimed '5-criteria Pass closed' (line 229)
- Phase 2.3 said 'criterion 5 may queue separately' (line 154)
- STOP-AND-ESCALATE bullet 1 ratified 4-of-5 carve-out (line 181)
Updated Provenance to consistently state criteria 1-4 close in this
slice; criterion 5 (v2-oracle parity) cementing test may carve to
separate slice. Single 4-of-5 framing throughout. INVARIANTS.md
'Documentation Describes Live State' satisfied.
Per feedback_construction_over_ratchets: don't claim closure that
isn't structurally complete. 4-of-5 + named carve is the honest
shape; 5/5 only on cementing-test land.
* briefs: cite Q-MachineConstraint sub-decision 6 supersession of T-V2-Retirement gate
Address Codex BLOCKING #1 on PR #1889: S3 Phase-2 + S9 Phase-1 Step 3
worker briefs consumed Q-MachineConstraint ratification without
reconciling the historical T-V2-Retirement-landing-first cascade
gate. Add an explicit precondition-framing section to each brief
citing Director ratification at gunbc#828 #issuecomment-4385530115
(Q-MachineConstraint sub-decision 6 UNIVERSAL substrate posture)
as the implicit supersession; PR #1856 landing without T-V2-Retirement
first confirms supersession in practice. STOP-AND-ESCALATE if
supersession contested at dispatch.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* slice-2.5 brief: mark pattern/format/content Path 2 acceptance pending Director ratification
Address Codex BLOCKING #2 on PR #1889: Path 2 deliverable + acceptance
language asserted fail-closed-with-named-dep semantics for
pattern/format/content, but Director rejected that shape (Option 4)
at gunbc#828 #issuecomment-4392.... Options A (consumer migration),
B (Q-Regex bundled into slice), C (typed fail-closed exception with
named trigger) await ratification per migration-cost catalog at
#issuecomment-4392116215. Brief now flags those three predicates +
PB-1 shim retirement coupling as PENDING DIRECTOR RATIFICATION; worker
STOP-AND-ESCALATE if dispatched pre-ratification.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* slice-2.5 brief: reconcile PB-1 retirement shape with Options A/B/C gating
Address Codex BLOCKING #4 on PR #1889: prior text claimed "PB-1
FULLY RETIRED" while preceding bullet said retirement gates on
chosen path — internal contradiction. Reframe to gate the
pattern/format/content half of shim retirement on Options A/B/C
ratification while preserving Director hard scope bar #3 binding
on the gt_zero/range/non_empty/brand half (cleanly retired via
Gap 1). No path leaves PB-1 alive as parallel-authority shim on
slice landing.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: tier-1 3/5 — S9 Phase-2 Float emission coordination worker brief
Tier-1 brief queue progress (3/5): S9 Phase-2 Float<N> emission
entries (Float<32>/Float<64> via Compose<ApproximateField<Real>,
MachineWidth<N>>) + Real<64> round-trip demonstration completing
§1.8 #67 Acceptance bullet. Coordination/synthesis brief — gates on
both S8 (ApproximateField<F> + Real base carrier) and S9 Phase-1
Step 3 (Int<N>/UInt<N> emission-entry pattern) landing before
dispatch. Inherits T-V2-Retirement supersession framing from
Phase-1 Step 3. STOP-AND-ESCALATE on missing preconditions or
unratified Q-ApproximateField-Axiom-Set.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: ledger advancement honesty (PRODUCER_LANDED not CONSUMER_LANDED) + restore exact issuecomment IDs
Address openai-pro BLOCKING + NON-BLOCKING findings on PR #1889:
BLOCKING (boundary discipline / same-PR consumer): S9 Phase-1 Step 3
+ Phase-2 Float briefs claimed DECLARED → CONSUMER_LANDED upon merge
while only requiring cross-program handoff receipt to Grounding G2.
For target-primitive substrate entries, handoff is not landed-consumer
proof. Reframed both briefs to advance rows DECLARED → PRODUCER_LANDED;
Grounding G2 follow-on PR (per-pair lowering rules + emitted-Rust-
primitive verification) advances to CONSUMER_LANDED. Default
split-PR producer-then-consumer per bundled-scope discipline ratified
at gunbc#1739 #issuecomment-4392225548 (parallel infrastructure
DISALLOWED in same PR).
NON-BLOCKING (citation hygiene): replaced ellipsized
"#issuecomment-4392..." stubs in slice-2.5 brief with the exact
Director Option-4-rejection comment ID #issuecomment-4392081719 so
the ratification claim is mechanically checkable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* slice-2.5 brief: absorb Director Option A revised RATIFICATION (final shape)
Director ratified Option A revised at gunbc#1739 #issuecomment-4392329709
(2026-05-06). Final brief shape:
- 4 predicates body-synthesize via Gap 1 (gt_zero / range / non_empty
/ brand)
- pattern / format / content STRUCTURALLY ABSENT from registry (not
pending-with-placeholder); user-code where-clause produces
Diagnostic::ResolveError naming Q-Regex-Primitive as separate
substrate-fact-introduction follow-on
- ~12 types.dag refined-type declarations migrate to drop
pattern/format/content where-clauses with inline-comment receipts
pointing to Q-Regex restoration path; 1 extdep (sts_endpoint)
adapts as needed
- PB-1 shim FULLY RETIRED — no exception, no preserved branch
- Q-Regex-Primitive as separate brief / worker pin / Director
ratification cycle when concretely needed (Option B XXL bundling
REJECTED)
Phase ordering rewritten to absorb migration as explicit phase
between predicate enrollment and shim retirement. proud-lynx-311
re-dispatches against revised brief; PR #1846 absorbs additional
scope per Director directive.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* s3-phase-2 brief: absorb Q-MC sub-decision 3 ratification (grammar shape was already answered)
Director observation at gunbc#1739 #issuecomment-4392382517: Q-MC
sub-decision 3 (Brian directive 2026-05-06 at gunbc#828
#issuecomment-4385530115) already ratified the grammar shape —
`Int<N>` user surface desugars to `Compose<Int, MachineWidth<N>>`
substrate parametrically. Earlier brief HOLD framing was redundant.
Brief revised:
- Replace bikeshed section with Q-MC sub-decision 3 quotation +
surface/substrate framing (Candidate C user-facing; Candidate D
substrate elaboration; A/B rejected per Director assessment)
- Phase 2.1 parser surface scoped to numeric-literal-position
recognition + parametric desugar
- Front-matter status: dispatchable per ratification (not HOLD)
- Provenance: ratification-state-grep discipline pin folded into
standing dispatch-checklist
valiant-ant-72 dispatch fires immediately per Director directive.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: remove truncated issuecomment-4392... stub from S9 Phase-1 Step 3 provenance
Address cursor NON-BLOCKING finding on PR #1889 sha 94292e48: line 144
truncated issuecomment ID. The originally-intended Director auto-nudge
comment doesn't map cleanly to a single ID, so replace with the
standing assignment reference (gunbc#1858) which is unambiguous.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: tier-1 4/5 + 5/5 — T-LBP complexity + cost lens cementing-test worker briefs
Tier-1 brief queue completes (4/5 + 5/5):
4/5 — T-LBP complexity-lens cementing test (§1.8 rows #79 + #87
first entry). Consumer-side cementing test against frozen v2-oracle
snapshot; producer-side Lens<Complexity> instance is upstream
precondition. Per Q-Lens-Behavioral-Parity-R3-Closeability option
(b) RATIFIED at gunbc#828 #issuecomment-4385329180 — T-LBP narrowed
to complexity + cost lenses only.
5/5 — T-LBP cost-lens cementing test + cost_lens_demonstration
(§1.8 rows #80 + #70 + #87 second entry). Absorbs row #70 demo
naturally — same source corpus + lens run; demonstration adds
observable-cost-bound assertions on top of parity check. ≥2
algebra-instances composed + ≥1 recursive call required per row #70.
Both briefs gate on:
- T-LBP per-lens substrate-producer brief landing (Lens<C> instance)
- Frozen v2-oracle snapshot capture (separate brief, pre-v2-retirement)
- DO NOT capture-against-live-oracle (violates
v2_oracle_no_remaining_test_consumers)
- DO NOT bundle producer/snapshot/algebra-instance authoring per
Director bundled-scope ratification at gunbc#1739
#issuecomment-4392225548
Tier-1 5/5 complete. Worker pins assigned at dispatch time once
preconditions land.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* canvas: emission-provenance substrate shape disposition (supersedes PR #1902 PROPOSAL)
Substrate Mgr canvas authored post-codex BLOCKING + PM disposition
handoff at gunbc#846 #issuecomment-4392510255.
Two real substrate-shape questions stacked:
Q1 (category mismatch): Lens<C>.read is per-Behavior; emission
provenance is per-emitted-line. 4 options surfaced — (a) per-Behavior
Lens-compatible / (b) per-line instrumentation NOT a lens / (c)
withdraw + re-canvas / (d) Mgr-surface both substrates distinct.
Mgr recommends (d) if visualization is load-bearing; (a) suffices
if per-Behavior aggregate works for the slide. Brian-channel
sub-question surfaced.
Q2 (fold-rule enumerability): T-Rule-Enumeration substrate-fact-
introduction is prerequisite for every Q1 path (rust_target.rs uses
inline &str template names; rules not enumerable). Mgr recommends
parallel-dispatch with Q1 deliberation — satisfies Brian's ASAP
framing while shape ratifies.
Q3 (§1.8 ledger retarget): #89 already taken by T-LAS; new cluster
location follows Q1 disposition.
Director ratification ask + Brian-channel sub-question surfaced.
PR #1902 close-superseded post-ratification.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: reconcile Compose<...> slot-1 spellings to algebraic-concept names per Q-MC sub-decision 3
Address cursor NON-BLOCKING on PR #1889 sha 2b49b30f: S3 Phase-2
brief mixed Compose<AbelianGroup, ...> with Compose<Int, ...>; S9
Phase-1 Step 3 used Compose<AbelianGroup, ...> + Compose<CommutativeSemiring,
...>; S9 Phase-2 used Compose<ApproximateField<Real>, ...>. All wrong
per Q-MC sub-decision 3 critical correction at gunbc#828
#issuecomment-4385530115:
"prior phrasing Compose<AbelianGroup, MachineWidth<64>> was wrong
— AbelianGroup<T> is a witness shape over carrier T, not a carrier
constructor; bare AbelianGroup in slot-1 composes witness rather
than concept."
Slot-1 is the fully-applied algebraic-concept name:
- Int<N> = Compose<Int, MachineWidth<N>> (Int = AbelianGroup<GroupCompletion<Nat>>)
- UInt<N> = Compose<UInt, MachineWidth<N>> (UInt = CommutativeSemiring<Nat>)
- Real<N> = Compose<Real, MachineWidth<N>> (Real = ApproximateField<Rational>)
Phase-2 Float brief also retargets user-facing surface from Float<N>
to Real<N> (Float is target-language name for the Rust primitive;
Real is the algebraic-concept name per Q-MC sub-decision 3 example).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* s9-p1-s3 brief: scope Deliverable 4 to producer-side verification (defer demonstration to Grounding G2)
Address Codex BLOCKING (line 82 inline + bundled review): brief
claimed end-to-end Int<32> → Rust i32 round-trip in Acceptance
while Deliverable 3 said lowering rules belong to Grounding G2
follow-on. Internal contradiction.
Reframe per Director bundled-scope discipline (gunbc#1739
#issuecomment-4392225548 — parallel infrastructure DISALLOWED
same-PR):
- Deliverable 4: producer-side substrate-axis verification only
(type-check + carrier acceptance + bootstrap snapshot/manifest
hold). NO end-to-end Rust emission in this PR.
- §1.8 #67 numeric_construction_demonstration receipts split:
producer half here (entries exist), consumer half on Grounding
G2 PR (full Int<32> → Rust i32 round-trip).
- Same-PR cross-program co-author available if Substrate Mgr
coordinates with Grounding Mgr at dispatch; default is split.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: pre-author T-Rule-Enumeration substrate-fact-introduction worker brief
Pre-authored per pre-authored-brief-queue discipline; dispatch fires
on Director Q2 ratification of emission-provenance canvas at
gunbc#828 #issuecomment-4392519713.
Brief covers LangSpec emission-rule names as enumerable substrate
(rust_target.rs uses inline &str template names today; 0 hits on
RuleName/FoldRule/EmitRule across src/v3/). Two carrier shape options
(α sum-type / β named-string list) with worker DFS at dispatch
deciding shape; Practice 4 checkpoint discipline + ledger-row receipt.
Prerequisite to every Q1 path (a/b/d) — Lens<EmissionProvenance>,
per-line instrumentation, and the both-substrates option all consume
the rule-name set. Path (c) leaves substrate-unblocked-but-unconsumed.
Mgr authoring authority per PM concur at gunbc#846
#issuecomment-4392510255 / #issuecomment-4392543633 (substrate-fact-
introduction is Mgr-tier, not PM tactical authoring).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* t-rule-enumeration: align gate name + status to Director ratification at #4392562911
* emission-provenance lens brief: revise per Director Q1 (a) RATIFICATION
Director Q1 (a) per-Behavior Lens<C>-compatible RATIFIED at gunbc#1739
#issuecomment-4392562911 (2026-05-06); (b) instrumentation REJECTED;
(d) parallel-substrates REJECTED.
Brief revised:
- Carrier shape: Lens<List<EmissionProvenance>> per (a)
- EmissionOrigin = SubstrateDeclMirror(SourceSpan) | FoldRuleAutoEmit(EmissionRule)
typed sum (closed; no third silent class; structural fail-closed)
- All 6 Lens<C> field bindings concrete (read/sequential/branch/iterate/validate)
- T-Rule-Enumeration named as PRECONDITION (must land first)
- §1.8 retarget under T-CostLens-Composition cluster per Director Q3
- Brian's slide visualization served by (a) projection (per-Behavior
list flattens to per-line view at visualization layer; no parallel
substrate needed)
Supersedes PM-authored proposal at PR #1902 (merged at 54419badf).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: bundled Rust-primitive-full-coverage substrate-prerequisite worker brief (Director Path A)
Author bundled substrate-prerequisite brief per Director Path A
RATIFICATION at gunbc#1739 #issuecomment-4392731264 (2026-05-06).
Three structurally-coupled gaps surfaced via Grounding loyal-stag-699
STOP at gunbc#1907 #issuecomment-4392585090 land as one bundled slice
per "necessary structural fix" exception (bundled-scope discipline):
1. IntervalInt::ExactInterval host-repr widening (α BigInt-based
recommended; β typed-variants if α infeasible)
2. RustPrimitive structural BoundDeclaration field (replaces static
range_min/max strings; consumes existing BoundDeclaration carrier
StaticBound + PlatformDependent variants)
3. spec/rust.dag PlatformDependent row population for isize/usize +
u128 row addition consuming widened ExactInterval
Closure gate: rust_primitive_full_coverage (Director-ratified name).
Worker pin: valiant-ibex-312 (freed-pool; substrate-authoring fresh;
numeric-domain-adjacent post-#1842).
Unblocks Grounding G2 Phase 2 full-coverage dispatch. G2 Phase 1
narrowed dispatch (i8-i64 + u8-u64) runs in parallel on existing
substrate; cleared at gunbc#1745 #issuecomment-4392795954.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: align Practice 4 classifications to GREEN/YELLOW rubric + add GREEN ledger receipts
Address codex BLOCKING on PR #1910 sha f36e1205: briefs used
PRIMITIVE/SCAFFOLD terminology that doesn't match docs/modeling-
discipline.md rubric. Per modeling-discipline.md lines 97/103/108:
the three classifications are 🟢 GREEN (terminal) / 🟡 YELLOW
(scaffold) / 🔴 RED (dissolvable-now); line 132 requires a ledger
entry if GREEN.
Replaced PRIMITIVE → GREEN, SCAFFOLD → YELLOW across all three
emission-provenance / Rust-primitive-full-coverage briefs. Added
explicit §1.8 ledger entry receipt for the GREEN classification on
EmissionOrigin (sibling row emission_origin_classification_green to
parent emission_provenance_lens_landed gate).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: Practice 4 terminal receipt for EmissionOrigin + bound for β scaffold
Address openai-pro BLOCKING (sha f36e1205, ran post-rubric-rename):
1. EmissionOrigin GREEN terminal classification needs documented
attempted-dissolutions per modeling-discipline.md "no richer
source exists" framing. Added 3 attempted dissolutions:
- collapse to optional-pair (rejected by codex Finding 1 prior)
- factor common richer source (orthogonal payload types prevent)
- third arm for unattributable lines (rejected per Slice 2.5
Option 4 placeholder anti-pattern)
2. T-Rule-Enumeration β (YELLOW) scaffold needs bound per three-part
bridge rule. Added: allowed call sites scoped to emission code only;
max lifetime gated on rule_name_typo_fail_closed_landed graduation;
new emission-introspection consumers MUST author against closed-sum
α form, not β.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* WIP: R3 Substrate
* briefs: address codex BLOCKING bundle (sha 2ed1046e) — record-form provenance + emit_model carrier alignment + T-Rule-Enumeration retirement flag
Three findings addressed:
1. Lens brief (Finding #1 — provenance shape conflates source vs rule
attribution): replaced typed-sum EmissionOrigin with record
EmissionProvenance { rule: EmissionRule [mandatory], source_span:
Option<SourceSpan> [optional] }. Co-inhabitance is real (e.g.,
#[derive(Debug)] line has both rule=derive_for_disj AND span
attribution to Foo decl). Rule-mandatory enforces fail-closed
structurally; Practice 4 N/A (record, not coproduct).
2. T-Rule-Enumeration brief (Finding #2 — rule authority lives in
spec/rust.dag + v3.std.emit_model, not call-site strings): brief
status flipped to FLAGGED FOR RETIREMENT. Same finding as
smart-ram-167 STOP at gunbc#1759 #issuecomment-4392696623; Reading
C surfaced to Director at gunbc#828 #issuecomment-4392780871.
3. Rust-primitive-full-coverage brief (Finding #3 — BoundDeclaration
not wired into TargetIntegerInhabitanceBound row surface):
reframed Deliverable 2 to target the actual carrier
`v3.std.emit_model::TargetIntegerInhabitanceBound` (used by
`TargetIntegerTypeInhabitance` rows in src/v3/spec/rust.dag lines
169-207). Two paths surfaced: Option (i) wire BoundDeclaration into
TargetIntegerInhabitanceBound; Option (ii) populate via existing
variants if PlatformDependent already exists on the live carrier.
Worker DFS at dispatch picks. Deliverable 3 retargeted to
TargetIntegerTypeInhabitance rows (the actual surface).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* rust-primitive-full-coverage: codex inline confirms TargetIntegerInhabitanceBound has no PlatformDependent variant — Option (ii) infeasible; carrier refactor REQUIRED
Codex BLOCKING inline at PR #1910 line 65 (sha 98507c432) verified at
HEAD: TargetIntegerInhabitanceBound = BoundUnspecified | StaticBoundFact(IntInterval).
No PlatformDependent variant. Earlier brief framing of Option (ii)
(populate via existing variants) was infeasible.
Brief reframed: carrier refactor is required (not optional). Two
structural shapes for the refactor — (i.a) embed BoundDeclaration
(Mgr recommendation; single-authority discipline) vs (i.b) extend
variant set with PlatformDependent. Existing 5 rows migrate either
way; bootstrap snapshot semantic-equivalence non-negotiable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* emission-provenance lane: T-Rule-Enumeration RETIRED + Lens brief revised per Director Reading C
Director Reading C RATIFIED at gunbc#1739 #issuecomment-4392797954
(2026-05-07):
- Q1: T-Rule-Enumeration retires (substrate gap was phantom; ~37
*SyntaxBinding/*OpsBinding field paths already provide enumeration
per smart-ram-167 substrate-state-grep at gunbc#1759 #issuecomment-4392696623)
- Q2: Lens<EmissionProvenance> dispatches now (no precondition gate)
- Q3: EmissionRule = String of field-path (per feedback_reason_not_label)
Actions:
- Delete docs/briefs/r3-substrate-t-rule-enumeration-worker.md
- Lens brief: replace T-Rule-Enumeration precondition gate with
Reading C absorption section; EmissionRule = field-path String;
cementing test verifies field-path round-trip; STOP triggers
retargeted to *SyntaxBinding field-set divergence
Worker pin: smart-ram-167 (Mgr discretion per Director ratification;
fresh context on field-path enumeration). valiant-ibex-312 stays on
Rust-primitive-full-coverage / T-Interval-Representation.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 7, 2026
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* fix(briefs): retire INVARIANTS.md line-number cites in r3-426 brief
Per codex review on PR #1782 (sha c2512cde): the new briefs
introduced fresh `INVARIANTS.md:<line>` citations — the same
stale-line-anchor shape the :425 sweep brief itself names as needing
dissolution. Replaces four line-number references in
r3-426-sfp-subdoc-rename-and-author-worker.md (lines 16, 28, 58, 85)
with prose-anchored locators citing the section anchor + bullet text.
The other PB-lane briefs codex flagged (r3-pb-runtime-equivalence-
corpus-seed-audit.md, r3-pb-t-fixedpoint-worker.md) belong to
neat-bear-351's lane and are not in this PR's scope; will route
separately.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): correct L6 type count + r3-435 dual-path retirement receipt
Per openai-pro review on PR #1782 (sha c2512cde):
1. L6 brief said acceptance ratchet asserts "five types" exist but
declares six (ShapeATarget, FormAxis, BehaviorAxis,
MethodTemplateContractKey, EmissionCell, EmissionPathProjection).
Updated both occurrences to enumerate all six explicitly.
2. r3-435 brief authorized both Path A (typed-AST reader, deletes
IntegrationRsScan) and Path B (scanner widening with Char state),
but retirement receipt language only covered Path B. Updated
ROADMAP/ledger retirement and authority-audit step 5 to cover
both authorized dissolution paths.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): L6 second 'five types' site + T-E-P lookup-gate authority
Per cursor review on PR #1782 (sha 2072b986):
1. L6 brief acceptance bullet at :131 still said "five types" while
the parenthetical and the Slice step at :117 said six. The prior
fix only caught one of two sites; corrected the second.
2. T-E-P brief `e_p_call_pattern_lookup_authoritative` acceptance
named only `lower_call_pattern` as "the only route from call
site to LoweringTarget" — but Phase 2 explicitly requires
lenses route through `per_call_pattern_at` (the L-7 single-
authority lookup), NOT `lower_call_pattern` (compiler-internal).
Acceptance text rewritten to name `per_call_pattern_at` as the
consumer-facing gate and clarify `lower_call_pattern`'s
internal-only scope.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): strengthen L6 ratchet from count-parity to per-row bijection
Per openai-pro REQUEST_CHANGES on PR #1782 (sha 2072b986): the L6
acceptance ratchet asserted only row-count parity between
emission_path_projections and the union of MethodTemplateContract
rows. Count parity permits duplicate projection keys and missing
source rows — the bridge from list-non-empty proxy to per-row
projection would not be mechanically proven. The whole point of
introducing MethodTemplateContractKey { target, dag_method } is to
provide 1:1 row identity.
Rewrites the ratchet to assert per-row bijection: every
MethodTemplateContract source row has exactly one matching
EmissionPathProjection row by key; every projection key resolves
to exactly one source row; duplicate projection keys fail closed.
(The other openai-pro BLOCKING finding — T-E-P
e_p_call_pattern_lookup_authoritative naming the wrong authority
— was already fixed at 6282a0704; the review citation was against
the pre-fix sha 2072b986.)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): split L6 ratchet into slice-active vs deferred-activation gates
Per BLOCKING inline review on PR #1782 at d92459423: the prior
bijection ratchet was unsatisfiable as authored. The
*_method_template_contracts lists at HEAD (rust/python/go) are
non-empty; shipping emission_path_projections: [] alongside a
strict bijection requirement means the bijection cannot pass when
the slice lands — source rows would be uncovered. Prior text
incorrectly said "vacuously true on the empty source set" — the
SOURCE set isn't empty, only the projection set is.
Splits the ratchet into two gate categories:
- Slice-active gates: types/data shape exist + empty-state
predicate (emission_path_projections == []).
- Deferred-activation gate: per-row key bijection, authored as a
test scaffold by this slice but #[ignore]'d (or feature-gated
on len > 0) until Grounding's follow-up populates rows. On
activation, asserts the bijection over the populated set.
This honors P1 doc faithfulness: the slice's acceptance contract
is satisfiable as authored, and the bijection becomes the
load-bearing gate the row-population PR satisfies.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): require L6 bootstrap enrollment to honor P2 facts-flow-forward
Per BLOCKING inline review on PR #1782 at L6:34: the slice authored
src/v3/std/cross_target_coverage.dag but never required enrolling
it in the bootstrap/load authority. Without enrollment the file
is dead-letter on disk and EmissionPathProjection can be absent
from the downstream Dag — silent P2 facts-flow-forward violation.
Adds a new Slice step 2 requiring a cross_target_coverage field
on BootstrapFixtures in src/v3/std/extdeps_bootstrap_fixtures.dag
mirroring the *_method_template_contracts enrollment pattern, plus
a loader-visibility ratchet. Renumbers subsequent Slice steps.
Adds matching Acceptance bullet.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* WIP: R3 Substrate
* WIP: R3 Substrate
* docs(briefs): normalize Practice 4 citation in S3 brief per codex review
Codex review on PR #1782 (sha 2614217e) flagged stale citation
"Practice 4 (Step 4 of the type-introduction checklist)" in
S3 brief — no such checklist step exists in the pinned rubric.
Live rule per docs/modeling-discipline.md:131-134 is the "What to
check" checkpoint-comment requirement under Practice 4 (coproduct
dissolution). Updated the S3 reference to cite that line directly
and drop the stale "Step 4 of type-introduction checklist" wording.
S2/S8/S9 already use bare "Practice 4" without the stale steps wording.
S1 (committed earlier today) does not contain the stale phrase.
S6 (pre-existing brief flagged by reviewer as carrying the same
pattern) is out of this branch's scope; flag for separate paydown.
* docs(briefs): sweep stale Practice 4 citations corpus-wide
Codex review on PR #1782 flagged the stale "Step 4 of the
type-introduction checklist" wording — no such checklist exists in
docs/modeling-discipline.md. The live rule is the "What to check"
checkpoint-comment requirement under Practice 4 (coproduct
dissolution) at docs/modeling-discipline.md:131.
Sweep across docs/ found 2 additional briefs carrying the stale
phrasing (S3 fixed previously at 9db003d40):
- docs/briefs/r3-l6-emission-path-projection-substrate-worker.md
- docs/briefs/r3-x1b-s1-transform-dispatch-substrate-worker.md
Both updated to cite the live rule directly. Verified zero
remaining "type-introduction checklist" or "Step 4 of the"
references in docs/ tree.
* docs(briefs): add Practice 4 citation discipline note to authoring checklist
Per parent inbox status flag at gunbc#828 #issuecomment-4385346221:
add a Citation discipline section to brief-authoring-checklist.md
naming the correct shape — Practice 4 (coproduct dissolution) +
"What to check" rule at docs/modeling-discipline.md:131 — and
calling out the stale "Step 4 of the type-introduction checklist"
phrasing as a failure mode. References the corpus sweep at
2dc92f34f.
Prevents future briefs that copy old templates from re-introducing
the stale citation pattern.
* docs(r3): cascade work post-Director ratification — S1 worker, R4 routing, Unit/Quantity canvas
Per Director ratification at gunbc#828 inbox response 2026-05-06
(zesty-bear-812):
1. S1 Q1+Q2 RATIFIED → author S1 gap-test representative worker brief
(`r3-substrate-s1-gap-test-representative-worker.md`). Dispatchable
on T-E-P Phase 1 + E6-G0d landing; closes ledger row #61.
2. S2 Q2 RATIFIED (option-(b) T-LBP narrowing) + S3 + S8 deferrals →
R4 carve-out routing ledger (`docs/r4-carve-out-routing.md`).
Captures C1 (parallelism lens), C2 (effect_enum lens), C3
(register narrowing), C4 (deferred MachineConstraint axes),
C5 (rounding-mode product extension), C6 (Unit/Quantity).
Cascade messages to Verification / Grounding / Debt-Paydown Mgrs.
3. S9 Phase-3 dimensional refinement boundary → Unit/Quantity carrier
canvas (`r3-substrate-s9-unit-quantity-carrier-canvas.md`).
Surfaces Q-Unit-1..5 for Director ratification: carrier name,
2-axis vs 3-axis Phase-1, Duration/Seconds collapse, Refined
composition order, Practice 4 classification.
Updates §1.8 ledger receipts for carved items; R3 closure receipts
treat carved items as explicitly-deferred (not drift).
* docs(briefs): citation-discipline fix-forward — anchors/quotes, not line numbers
Director redirect at gunbc#828 #issuecomment-... noted the citation
paydown's replacement form (`docs/modeling-discipline.md:131`) is the
same drift class as the original fabricated-step error. Bare line
numbers shift silently on any doc edit above the cited line.
Generalized fix:
- brief-authoring-checklist.md "Citation discipline" section
rewritten to prohibit bare line-numbers across ALL cross-doc
references (modeling-discipline.md, r3-program-plan.md,
r3-structure.md, INVARIANTS.md, MODELING.md, etc.). Use section
anchors (#section-name) or rule-text quotes.
- 3 briefs that swapped in :131 (S3, S6, X1B) re-converted:
- docs/modeling-discipline.md#4-coproduct-dissolution anchor
- "What to check" rule quoted inline (self-contained)
Brief-authoring-checklist failure-mode section names both vectors
(fabricated step + bare line numbers) and references both sweep
commits in audit trail.
Broader cross-doc citation sweep across briefs I authored on this
branch (frontmatter authority-docs lists, body prose) is recommended
per Director note — flagging as separate paydown task; this commit
closes the immediate recurrence vector.
* docs(briefs): fix unfaithful Int=AbelianGroup<Nat> + sweep stale 'Step 4' refs
Per openai-pro review on PR #1782 sha 0ecd4818 (gunbc#1782 #...):
BLOCKING fix (P1 modeling faithfulness):
- S9 + S3 directed workers to model `Int = AbelianGroup<Nat>`,
but Nat is not an Abelian-group carrier (additive inverses
not representable). Reframed:
- `Int` is the additive Abelian group on integers (carrier Z);
`Nat` is a commutative monoid under addition.
- Algebraic relationship is **explicit group-completion**:
`Int ≡ GroupCompletion<CommutativeMonoid<Nat>>` — NOT
`AbelianGroup<Nat>` parameterization.
- Worker either models `Int` directly as `AbelianGroup`
(terminal Abelian-group instance) OR introduces explicit
`GroupCompletion<M>` substrate carrier; decision lands in
Phase-1 PR per P1 procedure.
- S3 cross-reference + Phase-4 algebra-axis preservation note
reframed accordingly.
NON-BLOCKING in-PR cleanup (Practice 4 citation discipline):
- 6 briefs still carrying stale 'Practice 4 Step 4' wording
(r3-coproduct-1/2/3, r3-l6, r3-t-e-p, r3-x1b) updated to use
the section anchor (`docs/modeling-discipline.md#4-coproduct-dissolution`)
+ inline rule-text quote ("What to check") form per the
citation discipline established in `b4cae5299`.
Closes both findings from openai-pro review.
* docs(briefs): Q-Unit-1..5 RATIFIED — apply Scale.Unit→One rename + ratification banner
Director ratified Q-Unit-1..5 at gunbc#828 #issuecomment-... 2026-05-06
(zesty-bear-812). All 5 questions ratified; one rename required:
- Scale enum's `Unit` value collides with outer `Unit<Q, S>` carrier
name (`Unit<Time, Unit>` is shadowing-confusing). Rename to `One`.
- Duration ≡ Unit<Time, One> (collapses with Seconds per Q-Unit-3).
- Milliseconds ≡ Unit<Time, Milli>.
Canvas updated:
- Top-level RATIFICATION banner records all 5 ratifications + rename
- Scale enum: `Unit` → `One` (with inline rename receipt)
- Sketch product expansions use `Unit<Time, One>` / `Unit<Time, Milli>`
- Phase-3 reframe table products updated
- Q-Unit-3 prose collapse reference updated
Subsequent worker brief authoring (Unit/Quantity carrier landing
post-prerequisite) inherits the renamed shape.
* docs(briefs): Unit/Quantity carrier worker brief — Q-Unit-1..5 RATIFIED
Worker brief consumes the ratified Unit/Quantity canvas
(`5f22fd06e` ratification banner). Phase-1 lands `Unit<Q, S>` +
`Quantity` + `Scale` declarations at the ratified shape (Scale's
`One` value per Q-Unit-1 rename). Phase-2 reframes S9 Phase-3
dimensional refinements (Duration / Seconds / Milliseconds) to
outer-Refined / inner-Unit composition per Q-Unit-4 RATIFIED.
EpochMs deferred to R4 C6 (Aspect-axis follow-up).
Practice 4 marks pre-determined per Q-Unit-5: `Unit<Q, S>` 🟢
PRIMITIVE; `Quantity` + `Scale` 🟡 SCAFFOLD with "consumed by ≥1
emission rule" dissolution trigger.
Cross-program handoff to Grounding Mgr (#1745) for dimensional
emission consumption documented.
Worker pin TBD — dispatchable when idle pool refreshes.
* docs(briefs): S9 Phase-1 reframed — Int landed at #1466; UInt-only migration
Per proud-lynx-311 dispatch ack at gunbc#1739 #issuecomment-...
2026-05-06: re-grep at HEAD revealed substrate state diverges
materially from S9 brief's "Option A vs B decide" framing.
Reality check at HEAD:
- `dsl/std/integer.dag:55` already carries
`type Int = AbelianGroup<GroupCompletion<Nat>>` per Slice 3 PR
#1466 / commit `4ba0a6d04`. Single-authority Q6 audit doc at
`docs/audit/t-numeric-construction-group-completion-6q.md`
ratified this shape and explicitly rejected the compact form.
- `dsl/std/integer.dag:56` still on legacy `type UInt = UInt64`
alias; UInt migration is the actual Phase-1 worker scope.
Brief Phase-1 step 1 reframed: Int landing acknowledged; worker
confirms HEAD match. Phase-1 step 2 reframed: UInt migration
target named (`type UInt = CommutativeMonoid<Nat>`); fixed-width
rows preserved per existing legacy-rows policy.
Historical Option A vs B framing kept as note (superseded).
Confirmed S9 dispatch ack from worker; UInt-only Phase-1 PR is
in-scope. Emission entries deferred per S3 (valiant-ant-72
pre-landing); Phase-2 deferred per S8 (quiet-boar-160 pre-landing).
* docs(briefs): fix two BLOCKING citation drift items per codex review
Codex review on PR #1782 sha 5f22fd06 flagged two authority-receipt
gaps:
1. brief-authoring-checklist.md:118 — `<this-commit>` placeholder in
the failure-mode section was unverifiable post-merge. Replaced
with `b4cae5299` (the actual anchor/quote conversion commit).
2. r3-378-http-path-helper-fail-open-worker.md:84 — `INVARIANTS.md
C-8 fail-closed` cited without anchor or rule-text quote, in
violation of the citation discipline added in this PR.
Converted to `INVARIANTS.md#p3-fail-closed` (P3 heading anchor)
+ inline rule-text quote naming P3 + C-8.
PR's own discipline now consistently applied across the diff.
* docs(briefs): sweep bare .md:NNN citations in Substrate-Mgr-authored briefs
Per Director recommendation in citation-discipline thread (gunbc#828
follow-on): bare line-number references drift silently. Swept 19
remaining instances across 5 briefs I authored on this branch:
- docs/briefs/r3-substrate-s1-q-class-2-chain-break-gap-test-canvas.md
- docs/briefs/r3-substrate-s1-gap-test-representative-worker.md
- docs/briefs/r3-substrate-s2-t-lbp-scope-calibration-canvas.md
- docs/briefs/r3-substrate-s3-machine-constraint-carrier-worker.md
Conversion: bare `file.md:NNN` → `file.md §"Section name"` form.
Stable; survives doc edits; human-readable. Section names per
HEAD heading audit.
Verification: `grep -nE '.md:[0-9]+' docs/briefs/r3-substrate-*.md
docs/r4-carve-out-routing.md` returns zero results in
Substrate-Mgr-authored scope.
Broader corpus scope (~232 references across non-Substrate briefs)
remains as separate paydown task.
* docs(briefs): fix BLOCKING + NON-BLOCKING items per openai-pro review on a462f007
Three findings addressed; 2 BLOCKING + 1 NON-BLOCKING:
BLOCKING — R3/R4 dual authority for Unit/Quantity carrier:
R4 carve-out C6 narrowed from "Unit/Quantity carrier landing"
(deferred all of it) to **"Aspect-axis (PointKind) follow-on
for EpochMs and instant/rate-shaped refinements only"**. The
2-axis Measure<Q,S> carrier landing IS in R3 per Q-Unit-1..5
RATIFIED; only the Aspect axis (PointKind = Magnitude | Instant
| Rate) carves to R4. Cascade-message C6 reference + R4 program-
plan input list updated to match. Cross-Mgr message clarifies
Grounding consumes 2-axis dimensional refinements in R3.
BLOCKING — Locked design partially applied (Scale.Unit→One
rename in canvas alternate-sketch sections):
- Canvas line 105-108 alternate sketch ("Open question: Instant
vs Duration axis") still used pre-rename `Unit` Scale value;
added pre-ratification disclaimer marking the section as
deferred to R4 C6 + noting `Unit` is pre-rename
- Canvas line 163 Q-Unit-3 collapse note: `Scale = Unit` → `Scale = One`
(BLOCKING — bare line-numbers in S1 canvas frontmatter): NOT
APPLICABLE — already fixed at sweep commit 127287a66 prior to
review; reviewer was on sha a462f007 pre-sweep.
NON-BLOCKING — placeholder receipts:
- brief-authoring-checklist.md `#issuecomment-...` → concrete ID
4385393971 (Director's anchor-discipline reply)
- r3-substrate-unit-quantity-carrier-worker.md (3 occurrences)
+ canvas RATIFICATION banner: `#issuecomment-...` → concrete ID
4385412256 (Q-Unit ratification reply)
- `<this-commit>` placeholder in checklist already fixed at 5be1c99f2
Verification: zero `#issuecomment-...` or `<this-commit>` placeholders
in docs/briefs/ + docs/r4-carve-out-routing.md.
* docs(briefs): codex BLOCKING review fixes — Practice 4 source-level + policy scope
Three BLOCKING findings on PR #1782 (codex sha b913e2f4):
1. **Citation-discipline policy scope**: addressed by scoping
the rule prospectively to briefs landing after the policy
commit (`b4cae5299`). Pre-existing corpus references are
flagged as paydown but NOT individually blocking; coordinated
single-PR sweep is the right paydown shape. Substrate-Mgr scope
already swept at `127287a66`.
2. **R4 C6 reconciliation**: already fixed at `6e5a5481a` (R4 C6
narrowed to Aspect-axis follow-on for EpochMs only; 2-axis
carrier landing IS in R3). Codex sha b913e2f4 was pre-fix.
3. **Practice 4 source-level vs PR-body conflation**: corrected.
Acceptance bullet for Unit/Quantity worker brief now requires
inline checkpoint comments on the LIVE declarations in
`dsl/std/units.dag` (load-bearing artifact per
`docs/modeling-discipline.md#4-coproduct-dissolution` "What to
check"). PR-body summary is supplementary, not substitute.
Closes the review's reconciliation gates.
* docs(briefs): absorb Q-MachineConstraint + Q-Unit-1-Recanvas ratifications
Two major Director ratifications absorbed:
1. Q-MachineConstraint-Carrier RATIFIED at gunbc#828
#issuecomment-4385530115 (Brian directive 2026-05-06: "universal
substrate, ratify defaults"; PR #1817 queued). 6 sub-decisions:
- MachineWidth<bits> only for R3; defer RegisterClass / EndianMode
/ Alignment / signedness post-R3
- Parametric Compose<Algebra, MachineConstraint> type-level
interaction shape (lookup-maps REJECTED — AlgebraMachineProduct
record-keyed table form retracted)
- Int<64> ≡ Compose<AbelianGroup, MachineWidth<64>> (algebra-side
options A and B preserve interaction semantics)
- Real<64> ≡ Compose<ApproximateField<Rational>, MachineWidth<64>>
(algebra approx + machine approx as independent composing axes)
- ≥3 pairs is minimum, not target
- UNIVERSAL substrate: every target carries machine-constraint
facts; targets without native machine-width semantics handle
omission at Grounding-level discharge (target-conditioned
lowering NOT target-conditioned substrate)
Reframed: S3 brief Phase-1, S3 Phase-3/4 emission entries,
S8 Phase-1/2/3 (ApproximateField<Real> → ApproximateField<Rational>;
AlgebraMachineProduct → Compose<...>), S9 emission entries.
2. Q-Unit-1-Recanvas RATIFIED Measure<Q, S> at gunbc#828
#issuecomment-4385539791. valiant-ibex-312's substrate-state
grep caught Unit<Q, S> collision with kernel Unit at
dsl/std/types.dag:176; Director ratified Measure<Q, S> as
outer carrier name (conventional in scientific computing;
no dsl/std/ collision).
Reframed: Unit/Quantity worker brief throughout (Unit<Q,S> →
Measure<Q,S>; dsl/std/units.dag → dsl/std/measure.dag); canvas
spec sections + sketch products (Refined<Measure<Time, ...>...>);
R4 carve-out C6 with informal/formal name disambiguation note.
Both rename + reshape are now consistent across S3 / S8 / S9 /
Unit-Q briefs + canvas + R4 routing. Workers (valiant-ant-72 on S3,
quiet-boar-160 on S8, proud-lynx-311 on S9, valiant-ibex-312 on
Unit-Q) all dispatch against the ratified shape.
* docs(briefs): S11 Slice C worker brief — ROADMAP :425 retirement post-#1801
Cascade-clearance trigger fired post-#1801 (Slice B) merge
2026-05-06. S11 brief authored per B1 RATIFIED prose+regen
bundling discipline.
Phase 1: per-file edits across the post-A+B residual list (~10
files per smart-ram-167's Slice B post-rebase enumeration);
anchor + rule-text-quote pattern per current citation discipline;
bootstrap regen in same PR (no split).
Phase 2: ROADMAP :425 row PARTIAL → Retired with 3-slice receipt;
Q-Slice-C-Retirement-Receipt resolution.
Worker pin smart-ram-167 (#1759) — Slice B precedent owner.
Dispatch packet sent separately to #1759.
* docs(briefs): Q-Refined-Phantom-Composition (c) RATIFIED — defer literal Refined form
Director ratified option (c) at gunbc#828 #issuecomment-... 2026-05-06:
drop `non_negative` predicate at Measure (phantom) layer; rely on
`Quantity = Time` tagging for non-negative-magnitude semantics.
Q-Unit-4 outer-Refined / inner-Measure composition is preserved as
**conceptual model**; the literal `Refined<Measure<...>, predicate>`
form is **deferred to value-typed-integration follow-up** when
substrate gains Refined<> as first-class type expression and
dimensional refinements get value-typed integration.
Updates:
- Canvas RATIFICATION banner Q-Unit-4 row reframed to "conceptual
model; literal form deferred"
- Canvas Phase-3 reframe table: Duration / Milliseconds drop the
Refined<...> wrapper; rely on Quantity tag
- Worker brief Phase-2 prose + reframe table + Acceptance bullet
all reflect deferred-literal-form posture
valiant-ibex-312 unblocked to proceed with Phase-2 (Grounding
handoff doc) per the (c)-shaped Acceptance.
* docs(briefs): S12 F2 + F8 doc-sharpening worker brief — post-#1820 cascade
#1820 (Slice C) merged 2026-05-06; ROADMAP :425 row Retired closes
the prose-reference paydown program. S12 cascade-clearance trigger
fires per B6 RATIFIED single bundled Mgr-tier doc-sharpening PR.
Scope: F2 (.v3 filename-suffix grammar dispatch) + F8 (bootstrap
load-order/exclusion authority) — single bundled PR per B6.
Cross-program coordination with PB Mgr (#1742) per design schedule
§3 P5 (comment-only OR co-author; no duplicate PR).
Worker pin smart-ram-167 (#1759) — Slice C precedent owner;
pattern-familiar with prose+regen bundling discipline. Dispatch
packet sent separately to #1759.
* docs(briefs): S9 brief — UInt landed via #1818; algebra label corrected to CommutativeSemiring
Per proud-lynx-311 merge report at gunbc#1746 #issuecomment-...
2026-05-06: PR #1818 landed `type UInt = Nat` after codex BLOCKING
review on initial `CommutativeMonoid<Nat>` form (P2 facts-flow-forward
violation — wrap projects out Nat's multiplicative monoid + identity 1).
Final landed form preserves full `CommutativeSemiring<Magnitude>`
surface — Nat IS a commutative semiring (addition + multiplication
+ 0 + 1). Algebra-side label for UInt is CommutativeSemiring, NOT
CommutativeMonoid (corrects my earlier dispatch labeling).
S9 Phase-1 step 2 marked LANDED with merge ref. Phase-1 step 3
emission entries reframed:
- Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>>
- UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>>
Worker discipline + reviewer caught a substrate-state error in my
brief framing; brief now matches landed reality.
* docs(briefs): fix contradictory (c) guidance in Unit/Quantity worker brief
Codex BLOCKING review on PR #1825 sha 1bce0418 caught a real
contradiction: brief Phase-2 prose at lines 115-118 stated
"predicates do NOT attach at the Measure (phantom) layer" per
Q-Refined-Phantom-Composition (c) RATIFIED, but immediately
following paragraph at lines 130-132 still said "Refined predicate
non_negative applies at the Measure-typed level" — left over from
the original Q-Unit-4 framing.
Updated lines 130-137 to consistently apply (c) shape: predicates
deferred to value-typed-integration follow-up; dimensional semantics
ride on Quantity tag at Measure layer; worker authors 3 in-scope
refinements WITHOUT predicate at this layer.
P2 boundary discipline restored: brief now gives single-authority
guidance for Phase-2 substrate landing.
* docs(briefs): name predicate authority + concretize ratification ID — codex BLOCKING
Codex BLOCKING review on PR #1825 sha 32541a56:
1. (BLOCKING) Q-Refined-Phantom-Composition (c) ratification deferred
phantom-carrier predicates without naming the value-level predicate
authority. Updated canvas Q-Unit-4 RATIFICATION row to explicitly
name: 'predicates re-attach at value-typed Int/UInt level via
where-sugar (e.g., range(min: 0) per current dsl/std/types.dag
precedent)' when value-typed integration lands.
2. (BLOCKING) Worker brief option-c reframe didn't sweep the audit
receipt §5 (Carrier dissolves the bridge). Updated to consistently
apply (c) shape — dimensional semantics ride on Quantity tag at
Measure layer; value-level predicate authority deferred; until
then, range/gt_zero constraints attach at value-typed Int/UInt
via where-sugar per dsl/std/types.dag convention.
3. (NON-BLOCKING) Canvas had #issuecomment-... placeholder for the
(c) ratification message ID. Replaced with concrete
#issuecomment-4385687075 per dashboard relay header.
Brief now names a single value-level authority (Int/UInt where-sugar)
for value-range facts; phantom-layer Quantity tag carries only
dimensional semantics. P2 boundary discipline maintained.
* docs(briefs): S12 RETIRED per Director option (2) ratification
Director ratified option (2) at gunbc#828 #issuecomment-4385873...
2026-05-06 post smart-ram-167's substrate-state STOP-AND-ESCALATE:
- F2 absorbed by 2026-05-04 routing + #1662 closure
(`Result/DivError span.file-keyed` retired)
- F8 topical concerns distributed across live rows
(Class 5 Gap 3 `bootstrap.rs::std_fixtures()`; Character-level
load-set / bootstrap; Filename-sentinel-bridges) with own owners
- Doc-discipline goals absorbed by per-Mgr citation paydown
routing at gunbc#828 #issuecomment-4385583247
Frontmatter updated to RETIRED with absorption receipt. Worker
(smart-ram-167) pin freed.
Director-side acknowledgment of Substrate Mgr's option (2) recommendation:
"Substrate Mgr's option (2) recommendation is the right move
structurally. Most Mgrs default to 'synthesize work to fill the brief'
when canvas-vs-HEAD mismatch surfaces. Recommending retirement when
the work is genuinely absorbed is harder discipline."
* docs(briefs): S7 PR-F (BoundDeclaration consumer + ReferenceModel<T>) worker brief
Per cascade-clearance trigger fired post-#1782 merge 2026-05-06.
loyal-wolf-828 (#1764) acknowledged S7 dispatch with substrate-state
pre-flight grep (gunbc#1764 #issuecomment-4385872...) and surfaced
await-brief posture. This brief satisfies that prerequisite.
Worker pre-flight inventory absorbed:
- BoundDeclaration substrate landed (substrate.dag StaticBound +
PlatformDependent); partial BoundDeclarationView consumer in
fold.rs (ScratchIntExamples scope; doc-explicit partial)
- ReferenceModel<T> NOT yet in dsl/std/ (substrate-fact-introduction
required; design-emission-model.md §Q2 framing)
Phase-1: BoundDeclaration consumer broadening to full
design-emission-model.md surface (u128 / isize / usize / walker
arms / pilot mirror).
Phase-2: ReferenceModel<T> P1 substrate-fact-introduction with
DFS-of-concept-DAG + Practice 4 in-source checkpoint comment +
named consumer demand (Grounding G1).
Phase-3: cross-program handoff to Grounding Mgr (#1745) for
G1 (T-Ground-Rust Phase 1).
Closure predicate: unblocks Grounding T-Ground-Rust Phase 1.
Worker pin loyal-wolf-828 holds; dispatch fires immediately
on this brief landing.
* docs(briefs): S9 Slice 2.5 — predicate-registry + PB-1 shim retirement (Path 3 RATIFIED)
Director ratified Path 3 at gunbc#828 #issuecomment-4390333451
post proud-lynx-311's pre-flight DFS at gunbc#1746 #issuecomment-4390253616.
Brief reframes Slice 2.5 from S-sized (primitive + 1 consumer) to
L-sized (predicate registry + PB-1 shim retirement). 5 deliverables:
1. gt_zero primitive predicate
2. Predicate registry infrastructure (resolver-side substrate-fact-introduction)
3. PB-1 shim retirement (span.file=='dsl/std/types.dag' special case dissolves)
4. Migration of existing types.dag predicates (range, pattern, etc.)
5. PositiveInt = Refined<Nat, gt_zero> consumer update
Per Director hard scope bars: no parallel predicate-resolution path;
no per-call-site special-casing; PB-1 shim retirement non-negotiable
in same slice. Single PR per Path 3 RATIFIED.
Discipline anchors: feedback_construction_over_ratchets (PB-1 shim
is bridge); feedback_dissolve_bridges (registry is structural fix);
feedback_no_textual_enforcement_bridges (file-path special-case is
anti-pattern shape); P1 substrate-fact-introduction procedure.
Worker pin proud-lynx-311 — Director pre-cleared on pre-flight DFS
context. PR #1840 (partial Slice 2 — NonNegativeInt = Nat) MERGED
at 16:54Z; Slice 2.5 dispatchable immediately on this brief landing.
* docs(briefs): Slice 2.5 audit receipt — Mgr independent grep parallel-verifies worker DFS
Codex BLOCKING review on PR #1845 sha 2d230949 flagged: external PR
receipt absorbed as live substrate state without re-grepping local
std declarations at brief authoring time. Authority receipt §1
cited proud-lynx-311's pre-flight DFS as authority but didn't
record Mgr-side independent verification.
Updated audit receipt §1 + §4 to record Mgr's independent grep at
HEAD as parallel verification:
- gt_zero zero matches in dsl/std/ + src/v3/ (verified)
- NonNegativeInt = Nat at dsl/std/integer.dag:133 (#1840 land)
- PositiveInt = Int where range(min: 1) at dsl/std/types.dag:255
- PB-1 shim doc-block confirms bridge-text framing
- Predicate registry confirmed not present
Worker DFS remains the dispatch-time discipline (substrate-state
grep at dispatch); Mgr's independent grep is parallel verification,
not substitute. Per Director's pattern note at gunbc#828
#issuecomment-4390199218 + #issuecomment-4390333451: dispatch-packet
authoring discipline includes Mgr-side substrate-state-grep, not
just relay of worker findings.
* docs(briefs): Slice 2.5 brief — Int extensibility scope-bound; lower.rs cites anchored
openai-pro REQUEST_CHANGES on PR #1845 sha 2d230949 flagged 2 issues:
BLOCKING — P1 modeling faithfulness / substrate scope leak:
`gt_zero` Int-side extensibility was left to 'worker judgment if
generalization cheap'. Per feedback_construction_over_ratchets:
don't speculatively extend predicate-carrier compatibility ahead
of named consumer demand. Reframed to **EXPLICITLY OUT-OF-SCOPE**
for Slice 2.5; Nat-only registration. Future Int-side consumer
surfaces as separate substrate-fact-introduction (P1 procedure)
brief with own consumer-demand receipt.
NON-BLOCKING — citation discipline self-violation:
3 instances of bare `lower.rs:821-835` line-range citations
contradict the brief's own 'no bare :NNN' requirement at line 211.
Converted to function-name anchored form:
`src/v3/compiler/src/lower.rs` (`lower_type_alias_refinements_phase` doc-block).
Both findings absorbed; no parallel scope ambiguity remains;
brief now self-consistent on citation discipline.
* WIP: R3 Substrate
* docs(briefs): Slice 2.5 brief — audit-receipt cites converted to declaration-name anchors
cursor APPROVE_WITH_COMMENTS on PR #1845 sha ebf6ac12 noted citation
self-inconsistency: audit-receipt section §1 + §4 used bare
`dsl/std/integer.dag:133` and `dsl/std/types.dag:255` line-number
pins despite the brief's own no-bare-:NNN rule.
Converted to declaration-name-anchored form:
- 'declaration named NonNegativeInt' (with reproducible grep command)
- 'declaration named PositiveInt' (with reproducible grep command)
- 'lower_type_alias_refinements_phase doc-block' for PB-1 location
Brief now fully self-consistent on its own citation discipline.
Verified zero remaining .dag:NNN / .rs:NNN / .md:NNN bare cites.
* docs(briefs): Slice 2.5 brief — Path (a) RATIFIED carrier+argument contract verification
Director Path (a) RATIFIED at gunbc#828 #issuecomment-4390760353
(triage) + #issuecomment-4390794121 (convergence ack):
predicate-body lowering with carrier+argument contract
verification at lower-time IN SAME SLICE. Placeholder semantics
REMOVED from registry path.
Codex REQUEST_CHANGES surfaced that registered-predicate-without-
contract = silent acceptance per feedback_fail_closed_discipline.
Brief reframe absorbs Path (a):
Hard scope bar #5 added:
- Same-slice carrier+argument-contract verification
- Mismatched carrier OR malformed argument → Diagnostic::ResolveError
- No 'named but not checked' path remains
Scope expansion section added (L → XL):
- Predicate-body lowering for all 7 registered predicates
- Per-predicate arg-shape validation
- Multi-session work acknowledged
- Carrier-check at 788d6acb4 retained; predicate-body builds on top
Acceptance bullets updated: gt_zero with arg-shape validation +
all 7 predicates with carrier+argument contract verification.
Director-pinned brief-authoring discipline: dissolution trigger
named in a brief authored as substrate-fact-introduction must
specify same-slice acceptance, not deferred follow-up. Otherwise
the brief encodes 'land bridge + defer dissolution' anti-pattern,
the same shape feedback_construction_over_ratchets warns against
at implementation layer applied at brief authoring layer.
This is the 11th miss this session — but the pattern surfacing it
(direct sweep + Codex REQUEST_CHANGES + Director ratification)
produces structural fixes that prevent the next miss class. Brief
revision is the structural recovery.
* docs(briefs): Slice 2.5 brief — Path 2 RATIFIED (Gap 1 discharge + Q-Regex carve-out)
Director Path 2 RATIFIED at gunbc#828 #issuecomment-4391985613
post Rung 3 STOP at gunbc#1746 #issuecomment-4391946213.
Path 2 narrowing of Path (a):
- gt_zero + range body synthesis already landed (5f3c40c3a)
- Gap 1 discharge mechanism: extends scalar-literal predicate
evaluation at lower-time. L-class; lands SAME SLICE per
same-slice-dissolution discipline
- non_empty + brand body synthesis: lands via Gap 1 discharge
- pattern / format / content: fail-closed-with-named-dep at
user-code authoring layer (Diagnostic::ResolveError naming
Q-Regex-Primitive). NOT placeholder; structural rejection
- PB-1 shim FULLY RETIRED per hard scope bar #3
- Q-Regex-Primitive carved as follow-on substrate-fact-introduction
(XL substrate; not pre-authored per feedback_construction_over_ratchets)
Acceptance updated:
- 4 of 7 body-synthesized in same slice
- 3 of 7 fail-closed-with-named-dep (no placeholder semantics)
- Gap 1 discharge mechanism landed (no partial implementation)
Worker disposition: proud-lynx-311 continues Rung 3 with Gap 1
implementation + non_empty/brand body synthesis + pattern/format/
content fail-closed-with-named-dep diagnostic.
* docs(briefs): S9 Phase-1 Step 3 emission entries — Tier-1 brief authoring (1/5)
First Tier-1 brief landing post Director auto-nudge on assignment
#1858. Authored against Q-MachineConstraint sub-decisions ratified
shape now landed at #1856:
- Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>> → Rust i32/i64/i128
- UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>> → Rust u32/u64/u128
- 6 concrete instantiations (≥3 minimum per sub-decision 5)
- Cross-program handoff to Grounding Mgr (#1745) for G2 emission consumption
- numeric_construction_demonstration Acceptance bullet (Int<32> round-trip);
Real<64> half deferred to S9 Phase 2 / S8 Float cascade
Brief authored with substrate-state-grep discipline + same-slice-
dissolution discipline (no deferred trigger; demonstration in same
slice or explicitly carved). 5-question authority audit completed
in brief body.
Worker pin proud-lynx-311 (S9 holds); brief queues post-Slice-2.5
completion as natural follow-on.
Tier-1 brief queue progress:
- 1/5 authored (this brief)
- 4 remaining: S9 Phase-2 Float coordination; T-LBP complexity
cementing; T-LBP cost cementing; S3 Phase-2 parser-grammar
* docs(briefs): S3 Phase-2 parser-grammar surface — Tier-1 brief 2/5
Authored per Director freed-pool pressure at gunbc#828
#issuecomment-4392095857 + Tier-1 brief-queue commitment.
Brief covers parser-grammar surface for Compose<Algebra, MachineConstraint>
interaction syntax post-#1856 (Phase-1 carrier slice landed):
- Phase 2.1: Parser surface per ratified Q-MachineConstraint-Grammar-Shape
- Phase 2.2: Bootstrap demonstrator with >=3 algebra*machine-axis pairs
- Phase 2.3: Class 1 5-criteria Pass receipt (criteria 1-4; criterion 5
v2-oracle parity may carve)
- Phase 2.4: numeric_construction_demonstration co-receipt with S9 Phase-1
Step 3 emission entries brief
Director ratification HOLDS on grammar bikeshed:
- Candidate A: Int @ Width<32> (annotation form)
- Candidate B: Int with MachineWidth<32> (with-form)
- Candidate C: Int<32> (positional generic — fits existing convention)
- Candidate D: Compose<AbelianGroup, MachineWidth<32>> (no parser change;
type aliases provide convenience names)
Worker pin valiant-ant-72 holds (S3 Phase-1 precedent owner); freed-pool
until Director ratifies grammar shape.
Tier-1 brief queue progress: 2/5 authored (this brief + S9 Phase-1 Step 3).
Remaining: S9 Phase-2 Float coordination; T-LBP complexity cementing;
T-LBP cost cementing.
* docs(briefs): S3 Phase-2 brief — fix internal contradiction on Class 1 closure scope
Codex BLOCKING review on PR #1889 sha ec609b4a flagged real
contradiction:
- Provenance section claimed '5-criteria Pass closed' (line 229)
- Phase 2.3 said 'criterion 5 may queue separately' (line 154)
- STOP-AND-ESCALATE bullet 1 ratified 4-of-5 carve-out (line 181)
Updated Provenance to consistently state criteria 1-4 close in this
slice; criterion 5 (v2-oracle parity) cementing test may carve to
separate slice. Single 4-of-5 framing throughout. INVARIANTS.md
'Documentation Describes Live State' satisfied.
Per feedback_construction_over_ratchets: don't claim closure that
isn't structurally complete. 4-of-5 + named carve is the honest
shape; 5/5 only on cementing-test land.
* briefs: cite Q-MachineConstraint sub-decision 6 supersession of T-V2-Retirement gate
Address Codex BLOCKING #1 on PR #1889: S3 Phase-2 + S9 Phase-1 Step 3
worker briefs consumed Q-MachineConstraint ratification without
reconciling the historical T-V2-Retirement-landing-first cascade
gate. Add an explicit precondition-framing section to each brief
citing Director ratification at gunbc#828 #issuecomment-4385530115
(Q-MachineConstraint sub-decision 6 UNIVERSAL substrate posture)
as the implicit supersession; PR #1856 landing without T-V2-Retirement
first confirms supersession in practice. STOP-AND-ESCALATE if
supersession contested at dispatch.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* slice-2.5 brief: mark pattern/format/content Path 2 acceptance pending Director ratification
Address Codex BLOCKING #2 on PR #1889: Path 2 deliverable + acceptance
language asserted fail-closed-with-named-dep semantics for
pattern/format/content, but Director rejected that shape (Option 4)
at gunbc#828 #issuecomment-4392.... Options A (consumer migration),
B (Q-Regex bundled into slice), C (typed fail-closed exception with
named trigger) await ratification per migration-cost catalog at
#issuecomment-4392116215. Brief now flags those three predicates +
PB-1 shim retirement coupling as PENDING DIRECTOR RATIFICATION; worker
STOP-AND-ESCALATE if dispatched pre-ratification.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* slice-2.5 brief: reconcile PB-1 retirement shape with Options A/B/C gating
Address Codex BLOCKING #4 on PR #1889: prior text claimed "PB-1
FULLY RETIRED" while preceding bullet said retirement gates on
chosen path — internal contradiction. Reframe to gate the
pattern/format/content half of shim retirement on Options A/B/C
ratification while preserving Director hard scope bar #3 binding
on the gt_zero/range/non_empty/brand half (cleanly retired via
Gap 1). No path leaves PB-1 alive as parallel-authority shim on
slice landing.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: tier-1 3/5 — S9 Phase-2 Float emission coordination worker brief
Tier-1 brief queue progress (3/5): S9 Phase-2 Float<N> emission
entries (Float<32>/Float<64> via Compose<ApproximateField<Real>,
MachineWidth<N>>) + Real<64> round-trip demonstration completing
§1.8 #67 Acceptance bullet. Coordination/synthesis brief — gates on
both S8 (ApproximateField<F> + Real base carrier) and S9 Phase-1
Step 3 (Int<N>/UInt<N> emission-entry pattern) landing before
dispatch. Inherits T-V2-Retirement supersession framing from
Phase-1 Step 3. STOP-AND-ESCALATE on missing preconditions or
unratified Q-ApproximateField-Axiom-Set.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: ledger advancement honesty (PRODUCER_LANDED not CONSUMER_LANDED) + restore exact issuecomment IDs
Address openai-pro BLOCKING + NON-BLOCKING findings on PR #1889:
BLOCKING (boundary discipline / same-PR consumer): S9 Phase-1 Step 3
+ Phase-2 Float briefs claimed DECLARED → CONSUMER_LANDED upon merge
while only requiring cross-program handoff receipt to Grounding G2.
For target-primitive substrate entries, handoff is not landed-consumer
proof. Reframed both briefs to advance rows DECLARED → PRODUCER_LANDED;
Grounding G2 follow-on PR (per-pair lowering rules + emitted-Rust-
primitive verification) advances to CONSUMER_LANDED. Default
split-PR producer-then-consumer per bundled-scope discipline ratified
at gunbc#1739 #issuecomment-4392225548 (parallel infrastructure
DISALLOWED in same PR).
NON-BLOCKING (citation hygiene): replaced ellipsized
"#issuecomment-4392..." stubs in slice-2.5 brief with the exact
Director Option-4-rejection comment ID #issuecomment-4392081719 so
the ratification claim is mechanically checkable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* slice-2.5 brief: absorb Director Option A revised RATIFICATION (final shape)
Director ratified Option A revised at gunbc#1739 #issuecomment-4392329709
(2026-05-06). Final brief shape:
- 4 predicates body-synthesize via Gap 1 (gt_zero / range / non_empty
/ brand)
- pattern / format / content STRUCTURALLY ABSENT from registry (not
pending-with-placeholder); user-code where-clause produces
Diagnostic::ResolveError naming Q-Regex-Primitive as separate
substrate-fact-introduction follow-on
- ~12 types.dag refined-type declarations migrate to drop
pattern/format/content where-clauses with inline-comment receipts
pointing to Q-Regex restoration path; 1 extdep (sts_endpoint)
adapts as needed
- PB-1 shim FULLY RETIRED — no exception, no preserved branch
- Q-Regex-Primitive as separate brief / worker pin / Director
ratification cycle when concretely needed (Option B XXL bundling
REJECTED)
Phase ordering rewritten to absorb migration as explicit phase
between predicate enrollment and shim retirement. proud-lynx-311
re-dispatches against revised brief; PR #1846 absorbs additional
scope per Director directive.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* s3-phase-2 brief: absorb Q-MC sub-decision 3 ratification (grammar shape was already answered)
Director observation at gunbc#1739 #issuecomment-4392382517: Q-MC
sub-decision 3 (Brian directive 2026-05-06 at gunbc#828
#issuecomment-4385530115) already ratified the grammar shape —
`Int<N>` user surface desugars to `Compose<Int, MachineWidth<N>>`
substrate parametrically. Earlier brief HOLD framing was redundant.
Brief revised:
- Replace bikeshed section with Q-MC sub-decision 3 quotation +
surface/substrate framing (Candidate C user-facing; Candidate D
substrate elaboration; A/B rejected per Director assessment)
- Phase 2.1 parser surface scoped to numeric-literal-position
recognition + parametric desugar
- Front-matter status: dispatchable per ratification (not HOLD)
- Provenance: ratification-state-grep discipline pin folded into
standing dispatch-checklist
valiant-ant-72 dispatch fires immediately per Director directive.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: remove truncated issuecomment-4392... stub from S9 Phase-1 Step 3 provenance
Address cursor NON-BLOCKING finding on PR #1889 sha 94292e48: line 144
truncated issuecomment ID. The originally-intended Director auto-nudge
comment doesn't map cleanly to a single ID, so replace with the
standing assignment reference (gunbc#1858) which is unambiguous.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: tier-1 4/5 + 5/5 — T-LBP complexity + cost lens cementing-test worker briefs
Tier-1 brief queue completes (4/5 + 5/5):
4/5 — T-LBP complexity-lens cementing test (§1.8 rows #79 + #87
first entry). Consumer-side cementing test against frozen v2-oracle
snapshot; producer-side Lens<Complexity> instance is upstream
precondition. Per Q-Lens-Behavioral-Parity-R3-Closeability option
(b) RATIFIED at gunbc#828 #issuecomment-4385329180 — T-LBP narrowed
to complexity + cost lenses only.
5/5 — T-LBP cost-lens cementing test + cost_lens_demonstration
(§1.8 rows #80 + #70 + #87 second entry). Absorbs row #70 demo
naturally — same source corpus + lens run; demonstration adds
observable-cost-bound assertions on top of parity check. ≥2
algebra-instances composed + ≥1 recursive call required per row #70.
Both briefs gate on:
- T-LBP per-lens substrate-producer brief landing (Lens<C> instance)
- Frozen v2-oracle snapshot capture (separate brief, pre-v2-retirement)
- DO NOT capture-against-live-oracle (violates
v2_oracle_no_remaining_test_consumers)
- DO NOT bundle producer/snapshot/algebra-instance authoring per
Director bundled-scope ratification at gunbc#1739
#issuecomment-4392225548
Tier-1 5/5 complete. Worker pins assigned at dispatch time once
preconditions land.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* canvas: emission-provenance substrate shape disposition (supersedes PR #1902 PROPOSAL)
Substrate Mgr canvas authored post-codex BLOCKING + PM disposition
handoff at gunbc#846 #issuecomment-4392510255.
Two real substrate-shape questions stacked:
Q1 (category mismatch): Lens<C>.read is per-Behavior; emission
provenance is per-emitted-line. 4 options surfaced — (a) per-Behavior
Lens-compatible / (b) per-line instrumentation NOT a lens / (c)
withdraw + re-canvas / (d) Mgr-surface both substrates distinct.
Mgr recommends (d) if visualization is load-bearing; (a) suffices
if per-Behavior aggregate works for the slide. Brian-channel
sub-question surfaced.
Q2 (fold-rule enumerability): T-Rule-Enumeration substrate-fact-
introduction is prerequisite for every Q1 path (rust_target.rs uses
inline &str template names; rules not enumerable). Mgr recommends
parallel-dispatch with Q1 deliberation — satisfies Brian's ASAP
framing while shape ratifies.
Q3 (§1.8 ledger retarget): #89 already taken by T-LAS; new cluster
location follows Q1 disposition.
Director ratification ask + Brian-channel sub-question surfaced.
PR #1902 close-superseded post-ratification.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: reconcile Compose<...> slot-1 spellings to algebraic-concept names per Q-MC sub-decision 3
Address cursor NON-BLOCKING on PR #1889 sha 2b49b30f: S3 Phase-2
brief mixed Compose<AbelianGroup, ...> with Compose<Int, ...>; S9
Phase-1 Step 3 used Compose<AbelianGroup, ...> + Compose<CommutativeSemiring,
...>; S9 Phase-2 used Compose<ApproximateField<Real>, ...>. All wrong
per Q-MC sub-decision 3 critical correction at gunbc#828
#issuecomment-4385530115:
"prior phrasing Compose<AbelianGroup, MachineWidth<64>> was wrong
— AbelianGroup<T> is a witness shape over carrier T, not a carrier
constructor; bare AbelianGroup in slot-1 composes witness rather
than concept."
Slot-1 is the fully-applied algebraic-concept name:
- Int<N> = Compose<Int, MachineWidth<N>> (Int = AbelianGroup<GroupCompletion<Nat>>)
- UInt<N> = Compose<UInt, MachineWidth<N>> (UInt = CommutativeSemiring<Nat>)
- Real<N> = Compose<Real, MachineWidth<N>> (Real = ApproximateField<Rational>)
Phase-2 Float brief also retargets user-facing surface from Float<N>
to Real<N> (Float is target-language name for the Rust primitive;
Real is the algebraic-concept name per Q-MC sub-decision 3 example).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* s9-p1-s3 brief: scope Deliverable 4 to producer-side verification (defer demonstration to Grounding G2)
Address Codex BLOCKING (line 82 inline + bundled review): brief
claimed end-to-end Int<32> → Rust i32 round-trip in Acceptance
while Deliverable 3 said lowering rules belong to Grounding G2
follow-on. Internal contradiction.
Reframe per Director bundled-scope discipline (gunbc#1739
#issuecomment-4392225548 — parallel infrastructure DISALLOWED
same-PR):
- Deliverable 4: producer-side substrate-axis verification only
(type-check + carrier acceptance + bootstrap snapshot/manifest
hold). NO end-to-end Rust emission in this PR.
- §1.8 #67 numeric_construction_demonstration receipts split:
producer half here (entries exist), consumer half on Grounding
G2 PR (full Int<32> → Rust i32 round-trip).
- Same-PR cross-program co-author available if Substrate Mgr
coordinates with Grounding Mgr at dispatch; default is split.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: pre-author T-Rule-Enumeration substrate-fact-introduction worker brief
Pre-authored per pre-authored-brief-queue discipline; dispatch fires
on Director Q2 ratification of emission-provenance canvas at
gunbc#828 #issuecomment-4392519713.
Brief covers LangSpec emission-rule names as enumerable substrate
(rust_target.rs uses inline &str template names today; 0 hits on
RuleName/FoldRule/EmitRule across src/v3/). Two carrier shape options
(α sum-type / β named-string list) with worker DFS at dispatch
deciding shape; Practice 4 checkpoint discipline + ledger-row receipt.
Prerequisite to every Q1 path (a/b/d) — Lens<EmissionProvenance>,
per-line instrumentation, and the both-substrates option all consume
the rule-name set. Path (c) leaves substrate-unblocked-but-unconsumed.
Mgr authoring authority per PM concur at gunbc#846
#issuecomment-4392510255 / #issuecomment-4392543633 (substrate-fact-
introduction is Mgr-tier, not PM tactical authoring).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* t-rule-enumeration: align gate name + status to Director ratification at #4392562911
* emission-provenance lens brief: revise per Director Q1 (a) RATIFICATION
Director Q1 (a) per-Behavior Lens<C>-compatible RATIFIED at gunbc#1739
#issuecomment-4392562911 (2026-05-06); (b) instrumentation REJECTED;
(d) parallel-substrates REJECTED.
Brief revised:
- Carrier shape: Lens<List<EmissionProvenance>> per (a)
- EmissionOrigin = SubstrateDeclMirror(SourceSpan) | FoldRuleAutoEmit(EmissionRule)
typed sum (closed; no third silent class; structural fail-closed)
- All 6 Lens<C> field bindings concrete (read/sequential/branch/iterate/validate)
- T-Rule-Enumeration named as PRECONDITION (must land first)
- §1.8 retarget under T-CostLens-Composition cluster per Director Q3
- Brian's slide visualization served by (a) projection (per-Behavior
list flattens to per-line view at visualization layer; no parallel
substrate needed)
Supersedes PM-authored proposal at PR #1902 (merged at 54419badf).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: bundled Rust-primitive-full-coverage substrate-prerequisite worker brief (Director Path A)
Author bundled substrate-prerequisite brief per Director Path A
RATIFICATION at gunbc#1739 #issuecomment-4392731264 (2026-05-06).
Three structurally-coupled gaps surfaced via Grounding loyal-stag-699
STOP at gunbc#1907 #issuecomment-4392585090 land as one bundled slice
per "necessary structural fix" exception (bundled-scope discipline):
1. IntervalInt::ExactInterval host-repr widening (α BigInt-based
recommended; β typed-variants if α infeasible)
2. RustPrimitive structural BoundDeclaration field (replaces static
range_min/max strings; consumes existing BoundDeclaration carrier
StaticBound + PlatformDependent variants)
3. spec/rust.dag PlatformDependent row population for isize/usize +
u128 row addition consuming widened ExactInterval
Closure gate: rust_primitive_full_coverage (Director-ratified name).
Worker pin: valiant-ibex-312 (freed-pool; substrate-authoring fresh;
numeric-domain-adjacent post-#1842).
Unblocks Grounding G2 Phase 2 full-coverage dispatch. G2 Phase 1
narrowed dispatch (i8-i64 + u8-u64) runs in parallel on existing
substrate; cleared at gunbc#1745 #issuecomment-4392795954.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: align Practice 4 classifications to GREEN/YELLOW rubric + add GREEN ledger receipts
Address codex BLOCKING on PR #1910 sha f36e1205: briefs used
PRIMITIVE/SCAFFOLD terminology that doesn't match docs/modeling-
discipline.md rubric. Per modeling-discipline.md lines 97/103/108:
the three classifications are 🟢 GREEN (terminal) / 🟡 YELLOW
(scaffold) / 🔴 RED (dissolvable-now); line 132 requires a ledger
entry if GREEN.
Replaced PRIMITIVE → GREEN, SCAFFOLD → YELLOW across all three
emission-provenance / Rust-primitive-full-coverage briefs. Added
explicit §1.8 ledger entry receipt for the GREEN classification on
EmissionOrigin (sibling row emission_origin_classification_green to
parent emission_provenance_lens_landed gate).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: Practice 4 terminal receipt for EmissionOrigin + bound for β scaffold
Address openai-pro BLOCKING (sha f36e1205, ran post-rubric-rename):
1. EmissionOrigin GREEN terminal classification needs documented
attempted-dissolutions per modeling-discipline.md "no richer
source exists" framing. Added 3 attempted dissolutions:
- collapse to optional-pair (rejected by codex Finding 1 prior)
- factor common richer source (orthogonal payload types prevent)
- third arm for unattributable lines (rejected per Slice 2.5
Option 4 placeholder anti-pattern)
2. T-Rule-Enumeration β (YELLOW) scaffold needs bound per three-part
bridge rule. Added: allowed call sites scoped to emission code only;
max lifetime gated on rule_name_typo_fail_closed_landed graduation;
new emission-introspection consumers MUST author against closed-sum
α form, not β.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* WIP: R3 Substrate
* briefs: address codex BLOCKING bundle (sha 2ed1046e) — record-form provenance + emit_model carrier alignment + T-Rule-Enumeration retirement flag
Three findings addressed:
1. Lens brief (Finding #1 — provenance shape conflates source vs rule
attribution): replaced typed-sum EmissionOrigin with record
EmissionProvenance { rule: EmissionRule [mandatory], source_span:
Option<SourceSpan> [optional] }. Co-inhabitance is real (e.g.,
#[derive(Debug)] line has both rule=derive_for_disj AND span
attribution to Foo decl). Rule-mandatory enforces fail-closed
structurally; Practice 4 N/A (record, not coproduct).
2. T-Rule-Enumeration brief (Finding #2 — rule authority lives in
spec/rust.dag + v3.std.emit_model, not call-site strings): brief
status flipped to FLAGGED FOR RETIREMENT. Same finding as
smart-ram-167 STOP at gunbc#1759 #issuecomment-4392696623; Reading
C surfaced to Director at gunbc#828 #issuecomment-4392780871.
3. Rust-primitive-full-coverage brief (Finding #3 — BoundDeclaration
not wired into TargetIntegerInhabitanceBound row surface):
reframed Deliverable 2 to target the actual carrier
`v3.std.emit_model::TargetIntegerInhabitanceBound` (used by
`TargetIntegerTypeInhabitance` rows in src/v3/spec/rust.dag lines
169-207). Two paths surfaced: Option (i) wire BoundDeclaration into
TargetIntegerInhabitanceBound; Option (ii) populate via existing
variants if PlatformDependent already exists on the live carrier.
Worker DFS at dispatch picks. Deliverable 3 retargeted to
TargetIntegerTypeInhabitance rows (the actual surface).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* rust-primitive-full-coverage: codex inline confirms TargetIntegerInhabitanceBound has no PlatformDependent variant — Option (ii) infeasible; carrier refactor REQUIRED
Codex BLOCKING inline at PR #1910 line 65 (sha 98507c432) verified at
HEAD: TargetIntegerInhabitanceBound = BoundUnspecified | StaticBoundFact(IntInterval).
No PlatformDependent variant. Earlier brief framing of Option (ii)
(populate via existing variants) was infeasible.
Brief reframed: carrier refactor is required (not optional). Two
structural shapes for the refactor — (i.a) embed BoundDeclaration
(Mgr recommendation; single-authority discipline) vs (i.b) extend
variant set with PlatformDependent. Existing 5 rows migrate either
way; bootstrap snapshot semantic-equivalence non-negotiable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* emission-provenance lane: T-Rule-Enumeration RETIRED + Lens brief revised per Director Reading C
Director Reading C RATIFIED at gunbc#1739 #issuecomment-4392797954
(2026-05-07):
- Q1: T-Rule-Enumeration retires (substrate gap was phantom; ~37
*SyntaxBinding/*OpsBinding field paths already provide enumeration
per smart-ram-167 substrate-state-grep at gunbc#1759 #issuecomment-4392696623)
- Q2: Lens<EmissionProvenance> dispatches now (no precondition gate)
- Q3: EmissionRule = String of field-path (per feedback_reason_not_label)
Actions:
- Delete docs/briefs/r3-substrate-t-rule-enumeration-worker.md
- Lens brief: replace T-Rule-Enumeration precondition gate with
Reading C absorption section; EmissionRule = field-path String;
cementing test verifies field-path round-trip; STOP triggers
retargeted to *SyntaxBinding field-set divergence
Worker pin: smart-ram-167 (Mgr discretion per Director ratification;
fresh context on field-path enumeration). valiant-ibex-312 stays on
Rust-primitive-full-coverage / T-Interval-Representation.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* emission-provenance lens brief: absorb codex BLOCKING bundle (sha 4229cd09 post-merge findings)
Two findings from codex bundled review on now-merged PR #1910 sha
4229cd09 land as brief revisions for the in-flight smart-ram-167
authoring cycle (PR #1928):
1. Field-path identity threading: render_named_template call sites
receive template values extracted from binding structs; field-path
may not be carried alongside the value at emission time. New STOP
trigger flags this as substrate-cascade if threading isn't
structurally possible (typed template-with-rule-path or
DeclarationRef carrier needed).
2. Read domain coverage gap: Lens<C>.read is per-Behavior; emitted
output also includes declarations + program scaffold. Acceptance
scope explicitly narrowed to Behavior-attributable lines only;
declarations + scaffold provenance out-of-scope (separate substrate
cascade if surfaced). Acceptance bullet retargeted to
"Behavior-attributable span-Some + Behavior-attributable span-None"
rather than full-emission coverage.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* rust-primitive-full-coverage: revert to Option (ii) — Mgr ratification-state-grep miss caught by valiant-ibex-312
valiant-ibex-312 ratification-state-grep at gunbc#1761
#issuecomment-4393074602 surfaced that origin/main HEAD has
TargetIntegerInhabitanceBound with 3 variants including
PlatformDependentFact (mirrors BoundDeclaration::PlatformDependent).
Earlier brief framing at 4229cd09d was against pre-PlatformDependentFact
state per codex BLOCKING reading; main moved since.
Option (ii) IS feasible: isize/usize rows populate via existing
PlatformDependentFact variant directly; u128 row populates StaticBoundFact
with BigInt-widened ExactInterval (Deliverable 1 unchanged). No
carrier refactor needed; 5 existing rows untouched.
Second Mgr-tier ratification-state-grep miss today (first was symbol-only
grep on T-Rule-Enumeration; now stale-codex-vs-current-HEAD on
TargetIntegerInhabitanceBound). Worker-tier discipline caught it before
sinking time into invasive Option (i.a) refactor.
Single-authority concern (parallel-named variants on
TargetIntegerInhabitanceBound vs BoundDeclaration) noted as future
substrate-fact-introduction; not blocking this slice — explicit-mirror
is the existing convention.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* rust-primitive-full-coverage: align Acceptance + Phase ordering + STOP triggers to Option (ii) (resolve codex internal-contradiction finding)
codex BLOCKING on PR #1929 sha 14f68b63 caught real internal contradiction: Context section flipped to Option (ii) feasible at 57ae4349b but Acceptance + Phase ordering still referenced BoundDeclaration migration / RustPrimitive bound field refactor from earlier Option (i.a) framing.
All sections now consistent with Option (ii):
- Phase 4 = no-op (TargetIntegerInhabitanceBound carrier untouched)
- Phase 5 = spec/rust.dag rows for isize/usize via existing PlatformDependentFact variant
- Phase 6 = u128 row via StaticBoundFact(IntInterval) + ratchet update
- Acceptance retargeted to row-population shape (no carrier refactor; existing 5 TargetIntegerTypeInhabitance rows untouched)
- STOP triggers retargeted to TargetIntegerInhabitanceBound variant-set sufficiency at HEAD
- Why-bundled rationale rewritten to Option (ii) shape (Phase A widening unblocks u128 StaticBoundFact; isize/usize use existing PlatformDependentFact)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* lens brief: narrow Deliverable 3 verification step 4 to Behavior-attributable lines (resolve codex #2 consistency)
codex BLOCKING #2 on PR #1929 sha 812602b4 caught: Acceptance + STOP triggers narrowed to Behavior-attributable lines but Deliverable 3 cementing-test step 4 still said 'every emitted line'. Now consistent — step 4 explicitly scopes to Behavior-attributable subset; declarations + program scaffold OUT-OF-SCOPE per same Finding #2 framing.
(codex BLOCKING #1 — claim that PlatformDependentFact NOT on main — is wrong; verified at origin/main HEAD: TargetIntegerInhabitanceBound has 3 variants including PlatformDependentFact. Rebut on PR.)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: pre-author IntPlatform/UIntPlatform substrate-fact-introduction follow-on (Director ratification at #4393248961)
Per Director Q1+Q2+Q3+Q4 RATIFICATION at gunbc#1739 #issuecomment-4393248961 (2026-05-07): pre-author follow-on brief for the substrate cascade surfaced by valiant-ibex-312 STOP at gunbc#1761 #issuecomment-4393222862.
Scope:
- Q1 naming: IntPlatform / UIntPlatform per feedback_reason_not_label
- Q2 algebra: Compose<Int, MachineWidth<Platform>> + Compose<UInt, MachineWidth<Platform>>; Platform is substrate token
- Q3 substrate-concept layer (target-agnostic per feedback_target_agnostic_ir)
- Q4 worker pin: valiant-ibex-312 post-PR-#1914 close (freshest substrate context)
Brief covers Platform substrate token + IntPlatform/UIntPlatform declarations + spec/rust.dag isize/usize row population + co-receipt advancing PR #1914's rust_primitive_full_coverage gate PARTIAL → CONSUMER_LANDED.
Closure gate: int_platform_uint_platform_substrate_landed.
Dispatchable post-PR-#1914 close per pre-authored-brief-queue discipline.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
briansrls
added a commit
that referenced
this pull request
May 7, 2026
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* WIP: R3 Substrate
* fix(briefs): retire INVARIANTS.md line-number cites in r3-426 brief
Per codex review on PR #1782 (sha c2512cde): the new briefs
introduced fresh `INVARIANTS.md:<line>` citations — the same
stale-line-anchor shape the :425 sweep brief itself names as needing
dissolution. Replaces four line-number references in
r3-426-sfp-subdoc-rename-and-author-worker.md (lines 16, 28, 58, 85)
with prose-anchored locators citing the section anchor + bullet text.
The other PB-lane briefs codex flagged (r3-pb-runtime-equivalence-
corpus-seed-audit.md, r3-pb-t-fixedpoint-worker.md) belong to
neat-bear-351's lane and are not in this PR's scope; will route
separately.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): correct L6 type count + r3-435 dual-path retirement receipt
Per openai-pro review on PR #1782 (sha c2512cde):
1. L6 brief said acceptance ratchet asserts "five types" exist but
declares six (ShapeATarget, FormAxis, BehaviorAxis,
MethodTemplateContractKey, EmissionCell, EmissionPathProjection).
Updated both occurrences to enumerate all six explicitly.
2. r3-435 brief authorized both Path A (typed-AST reader, deletes
IntegrationRsScan) and Path B (scanner widening with Char state),
but retirement receipt language only covered Path B. Updated
ROADMAP/ledger retirement and authority-audit step 5 to cover
both authorized dissolution paths.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): L6 second 'five types' site + T-E-P lookup-gate authority
Per cursor review on PR #1782 (sha 2072b986):
1. L6 brief acceptance bullet at :131 still said "five types" while
the parenthetical and the Slice step at :117 said six. The prior
fix only caught one of two sites; corrected the second.
2. T-E-P brief `e_p_call_pattern_lookup_authoritative` acceptance
named only `lower_call_pattern` as "the only route from call
site to LoweringTarget" — but Phase 2 explicitly requires
lenses route through `per_call_pattern_at` (the L-7 single-
authority lookup), NOT `lower_call_pattern` (compiler-internal).
Acceptance text rewritten to name `per_call_pattern_at` as the
consumer-facing gate and clarify `lower_call_pattern`'s
internal-only scope.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): strengthen L6 ratchet from count-parity to per-row bijection
Per openai-pro REQUEST_CHANGES on PR #1782 (sha 2072b986): the L6
acceptance ratchet asserted only row-count parity between
emission_path_projections and the union of MethodTemplateContract
rows. Count parity permits duplicate projection keys and missing
source rows — the bridge from list-non-empty proxy to per-row
projection would not be mechanically proven. The whole point of
introducing MethodTemplateContractKey { target, dag_method } is to
provide 1:1 row identity.
Rewrites the ratchet to assert per-row bijection: every
MethodTemplateContract source row has exactly one matching
EmissionPathProjection row by key; every projection key resolves
to exactly one source row; duplicate projection keys fail closed.
(The other openai-pro BLOCKING finding — T-E-P
e_p_call_pattern_lookup_authoritative naming the wrong authority
— was already fixed at 6282a0704; the review citation was against
the pre-fix sha 2072b986.)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): split L6 ratchet into slice-active vs deferred-activation gates
Per BLOCKING inline review on PR #1782 at d92459423: the prior
bijection ratchet was unsatisfiable as authored. The
*_method_template_contracts lists at HEAD (rust/python/go) are
non-empty; shipping emission_path_projections: [] alongside a
strict bijection requirement means the bijection cannot pass when
the slice lands — source rows would be uncovered. Prior text
incorrectly said "vacuously true on the empty source set" — the
SOURCE set isn't empty, only the projection set is.
Splits the ratchet into two gate categories:
- Slice-active gates: types/data shape exist + empty-state
predicate (emission_path_projections == []).
- Deferred-activation gate: per-row key bijection, authored as a
test scaffold by this slice but #[ignore]'d (or feature-gated
on len > 0) until Grounding's follow-up populates rows. On
activation, asserts the bijection over the populated set.
This honors P1 doc faithfulness: the slice's acceptance contract
is satisfiable as authored, and the bijection becomes the
load-bearing gate the row-population PR satisfies.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(briefs): require L6 bootstrap enrollment to honor P2 facts-flow-forward
Per BLOCKING inline review on PR #1782 at L6:34: the slice authored
src/v3/std/cross_target_coverage.dag but never required enrolling
it in the bootstrap/load authority. Without enrollment the file
is dead-letter on disk and EmissionPathProjection can be absent
from the downstream Dag — silent P2 facts-flow-forward violation.
Adds a new Slice step 2 requiring a cross_target_coverage field
on BootstrapFixtures in src/v3/std/extdeps_bootstrap_fixtures.dag
mirroring the *_method_template_contracts enrollment pattern, plus
a loader-visibility ratchet. Renumbers subsequent Slice steps.
Adds matching Acceptance bullet.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* WIP: R3 Substrate
* WIP: R3 Substrate
* docs(briefs): normalize Practice 4 citation in S3 brief per codex review
Codex review on PR #1782 (sha 2614217e) flagged stale citation
"Practice 4 (Step 4 of the type-introduction checklist)" in
S3 brief — no such checklist step exists in the pinned rubric.
Live rule per docs/modeling-discipline.md:131-134 is the "What to
check" checkpoint-comment requirement under Practice 4 (coproduct
dissolution). Updated the S3 reference to cite that line directly
and drop the stale "Step 4 of type-introduction checklist" wording.
S2/S8/S9 already use bare "Practice 4" without the stale steps wording.
S1 (committed earlier today) does not contain the stale phrase.
S6 (pre-existing brief flagged by reviewer as carrying the same
pattern) is out of this branch's scope; flag for separate paydown.
* docs(briefs): sweep stale Practice 4 citations corpus-wide
Codex review on PR #1782 flagged the stale "Step 4 of the
type-introduction checklist" wording — no such checklist exists in
docs/modeling-discipline.md. The live rule is the "What to check"
checkpoint-comment requirement under Practice 4 (coproduct
dissolution) at docs/modeling-discipline.md:131.
Sweep across docs/ found 2 additional briefs carrying the stale
phrasing (S3 fixed previously at 9db003d40):
- docs/briefs/r3-l6-emission-path-projection-substrate-worker.md
- docs/briefs/r3-x1b-s1-transform-dispatch-substrate-worker.md
Both updated to cite the live rule directly. Verified zero
remaining "type-introduction checklist" or "Step 4 of the"
references in docs/ tree.
* docs(briefs): add Practice 4 citation discipline note to authoring checklist
Per parent inbox status flag at gunbc#828 #issuecomment-4385346221:
add a Citation discipline section to brief-authoring-checklist.md
naming the correct shape — Practice 4 (coproduct dissolution) +
"What to check" rule at docs/modeling-discipline.md:131 — and
calling out the stale "Step 4 of the type-introduction checklist"
phrasing as a failure mode. References the corpus sweep at
2dc92f34f.
Prevents future briefs that copy old templates from re-introducing
the stale citation pattern.
* docs(r3): cascade work post-Director ratification — S1 worker, R4 routing, Unit/Quantity canvas
Per Director ratification at gunbc#828 inbox response 2026-05-06
(zesty-bear-812):
1. S1 Q1+Q2 RATIFIED → author S1 gap-test representative worker brief
(`r3-substrate-s1-gap-test-representative-worker.md`). Dispatchable
on T-E-P Phase 1 + E6-G0d landing; closes ledger row #61.
2. S2 Q2 RATIFIED (option-(b) T-LBP narrowing) + S3 + S8 deferrals →
R4 carve-out routing ledger (`docs/r4-carve-out-routing.md`).
Captures C1 (parallelism lens), C2 (effect_enum lens), C3
(register narrowing), C4 (deferred MachineConstraint axes),
C5 (rounding-mode product extension), C6 (Unit/Quantity).
Cascade messages to Verification / Grounding / Debt-Paydown Mgrs.
3. S9 Phase-3 dimensional refinement boundary → Unit/Quantity carrier
canvas (`r3-substrate-s9-unit-quantity-carrier-canvas.md`).
Surfaces Q-Unit-1..5 for Director ratification: carrier name,
2-axis vs 3-axis Phase-1, Duration/Seconds collapse, Refined
composition order, Practice 4 classification.
Updates §1.8 ledger receipts for carved items; R3 closure receipts
treat carved items as explicitly-deferred (not drift).
* docs(briefs): citation-discipline fix-forward — anchors/quotes, not line numbers
Director redirect at gunbc#828 #issuecomment-... noted the citation
paydown's replacement form (`docs/modeling-discipline.md:131`) is the
same drift class as the original fabricated-step error. Bare line
numbers shift silently on any doc edit above the cited line.
Generalized fix:
- brief-authoring-checklist.md "Citation discipline" section
rewritten to prohibit bare line-numbers across ALL cross-doc
references (modeling-discipline.md, r3-program-plan.md,
r3-structure.md, INVARIANTS.md, MODELING.md, etc.). Use section
anchors (#section-name) or rule-text quotes.
- 3 briefs that swapped in :131 (S3, S6, X1B) re-converted:
- docs/modeling-discipline.md#4-coproduct-dissolution anchor
- "What to check" rule quoted inline (self-contained)
Brief-authoring-checklist failure-mode section names both vectors
(fabricated step + bare line numbers) and references both sweep
commits in audit trail.
Broader cross-doc citation sweep across briefs I authored on this
branch (frontmatter authority-docs lists, body prose) is recommended
per Director note — flagging as separate paydown task; this commit
closes the immediate recurrence vector.
* docs(briefs): fix unfaithful Int=AbelianGroup<Nat> + sweep stale 'Step 4' refs
Per openai-pro review on PR #1782 sha 0ecd4818 (gunbc#1782 #...):
BLOCKING fix (P1 modeling faithfulness):
- S9 + S3 directed workers to model `Int = AbelianGroup<Nat>`,
but Nat is not an Abelian-group carrier (additive inverses
not representable). Reframed:
- `Int` is the additive Abelian group on integers (carrier Z);
`Nat` is a commutative monoid under addition.
- Algebraic relationship is **explicit group-completion**:
`Int ≡ GroupCompletion<CommutativeMonoid<Nat>>` — NOT
`AbelianGroup<Nat>` parameterization.
- Worker either models `Int` directly as `AbelianGroup`
(terminal Abelian-group instance) OR introduces explicit
`GroupCompletion<M>` substrate carrier; decision lands in
Phase-1 PR per P1 procedure.
- S3 cross-reference + Phase-4 algebra-axis preservation note
reframed accordingly.
NON-BLOCKING in-PR cleanup (Practice 4 citation discipline):
- 6 briefs still carrying stale 'Practice 4 Step 4' wording
(r3-coproduct-1/2/3, r3-l6, r3-t-e-p, r3-x1b) updated to use
the section anchor (`docs/modeling-discipline.md#4-coproduct-dissolution`)
+ inline rule-text quote ("What to check") form per the
citation discipline established in `b4cae5299`.
Closes both findings from openai-pro review.
* docs(briefs): Q-Unit-1..5 RATIFIED — apply Scale.Unit→One rename + ratification banner
Director ratified Q-Unit-1..5 at gunbc#828 #issuecomment-... 2026-05-06
(zesty-bear-812). All 5 questions ratified; one rename required:
- Scale enum's `Unit` value collides with outer `Unit<Q, S>` carrier
name (`Unit<Time, Unit>` is shadowing-confusing). Rename to `One`.
- Duration ≡ Unit<Time, One> (collapses with Seconds per Q-Unit-3).
- Milliseconds ≡ Unit<Time, Milli>.
Canvas updated:
- Top-level RATIFICATION banner records all 5 ratifications + rename
- Scale enum: `Unit` → `One` (with inline rename receipt)
- Sketch product expansions use `Unit<Time, One>` / `Unit<Time, Milli>`
- Phase-3 reframe table products updated
- Q-Unit-3 prose collapse reference updated
Subsequent worker brief authoring (Unit/Quantity carrier landing
post-prerequisite) inherits the renamed shape.
* docs(briefs): Unit/Quantity carrier worker brief — Q-Unit-1..5 RATIFIED
Worker brief consumes the ratified Unit/Quantity canvas
(`5f22fd06e` ratification banner). Phase-1 lands `Unit<Q, S>` +
`Quantity` + `Scale` declarations at the ratified shape (Scale's
`One` value per Q-Unit-1 rename). Phase-2 reframes S9 Phase-3
dimensional refinements (Duration / Seconds / Milliseconds) to
outer-Refined / inner-Unit composition per Q-Unit-4 RATIFIED.
EpochMs deferred to R4 C6 (Aspect-axis follow-up).
Practice 4 marks pre-determined per Q-Unit-5: `Unit<Q, S>` 🟢
PRIMITIVE; `Quantity` + `Scale` 🟡 SCAFFOLD with "consumed by ≥1
emission rule" dissolution trigger.
Cross-program handoff to Grounding Mgr (#1745) for dimensional
emission consumption documented.
Worker pin TBD — dispatchable when idle pool refreshes.
* docs(briefs): S9 Phase-1 reframed — Int landed at #1466; UInt-only migration
Per proud-lynx-311 dispatch ack at gunbc#1739 #issuecomment-...
2026-05-06: re-grep at HEAD revealed substrate state diverges
materially from S9 brief's "Option A vs B decide" framing.
Reality check at HEAD:
- `dsl/std/integer.dag:55` already carries
`type Int = AbelianGroup<GroupCompletion<Nat>>` per Slice 3 PR
#1466 / commit `4ba0a6d04`. Single-authority Q6 audit doc at
`docs/audit/t-numeric-construction-group-completion-6q.md`
ratified this shape and explicitly rejected the compact form.
- `dsl/std/integer.dag:56` still on legacy `type UInt = UInt64`
alias; UInt migration is the actual Phase-1 worker scope.
Brief Phase-1 step 1 reframed: Int landing acknowledged; worker
confirms HEAD match. Phase-1 step 2 reframed: UInt migration
target named (`type UInt = CommutativeMonoid<Nat>`); fixed-width
rows preserved per existing legacy-rows policy.
Historical Option A vs B framing kept as note (superseded).
Confirmed S9 dispatch ack from worker; UInt-only Phase-1 PR is
in-scope. Emission entries deferred per S3 (valiant-ant-72
pre-landing); Phase-2 deferred per S8 (quiet-boar-160 pre-landing).
* docs(briefs): fix two BLOCKING citation drift items per codex review
Codex review on PR #1782 sha 5f22fd06 flagged two authority-receipt
gaps:
1. brief-authoring-checklist.md:118 — `<this-commit>` placeholder in
the failure-mode section was unverifiable post-merge. Replaced
with `b4cae5299` (the actual anchor/quote conversion commit).
2. r3-378-http-path-helper-fail-open-worker.md:84 — `INVARIANTS.md
C-8 fail-closed` cited without anchor or rule-text quote, in
violation of the citation discipline added in this PR.
Converted to `INVARIANTS.md#p3-fail-closed` (P3 heading anchor)
+ inline rule-text quote naming P3 + C-8.
PR's own discipline now consistently applied across the diff.
* docs(briefs): sweep bare .md:NNN citations in Substrate-Mgr-authored briefs
Per Director recommendation in citation-discipline thread (gunbc#828
follow-on): bare line-number references drift silently. Swept 19
remaining instances across 5 briefs I authored on this branch:
- docs/briefs/r3-substrate-s1-q-class-2-chain-break-gap-test-canvas.md
- docs/briefs/r3-substrate-s1-gap-test-representative-worker.md
- docs/briefs/r3-substrate-s2-t-lbp-scope-calibration-canvas.md
- docs/briefs/r3-substrate-s3-machine-constraint-carrier-worker.md
Conversion: bare `file.md:NNN` → `file.md §"Section name"` form.
Stable; survives doc edits; human-readable. Section names per
HEAD heading audit.
Verification: `grep -nE '.md:[0-9]+' docs/briefs/r3-substrate-*.md
docs/r4-carve-out-routing.md` returns zero results in
Substrate-Mgr-authored scope.
Broader corpus scope (~232 references across non-Substrate briefs)
remains as separate paydown task.
* docs(briefs): fix BLOCKING + NON-BLOCKING items per openai-pro review on a462f007
Three findings addressed; 2 BLOCKING + 1 NON-BLOCKING:
BLOCKING — R3/R4 dual authority for Unit/Quantity carrier:
R4 carve-out C6 narrowed from "Unit/Quantity carrier landing"
(deferred all of it) to **"Aspect-axis (PointKind) follow-on
for EpochMs and instant/rate-shaped refinements only"**. The
2-axis Measure<Q,S> carrier landing IS in R3 per Q-Unit-1..5
RATIFIED; only the Aspect axis (PointKind = Magnitude | Instant
| Rate) carves to R4. Cascade-message C6 reference + R4 program-
plan input list updated to match. Cross-Mgr message clarifies
Grounding consumes 2-axis dimensional refinements in R3.
BLOCKING — Locked design partially applied (Scale.Unit→One
rename in canvas alternate-sketch sections):
- Canvas line 105-108 alternate sketch ("Open question: Instant
vs Duration axis") still used pre-rename `Unit` Scale value;
added pre-ratification disclaimer marking the section as
deferred to R4 C6 + noting `Unit` is pre-rename
- Canvas line 163 Q-Unit-3 collapse note: `Scale = Unit` → `Scale = One`
(BLOCKING — bare line-numbers in S1 canvas frontmatter): NOT
APPLICABLE — already fixed at sweep commit 127287a66 prior to
review; reviewer was on sha a462f007 pre-sweep.
NON-BLOCKING — placeholder receipts:
- brief-authoring-checklist.md `#issuecomment-...` → concrete ID
4385393971 (Director's anchor-discipline reply)
- r3-substrate-unit-quantity-carrier-worker.md (3 occurrences)
+ canvas RATIFICATION banner: `#issuecomment-...` → concrete ID
4385412256 (Q-Unit ratification reply)
- `<this-commit>` placeholder in checklist already fixed at 5be1c99f2
Verification: zero `#issuecomment-...` or `<this-commit>` placeholders
in docs/briefs/ + docs/r4-carve-out-routing.md.
* docs(briefs): codex BLOCKING review fixes — Practice 4 source-level + policy scope
Three BLOCKING findings on PR #1782 (codex sha b913e2f4):
1. **Citation-discipline policy scope**: addressed by scoping
the rule prospectively to briefs landing after the policy
commit (`b4cae5299`). Pre-existing corpus references are
flagged as paydown but NOT individually blocking; coordinated
single-PR sweep is the right paydown shape. Substrate-Mgr scope
already swept at `127287a66`.
2. **R4 C6 reconciliation**: already fixed at `6e5a5481a` (R4 C6
narrowed to Aspect-axis follow-on for EpochMs only; 2-axis
carrier landing IS in R3). Codex sha b913e2f4 was pre-fix.
3. **Practice 4 source-level vs PR-body conflation**: corrected.
Acceptance bullet for Unit/Quantity worker brief now requires
inline checkpoint comments on the LIVE declarations in
`dsl/std/units.dag` (load-bearing artifact per
`docs/modeling-discipline.md#4-coproduct-dissolution` "What to
check"). PR-body summary is supplementary, not substitute.
Closes the review's reconciliation gates.
* docs(briefs): absorb Q-MachineConstraint + Q-Unit-1-Recanvas ratifications
Two major Director ratifications absorbed:
1. Q-MachineConstraint-Carrier RATIFIED at gunbc#828
#issuecomment-4385530115 (Brian directive 2026-05-06: "universal
substrate, ratify defaults"; PR #1817 queued). 6 sub-decisions:
- MachineWidth<bits> only for R3; defer RegisterClass / EndianMode
/ Alignment / signedness post-R3
- Parametric Compose<Algebra, MachineConstraint> type-level
interaction shape (lookup-maps REJECTED — AlgebraMachineProduct
record-keyed table form retracted)
- Int<64> ≡ Compose<AbelianGroup, MachineWidth<64>> (algebra-side
options A and B preserve interaction semantics)
- Real<64> ≡ Compose<ApproximateField<Rational>, MachineWidth<64>>
(algebra approx + machine approx as independent composing axes)
- ≥3 pairs is minimum, not target
- UNIVERSAL substrate: every target carries machine-constraint
facts; targets without native machine-width semantics handle
omission at Grounding-level discharge (target-conditioned
lowering NOT target-conditioned substrate)
Reframed: S3 brief Phase-1, S3 Phase-3/4 emission entries,
S8 Phase-1/2/3 (ApproximateField<Real> → ApproximateField<Rational>;
AlgebraMachineProduct → Compose<...>), S9 emission entries.
2. Q-Unit-1-Recanvas RATIFIED Measure<Q, S> at gunbc#828
#issuecomment-4385539791. valiant-ibex-312's substrate-state
grep caught Unit<Q, S> collision with kernel Unit at
dsl/std/types.dag:176; Director ratified Measure<Q, S> as
outer carrier name (conventional in scientific computing;
no dsl/std/ collision).
Reframed: Unit/Quantity worker brief throughout (Unit<Q,S> →
Measure<Q,S>; dsl/std/units.dag → dsl/std/measure.dag); canvas
spec sections + sketch products (Refined<Measure<Time, ...>...>);
R4 carve-out C6 with informal/formal name disambiguation note.
Both rename + reshape are now consistent across S3 / S8 / S9 /
Unit-Q briefs + canvas + R4 routing. Workers (valiant-ant-72 on S3,
quiet-boar-160 on S8, proud-lynx-311 on S9, valiant-ibex-312 on
Unit-Q) all dispatch against the ratified shape.
* docs(briefs): S11 Slice C worker brief — ROADMAP :425 retirement post-#1801
Cascade-clearance trigger fired post-#1801 (Slice B) merge
2026-05-06. S11 brief authored per B1 RATIFIED prose+regen
bundling discipline.
Phase 1: per-file edits across the post-A+B residual list (~10
files per smart-ram-167's Slice B post-rebase enumeration);
anchor + rule-text-quote pattern per current citation discipline;
bootstrap regen in same PR (no split).
Phase 2: ROADMAP :425 row PARTIAL → Retired with 3-slice receipt;
Q-Slice-C-Retirement-Receipt resolution.
Worker pin smart-ram-167 (#1759) — Slice B precedent owner.
Dispatch packet sent separately to #1759.
* docs(briefs): Q-Refined-Phantom-Composition (c) RATIFIED — defer literal Refined form
Director ratified option (c) at gunbc#828 #issuecomment-... 2026-05-06:
drop `non_negative` predicate at Measure (phantom) layer; rely on
`Quantity = Time` tagging for non-negative-magnitude semantics.
Q-Unit-4 outer-Refined / inner-Measure composition is preserved as
**conceptual model**; the literal `Refined<Measure<...>, predicate>`
form is **deferred to value-typed-integration follow-up** when
substrate gains Refined<> as first-class type expression and
dimensional refinements get value-typed integration.
Updates:
- Canvas RATIFICATION banner Q-Unit-4 row reframed to "conceptual
model; literal form deferred"
- Canvas Phase-3 reframe table: Duration / Milliseconds drop the
Refined<...> wrapper; rely on Quantity tag
- Worker brief Phase-2 prose + reframe table + Acceptance bullet
all reflect deferred-literal-form posture
valiant-ibex-312 unblocked to proceed with Phase-2 (Grounding
handoff doc) per the (c)-shaped Acceptance.
* docs(briefs): S12 F2 + F8 doc-sharpening worker brief — post-#1820 cascade
#1820 (Slice C) merged 2026-05-06; ROADMAP :425 row Retired closes
the prose-reference paydown program. S12 cascade-clearance trigger
fires per B6 RATIFIED single bundled Mgr-tier doc-sharpening PR.
Scope: F2 (.v3 filename-suffix grammar dispatch) + F8 (bootstrap
load-order/exclusion authority) — single bundled PR per B6.
Cross-program coordination with PB Mgr (#1742) per design schedule
§3 P5 (comment-only OR co-author; no duplicate PR).
Worker pin smart-ram-167 (#1759) — Slice C precedent owner;
pattern-familiar with prose+regen bundling discipline. Dispatch
packet sent separately to #1759.
* docs(briefs): S9 brief — UInt landed via #1818; algebra label corrected to CommutativeSemiring
Per proud-lynx-311 merge report at gunbc#1746 #issuecomment-...
2026-05-06: PR #1818 landed `type UInt = Nat` after codex BLOCKING
review on initial `CommutativeMonoid<Nat>` form (P2 facts-flow-forward
violation — wrap projects out Nat's multiplicative monoid + identity 1).
Final landed form preserves full `CommutativeSemiring<Magnitude>`
surface — Nat IS a commutative semiring (addition + multiplication
+ 0 + 1). Algebra-side label for UInt is CommutativeSemiring, NOT
CommutativeMonoid (corrects my earlier dispatch labeling).
S9 Phase-1 step 2 marked LANDED with merge ref. Phase-1 step 3
emission entries reframed:
- Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>>
- UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>>
Worker discipline + reviewer caught a substrate-state error in my
brief framing; brief now matches landed reality.
* docs(briefs): fix contradictory (c) guidance in Unit/Quantity worker brief
Codex BLOCKING review on PR #1825 sha 1bce0418 caught a real
contradiction: brief Phase-2 prose at lines 115-118 stated
"predicates do NOT attach at the Measure (phantom) layer" per
Q-Refined-Phantom-Composition (c) RATIFIED, but immediately
following paragraph at lines 130-132 still said "Refined predicate
non_negative applies at the Measure-typed level" — left over from
the original Q-Unit-4 framing.
Updated lines 130-137 to consistently apply (c) shape: predicates
deferred to value-typed-integration follow-up; dimensional semantics
ride on Quantity tag at Measure layer; worker authors 3 in-scope
refinements WITHOUT predicate at this layer.
P2 boundary discipline restored: brief now gives single-authority
guidance for Phase-2 substrate landing.
* docs(briefs): name predicate authority + concretize ratification ID — codex BLOCKING
Codex BLOCKING review on PR #1825 sha 32541a56:
1. (BLOCKING) Q-Refined-Phantom-Composition (c) ratification deferred
phantom-carrier predicates without naming the value-level predicate
authority. Updated canvas Q-Unit-4 RATIFICATION row to explicitly
name: 'predicates re-attach at value-typed Int/UInt level via
where-sugar (e.g., range(min: 0) per current dsl/std/types.dag
precedent)' when value-typed integration lands.
2. (BLOCKING) Worker brief option-c reframe didn't sweep the audit
receipt §5 (Carrier dissolves the bridge). Updated to consistently
apply (c) shape — dimensional semantics ride on Quantity tag at
Measure layer; value-level predicate authority deferred; until
then, range/gt_zero constraints attach at value-typed Int/UInt
via where-sugar per dsl/std/types.dag convention.
3. (NON-BLOCKING) Canvas had #issuecomment-... placeholder for the
(c) ratification message ID. Replaced with concrete
#issuecomment-4385687075 per dashboard relay header.
Brief now names a single value-level authority (Int/UInt where-sugar)
for value-range facts; phantom-layer Quantity tag carries only
dimensional semantics. P2 boundary discipline maintained.
* docs(briefs): S12 RETIRED per Director option (2) ratification
Director ratified option (2) at gunbc#828 #issuecomment-4385873...
2026-05-06 post smart-ram-167's substrate-state STOP-AND-ESCALATE:
- F2 absorbed by 2026-05-04 routing + #1662 closure
(`Result/DivError span.file-keyed` retired)
- F8 topical concerns distributed across live rows
(Class 5 Gap 3 `bootstrap.rs::std_fixtures()`; Character-level
load-set / bootstrap; Filename-sentinel-bridges) with own owners
- Doc-discipline goals absorbed by per-Mgr citation paydown
routing at gunbc#828 #issuecomment-4385583247
Frontmatter updated to RETIRED with absorption receipt. Worker
(smart-ram-167) pin freed.
Director-side acknowledgment of Substrate Mgr's option (2) recommendation:
"Substrate Mgr's option (2) recommendation is the right move
structurally. Most Mgrs default to 'synthesize work to fill the brief'
when canvas-vs-HEAD mismatch surfaces. Recommending retirement when
the work is genuinely absorbed is harder discipline."
* docs(briefs): S7 PR-F (BoundDeclaration consumer + ReferenceModel<T>) worker brief
Per cascade-clearance trigger fired post-#1782 merge 2026-05-06.
loyal-wolf-828 (#1764) acknowledged S7 dispatch with substrate-state
pre-flight grep (gunbc#1764 #issuecomment-4385872...) and surfaced
await-brief posture. This brief satisfies that prerequisite.
Worker pre-flight inventory absorbed:
- BoundDeclaration substrate landed (substrate.dag StaticBound +
PlatformDependent); partial BoundDeclarationView consumer in
fold.rs (ScratchIntExamples scope; doc-explicit partial)
- ReferenceModel<T> NOT yet in dsl/std/ (substrate-fact-introduction
required; design-emission-model.md §Q2 framing)
Phase-1: BoundDeclaration consumer broadening to full
design-emission-model.md surface (u128 / isize / usize / walker
arms / pilot mirror).
Phase-2: ReferenceModel<T> P1 substrate-fact-introduction with
DFS-of-concept-DAG + Practice 4 in-source checkpoint comment +
named consumer demand (Grounding G1).
Phase-3: cross-program handoff to Grounding Mgr (#1745) for
G1 (T-Ground-Rust Phase 1).
Closure predicate: unblocks Grounding T-Ground-Rust Phase 1.
Worker pin loyal-wolf-828 holds; dispatch fires immediately
on this brief landing.
* docs(briefs): S9 Slice 2.5 — predicate-registry + PB-1 shim retirement (Path 3 RATIFIED)
Director ratified Path 3 at gunbc#828 #issuecomment-4390333451
post proud-lynx-311's pre-flight DFS at gunbc#1746 #issuecomment-4390253616.
Brief reframes Slice 2.5 from S-sized (primitive + 1 consumer) to
L-sized (predicate registry + PB-1 shim retirement). 5 deliverables:
1. gt_zero primitive predicate
2. Predicate registry infrastructure (resolver-side substrate-fact-introduction)
3. PB-1 shim retirement (span.file=='dsl/std/types.dag' special case dissolves)
4. Migration of existing types.dag predicates (range, pattern, etc.)
5. PositiveInt = Refined<Nat, gt_zero> consumer update
Per Director hard scope bars: no parallel predicate-resolution path;
no per-call-site special-casing; PB-1 shim retirement non-negotiable
in same slice. Single PR per Path 3 RATIFIED.
Discipline anchors: feedback_construction_over_ratchets (PB-1 shim
is bridge); feedback_dissolve_bridges (registry is structural fix);
feedback_no_textual_enforcement_bridges (file-path special-case is
anti-pattern shape); P1 substrate-fact-introduction procedure.
Worker pin proud-lynx-311 — Director pre-cleared on pre-flight DFS
context. PR #1840 (partial Slice 2 — NonNegativeInt = Nat) MERGED
at 16:54Z; Slice 2.5 dispatchable immediately on this brief landing.
* docs(briefs): Slice 2.5 audit receipt — Mgr independent grep parallel-verifies worker DFS
Codex BLOCKING review on PR #1845 sha 2d230949 flagged: external PR
receipt absorbed as live substrate state without re-grepping local
std declarations at brief authoring time. Authority receipt §1
cited proud-lynx-311's pre-flight DFS as authority but didn't
record Mgr-side independent verification.
Updated audit receipt §1 + §4 to record Mgr's independent grep at
HEAD as parallel verification:
- gt_zero zero matches in dsl/std/ + src/v3/ (verified)
- NonNegativeInt = Nat at dsl/std/integer.dag:133 (#1840 land)
- PositiveInt = Int where range(min: 1) at dsl/std/types.dag:255
- PB-1 shim doc-block confirms bridge-text framing
- Predicate registry confirmed not present
Worker DFS remains the dispatch-time discipline (substrate-state
grep at dispatch); Mgr's independent grep is parallel verification,
not substitute. Per Director's pattern note at gunbc#828
#issuecomment-4390199218 + #issuecomment-4390333451: dispatch-packet
authoring discipline includes Mgr-side substrate-state-grep, not
just relay of worker findings.
* docs(briefs): Slice 2.5 brief — Int extensibility scope-bound; lower.rs cites anchored
openai-pro REQUEST_CHANGES on PR #1845 sha 2d230949 flagged 2 issues:
BLOCKING — P1 modeling faithfulness / substrate scope leak:
`gt_zero` Int-side extensibility was left to 'worker judgment if
generalization cheap'. Per feedback_construction_over_ratchets:
don't speculatively extend predicate-carrier compatibility ahead
of named consumer demand. Reframed to **EXPLICITLY OUT-OF-SCOPE**
for Slice 2.5; Nat-only registration. Future Int-side consumer
surfaces as separate substrate-fact-introduction (P1 procedure)
brief with own consumer-demand receipt.
NON-BLOCKING — citation discipline self-violation:
3 instances of bare `lower.rs:821-835` line-range citations
contradict the brief's own 'no bare :NNN' requirement at line 211.
Converted to function-name anchored form:
`src/v3/compiler/src/lower.rs` (`lower_type_alias_refinements_phase` doc-block).
Both findings absorbed; no parallel scope ambiguity remains;
brief now self-consistent on citation discipline.
* WIP: R3 Substrate
* docs(briefs): Slice 2.5 brief — audit-receipt cites converted to declaration-name anchors
cursor APPROVE_WITH_COMMENTS on PR #1845 sha ebf6ac12 noted citation
self-inconsistency: audit-receipt section §1 + §4 used bare
`dsl/std/integer.dag:133` and `dsl/std/types.dag:255` line-number
pins despite the brief's own no-bare-:NNN rule.
Converted to declaration-name-anchored form:
- 'declaration named NonNegativeInt' (with reproducible grep command)
- 'declaration named PositiveInt' (with reproducible grep command)
- 'lower_type_alias_refinements_phase doc-block' for PB-1 location
Brief now fully self-consistent on its own citation discipline.
Verified zero remaining .dag:NNN / .rs:NNN / .md:NNN bare cites.
* docs(briefs): Slice 2.5 brief — Path (a) RATIFIED carrier+argument contract verification
Director Path (a) RATIFIED at gunbc#828 #issuecomment-4390760353
(triage) + #issuecomment-4390794121 (convergence ack):
predicate-body lowering with carrier+argument contract
verification at lower-time IN SAME SLICE. Placeholder semantics
REMOVED from registry path.
Codex REQUEST_CHANGES surfaced that registered-predicate-without-
contract = silent acceptance per feedback_fail_closed_discipline.
Brief reframe absorbs Path (a):
Hard scope bar #5 added:
- Same-slice carrier+argument-contract verification
- Mismatched carrier OR malformed argument → Diagnostic::ResolveError
- No 'named but not checked' path remains
Scope expansion section added (L → XL):
- Predicate-body lowering for all 7 registered predicates
- Per-predicate arg-shape validation
- Multi-session work acknowledged
- Carrier-check at 788d6acb4 retained; predicate-body builds on top
Acceptance bullets updated: gt_zero with arg-shape validation +
all 7 predicates with carrier+argument contract verification.
Director-pinned brief-authoring discipline: dissolution trigger
named in a brief authored as substrate-fact-introduction must
specify same-slice acceptance, not deferred follow-up. Otherwise
the brief encodes 'land bridge + defer dissolution' anti-pattern,
the same shape feedback_construction_over_ratchets warns against
at implementation layer applied at brief authoring layer.
This is the 11th miss this session — but the pattern surfacing it
(direct sweep + Codex REQUEST_CHANGES + Director ratification)
produces structural fixes that prevent the next miss class. Brief
revision is the structural recovery.
* docs(briefs): Slice 2.5 brief — Path 2 RATIFIED (Gap 1 discharge + Q-Regex carve-out)
Director Path 2 RATIFIED at gunbc#828 #issuecomment-4391985613
post Rung 3 STOP at gunbc#1746 #issuecomment-4391946213.
Path 2 narrowing of Path (a):
- gt_zero + range body synthesis already landed (5f3c40c3a)
- Gap 1 discharge mechanism: extends scalar-literal predicate
evaluation at lower-time. L-class; lands SAME SLICE per
same-slice-dissolution discipline
- non_empty + brand body synthesis: lands via Gap 1 discharge
- pattern / format / content: fail-closed-with-named-dep at
user-code authoring layer (Diagnostic::ResolveError naming
Q-Regex-Primitive). NOT placeholder; structural rejection
- PB-1 shim FULLY RETIRED per hard scope bar #3
- Q-Regex-Primitive carved as follow-on substrate-fact-introduction
(XL substrate; not pre-authored per feedback_construction_over_ratchets)
Acceptance updated:
- 4 of 7 body-synthesized in same slice
- 3 of 7 fail-closed-with-named-dep (no placeholder semantics)
- Gap 1 discharge mechanism landed (no partial implementation)
Worker disposition: proud-lynx-311 continues Rung 3 with Gap 1
implementation + non_empty/brand body synthesis + pattern/format/
content fail-closed-with-named-dep diagnostic.
* docs(briefs): S9 Phase-1 Step 3 emission entries — Tier-1 brief authoring (1/5)
First Tier-1 brief landing post Director auto-nudge on assignment
#1858. Authored against Q-MachineConstraint sub-decisions ratified
shape now landed at #1856:
- Int<N> ≡ Compose<AbelianGroup, MachineWidth<N>> → Rust i32/i64/i128
- UInt<N> ≡ Compose<CommutativeSemiring, MachineWidth<N>> → Rust u32/u64/u128
- 6 concrete instantiations (≥3 minimum per sub-decision 5)
- Cross-program handoff to Grounding Mgr (#1745) for G2 emission consumption
- numeric_construction_demonstration Acceptance bullet (Int<32> round-trip);
Real<64> half deferred to S9 Phase 2 / S8 Float cascade
Brief authored with substrate-state-grep discipline + same-slice-
dissolution discipline (no deferred trigger; demonstration in same
slice or explicitly carved). 5-question authority audit completed
in brief body.
Worker pin proud-lynx-311 (S9 holds); brief queues post-Slice-2.5
completion as natural follow-on.
Tier-1 brief queue progress:
- 1/5 authored (this brief)
- 4 remaining: S9 Phase-2 Float coordination; T-LBP complexity
cementing; T-LBP cost cementing; S3 Phase-2 parser-grammar
* docs(briefs): S3 Phase-2 parser-grammar surface — Tier-1 brief 2/5
Authored per Director freed-pool pressure at gunbc#828
#issuecomment-4392095857 + Tier-1 brief-queue commitment.
Brief covers parser-grammar surface for Compose<Algebra, MachineConstraint>
interaction syntax post-#1856 (Phase-1 carrier slice landed):
- Phase 2.1: Parser surface per ratified Q-MachineConstraint-Grammar-Shape
- Phase 2.2: Bootstrap demonstrator with >=3 algebra*machine-axis pairs
- Phase 2.3: Class 1 5-criteria Pass receipt (criteria 1-4; criterion 5
v2-oracle parity may carve)
- Phase 2.4: numeric_construction_demonstration co-receipt with S9 Phase-1
Step 3 emission entries brief
Director ratification HOLDS on grammar bikeshed:
- Candidate A: Int @ Width<32> (annotation form)
- Candidate B: Int with MachineWidth<32> (with-form)
- Candidate C: Int<32> (positional generic — fits existing convention)
- Candidate D: Compose<AbelianGroup, MachineWidth<32>> (no parser change;
type aliases provide convenience names)
Worker pin valiant-ant-72 holds (S3 Phase-1 precedent owner); freed-pool
until Director ratifies grammar shape.
Tier-1 brief queue progress: 2/5 authored (this brief + S9 Phase-1 Step 3).
Remaining: S9 Phase-2 Float coordination; T-LBP complexity cementing;
T-LBP cost cementing.
* docs(briefs): S3 Phase-2 brief — fix internal contradiction on Class 1 closure scope
Codex BLOCKING review on PR #1889 sha ec609b4a flagged real
contradiction:
- Provenance section claimed '5-criteria Pass closed' (line 229)
- Phase 2.3 said 'criterion 5 may queue separately' (line 154)
- STOP-AND-ESCALATE bullet 1 ratified 4-of-5 carve-out (line 181)
Updated Provenance to consistently state criteria 1-4 close in this
slice; criterion 5 (v2-oracle parity) cementing test may carve to
separate slice. Single 4-of-5 framing throughout. INVARIANTS.md
'Documentation Describes Live State' satisfied.
Per feedback_construction_over_ratchets: don't claim closure that
isn't structurally complete. 4-of-5 + named carve is the honest
shape; 5/5 only on cementing-test land.
* briefs: cite Q-MachineConstraint sub-decision 6 supersession of T-V2-Retirement gate
Address Codex BLOCKING #1 on PR #1889: S3 Phase-2 + S9 Phase-1 Step 3
worker briefs consumed Q-MachineConstraint ratification without
reconciling the historical T-V2-Retirement-landing-first cascade
gate. Add an explicit precondition-framing section to each brief
citing Director ratification at gunbc#828 #issuecomment-4385530115
(Q-MachineConstraint sub-decision 6 UNIVERSAL substrate posture)
as the implicit supersession; PR #1856 landing without T-V2-Retirement
first confirms supersession in practice. STOP-AND-ESCALATE if
supersession contested at dispatch.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* slice-2.5 brief: mark pattern/format/content Path 2 acceptance pending Director ratification
Address Codex BLOCKING #2 on PR #1889: Path 2 deliverable + acceptance
language asserted fail-closed-with-named-dep semantics for
pattern/format/content, but Director rejected that shape (Option 4)
at gunbc#828 #issuecomment-4392.... Options A (consumer migration),
B (Q-Regex bundled into slice), C (typed fail-closed exception with
named trigger) await ratification per migration-cost catalog at
#issuecomment-4392116215. Brief now flags those three predicates +
PB-1 shim retirement coupling as PENDING DIRECTOR RATIFICATION; worker
STOP-AND-ESCALATE if dispatched pre-ratification.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* slice-2.5 brief: reconcile PB-1 retirement shape with Options A/B/C gating
Address Codex BLOCKING #4 on PR #1889: prior text claimed "PB-1
FULLY RETIRED" while preceding bullet said retirement gates on
chosen path — internal contradiction. Reframe to gate the
pattern/format/content half of shim retirement on Options A/B/C
ratification while preserving Director hard scope bar #3 binding
on the gt_zero/range/non_empty/brand half (cleanly retired via
Gap 1). No path leaves PB-1 alive as parallel-authority shim on
slice landing.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: tier-1 3/5 — S9 Phase-2 Float emission coordination worker brief
Tier-1 brief queue progress (3/5): S9 Phase-2 Float<N> emission
entries (Float<32>/Float<64> via Compose<ApproximateField<Real>,
MachineWidth<N>>) + Real<64> round-trip demonstration completing
§1.8 #67 Acceptance bullet. Coordination/synthesis brief — gates on
both S8 (ApproximateField<F> + Real base carrier) and S9 Phase-1
Step 3 (Int<N>/UInt<N> emission-entry pattern) landing before
dispatch. Inherits T-V2-Retirement supersession framing from
Phase-1 Step 3. STOP-AND-ESCALATE on missing preconditions or
unratified Q-ApproximateField-Axiom-Set.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: ledger advancement honesty (PRODUCER_LANDED not CONSUMER_LANDED) + restore exact issuecomment IDs
Address openai-pro BLOCKING + NON-BLOCKING findings on PR #1889:
BLOCKING (boundary discipline / same-PR consumer): S9 Phase-1 Step 3
+ Phase-2 Float briefs claimed DECLARED → CONSUMER_LANDED upon merge
while only requiring cross-program handoff receipt to Grounding G2.
For target-primitive substrate entries, handoff is not landed-consumer
proof. Reframed both briefs to advance rows DECLARED → PRODUCER_LANDED;
Grounding G2 follow-on PR (per-pair lowering rules + emitted-Rust-
primitive verification) advances to CONSUMER_LANDED. Default
split-PR producer-then-consumer per bundled-scope discipline ratified
at gunbc#1739 #issuecomment-4392225548 (parallel infrastructure
DISALLOWED in same PR).
NON-BLOCKING (citation hygiene): replaced ellipsized
"#issuecomment-4392..." stubs in slice-2.5 brief with the exact
Director Option-4-rejection comment ID #issuecomment-4392081719 so
the ratification claim is mechanically checkable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* slice-2.5 brief: absorb Director Option A revised RATIFICATION (final shape)
Director ratified Option A revised at gunbc#1739 #issuecomment-4392329709
(2026-05-06). Final brief shape:
- 4 predicates body-synthesize via Gap 1 (gt_zero / range / non_empty
/ brand)
- pattern / format / content STRUCTURALLY ABSENT from registry (not
pending-with-placeholder); user-code where-clause produces
Diagnostic::ResolveError naming Q-Regex-Primitive as separate
substrate-fact-introduction follow-on
- ~12 types.dag refined-type declarations migrate to drop
pattern/format/content where-clauses with inline-comment receipts
pointing to Q-Regex restoration path; 1 extdep (sts_endpoint)
adapts as needed
- PB-1 shim FULLY RETIRED — no exception, no preserved branch
- Q-Regex-Primitive as separate brief / worker pin / Director
ratification cycle when concretely needed (Option B XXL bundling
REJECTED)
Phase ordering rewritten to absorb migration as explicit phase
between predicate enrollment and shim retirement. proud-lynx-311
re-dispatches against revised brief; PR #1846 absorbs additional
scope per Director directive.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* s3-phase-2 brief: absorb Q-MC sub-decision 3 ratification (grammar shape was already answered)
Director observation at gunbc#1739 #issuecomment-4392382517: Q-MC
sub-decision 3 (Brian directive 2026-05-06 at gunbc#828
#issuecomment-4385530115) already ratified the grammar shape —
`Int<N>` user surface desugars to `Compose<Int, MachineWidth<N>>`
substrate parametrically. Earlier brief HOLD framing was redundant.
Brief revised:
- Replace bikeshed section with Q-MC sub-decision 3 quotation +
surface/substrate framing (Candidate C user-facing; Candidate D
substrate elaboration; A/B rejected per Director assessment)
- Phase 2.1 parser surface scoped to numeric-literal-position
recognition + parametric desugar
- Front-matter status: dispatchable per ratification (not HOLD)
- Provenance: ratification-state-grep discipline pin folded into
standing dispatch-checklist
valiant-ant-72 dispatch fires immediately per Director directive.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: remove truncated issuecomment-4392... stub from S9 Phase-1 Step 3 provenance
Address cursor NON-BLOCKING finding on PR #1889 sha 94292e48: line 144
truncated issuecomment ID. The originally-intended Director auto-nudge
comment doesn't map cleanly to a single ID, so replace with the
standing assignment reference (gunbc#1858) which is unambiguous.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: tier-1 4/5 + 5/5 — T-LBP complexity + cost lens cementing-test worker briefs
Tier-1 brief queue completes (4/5 + 5/5):
4/5 — T-LBP complexity-lens cementing test (§1.8 rows #79 + #87
first entry). Consumer-side cementing test against frozen v2-oracle
snapshot; producer-side Lens<Complexity> instance is upstream
precondition. Per Q-Lens-Behavioral-Parity-R3-Closeability option
(b) RATIFIED at gunbc#828 #issuecomment-4385329180 — T-LBP narrowed
to complexity + cost lenses only.
5/5 — T-LBP cost-lens cementing test + cost_lens_demonstration
(§1.8 rows #80 + #70 + #87 second entry). Absorbs row #70 demo
naturally — same source corpus + lens run; demonstration adds
observable-cost-bound assertions on top of parity check. ≥2
algebra-instances composed + ≥1 recursive call required per row #70.
Both briefs gate on:
- T-LBP per-lens substrate-producer brief landing (Lens<C> instance)
- Frozen v2-oracle snapshot capture (separate brief, pre-v2-retirement)
- DO NOT capture-against-live-oracle (violates
v2_oracle_no_remaining_test_consumers)
- DO NOT bundle producer/snapshot/algebra-instance authoring per
Director bundled-scope ratification at gunbc#1739
#issuecomment-4392225548
Tier-1 5/5 complete. Worker pins assigned at dispatch time once
preconditions land.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* canvas: emission-provenance substrate shape disposition (supersedes PR #1902 PROPOSAL)
Substrate Mgr canvas authored post-codex BLOCKING + PM disposition
handoff at gunbc#846 #issuecomment-4392510255.
Two real substrate-shape questions stacked:
Q1 (category mismatch): Lens<C>.read is per-Behavior; emission
provenance is per-emitted-line. 4 options surfaced — (a) per-Behavior
Lens-compatible / (b) per-line instrumentation NOT a lens / (c)
withdraw + re-canvas / (d) Mgr-surface both substrates distinct.
Mgr recommends (d) if visualization is load-bearing; (a) suffices
if per-Behavior aggregate works for the slide. Brian-channel
sub-question surfaced.
Q2 (fold-rule enumerability): T-Rule-Enumeration substrate-fact-
introduction is prerequisite for every Q1 path (rust_target.rs uses
inline &str template names; rules not enumerable). Mgr recommends
parallel-dispatch with Q1 deliberation — satisfies Brian's ASAP
framing while shape ratifies.
Q3 (§1.8 ledger retarget): #89 already taken by T-LAS; new cluster
location follows Q1 disposition.
Director ratification ask + Brian-channel sub-question surfaced.
PR #1902 close-superseded post-ratification.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: reconcile Compose<...> slot-1 spellings to algebraic-concept names per Q-MC sub-decision 3
Address cursor NON-BLOCKING on PR #1889 sha 2b49b30f: S3 Phase-2
brief mixed Compose<AbelianGroup, ...> with Compose<Int, ...>; S9
Phase-1 Step 3 used Compose<AbelianGroup, ...> + Compose<CommutativeSemiring,
...>; S9 Phase-2 used Compose<ApproximateField<Real>, ...>. All wrong
per Q-MC sub-decision 3 critical correction at gunbc#828
#issuecomment-4385530115:
"prior phrasing Compose<AbelianGroup, MachineWidth<64>> was wrong
— AbelianGroup<T> is a witness shape over carrier T, not a carrier
constructor; bare AbelianGroup in slot-1 composes witness rather
than concept."
Slot-1 is the fully-applied algebraic-concept name:
- Int<N> = Compose<Int, MachineWidth<N>> (Int = AbelianGroup<GroupCompletion<Nat>>)
- UInt<N> = Compose<UInt, MachineWidth<N>> (UInt = CommutativeSemiring<Nat>)
- Real<N> = Compose<Real, MachineWidth<N>> (Real = ApproximateField<Rational>)
Phase-2 Float brief also retargets user-facing surface from Float<N>
to Real<N> (Float is target-language name for the Rust primitive;
Real is the algebraic-concept name per Q-MC sub-decision 3 example).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* s9-p1-s3 brief: scope Deliverable 4 to producer-side verification (defer demonstration to Grounding G2)
Address Codex BLOCKING (line 82 inline + bundled review): brief
claimed end-to-end Int<32> → Rust i32 round-trip in Acceptance
while Deliverable 3 said lowering rules belong to Grounding G2
follow-on. Internal contradiction.
Reframe per Director bundled-scope discipline (gunbc#1739
#issuecomment-4392225548 — parallel infrastructure DISALLOWED
same-PR):
- Deliverable 4: producer-side substrate-axis verification only
(type-check + carrier acceptance + bootstrap snapshot/manifest
hold). NO end-to-end Rust emission in this PR.
- §1.8 #67 numeric_construction_demonstration receipts split:
producer half here (entries exist), consumer half on Grounding
G2 PR (full Int<32> → Rust i32 round-trip).
- Same-PR cross-program co-author available if Substrate Mgr
coordinates with Grounding Mgr at dispatch; default is split.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: pre-author T-Rule-Enumeration substrate-fact-introduction worker brief
Pre-authored per pre-authored-brief-queue discipline; dispatch fires
on Director Q2 ratification of emission-provenance canvas at
gunbc#828 #issuecomment-4392519713.
Brief covers LangSpec emission-rule names as enumerable substrate
(rust_target.rs uses inline &str template names today; 0 hits on
RuleName/FoldRule/EmitRule across src/v3/). Two carrier shape options
(α sum-type / β named-string list) with worker DFS at dispatch
deciding shape; Practice 4 checkpoint discipline + ledger-row receipt.
Prerequisite to every Q1 path (a/b/d) — Lens<EmissionProvenance>,
per-line instrumentation, and the both-substrates option all consume
the rule-name set. Path (c) leaves substrate-unblocked-but-unconsumed.
Mgr authoring authority per PM concur at gunbc#846
#issuecomment-4392510255 / #issuecomment-4392543633 (substrate-fact-
introduction is Mgr-tier, not PM tactical authoring).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* t-rule-enumeration: align gate name + status to Director ratification at #4392562911
* emission-provenance lens brief: revise per Director Q1 (a) RATIFICATION
Director Q1 (a) per-Behavior Lens<C>-compatible RATIFIED at gunbc#1739
#issuecomment-4392562911 (2026-05-06); (b) instrumentation REJECTED;
(d) parallel-substrates REJECTED.
Brief revised:
- Carrier shape: Lens<List<EmissionProvenance>> per (a)
- EmissionOrigin = SubstrateDeclMirror(SourceSpan) | FoldRuleAutoEmit(EmissionRule)
typed sum (closed; no third silent class; structural fail-closed)
- All 6 Lens<C> field bindings concrete (read/sequential/branch/iterate/validate)
- T-Rule-Enumeration named as PRECONDITION (must land first)
- §1.8 retarget under T-CostLens-Composition cluster per Director Q3
- Brian's slide visualization served by (a) projection (per-Behavior
list flattens to per-line view at visualization layer; no parallel
substrate needed)
Supersedes PM-authored proposal at PR #1902 (merged at 54419badf).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: bundled Rust-primitive-full-coverage substrate-prerequisite worker brief (Director Path A)
Author bundled substrate-prerequisite brief per Director Path A
RATIFICATION at gunbc#1739 #issuecomment-4392731264 (2026-05-06).
Three structurally-coupled gaps surfaced via Grounding loyal-stag-699
STOP at gunbc#1907 #issuecomment-4392585090 land as one bundled slice
per "necessary structural fix" exception (bundled-scope discipline):
1. IntervalInt::ExactInterval host-repr widening (α BigInt-based
recommended; β typed-variants if α infeasible)
2. RustPrimitive structural BoundDeclaration field (replaces static
range_min/max strings; consumes existing BoundDeclaration carrier
StaticBound + PlatformDependent variants)
3. spec/rust.dag PlatformDependent row population for isize/usize +
u128 row addition consuming widened ExactInterval
Closure gate: rust_primitive_full_coverage (Director-ratified name).
Worker pin: valiant-ibex-312 (freed-pool; substrate-authoring fresh;
numeric-domain-adjacent post-#1842).
Unblocks Grounding G2 Phase 2 full-coverage dispatch. G2 Phase 1
narrowed dispatch (i8-i64 + u8-u64) runs in parallel on existing
substrate; cleared at gunbc#1745 #issuecomment-4392795954.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: align Practice 4 classifications to GREEN/YELLOW rubric + add GREEN ledger receipts
Address codex BLOCKING on PR #1910 sha f36e1205: briefs used
PRIMITIVE/SCAFFOLD terminology that doesn't match docs/modeling-
discipline.md rubric. Per modeling-discipline.md lines 97/103/108:
the three classifications are 🟢 GREEN (terminal) / 🟡 YELLOW
(scaffold) / 🔴 RED (dissolvable-now); line 132 requires a ledger
entry if GREEN.
Replaced PRIMITIVE → GREEN, SCAFFOLD → YELLOW across all three
emission-provenance / Rust-primitive-full-coverage briefs. Added
explicit §1.8 ledger entry receipt for the GREEN classification on
EmissionOrigin (sibling row emission_origin_classification_green to
parent emission_provenance_lens_landed gate).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: Practice 4 terminal receipt for EmissionOrigin + bound for β scaffold
Address openai-pro BLOCKING (sha f36e1205, ran post-rubric-rename):
1. EmissionOrigin GREEN terminal classification needs documented
attempted-dissolutions per modeling-discipline.md "no richer
source exists" framing. Added 3 attempted dissolutions:
- collapse to optional-pair (rejected by codex Finding 1 prior)
- factor common richer source (orthogonal payload types prevent)
- third arm for unattributable lines (rejected per Slice 2.5
Option 4 placeholder anti-pattern)
2. T-Rule-Enumeration β (YELLOW) scaffold needs bound per three-part
bridge rule. Added: allowed call sites scoped to emission code only;
max lifetime gated on rule_name_typo_fail_closed_landed graduation;
new emission-introspection consumers MUST author against closed-sum
α form, not β.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* WIP: R3 Substrate
* briefs: address codex BLOCKING bundle (sha 2ed1046e) — record-form provenance + emit_model carrier alignment + T-Rule-Enumeration retirement flag
Three findings addressed:
1. Lens brief (Finding #1 — provenance shape conflates source vs rule
attribution): replaced typed-sum EmissionOrigin with record
EmissionProvenance { rule: EmissionRule [mandatory], source_span:
Option<SourceSpan> [optional] }. Co-inhabitance is real (e.g.,
#[derive(Debug)] line has both rule=derive_for_disj AND span
attribution to Foo decl). Rule-mandatory enforces fail-closed
structurally; Practice 4 N/A (record, not coproduct).
2. T-Rule-Enumeration brief (Finding #2 — rule authority lives in
spec/rust.dag + v3.std.emit_model, not call-site strings): brief
status flipped to FLAGGED FOR RETIREMENT. Same finding as
smart-ram-167 STOP at gunbc#1759 #issuecomment-4392696623; Reading
C surfaced to Director at gunbc#828 #issuecomment-4392780871.
3. Rust-primitive-full-coverage brief (Finding #3 — BoundDeclaration
not wired into TargetIntegerInhabitanceBound row surface):
reframed Deliverable 2 to target the actual carrier
`v3.std.emit_model::TargetIntegerInhabitanceBound` (used by
`TargetIntegerTypeInhabitance` rows in src/v3/spec/rust.dag lines
169-207). Two paths surfaced: Option (i) wire BoundDeclaration into
TargetIntegerInhabitanceBound; Option (ii) populate via existing
variants if PlatformDependent already exists on the live carrier.
Worker DFS at dispatch picks. Deliverable 3 retargeted to
TargetIntegerTypeInhabitance rows (the actual surface).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* rust-primitive-full-coverage: codex inline confirms TargetIntegerInhabitanceBound has no PlatformDependent variant — Option (ii) infeasible; carrier refactor REQUIRED
Codex BLOCKING inline at PR #1910 line 65 (sha 98507c432) verified at
HEAD: TargetIntegerInhabitanceBound = BoundUnspecified | StaticBoundFact(IntInterval).
No PlatformDependent variant. Earlier brief framing of Option (ii)
(populate via existing variants) was infeasible.
Brief reframed: carrier refactor is required (not optional). Two
structural shapes for the refactor — (i.a) embed BoundDeclaration
(Mgr recommendation; single-authority discipline) vs (i.b) extend
variant set with PlatformDependent. Existing 5 rows migrate either
way; bootstrap snapshot semantic-equivalence non-negotiable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* emission-provenance lane: T-Rule-Enumeration RETIRED + Lens brief revised per Director Reading C
Director Reading C RATIFIED at gunbc#1739 #issuecomment-4392797954
(2026-05-07):
- Q1: T-Rule-Enumeration retires (substrate gap was phantom; ~37
*SyntaxBinding/*OpsBinding field paths already provide enumeration
per smart-ram-167 substrate-state-grep at gunbc#1759 #issuecomment-4392696623)
- Q2: Lens<EmissionProvenance> dispatches now (no precondition gate)
- Q3: EmissionRule = String of field-path (per feedback_reason_not_label)
Actions:
- Delete docs/briefs/r3-substrate-t-rule-enumeration-worker.md
- Lens brief: replace T-Rule-Enumeration precondition gate with
Reading C absorption section; EmissionRule = field-path String;
cementing test verifies field-path round-trip; STOP triggers
retargeted to *SyntaxBinding field-set divergence
Worker pin: smart-ram-167 (Mgr discretion per Director ratification;
fresh context on field-path enumeration). valiant-ibex-312 stays on
Rust-primitive-full-coverage / T-Interval-Representation.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* emission-provenance lens brief: absorb codex BLOCKING bundle (sha 4229cd09 post-merge findings)
Two findings from codex bundled review on now-merged PR #1910 sha
4229cd09 land as brief revisions for the in-flight smart-ram-167
authoring cycle (PR #1928):
1. Field-path identity threading: render_named_template call sites
receive template values extracted from binding structs; field-path
may not be carried alongside the value at emission time. New STOP
trigger flags this as substrate-cascade if threading isn't
structurally possible (typed template-with-rule-path or
DeclarationRef carrier needed).
2. Read domain coverage gap: Lens<C>.read is per-Behavior; emitted
output also includes declarations + program scaffold. Acceptance
scope explicitly narrowed to Behavior-attributable lines only;
declarations + scaffold provenance out-of-scope (separate substrate
cascade if surfaced). Acceptance bullet retargeted to
"Behavior-attributable span-Some + Behavior-attributable span-None"
rather than full-emission coverage.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* rust-primitive-full-coverage: revert to Option (ii) — Mgr ratification-state-grep miss caught by valiant-ibex-312
valiant-ibex-312 ratification-state-grep at gunbc#1761
#issuecomment-4393074602 surfaced that origin/main HEAD has
TargetIntegerInhabitanceBound with 3 variants including
PlatformDependentFact (mirrors BoundDeclaration::PlatformDependent).
Earlier brief framing at 4229cd09d was against pre-PlatformDependentFact
state per codex BLOCKING reading; main moved since.
Option (ii) IS feasible: isize/usize rows populate via existing
PlatformDependentFact variant directly; u128 row populates StaticBoundFact
with BigInt-widened ExactInterval (Deliverable 1 unchanged). No
carrier refactor needed; 5 existing rows untouched.
Second Mgr-tier ratification-state-grep miss today (first was symbol-only
grep on T-Rule-Enumeration; now stale-codex-vs-current-HEAD on
TargetIntegerInhabitanceBound). Worker-tier discipline caught it before
sinking time into invasive Option (i.a) refactor.
Single-authority concern (parallel-named variants on
TargetIntegerInhabitanceBound vs BoundDeclaration) noted as future
substrate-fact-introduction; not blocking this slice — explicit-mirror
is the existing convention.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* rust-primitive-full-coverage: align Acceptance + Phase ordering + STOP triggers to Option (ii) (resolve codex internal-contradiction finding)
codex BLOCKING on PR #1929 sha 14f68b63 caught real internal contradiction: Context section flipped to Option (ii) feasible at 57ae4349b but Acceptance + Phase ordering still referenced BoundDeclaration migration / RustPrimitive bound field refactor from earlier Option (i.a) framing.
All sections now consistent with Option (ii):
- Phase 4 = no-op (TargetIntegerInhabitanceBound carrier untouched)
- Phase 5 = spec/rust.dag rows for isize/usize via existing PlatformDependentFact variant
- Phase 6 = u128 row via StaticBoundFact(IntInterval) + ratchet update
- Acceptance retargeted to row-population shape (no carrier refactor; existing 5 TargetIntegerTypeInhabitance rows untouched)
- STOP triggers retargeted to TargetIntegerInhabitanceBound variant-set sufficiency at HEAD
- Why-bundled rationale rewritten to Option (ii) shape (Phase A widening unblocks u128 StaticBoundFact; isize/usize use existing PlatformDependentFact)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* lens brief: narrow Deliverable 3 verification step 4 to Behavior-attributable lines (resolve codex #2 consistency)
codex BLOCKING #2 on PR #1929 sha 812602b4 caught: Acceptance + STOP triggers narrowed to Behavior-attributable lines but Deliverable 3 cementing-test step 4 still said 'every emitted line'. Now consistent — step 4 explicitly scopes to Behavior-attributable subset; declarations + program scaffold OUT-OF-SCOPE per same Finding #2 framing.
(codex BLOCKING #1 — claim that PlatformDependentFact NOT on main — is wrong; verified at origin/main HEAD: TargetIntegerInhabitanceBound has 3 variants including PlatformDependentFact. Rebut on PR.)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* briefs: pre-author IntPlatform/UIntPlatform substrate-fact-introduction follow-on (Director ratification at #4393248961)
Per Director Q1+Q2+Q3+Q4 RATIFICATION at gunbc#1739 #issuecomment-4393248961 (2026-05-07): pre-author follow-on brief for the substrate cascade surfaced by valiant-ibex-312 STOP at gunbc#1761 #issuecomment-4393222862.
Scope:
- Q1 naming: IntPlatform / UIntPlatform per feedback_reason_not_label
- Q2 algebra: Compose<Int, MachineWidth<Platform>> + Compose<UInt, MachineWidth<Platform>>; Platform is substrate token
- Q3 substrate-concept layer (target-agnostic per feedback_target_agnostic_ir)
- Q4 worker pin: valiant-ibex-312 post-PR-#1914 close (freshest substrate context)
Brief covers Platform substrate token + IntPlatform/UIntPlatform declarations + spec/rust.dag isize/usize row population + co-receipt advancing PR #1914's rust_primitive_full_coverage gate PARTIAL → CONSUMER_LANDED.
Closure gate: int_platform_uint_platform_substrate_landed.
Dispatchable post-PR-#1914 close per pre-authored-brief-queue discipline.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* intplatform brief: rename inner token Platform → PointerWidth per Director ack at…
This was referenced May 7, 2026
briansrls
added a commit
that referenced
this pull request
May 9, 2026
…-authority HEAD state (#2429) * docs(r3): gate #24 partial-CONSUMER_LANDED — Int+UInt arms via single-authority HEAD state Mgr-ratified scope-narrow disposition at gunbc#2068 c#4413651570 (same shape as gate #18 quiet-ibex-647 precedent, Director (c) structural- coverage at gunbc#828 c#4413633535): gate #24 numeric_reframe_no_parallel_authority closes for Int + UInt arms by HEAD construction; Float arm tracked separately on S8 ApproximateField<F> cascade. Status moves DECLARED -> PARTIAL CONSUMER_LANDED 2026-05-09: - Int arm: type Int = AbelianGroup<GroupCompletion<Nat>> at dsl/std/integer.dag:147 (Slice 3 PR #1466 Q6 single-authority form); no type Int = Int64 residue. - UInt arm: type UInt = Nat at dsl/std/integer.dag:148 (PR #1818 canonical-instance form); no type UInt = UInt64 residue. - Float arm: HELD on S8 cascade per S9 brief Phase-2 STOP-AND-ESCALATE (docs/briefs/r3-substrate-s9-t-numeric-construction-worker.md:95-100). dsl/std/float.dag:18 type Float = Float64 persists pending S8 ApproximateField<F> import; hand-rewrite ahead of S8 precluded by feedback_import_not_redeclare_carriers. Advances to full CONSUMER_LANDED upon S8 land. Audit receipt: fierce-fox-86 grep at gunbc#2068 c#4413645211. Doc/ledger-closure scope only — no code edits to dsl/std/float.dag or substrate carriers. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(r3): gate #24 — keep Status DECLARED, move per-arm evidence to Notes (§1.7 schema fix) Codex BLOCKING review on PR #2429: §1.7 status taxonomy defines only DECLARED / CONSUMER_LANDED / PASSING. The prior "PARTIAL CONSUMER_LANDED" label introduced a non-canonical status not in the schema. Per reviewer's option (1): keep Status=DECLARED; per-arm HEAD evidence (Int/UInt arms structurally clear at HEAD; Float arm HELD on S8) lives in Notes column as non-closure receipt. Status → CONSUMER_LANDED when an executable HEAD-grep consumer lands covering all three arms. Mirrors the §1.7 corpus rule: slice receipts are evidence in Notes, not closure-status promotions. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
T-Numeric-Construction Slice 3 (R3 lane #6, dispatch on inbox #1288 #4362239705). Pivots the default
Intalias to the canonical Q6 single-authority form per the audit atdocs/audit/t-numeric-construction-group-completion-6q.md. Builds on the now-mergedGroupCompletion<M>substrate carrier (#1448) plus Slices 1/2 (#1378, #1399).Construction-chain progression
What changed
dsl/std/integer.dag:45Plus imports for
AbelianGroup/GroupCompletion(fromstd.algebra) andNat(from Slice 2'sstd.nat). Header MODELING NOTE updated to document the construction-chain pivot and the Q6 single-authority resolution.Type-correctness under standard parametric reading
T = GroupCompletion<Nat>— opaque atom denoting "the abelian-group carrier derived fromNatvia the Grothendieck construction." Substrate-introduced in feat(std): GroupCompletion<M> opaque-atom substrate-introduction #1448.AbelianGroup<T>over that carrier typesinverse: fn(GroupCompletion<Nat>) -> GroupCompletion<Nat>— honest by construction (every element of the group-completion carrier has an additive inverse). The earlier rejectedAbelianGroup<Nat>form was M9-false becauseNat's elements have no additive inverses; the carrier-construction layer fixes that.What did NOT change (per dispatch guardrails)
Int8..Int128 = OrderedRing<Word*>,UInt8..UInt128 = Semiring<Word*>).UInt = UInt64alias unchanged (per design doc §6 cascade scope).LitInt/ literal-grammar / refinement-syntax work.Word*storage carriers stay storage refinements atdsl/std/bit.dag— not alternateInt/Nat/Magnitude/GroupCompletionauthorities.Downstream impact verification
The dispatch's STOP+PING condition was: "if changing the alias triggers broad consumer migration beyond ratchets/generated outputs." Verified locally on the full integration suite — only 2 tests were affected by the alias pivot, both within "ratchets/generated outputs" scope:
m1_substrate_test::bootstrap_int_add_walk_reaches_ordered_ring_add_nobody_arrow— legacyInt → OrderedRingratchet. Repointed to walkInt64directly (theOrderedRing<Word64>chain it structurally tests now lives at the fixed-width row, not the default alias). Renamed tobootstrap_int64_add_walk_reaches_ordered_ring_add_nobody_arrow. The default-alias ratchet now lives inint_default_alias_resolves_to_abelian_group_over_group_completion_of_nat(new, see below).parse_stage4_prep::handwritten_parse_snapshot_matches_manifest—integer.dagbyte/hash drift from the alias edit; refreshed viarefresh_handwritten_parse_snapshot_manifest.Final: 740 passed / 0 failed / 26 ignored in the integration suite; 225 passed in lib; clippy clean; fmt clean.
New ratchet
src/v3/compiler/tests/integration/m2_substrate_inhabitance_test.rs::int_default_alias_resolves_to_abelian_group_over_group_completion_of_nat:Int.span.file == "dsl/std/integer.dag".AbelianGroup(NOTOrderedRing/Int64, NOT the rejected compactGroupCompletion<Nat>form, NOT aWord*storage carrier).Instantiation { template: GroupCompletion, arguments: [Nat] }— proving the two-step construction.Files
dsl/std/integer.dag— alias pivot + header update.src/v3/compiler/tests/integration/m2_substrate_inhabitance_test.rs— new structural ratchet for the canonical Slice 3 form.src/v3/compiler/tests/integration/common/substrate_receipts.rs+m1_substrate_test.rs— repurposed legacy receipt to walkInt64directly + renamed test.src/v3/compiler/tests/integration/parse_corpus_manifest.txt— refreshed.src/v3/compiler/src/bootstrap_{std_,}generated{,_without_parse_surface}.rs— regenerated.Test plan
cargo run -p v3-compiler --features bootstrap-regen-fresh --bin regen_bootstrap(clean).cargo test -p v3-compiler --lib -- --skip execute_command_timebound(225/0).cargo test -p v3-compiler --test integration(740/0/26 ignored).cargo clippy --all-targets -- -D warningsclean.cargo fmt --all --checkclean.