Compiler pipeline correctness and scaffold dissolution (AI-5 InferredFacts illegal states, AI-7 TestClaim cache key, AI-13 SurfaceSugarKind local enum, AI-14 LexPattern/GrammarExpr parallel ASTs, AI-15/R1-2 translate source gap, R1-3 constraints singleton-MVP, R1-4 Lens apply_lens stub, R1-5 eval lo by briansrls · Pull Request #3587 · gunb-ai/gunbc

briansrls · 2026-05-23T22:58:20Z

Auto-opened by session-dashboard for session deep-bear-704.
Pushing to session/deep-bear-704 advances this PR.

Worker attestation

Before flipping this PR to ready for review, confirm each item:

Title describes the change (not the session id or branch).
PR body summarises what and why (replace the TODO below).
Tests run: name the command (e.g. npm test, cargo test) and the result.
If this closes a work item, the body contains a Closes #N directive.
No commits on this branch are surprises (no fork/cherry-pick I did not make).
No secrets / credentials / large binaries staged.

Summary

TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.

Test plan

TODO: list the commands that ran (or "no tests changed; relied on CI") and the outcome.

…erredFac

briansrls

Review metadata

Provider / model: codex / unknown
Commit: 230eef2b · Trigger: schedule
Thinking: 330s wall

BLOCKING (1)

Root Cause

src/v4/std/constraints.dag CanonicalGrounding is being treated as a proof-by-construction carrier while its coherence is the upstream authority -> enforce node/witness/closedness alignment in that carrier/API or keep the infer boundary check until it does.

⚠️ The InferredFacts consolidation is the right direction, but the boundary still trusts grounding coherence by convention.

Sign in to view

  facts: InferredFacts,
 ) -> Outcome<CanonicalGrounding> {
-  if inferred_facts_canonical_grounding_coherent(facts: facts) {
+  if well_formed(n: facts.grounding.node) {


…erredFac

…ority) Fixtures and the R2-7 shape ratchet assumed PR #3587's collapsed {grounding, descent} type had landed in 04_infer.dag, but authority still declares {resolved_type, inhabits, canonical, descent} on main. Restore test/claim fixtures to the four-field shape, drop the drift guard, and keep R2-1 / AI-16 scope intact. Co-authored-by: Cursor <cursoragent@cursor.com>

…erredFac

…ing production path Clarify construction authority on CanonicalGrounding: production callers must use solve_constraints; direct construction is test-fixture-only until T-9 closedness bridge. Co-authored-by: Cursor <cursoragent@cursor.com>

briansrls · 2026-05-23T23:41:10Z

Addressed in 202fb1c: added construction-authority comment to constraints.dag documenting solve_constraints as sole production path; direct construction is now documented-forbidden in production, with explicit test-fixture exception. Coherence-by-construction until T-9 dissolution gate fires.

(Re: api-review review 4351534588 / codex blocking finding on CanonicalGrounding convention-tier coherence.)

…ize/parse Gate LexCharClass, LexRule, LexRules, and Grammar with feature:CP-1b-bidirectional- grammar-carrier; remove duplicate Grammar mark on 02_parse.dag. Co-authored-by: Cursor <cursoragent@cursor.com>

Keep main's apply_compile_lens/CompileLens rename; preserve T-23 gated migration stub comment from session branch. Co-authored-by: Cursor <cursoragent@cursor.com>

…erredFac

Preempt #3598 removal of non_empty from collection.dag; canonical owner is algebra.dag. Co-authored-by: Cursor <cursoragent@cursor.com>

…erredFac

…ding Re-validate canonical_grounding_admits_infer_facts at inferred_facts_for_eval and eval_accept_runtime_value_with_facts so directly-constructed test facts cannot bypass infer admission at the eval consumer boundary. Co-authored-by: Cursor <cursoragent@cursor.com>

briansrls

Review metadata

Provider / model: codex / unknown
Commit: 7e140448 · Trigger: schedule
Thinking: 394s wall

Non-blocking — Strengths

src/v4/compiler/04_infer.dag The changed infer boundary now fails closed on incoherent grounding instead of exporting a convention-only fact bundle.

✅ No blocking concerns in the changed lines.

briansrls · 2026-05-24T00:39:48Z

Verified on current HEAD 5d5a2dd64: infer fail-closed admission via canonical_grounding_admits_infer_facts at construction/re-export (04_infer.dag) plus eval consumer re-validation at inferred_facts_for_eval and eval_accept_runtime_value_with_facts (05_eval.dag). Agree with codex sha 7e140448 non-blocking assessment — no further changes needed for this finding.

— sent from deep-bear-704

briansrls · 2026-05-24T00:44:43Z

Verified on HEAD 5d5a2dd64: composer APPROVE aligns with current tree — infer/eval fail-closed via canonical_grounding_admits_infer_facts, constraints construction-authority comment, fixture centralization, and gated scaffold marks all present. Optional cleanup debt (sr_mod_canonical_witness / sr_unbound_canonical_witness in two structural-resolution claims) noted; not treating as blocking per review.

— sent from deep-bear-704

Keep T-23 gated migration stub comment on apply_compile_lens; integrate main through aea4481 (T-31 de-prose mop-up, T-4.6 sql, T-23 lens app). Co-authored-by: Cursor <cursoragent@cursor.com>

briansrls · 2026-05-24T01:09:39Z

Verified on HEAD 16ccbfbb8 (post-main merge): claude-opus APPROVE findings hold — InferredFacts illegal-states collapse + single canonical_grounding_admits_infer_facts gate in infer/eval, constraints construction-authority comment, and gated scaffold receipts all present; no dangling removed-field references. No further changes for this review item.

— sent from deep-bear-704

briansrls

Review metadata

Provider / model: codex / unknown
Commit: 5d5a2dd6 · Trigger: schedule
Thinking: 3618s wall

BLOCKING (1)

Root Cause

src/v4/compiler/00_compile.dag T-23 application.dag is already filled, but the compile-side adapter receipt still treats apply_lens as future and leaves the ratified apply_lens surface unavailable → add/rename the compile gate to the ratified apply_lens adapter or update the changed receipt and smoke contract with an explicit higher-authority reconciliation.

Non-blocking — Strengths

src/v4/compiler/04_infer.dag The InferredFacts reshape removes the prior independent resolved_type/inhabits/canonical slots and makes eval fail closed on incoherent CanonicalGrounding.

⚠️ One blocking T-23 gate/surface mismatch remains.

briansrls · 2026-05-24T02:59:13Z

Violations (could not place on specific lines):

src/v4/compiler/00_compile.dag:117 BLOCKING: The new T-23 migration gate is already stale: v4.lens.application declares canonical apply_lens today, while compile still exports only apply_compile_lens and the existing T-23 smoke gate fails, so P5 Progress Is Dissolution requires dissolving or explicitly reconciling this opened gate now.

briansrls

Review metadata

Provider / model: codex / unknown
Commit: 16ccbfbb · Trigger: schedule
Thinking: 420s wall

BLOCKING (1)

Root Cause

src/v4/std/coercion.dag coercion_not_realized_diagnostic only reads source.node but is typed as CanonicalGrounding, forcing a fabricated grounding for a diagnostic → change that diagnostic boundary to accept Node so solve_constraints remains the single production constructor.

Non-blocking — Strengths

src/v4/compiler/04_infer.dag The InferredFacts consolidation reduces the prior illegal-state surface and eval now fails closed on incoherent grounding.

⚠️ The new CanonicalGrounding production-path invariant needs to be reconciled with the remaining production-side direct construction.

briansrls · 2026-05-24T04:10:23Z

 type ConstraintGraph {
  root: Node
 }
+// Sole production path: solve_constraints is the ONLY permitted non-test construction of


BLOCKING: P2 single-authority is not actually enforced: the new “ONLY permitted non-test construction” receipt is false because src/v4/std/coercion.dag still constructs CanonicalGrounding directly in production code.

briansrls · 2026-05-24T18:13:48Z

Review metadata

Provider / model: openai-pro / gpt-5-5-pro
Commit: 16ccbfbb · Trigger: manual
Comparison: main @ aea44815 ... session/deep-bear-704 @ 16ccbfbb
Conversation: View conversation

1. Story of the diff

This PR tightens the v4 compiler pipeline around inferred facts and scaffold tracking. The load-bearing production change is that InferredFacts no longer carries independently constructible resolved_type, inhabits, and canonical fields; it now carries a single CanonicalGrounding plus descent evidence, with 04_infer.dag adding constructor/admission helpers and 05_eval.dag revalidating grounding before evaluation. Around that, the PR centralizes many test-claim fact fixtures through src/v4/test/claim/lens_common/infer_fixture.dag, updates claim files to the new fact shape, and retags several known scaffold surfaces with bounded yellow gates: lens application, lexer/parser grammar duplication, local surface-sugar classification, eval cache hashes, eval-local Outcome/control/edge helpers, and translate source anchoring. The intent is directionally right: reduce parallel fact fields, make eval fail closed on incoherent grounding, and make remaining bridges explicit rather than silent.

2. Invariant categories

LAYER MODEL (substrate vs implementation).

Compliant — the production substrate-facing InferredFacts shape now uses one CanonicalGrounding authority at src/v4/compiler/04_infer.dag:64 and eval consumes the public accessors at src/v4/compiler/05_eval.dag:14-16, rather than reading the old parallel fields.

INVARIANTS.md + modeling-discipline.md.

Finding — src/v4/test/claim/lens_common/infer_fixture.dag:46 / src/v4/test/claim/lens_common/infer_fixture.dag:50, fail-closed + illegal states unrepresentable. The new production admission gate requires CanonicalGrounding.witness.structural.property == constraint_property_being_canonical and closedness.property == constraint_property_candidate_set_closedness at src/v4/compiler/04_infer.dag:158-159, but the shared “coherent” fixture helper still fills those properties from arbitrary caller-supplied symbols at infer_fixture.dag:46 and infer_fixture.dag:50. Several updated positive fixtures continue passing local symbols, for example src/v4/test/claim/lens_affected_set/node_precision.dag:53-54, so claim_inferred_facts(...) can manufacture facts that production canonical_grounding_admits_infer_facts would reject. That preserves the illegal test state this PR is trying to dissolve. The coherent helper should hard-code the canonical constraint-property symbols, and any intentionally incoherent case should use a separately named negative/staging helper.

CODING.md.

Compliant — the production refactor stays in data + free-function form: inferred_facts_resolved_type, inferred_facts_algebra_ref, and inferred_facts_descent are explicit accessors at src/v4/compiler/04_infer.dag:69, src/v4/compiler/04_infer.dag:74, and src/v4/compiler/04_infer.dag:84, and the constructor returns Outcome<InferredFacts> at src/v4/compiler/04_infer.dag:164-188.

TESTING.md.

Finding — src/v4/test/claim/lens_common/infer_fixture.dag:46 / src/v4/test/claim/lens_affected_set/node_precision.dag:53-54, behavior-driven regression shape. The test updates centralize fixture construction, which is good, but the central helper currently lets positive claims build non-admissible canonical groundings while the production path now rejects those exact shapes. That means the fixture layer is no longer a trustworthy behavior-driven stand-in for production inference. Add a regression claim that intentionally fails on incoherent canonical-property symbols, and make the normal helper produce only admissible facts.

LOCKED DESIGN DECISIONS.

Compliant — the translate source-anchor gap is explicitly kept as a scaffold, not promoted into emit authority: src/v4/compiler/06_translate.dag:16 says TargetModel.authority_source_text remains a self-host anchor only, and src/v4/compiler/06_translate.dag:467 forbids substituting it for validated emitted-node serialization.

TRACKED vs UNTRACKED DEBT.

Compliant — the new bridge markers generally include documentation, bounds, and dissolution triggers. Examples: the compile lens stub binds to T-23 and names its deletion trigger at src/v4/compiler/00_compile.dag:117, and the eval cache hash placeholder binds to T-22/IRT-4 with an explicit content-hash dissolution trigger at src/v4/compiler/05_eval.dag:109.

2.5. Top-down PM intent review

Compliant — the PR preserves the PM-level direction: v4 compiler facts move toward a single canonical grounding authority, remaining hand-authored/scaffolded surfaces are tagged with named dissolution gates, and translate/emit is not diluted into string/template authority. The blocking issue above is an implementation/test-fixture coherence problem, not a semantic change to the top-level plan.

3. Verdict

REQUEST_CHANGES

The production direction is sound, but the shared test fixture now constructs “canonical” InferredFacts that the new production admission gate rejects. Fix the fixture split so positive claims are admissible and negative incoherence is explicit, then this should be in good shape.

…ority) Fixtures and the R2-7 shape ratchet assumed PR #3587's collapsed {grounding, descent} type had landed in 04_infer.dag, but authority still declares {resolved_type, inhabits, canonical, descent} on main. Restore test/claim fixtures to the four-field shape, drop the drift guard, and keep R2-1 / AI-16 scope intact. Co-authored-by: Cursor <cursoragent@cursor.com>

…n, affected_set markers) (#3589) * WIP: Lens, CI, and integration integrity (R2-1 P9 llvm cost owner gap, AI-16/ * WIP: Lens, CI, and integration integrity (R2-1 P9 llvm cost owner gap, AI-16/ * WIP: Lens, CI, and integration integrity (R2-1 P9 llvm cost owner gap, AI-16/ * WIP: Lens, CI, and integration integrity (R2-1 P9 llvm cost owner gap, AI-16/ * WIP: Lens, CI, and integration integrity (R2-1 P9 llvm cost owner gap, AI-16/ * WIP: Lens, CI, and integration integrity (R2-1 P9 llvm cost owner gap, AI-16/ * fix: revert premature InferredFacts fixture migration (P2 single-authority) Fixtures and the R2-7 shape ratchet assumed PR #3587's collapsed {grounding, descent} type had landed in 04_infer.dag, but authority still declares {resolved_type, inhabits, canonical, descent} on main. Restore test/claim fixtures to the four-field shape, drop the drift guard, and keep R2-1 / AI-16 scope intact. Co-authored-by: Cursor <cursoragent@cursor.com> * chore: re-trigger CI after SG-0 PR-body discipline fix PR body now declares SG-0 hand-path delta +1 and pairing (b) for the P9 census receipt; prior ci job failed SG-0 before bootstrap steps ran. Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Lens, CI, and integration integrity (R2-1 P9 llvm cost owner gap, AI-16/ * fix: remove body prose from ci.dag fail-closed selector (CODING.md) openai-pro REQUEST_CHANGES: the four-line narrative before ci_select_from_rerun_nodes violated .dag no-prose discipline. Contract is already encoded in the function body plus v4_workflow_ci_runner smoke assertions and INVARIANTS §P5(b) receipt. Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Lens, CI, and integration integrity (R2-1 P9 llvm cost owner gap, AI-16/ * WIP: PR #3589 openai-pro RC fix (Lens/CI integrity, branch session/tidy-otter * fix: tag CI gate run policy coproduct * fix: migrate pipeline rejection CiGate fixtures --------- Co-authored-by: Cursor <cursoragent@cursor.com>

briansrls · 2026-05-25T00:05:07Z

Superseded by consolidated PR #3636 (consolidated/correctness-sweep). All changes from this branch are included; eval conflicts resolved in favour of T-22 Wave 2 HEAD (#3611).

Consolidates session/deep-bear-704 into consolidated/correctness-sweep. AI-5 InferredFacts illegal-state dissolution: - src/v4/compiler/04_infer.dag: InferredFacts restructured from {resolved_type, inhabits, descent, canonical} to {grounding: CanonicalGrounding, descent}; new gate fn canonical_grounding_admits_infer_facts; new accessors inferred_facts_resolved_type/algebra_ref/canonical_witness/descent; inferred_facts_from_grounding returns Outcome<InferredFacts> - src/v4/std/constraints.dag: CanonicalGrounding as sole production path AI-7 TestClaimCacheKey restructuring: - src/v4/compiler/05_eval.dag: TestClaimCacheKey<T> carries hash_staging: TestClaimCacheKeyStagingPlaceholder; import non_empty from v4.std.algebra AI-13 SurfaceSugarKind scaffold marker upgrades: - src/v4/compiler/03_normalize.dag, 00_compile.dag: updated scaffold markers AI-14 LexPattern/GrammarExpr parallel ASTs: - src/v4/compiler/01_tokenize.dag: scaffold comments updated to HEAD standard AI-15/R1-2 translate source gap, R1-3 constraints singleton-MVP, R1-4 Lens apply_lens stub, R1-5 eval loop: - src/v4/compiler/06_translate.dag, 02_parse.dag: updates - src/v4/test/claim/lens_common/infer_fixture.dag: removes algebra_witness_symbol and canonical_evidence_symbol params; adds claim_canonical_grounding, claim_inferred_facts_witness, claim_inferred_facts_from_nodes Conflict resolution: eval conflicts resolved in favour of HEAD (T-22 Wave 2 #3611 is more complete); 01_tokenize.dag comment conflicts took HEAD's dissolution markers; 04_infer.dag took branch's explanatory comment for InferredFacts shape. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

…iler pipeline correctness (#3636) * squash-merge PR #3557: Generative TestClaim categories lane Consolidates session/quiet-moth-603 into consolidated/correctness-sweep. Changes from PR #3557: - src/v4/lens/testgen.dag: add AlgebraLawCase type + algebra_law_manual_claim_case helper, testgen_emit_algebra_law_claim, add algebra_law_subject_for_manual_anchor (subset not already on main), add Generator.kind field - src/v4/test/claim/generated/algebra_law_conformance.dag: new generated corpus of Nat algebra law TestClaims (law_nat_add_left_identity, add_associativity, mul_annihilator) - src/v4/test/claim/generated/testgen_category_wishlist.dag: new dispatch ledger (3 pending + 4 dispatched non-tautological generator categories; TestgenOracleBasis type, dissolve-on-arrival) - docs/design-testgen-generative-categories.md: design doc for non-tautological generator categories - src/v4/std/verification.dag: add test_claim_coproduct_variant function - src/v3/compiler/tests/integration/v4_lens_testgen_dag_smoke_test.rs: expanded smoke receipts - src/v3/compiler/tests/integration/v4_test_bootstrap_infra_closeout_test.rs: new assertions for wishlist + algebra law case helpers - src/v4/TASKS.md: T-19 follow-up task entries - scripts/check_t19_testgen_activation.py: testgen activation check script - INVARIANTS.md: updated smoke-test receipt row for testgen - scripts/v4-mvp1-e2e-gate.sh: cargo run exit-status fix Conflict resolution: DiagnosticExhaustiveness variant shape conflict resolved in favour of HEAD (omitted_variant: TestClaimCoproductVariant from T-19 #3561). Branch's DiagnosticExhaustivenessSubject type removed; coproduct_exhaustiveness.dag and testgen_category_wishlist.dag updated to use HEAD's omitted_variant shape. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * squash-merge PR #3587: compiler pipeline correctness sweep Consolidates session/deep-bear-704 into consolidated/correctness-sweep. AI-5 InferredFacts illegal-state dissolution: - src/v4/compiler/04_infer.dag: InferredFacts restructured from {resolved_type, inhabits, descent, canonical} to {grounding: CanonicalGrounding, descent}; new gate fn canonical_grounding_admits_infer_facts; new accessors inferred_facts_resolved_type/algebra_ref/canonical_witness/descent; inferred_facts_from_grounding returns Outcome<InferredFacts> - src/v4/std/constraints.dag: CanonicalGrounding as sole production path AI-7 TestClaimCacheKey restructuring: - src/v4/compiler/05_eval.dag: TestClaimCacheKey<T> carries hash_staging: TestClaimCacheKeyStagingPlaceholder; import non_empty from v4.std.algebra AI-13 SurfaceSugarKind scaffold marker upgrades: - src/v4/compiler/03_normalize.dag, 00_compile.dag: updated scaffold markers AI-14 LexPattern/GrammarExpr parallel ASTs: - src/v4/compiler/01_tokenize.dag: scaffold comments updated to HEAD standard AI-15/R1-2 translate source gap, R1-3 constraints singleton-MVP, R1-4 Lens apply_lens stub, R1-5 eval loop: - src/v4/compiler/06_translate.dag, 02_parse.dag: updates - src/v4/test/claim/lens_common/infer_fixture.dag: removes algebra_witness_symbol and canonical_evidence_symbol params; adds claim_canonical_grounding, claim_inferred_facts_witness, claim_inferred_facts_from_nodes Conflict resolution: eval conflicts resolved in favour of HEAD (T-22 Wave 2 #3611 is more complete); 01_tokenize.dag comment conflicts took HEAD's dissolution markers; 04_infer.dag took branch's explanatory comment for InferredFacts shape. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: remove unused non_empty import from 05_eval.dag * fix: restore coproduct_exhaustiveness_missing_variant check (revert stale DiagnosticExhaustivenessSubject strings) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: drop stale t19_ prefix from wishlist check needle (symbol renamed in HEAD design) * fix: update coproduct_exhaustiveness check needles to omitted_variant design * fix: update 05_eval.dag to new InferredFacts API; drop t19_ prefix from algebra_law_anchor_mismatch - Replace dissolved inferred_facts_resolved_type_matches/inhabitance_well_grounded/canonical_well_grounded calls in eval_runtime_value_acceptance_witness with inferred_facts_resolved_type equality check and canonical_grounding_admits_infer_facts (new 04_infer.dag API) - Declare algebra_law_anchor_mismatch symbol and replace undefined t19_algebra_law_anchor_mismatch reference in testgen.dag Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: enforce InferredFactsEntry node==grounding.node invariant; add T-19 AlgebraLaw receipt to TASKS.md Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * WIP: PM May 22 * fix: add facts.grounding.node == node guard in inferred_facts_for_eval Eval consumption point must verify the map hit is for the correct node. `canonical_grounding_admits_infer_facts` only checks grounding internal consistency; it does not guard against a map returning facts whose grounding.node diverges from the lookup key. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: add P5 receipt for T-19 AlgebraLaw expansion in INVARIANTS.md SG-0 table Per INVARIANTS.md P5 Mechanism (b): expanded hand-Rust under src/v3/ requires a checkable receipt. The AlgebraLaw Nat corpus assertions in v4_test_bootstrap_infra_closeout_test.rs land in this PR; the deferral lane is TASKS.md §T-19 (algebra_law_conformance.dag corpus) + ROADMAP.md Nine lanes row T-PB-B. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: make facts_map_from_entries fail-closed via Outcome Previously, facts_map_from_entries silently dropped InferredFactsEntry items where entry.node != entry.facts.grounding.node. This compensated for the free constructibility of InferredFactsEntry (P2 boundary gap) without surfacing the violation diagnostically. Now: the fold rejects with infer_canonical_grounding_incoherent_diagnostic on any invariant-violating entry, and infer() threads the Outcome. admitted_inferred_facts_entry remains the primary enforcement point; facts_map_from_entries is the secondary fail-closed gate that makes invariant violations visible as Rejected rather than silent drops. Addresses codex RC on bc16ae7: InferredFactsEntry remained freely constructible and map construction compensated with a nondiagnostic filter. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: replace bare [] with typed helper in facts_map_from_entries The v4 compiler cannot infer the element type of [] when it appears as a record field value inside Accepted { value: [], ... }. Introduce empty_inferred_facts_entry_list() -> List<InferredFactsEntry> (matching the existing empty_edge_list / empty_canonical_symbol_set pattern) and use it as the fold init. Fixes Lens-CI semantic compile failure on d1299ea. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: add P5 receipt for PR #3636 anchor-strip + AlgebraLaw subject helper guard in INVARIANTS.md - Updates Generator field name from t19_anchor to anchor in existing ratchet description - Appends PR #3636 same-file receipt: renamed tests, new kind/anchor assertions, assert_six_algebra_law_subject_paths_in_testgen, forbidden nat_algebra_law_subject_symbol_* checks, and generalized generator_field_ty / between helpers Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: update stale T19-prefix type names in INVARIANTS.md testgen row - Rename test fn refs: t19_anchor_field_is_manual_key → anchor_field_is_claim_anchor_key, concept_projection_matches_claim_t19_anchor → _claim_anchor, generator_carries_claim_kind_and_classification → _classification_and_anchor - Update description: T19ManualAnchorKey → ManualAnchorKey, Generator.t19_anchor → Generator.anchor, T19ClaimAnchorKey → ClaimAnchorKey, TestClaim.t19_anchor → TestClaim.anchor, T19ManualClaimAnchor → ManualClaimAnchor, T19ManualAnchorAbsent → ManualAnchorAbsent, T19GeneratedClaimAnchor → GeneratedClaimAnchor - Update inline carrier description: Generator.t19_anchor: T19ClaimAnchorKey → Generator.anchor: ClaimAnchorKey Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * WIP: PM May 22 * fix: strip stale T19 names from closeout + testgen INVARIANTS rows (cursor RC #17566) Line 374 (v4_test_bootstrap_infra_closeout_test.rs): update 7 test fn names (t19_testgen_concept_surface_stays_closed_and_classified → testgen_..., etc.) and 6 type names (T19ManualAnchorKey → ManualAnchorKey, TestClaim.t19_anchor → .anchor, t19_manual_claim_anchor → manual_claim_anchor, T19GeneratedCoproductExhaustiveness, T19GeneratedClaimAnchor, T19ClaimAnchorKey). Line 383 (v4_lens_testgen_dag_smoke_test.rs): simplify PR #3636 receipt section to remove all from→to rename references; receipt now describes current ratchet state only. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(04_infer): adapt merge-conflict resolution to our InferredFacts struct layout The merge of main (PR #3646) erroneously took code that references `facts.inhabits.algebra` and `inferred_facts_canonical_grounding_coherent`, both of which belong to main's refactored InferredFacts struct. Our branch carries `InferredFacts { grounding, descent }`, so the guards must use `inferred_facts_algebra_ref(facts: facts).algebra` and `canonical_grounding_admits_infer_facts(grounding: facts.grounding)` instead. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * WIP: PM May 22 * WIP: PM May 22 * WIP: PM May 22 * WIP: PM May 22 --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

briansrls added 2 commits May 23, 2026 22:58

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

013f5e8

…erredFac

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

230eef2

…erredFac

briansrls marked this pull request as ready for review May 23, 2026 23:03

briansrls added 2 commits May 23, 2026 23:08

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

c586c5f

…erredFac

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

ea0ac43

…erredFac

briansrls commented May 23, 2026

View reviewed changes

Comment thread src/v4/compiler/04_infer.dag Outdated

facts: InferredFacts,

) -> Outcome<CanonicalGrounding> {

if inferred_facts_canonical_grounding_coherent(facts: facts) {

if well_formed(n: facts.grounding.node) {

This comment was marked as resolved.

Sign in to view

briansrls added 2 commits May 23, 2026 23:16

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

bb3c2d2

…erredFac

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

b9da4d7

…erredFac

briansrls mentioned this pull request May 23, 2026

Lens, CI, and integration integrity (P9 owner, InferredFacts migration, affected_set markers) #3589

Merged

6 of 8 tasks

briansrls added 2 commits May 23, 2026 23:29

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

4d4831a

…erredFac

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

d3bb794

…erredFac

briansrls and others added 2 commits May 23, 2026 23:39

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

7e14044

…erredFac

briansrls and others added 6 commits May 23, 2026 23:45

chore(compiler): upgrade remaining CP-1b bare scaffold marks in token…

ac4307d

…ize/parse Gate LexCharClass, LexRule, LexRules, and Grammar with feature:CP-1b-bidirectional- grammar-carrier; remove duplicate Grammar mark on 02_parse.dag. Co-authored-by: Cursor <cursoragent@cursor.com>

merge: resolve PR #3587 conflict with origin/main

28f10d6

Keep main's apply_compile_lens/CompileLens rename; preserve T-23 gated migration stub comment from session branch. Co-authored-by: Cursor <cursoragent@cursor.com>

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

98b7be5

…erredFac

fix(compiler): import non_empty from v4.std.algebra in 05_eval

b29268d

Preempt #3598 removal of non_empty from collection.dag; canonical owner is algebra.dag. Co-authored-by: Cursor <cursoragent@cursor.com>

WIP: Compiler pipeline correctness and scaffold dissolution (AI-5 Inf…

90322a5

…erredFac

briansrls commented May 24, 2026

View reviewed changes

merge: resolve PR #3587 conflict with origin/main

16ccbfb

Keep T-23 gated migration stub comment on apply_compile_lens; integrate main through aea4481 (T-31 de-prose mop-up, T-4.6 sql, T-23 lens app). Co-authored-by: Cursor <cursoragent@cursor.com>

briansrls commented May 24, 2026

View reviewed changes

briansrls mentioned this pull request May 25, 2026

Consolidate PRs #3557 + #3587: generative TestClaim categories + compiler pipeline correctness #3636

Merged

5 tasks

briansrls closed this May 25, 2026

briansrls deleted the session/deep-bear-704 branch June 1, 2026 18:42

Navigation Menu

Repository navigation

briansrls commented May 23, 2026

Uh oh!

briansrls left a comment

Uh oh!

This comment was marked as resolved.

Uh oh!

briansrls commented May 23, 2026

Uh oh!

briansrls left a comment

Uh oh!

briansrls commented May 24, 2026

Uh oh!

briansrls commented May 24, 2026

Uh oh!

briansrls commented May 24, 2026

Uh oh!

briansrls left a comment

Uh oh!

briansrls commented May 24, 2026

Uh oh!

briansrls left a comment

Uh oh!

briansrls May 24, 2026

Uh oh!

briansrls commented May 24, 2026

Uh oh!

briansrls commented May 25, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

briansrls commented May 23, 2026

Worker attestation

Summary

Test plan

Uh oh!

briansrls left a comment

Choose a reason for hiding this comment

Uh oh!

This comment was marked as resolved.

Uh oh!

briansrls commented May 23, 2026

Uh oh!

briansrls left a comment

Choose a reason for hiding this comment

Uh oh!

briansrls commented May 24, 2026

Uh oh!

briansrls commented May 24, 2026

Uh oh!

briansrls commented May 24, 2026

Uh oh!

briansrls left a comment

Choose a reason for hiding this comment

Uh oh!

briansrls commented May 24, 2026

Uh oh!

briansrls left a comment

Choose a reason for hiding this comment

Uh oh!

briansrls May 24, 2026

Choose a reason for hiding this comment

Uh oh!

briansrls commented May 24, 2026

1. Story of the diff

2. Invariant categories

2.5. Top-down PM intent review

3. Verdict

Uh oh!

briansrls commented May 25, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant