get main CI green by gunbai-bot[bot] · Pull Request #6352 · gunb-ai/gunbc

gunbai-bot · 2026-07-07T03:40:49Z

Auto-opened by session-dashboard for session crisp-bear-170.
Pushing to session/crisp-bear-170 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.

…ale-binary no-count refusal) The ci_regen_ratchet required job failed because ci_release_build_line built only claim_executor+gunbc, while regen_stage0 was left to ensure_regen_stage0_built's build-if-absent path. On the self-hosted runners target/release/ persists across jobs, so a stale regen_stage0 (pre-dating the regen_divergence_count contract) was reused, emitted no count, and tripped the fail-closed no-count refusal. cargo build is source-change-aware, so listing --bin regen_stage0 in the shared release step guarantees a fresh binary. Also surface the captured shell stdout in the gate's ExitFailure reason so the located refusal (which arm + observed count) is visible in the CI log rather than swallowed. Regenerated ci.yml + falsifier.yml. Receipt: ci_regen_ratchet job 85537100872, run 28841755357 @ 7927413. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

gunbai-bot · 2026-07-07T06:34:25Z

Status (not abandoned; intentionally parked as draft):

WHAT IS DONE + VERIFIED: the get-main-green two-job split (required green ci job + non-required honest-red ci_regen job) + the §5 counted-divergence ratchet (ci_regen_ratchet). The ratchet is GREEN by execution in CI (validated on 95930fe; stale-binary no-count bug fixed by building regen_stage0 in the shared release step). Compiles clean whole-corpus.

WHY STILL DRAFT (not ready): the required green ci job cannot pass yet because batch-1 dag_compile_clean_gate times out at the 10-min floor budget in CI. This is a PRE-EXISTING main-wide condition (recent main runs time out at the identical batch-1), NOT introduced by this PR. It is the compile-clean serial wall (roadmap §0 item 2-compile-clean-serial). Affected-set scoping (lever a) that would shrink batch-1 is transitively blocked on completing the whole-tree resolve-grounding substrate (lively-raven SymbolIndex lane, gated on #6348) — until that lands, every non-docs PR runs the whole corpus and times out. This PR + bright-heron #6335 (selection) are ready prep behind that substrate.

UN-PARK TRIGGERS: (1) when #6335 merges, rebase origin/main into this branch + resolve the ci_workflow.dag overlap; (2) when the resolve-grounding substrate lands, the green job fits budget and this flips ready. Coordinated with clever-koi (routing) + bright-heron (#6335). Closing would discard coordinated prep behind an active critical-path lane.

…sses The operator's 4-rule business logic generalized on one axis: the decision is a function of WHEN the redundancy is knowable and WHETHER what was knowable was prepared for. Errors fire only on knowable-but-unprepared; genuine emergence and declared triviality are typed acceptances, never silence. std.materialization_ladder (dag/std): - Frame/FrameKind: SharedStateFrame | IsolatedChildrenFrame | ReplayedFrame{attempts} | UnboundedSiblingsFrame — the nested-scope axis; a 'run' at any layer is a frame, never a different kind of thing. - DemandNature: Pure | IdempotentEffect | WorldRead{envelope_declared} | FreshEffect — the D3 effect gate; TTL = unmodeled dependency XOR declared staleness, never an eviction knob. - CacheProvider: keying (ContentKeyed|ExistenceKeyed) + EvictionPolicy (ScopeExit|SpacePacked) REQUIRED by construction — rule 3 unwritable-bad. - LadderVerdict (total): AuthoredDuplication | Discharged | RefusedNoProvider | RefusedScopeTooNarrow | RefusedExistenceKeyed (the build-if-absent #6352 wall) | RefusedUnmodeledWorldRead | ExemptFreshEffect | AcceptedBelowCostFloor | AcceptedSingleRecompute. - Declared-emergent frames obligate UP FRONT: retry frames derive checkpointing, unbounded-sibling frames (server loops, CI-runs-over-time) derive persistent caches — prepare-before-demand. 12/12 witnesses green by execution; each test is one cell of the table, fixtures mirror the live worked example (workflow/jobs/steps, retry, fleet). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * Analysis spine: spine_receipt composes dependency_view + materialize (increment 1) The Rc-safe half of run ≜ realize ∘ materialize ∘ dependency_view: compose the two catamorphisms that need no interpreter thread-safety. - src/v2/std/spine.dag: level-profile fold over the DependencyView DAG → critical_path_depth (longest dependency chain / the reduce spine) + independence_width (max nodes at one level / what parallelizes to hw width). spine_receipt emits the §9 numbers by COMPUTATION, not assertion, joining materialize's content-hash Share/dedup counts. - Discriminating RED (§9): single chain A→B→C → depth 3 / width 1 (serial); root over N independent leaves → depth 2 / width N. - Witnesses 9/9 (spine) green by execution; materialize 6/6 unchanged. The RUNNER (wall-clock parallel execution, §8 steps 3-4) stays increment 2+, gated on interpreter Rc→Arc (v1 interp is !Send). This increment measures the achievable width/critical-path floor the runner targets; it does not fabricate wall-clock N×. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Unbundle recompute-trace interpreter extension from the analysis-spine PR CI root cause: the recompute-trace extension edited v1_interpreter.rs, which hosts the atom_identity_hash/hash_combine fnv1a64 intrinsics that dag_collect_fingerprint_witness executes. That correctly invalidated the re-verify 'assumed-green node-frontier' SKIP of that witness (cone-touch, fail-closed) — but the witness is a pre-existing 806s (13min) fnv1a64 whole-corpus walk, so un-skipping it blew the 30-min floor budget. Main stays green only because its interpreter cone is untouched (witness stays skipped). The interpreter extension is a separable, env-gated (GUNBC_RECOMPUTE_TRACE=1), READ-mode diagnostic — not a dependency of the spine/materialize analysis substrate (pure .dag, content_hash-based). Reverting v1_interpreter.rs to main restores byte-identical interpreter content → the fingerprint witness returns to assumed-green SKIP → floor budget restored. The recompute-trace extension lands as its own follow-up PR where the fingerprint-witness re-run is expected. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Materialization ladder: the state x decision law, as executable witnesses The operator's 4-rule business logic generalized on one axis: the decision is a function of WHEN the redundancy is knowable and WHETHER what was knowable was prepared for. Errors fire only on knowable-but-unprepared; genuine emergence and declared triviality are typed acceptances, never silence. std.materialization_ladder (dag/std): - Frame/FrameKind: SharedStateFrame | IsolatedChildrenFrame | ReplayedFrame{attempts} | UnboundedSiblingsFrame — the nested-scope axis; a 'run' at any layer is a frame, never a different kind of thing. - DemandNature: Pure | IdempotentEffect | WorldRead{envelope_declared} | FreshEffect — the D3 effect gate; TTL = unmodeled dependency XOR declared staleness, never an eviction knob. - CacheProvider: keying (ContentKeyed|ExistenceKeyed) + EvictionPolicy (ScopeExit|SpacePacked) REQUIRED by construction — rule 3 unwritable-bad. - LadderVerdict (total): AuthoredDuplication | Discharged | RefusedNoProvider | RefusedScopeTooNarrow | RefusedExistenceKeyed (the build-if-absent #6352 wall) | RefusedUnmodeledWorldRead | ExemptFreshEffect | AcceptedBelowCostFloor | AcceptedSingleRecompute. - Declared-emergent frames obligate UP FRONT: retry frames derive checkpointing, unbounded-sibling frames (server loops, CI-runs-over-time) derive persistent caches — prepare-before-demand. 12/12 witnesses green by execution; each test is one cell of the table, fixtures mirror the live worked example (workflow/jobs/steps, retry, fleet). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Docs crystallization: derived materialization placement into roadmap ④ + the state x decision table - roadmap_authority ④ (ROADMAP.md regenerated via main_wet, drift-consistent): the nested-frame law, plurality cells, declared-emergent prepare-up-front, keying/staleness walls, the caching-completion un-shelve, the ①-vs-§2 cross-run-caching contradiction subsumed into per-node derived verdicts, and the v1.compiler.ownership §3-convergence row. - duplicate-work design doc: the state x decision table as the qualifier's final verdict logic + the ownership consolidation subsection. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Consolidation plan: realization + materialization + ownership are one law Operator synthesis (2026-07-09): Materialization = the verdict vocabulary; ownership = the verdict computation at the eval frame; realization = provider selection discharging the verdict at each frame's carrier. Share's handler is layer-aware — the cheapest cache is a REFERENCE (same program + ref-capable carrier + readonly => point-at-it, no copy, no destroyed memory), demoting through HAMT structural share / process memo / artifact / CAS, and demotion must be priced, never silent (#6249 clone-fallback is the receipt). Census of every hand-rolled instance found in-tree (16 rows): each named with its frame, ladder cell, action, and dissolution trigger — ownership (consolidated), Rc/HAMT (reference tier), clone-fallback (demotion wall), ParseTable + cached_stage (C5 grounding), M1 memo / intern / assumed-green / sccache / cargo-cache (C2 provider rows), PROCESS_RESOLVE_STORE (C3 = the 9GB scope fix), build-if-absent (landed as the keying wall), resolve-cache #5789 (rule-4 requirements), recompute-trace (state-4 finding source). Peers kept distinct: affected-set, Independence/Placement, mutable state. Sequence C1-C5 with C1 = this PR (ladder + live CI gate). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Ownership refactored onto materialization: the value tier, tested as providers First consolidation increment (operator-directed): v1.compiler.ownership's decision IS provider selection at the value grain, now expressed in the ladder and proven by mirror witnesses. - ProviderTier: Reference | Copy | Memo{keying} | Artifact{keying} | Cas{keying} — keying folded INTO store tiers, so an existence-keyed reference is unwritable BY SHAPE (§5 construction); frame obligations at isolation LCAs are dischargeable only by store tiers (a reference cannot cross an isolation boundary — witnessed). - Value tier: AccessMode (Read|Consume|Carry|Project), ValueUsage, value_materialization -> ValueDead | ValueMoved | ValueSharedByReference | ValueDemotedToCopy (PRICED, requires a declared CopyTier provider — the #6249 silent clone-fallback made refusable) | ValueRefusedAffine | ValueRefusedNoCarrierProvider. - Faithful to v1's THREE distinct plurality readings, kept separate: take_count = semantic_consumer_count (Consumed only, affine axis); value_access_plurality = binding_fan_out (Carry excluded, reference axis); borrow_count = whole_value_borrow_count (Read+Carry not Project, movability). - Fold-accumulator case mirrored at conservative pre-FoldAccUnwrap semantics: carries excluded from plurality but blocking the move -> demoted-to-copy. 23/23 witnesses green (14 frame cells + 9 value cells). Corpus-level equivalence against live v1 ownership folds = the C4 receipt. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * WIP: Duplicate Computation/Materialization * G: exhaustive cache/share census (52 mechanisms, 6 groups) + C0 catalog-grounding of provider rows + forward-wiring walls The sweep found the tree already models caches: extdeps/cache cache_catalog (CacheInterfaceCatalogFacts, 9 rows) + extdeps/realization carrier facts + the warm==cold purity oracle. C0 grounds CacheProvider on that catalog (provider_from_catalog projection; CacheProvider.id -> CacheInterfaceId) so the ladder and the catalog never restate each other's half. Three forward walls (demands-from-DependencyView, hand-cache shape lens, live-provider-or-red) make new caches born as provider rows only. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * C0: provider rows derive from the cited cache catalog (extdeps/cache/materialization.dag projection) Operator ruling 2026-07-09: extdeps stays the ground-truth/interface layer; every USE of a cache goes through materialization so memo never gets re-invented. provider_from_catalog is the only door from a catalog row into the ladder: keying/tier/eviction derived from cited facts (mechanism->class: InProcess=>ScopeExit; Ttl/Lru/SizeBounded=>SpacePacked; Never/Manual outside a process => typed ProjectionRefused, counted by the new enrolled witness; HandAuthored/prefix-fallback keys => ExistenceKeyed until key-completeness upgrades them). CI sccache row now derived, hand-typed tier facts deleted, 6/6 + 23/23 witnesses green. 10b re-scoped: v1-internal caches = inventory only (retire with seed); ParseTable/cached_stage promoted to next. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Fail-closed hardening: RefusedNatureConflict arm, weakest-claim nature default, UsesStep silence counted, ladder cells enrolled in CI The complexity-lens false-clean lesson applied to this gate before anyone relies on it: (1) heterogeneous natures on one identity were first-wins — now a typed RefusedNatureConflict (24th ladder cell, RED witness); (2) an unrostered run-step claimed PureComputation — now IdempotentEffect, the weakest claim (verdict-identical, no purity overclaim; a misdeclared WorldRead can no longer be legally memoized by default); (3) UsesStep foreign actions produced silence — now a pinned denominator (==9; a new foreign action must consciously bump it); (4) the ladder's 24 cells enroll on the CI surface so the law's own witnesses gate every run. 24/24 + 7/7. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * M1 ratchet bump: declared resolve count 1 -> 3 (gate's first live catch — enrollment pays counted resolves) Run 29058798771: enrolling the ladder witness file moved resolves_total 1 -> 3 (each enrolled entry file pays one closure resolve against the shared index) and the receipt gate redded that exact run — the designed semantics, receipted in the note. Bump acknowledges the two witness-entry resolves as declared debt; M2 (one closure resolve per roots set) ratchets back toward 1. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Retire the M1/M2 labels from carriers (no-codename rule; collided with DESIGN's floor-memoization M1/M2) Step renamed to 'Floor resolve receipt gate (declared cold-resolve count)'; notes and the gate echo now say what the things are: the counted cold-resolve receipt, and the shared-resolve rewire (one closure resolve per source-roots set). Naming-retirement note left on the carrier. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * Run-step natures pinned as counted claims, not affordances Operator ruling 2026-07-10: hardcoded shell scripts get no affordances — run-steps are supposed to be modeled as typed intents and emitted (shell-emission-model slice 4 covers ci_workflow RunSteps). Until that slice lands, every nature in job_run_demands is a claim (rostered FreshEffect / weakest-default IdempotentEffect), never derived from a modeled effect row. Count the whole claimed-nature surface (16 by execution) and pin it in the enrolled witness: a new raw-script step must consciously bump the pin, and slice-4 migration becomes a countable ratchet ending at zero, where the pin becomes a wall. Rejected alternative recorded in the carrier note: a per-step declared-natures roster (a second parallel ledger over scripts the shell-emission plan already governs). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Regen DESIGN.md after main merge (duplicate-work row + #6373 body-lowering row coexist) Conflict was both sides editing the open-threads list in design_document.dag: ours added the duplicate-work thread, main's #6373 rewrote body-lowering to Stages 1-3 LANDED. Resolution keeps both rows in the .dag authority; DESIGN.md regenerated from it, never hand-edited. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ealization carrier, provider rows via provider_from_catalog (design: docs/plans/duplicate-work-graph-lens-design.md section 10b; the .dag-substrate demonstration lane) (#6422) * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Ground v2 ParseTable + cached_stage on materialization: inhabit the Real * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * Analysis spine: spine_receipt composes dependency_view + materialize (increment 1) The Rc-safe half of run ≜ realize ∘ materialize ∘ dependency_view: compose the two catamorphisms that need no interpreter thread-safety. - src/v2/std/spine.dag: level-profile fold over the DependencyView DAG → critical_path_depth (longest dependency chain / the reduce spine) + independence_width (max nodes at one level / what parallelizes to hw width). spine_receipt emits the §9 numbers by COMPUTATION, not assertion, joining materialize's content-hash Share/dedup counts. - Discriminating RED (§9): single chain A→B→C → depth 3 / width 1 (serial); root over N independent leaves → depth 2 / width N. - Witnesses 9/9 (spine) green by execution; materialize 6/6 unchanged. The RUNNER (wall-clock parallel execution, §8 steps 3-4) stays increment 2+, gated on interpreter Rc→Arc (v1 interp is !Send). This increment measures the achievable width/critical-path floor the runner targets; it does not fabricate wall-clock N×. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Unbundle recompute-trace interpreter extension from the analysis-spine PR CI root cause: the recompute-trace extension edited v1_interpreter.rs, which hosts the atom_identity_hash/hash_combine fnv1a64 intrinsics that dag_collect_fingerprint_witness executes. That correctly invalidated the re-verify 'assumed-green node-frontier' SKIP of that witness (cone-touch, fail-closed) — but the witness is a pre-existing 806s (13min) fnv1a64 whole-corpus walk, so un-skipping it blew the 30-min floor budget. Main stays green only because its interpreter cone is untouched (witness stays skipped). The interpreter extension is a separable, env-gated (GUNBC_RECOMPUTE_TRACE=1), READ-mode diagnostic — not a dependency of the spine/materialize analysis substrate (pure .dag, content_hash-based). Reverting v1_interpreter.rs to main restores byte-identical interpreter content → the fingerprint witness returns to assumed-green SKIP → floor budget restored. The recompute-trace extension lands as its own follow-up PR where the fingerprint-witness re-run is expected. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Materialization ladder: the state x decision law, as executable witnesses The operator's 4-rule business logic generalized on one axis: the decision is a function of WHEN the redundancy is knowable and WHETHER what was knowable was prepared for. Errors fire only on knowable-but-unprepared; genuine emergence and declared triviality are typed acceptances, never silence. std.materialization_ladder (dag/std): - Frame/FrameKind: SharedStateFrame | IsolatedChildrenFrame | ReplayedFrame{attempts} | UnboundedSiblingsFrame — the nested-scope axis; a 'run' at any layer is a frame, never a different kind of thing. - DemandNature: Pure | IdempotentEffect | WorldRead{envelope_declared} | FreshEffect — the D3 effect gate; TTL = unmodeled dependency XOR declared staleness, never an eviction knob. - CacheProvider: keying (ContentKeyed|ExistenceKeyed) + EvictionPolicy (ScopeExit|SpacePacked) REQUIRED by construction — rule 3 unwritable-bad. - LadderVerdict (total): AuthoredDuplication | Discharged | RefusedNoProvider | RefusedScopeTooNarrow | RefusedExistenceKeyed (the build-if-absent #6352 wall) | RefusedUnmodeledWorldRead | ExemptFreshEffect | AcceptedBelowCostFloor | AcceptedSingleRecompute. - Declared-emergent frames obligate UP FRONT: retry frames derive checkpointing, unbounded-sibling frames (server loops, CI-runs-over-time) derive persistent caches — prepare-before-demand. 12/12 witnesses green by execution; each test is one cell of the table, fixtures mirror the live worked example (workflow/jobs/steps, retry, fleet). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Docs crystallization: derived materialization placement into roadmap ④ + the state x decision table - roadmap_authority ④ (ROADMAP.md regenerated via main_wet, drift-consistent): the nested-frame law, plurality cells, declared-emergent prepare-up-front, keying/staleness walls, the caching-completion un-shelve, the ①-vs-§2 cross-run-caching contradiction subsumed into per-node derived verdicts, and the v1.compiler.ownership §3-convergence row. - duplicate-work design doc: the state x decision table as the qualifier's final verdict logic + the ownership consolidation subsection. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Consolidation plan: realization + materialization + ownership are one law Operator synthesis (2026-07-09): Materialization = the verdict vocabulary; ownership = the verdict computation at the eval frame; realization = provider selection discharging the verdict at each frame's carrier. Share's handler is layer-aware — the cheapest cache is a REFERENCE (same program + ref-capable carrier + readonly => point-at-it, no copy, no destroyed memory), demoting through HAMT structural share / process memo / artifact / CAS, and demotion must be priced, never silent (#6249 clone-fallback is the receipt). Census of every hand-rolled instance found in-tree (16 rows): each named with its frame, ladder cell, action, and dissolution trigger — ownership (consolidated), Rc/HAMT (reference tier), clone-fallback (demotion wall), ParseTable + cached_stage (C5 grounding), M1 memo / intern / assumed-green / sccache / cargo-cache (C2 provider rows), PROCESS_RESOLVE_STORE (C3 = the 9GB scope fix), build-if-absent (landed as the keying wall), resolve-cache #5789 (rule-4 requirements), recompute-trace (state-4 finding source). Peers kept distinct: affected-set, Independence/Placement, mutable state. Sequence C1-C5 with C1 = this PR (ladder + live CI gate). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Ownership refactored onto materialization: the value tier, tested as providers First consolidation increment (operator-directed): v1.compiler.ownership's decision IS provider selection at the value grain, now expressed in the ladder and proven by mirror witnesses. - ProviderTier: Reference | Copy | Memo{keying} | Artifact{keying} | Cas{keying} — keying folded INTO store tiers, so an existence-keyed reference is unwritable BY SHAPE (§5 construction); frame obligations at isolation LCAs are dischargeable only by store tiers (a reference cannot cross an isolation boundary — witnessed). - Value tier: AccessMode (Read|Consume|Carry|Project), ValueUsage, value_materialization -> ValueDead | ValueMoved | ValueSharedByReference | ValueDemotedToCopy (PRICED, requires a declared CopyTier provider — the #6249 silent clone-fallback made refusable) | ValueRefusedAffine | ValueRefusedNoCarrierProvider. - Faithful to v1's THREE distinct plurality readings, kept separate: take_count = semantic_consumer_count (Consumed only, affine axis); value_access_plurality = binding_fan_out (Carry excluded, reference axis); borrow_count = whole_value_borrow_count (Read+Carry not Project, movability). - Fold-accumulator case mirrored at conservative pre-FoldAccUnwrap semantics: carries excluded from plurality but blocking the move -> demoted-to-copy. 23/23 witnesses green (14 frame cells + 9 value cells). Corpus-level equivalence against live v1 ownership folds = the C4 receipt. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * WIP: Duplicate Computation/Materialization * G: exhaustive cache/share census (52 mechanisms, 6 groups) + C0 catalog-grounding of provider rows + forward-wiring walls The sweep found the tree already models caches: extdeps/cache cache_catalog (CacheInterfaceCatalogFacts, 9 rows) + extdeps/realization carrier facts + the warm==cold purity oracle. C0 grounds CacheProvider on that catalog (provider_from_catalog projection; CacheProvider.id -> CacheInterfaceId) so the ladder and the catalog never restate each other's half. Three forward walls (demands-from-DependencyView, hand-cache shape lens, live-provider-or-red) make new caches born as provider rows only. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * C0: provider rows derive from the cited cache catalog (extdeps/cache/materialization.dag projection) Operator ruling 2026-07-09: extdeps stays the ground-truth/interface layer; every USE of a cache goes through materialization so memo never gets re-invented. provider_from_catalog is the only door from a catalog row into the ladder: keying/tier/eviction derived from cited facts (mechanism->class: InProcess=>ScopeExit; Ttl/Lru/SizeBounded=>SpacePacked; Never/Manual outside a process => typed ProjectionRefused, counted by the new enrolled witness; HandAuthored/prefix-fallback keys => ExistenceKeyed until key-completeness upgrades them). CI sccache row now derived, hand-typed tier facts deleted, 6/6 + 23/23 witnesses green. 10b re-scoped: v1-internal caches = inventory only (retire with seed); ParseTable/cached_stage promoted to next. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Ground v2 ParseTable + cached_stage on materialization: inhabit the Real * Fix compiler materialization witnesses: RealizedStep carrier, plural RED fixture. Use RealizedStep<Nano> (not unit), drop 02_parse import from witnesses in favor of parse_table_carrier_grounded_on_catalog, and model cross-run memo plurality with an isolation-boundary LCA matching the CI ladder pattern. Co-authored-by: Cursor <cursoragent@cursor.com> * Fix CI: move materialization_carriers to compiler layer (no std→extdeps import). Resolves layering_imports_gate and compile-clean failures from v2.std importing extdeps via materialization_carriers. CachedStageRealization lives in v2.compiler.materialization_carriers; staging.dag stays transport-only. Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Ground v2 ParseTable + cached_stage on materialization: inhabit the Real * WIP: Ground v2 ParseTable + cached_stage on materialization: inhabit the Real * Fix governed memo door: resolve_probe in cached_stage_governed, parse lookup match braces. Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Ground v2 ParseTable + cached_stage on materialization: inhabit the Real * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Fail-closed hardening: RefusedNatureConflict arm, weakest-claim nature default, UsesStep silence counted, ladder cells enrolled in CI The complexity-lens false-clean lesson applied to this gate before anyone relies on it: (1) heterogeneous natures on one identity were first-wins — now a typed RefusedNatureConflict (24th ladder cell, RED witness); (2) an unrostered run-step claimed PureComputation — now IdempotentEffect, the weakest claim (verdict-identical, no purity overclaim; a misdeclared WorldRead can no longer be legally memoized by default); (3) UsesStep foreign actions produced silence — now a pinned denominator (==9; a new foreign action must consciously bump it); (4) the ladder's 24 cells enroll on the CI surface so the law's own witnesses gate every run. 24/24 + 7/7. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Merge merry-moth-539 (main-integrated); harden governed memo door witnesses. Merge origin/session/merry-moth-539 for post-main semantics alignment. Export parse_table_memo_lookup_refuses_store / parse_table_memo_insert_refused door oracles in 02_parse; witnesses call them directly (reason atom, not full Diagnostic equality). Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Ground v2 ParseTable + cached_stage on materialization: inhabit the Real * Regenerate ci.yml for floor resolve receipt pin at 3. CI run 29059860791 measured resolves_total=3 after enrolling the two new v2 witness entry classes; sync the emitted gate with ci_floor_declared_resolve_count in ci_materialization.dag. Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Ground v2 ParseTable + cached_stage on materialization: inhabit the Real * fix: escape ComputationIdentity braces in design_document.dag Rebase conflict left unescaped {bound}/{cause} in a li() string, which broke dag compile and the generated-artifact drift gate. Regenerate ci.yml. Co-authored-by: Cursor <cursoragent@cursor.com> * merge origin/main and resolve conflict markers; fix ParseTableMemo Case-A leak Integrate main (#6431) atop #6375 base. Remove leftover merge conflict markers from ci_materialization, commit_workflow, design_document, ci.yml. Gate seed ParseTableMemo insert/serve on Memoize only (after governed door) so insert-then-lookup door observables are order-independent; enroll parse_table_memo_door_order_independent witness. Co-authored-by: Cursor <cursoragent@cursor.com> * fix: add Terminal disposition contracts for carrier predicate helpers Land v2_compiler_materialization_memo_gate_predicate_contract and v2_compiler_catalog_projection_predicate_contract on materialization_allows_memo_store / projection_is_projected — matching the Terminal predicate discipline already in materialization_ladder.dag. Co-authored-by: Cursor <cursoragent@cursor.com> * fix: correct resolve-receipt note attribution for #6422 enrollments Pin was already 3 before #6422 witness entries; run 29059860791 held at 3 with those entries enrolled because they pooled warm on the shared index (zero additional cold resolves). Remove false per-entry bump claim. Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Ground v2 ParseTable + cached_stage on materialization: inhabit the Real * fix: drop duplicate projection_is_projected; use extdeps authority Remove forked CatalogProviderProjection predicate from materialization_carriers.dag; projection_ok cells now call !projection_is_refused from extdeps.cache.materialization (§3 single authority). Remove local Terminal contract that documented the fork. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(ci): raise floor step timeout to 45m for #6422 witness cone Receipt run 29065683045 @ a2bcd6f: batches 1-3 ~20m, batch 4 rust_monolith_gate killed at 30m step cap. Bump floor step and ci/ci_regen job backstop (85m→100m) to match rust gate step budget. Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Cursor <cursoragent@cursor.com>

…rier (#6435) * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * Analysis spine: spine_receipt composes dependency_view + materialize (increment 1) The Rc-safe half of run ≜ realize ∘ materialize ∘ dependency_view: compose the two catamorphisms that need no interpreter thread-safety. - src/v2/std/spine.dag: level-profile fold over the DependencyView DAG → critical_path_depth (longest dependency chain / the reduce spine) + independence_width (max nodes at one level / what parallelizes to hw width). spine_receipt emits the §9 numbers by COMPUTATION, not assertion, joining materialize's content-hash Share/dedup counts. - Discriminating RED (§9): single chain A→B→C → depth 3 / width 1 (serial); root over N independent leaves → depth 2 / width N. - Witnesses 9/9 (spine) green by execution; materialize 6/6 unchanged. The RUNNER (wall-clock parallel execution, §8 steps 3-4) stays increment 2+, gated on interpreter Rc→Arc (v1 interp is !Send). This increment measures the achievable width/critical-path floor the runner targets; it does not fabricate wall-clock N×. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Unbundle recompute-trace interpreter extension from the analysis-spine PR CI root cause: the recompute-trace extension edited v1_interpreter.rs, which hosts the atom_identity_hash/hash_combine fnv1a64 intrinsics that dag_collect_fingerprint_witness executes. That correctly invalidated the re-verify 'assumed-green node-frontier' SKIP of that witness (cone-touch, fail-closed) — but the witness is a pre-existing 806s (13min) fnv1a64 whole-corpus walk, so un-skipping it blew the 30-min floor budget. Main stays green only because its interpreter cone is untouched (witness stays skipped). The interpreter extension is a separable, env-gated (GUNBC_RECOMPUTE_TRACE=1), READ-mode diagnostic — not a dependency of the spine/materialize analysis substrate (pure .dag, content_hash-based). Reverting v1_interpreter.rs to main restores byte-identical interpreter content → the fingerprint witness returns to assumed-green SKIP → floor budget restored. The recompute-trace extension lands as its own follow-up PR where the fingerprint-witness re-run is expected. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Materialization ladder: the state x decision law, as executable witnesses The operator's 4-rule business logic generalized on one axis: the decision is a function of WHEN the redundancy is knowable and WHETHER what was knowable was prepared for. Errors fire only on knowable-but-unprepared; genuine emergence and declared triviality are typed acceptances, never silence. std.materialization_ladder (dag/std): - Frame/FrameKind: SharedStateFrame | IsolatedChildrenFrame | ReplayedFrame{attempts} | UnboundedSiblingsFrame — the nested-scope axis; a 'run' at any layer is a frame, never a different kind of thing. - DemandNature: Pure | IdempotentEffect | WorldRead{envelope_declared} | FreshEffect — the D3 effect gate; TTL = unmodeled dependency XOR declared staleness, never an eviction knob. - CacheProvider: keying (ContentKeyed|ExistenceKeyed) + EvictionPolicy (ScopeExit|SpacePacked) REQUIRED by construction — rule 3 unwritable-bad. - LadderVerdict (total): AuthoredDuplication | Discharged | RefusedNoProvider | RefusedScopeTooNarrow | RefusedExistenceKeyed (the build-if-absent #6352 wall) | RefusedUnmodeledWorldRead | ExemptFreshEffect | AcceptedBelowCostFloor | AcceptedSingleRecompute. - Declared-emergent frames obligate UP FRONT: retry frames derive checkpointing, unbounded-sibling frames (server loops, CI-runs-over-time) derive persistent caches — prepare-before-demand. 12/12 witnesses green by execution; each test is one cell of the table, fixtures mirror the live worked example (workflow/jobs/steps, retry, fleet). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Docs crystallization: derived materialization placement into roadmap ④ + the state x decision table - roadmap_authority ④ (ROADMAP.md regenerated via main_wet, drift-consistent): the nested-frame law, plurality cells, declared-emergent prepare-up-front, keying/staleness walls, the caching-completion un-shelve, the ①-vs-§2 cross-run-caching contradiction subsumed into per-node derived verdicts, and the v1.compiler.ownership §3-convergence row. - duplicate-work design doc: the state x decision table as the qualifier's final verdict logic + the ownership consolidation subsection. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Consolidation plan: realization + materialization + ownership are one law Operator synthesis (2026-07-09): Materialization = the verdict vocabulary; ownership = the verdict computation at the eval frame; realization = provider selection discharging the verdict at each frame's carrier. Share's handler is layer-aware — the cheapest cache is a REFERENCE (same program + ref-capable carrier + readonly => point-at-it, no copy, no destroyed memory), demoting through HAMT structural share / process memo / artifact / CAS, and demotion must be priced, never silent (#6249 clone-fallback is the receipt). Census of every hand-rolled instance found in-tree (16 rows): each named with its frame, ladder cell, action, and dissolution trigger — ownership (consolidated), Rc/HAMT (reference tier), clone-fallback (demotion wall), ParseTable + cached_stage (C5 grounding), M1 memo / intern / assumed-green / sccache / cargo-cache (C2 provider rows), PROCESS_RESOLVE_STORE (C3 = the 9GB scope fix), build-if-absent (landed as the keying wall), resolve-cache #5789 (rule-4 requirements), recompute-trace (state-4 finding source). Peers kept distinct: affected-set, Independence/Placement, mutable state. Sequence C1-C5 with C1 = this PR (ladder + live CI gate). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Ownership refactored onto materialization: the value tier, tested as providers First consolidation increment (operator-directed): v1.compiler.ownership's decision IS provider selection at the value grain, now expressed in the ladder and proven by mirror witnesses. - ProviderTier: Reference | Copy | Memo{keying} | Artifact{keying} | Cas{keying} — keying folded INTO store tiers, so an existence-keyed reference is unwritable BY SHAPE (§5 construction); frame obligations at isolation LCAs are dischargeable only by store tiers (a reference cannot cross an isolation boundary — witnessed). - Value tier: AccessMode (Read|Consume|Carry|Project), ValueUsage, value_materialization -> ValueDead | ValueMoved | ValueSharedByReference | ValueDemotedToCopy (PRICED, requires a declared CopyTier provider — the #6249 silent clone-fallback made refusable) | ValueRefusedAffine | ValueRefusedNoCarrierProvider. - Faithful to v1's THREE distinct plurality readings, kept separate: take_count = semantic_consumer_count (Consumed only, affine axis); value_access_plurality = binding_fan_out (Carry excluded, reference axis); borrow_count = whole_value_borrow_count (Read+Carry not Project, movability). - Fold-accumulator case mirrored at conservative pre-FoldAccUnwrap semantics: carries excluded from plurality but blocking the move -> demoted-to-copy. 23/23 witnesses green (14 frame cells + 9 value cells). Corpus-level equivalence against live v1 ownership folds = the C4 receipt. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * WIP: Duplicate Computation/Materialization * G: exhaustive cache/share census (52 mechanisms, 6 groups) + C0 catalog-grounding of provider rows + forward-wiring walls The sweep found the tree already models caches: extdeps/cache cache_catalog (CacheInterfaceCatalogFacts, 9 rows) + extdeps/realization carrier facts + the warm==cold purity oracle. C0 grounds CacheProvider on that catalog (provider_from_catalog projection; CacheProvider.id -> CacheInterfaceId) so the ladder and the catalog never restate each other's half. Three forward walls (demands-from-DependencyView, hand-cache shape lens, live-provider-or-red) make new caches born as provider rows only. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * C0: provider rows derive from the cited cache catalog (extdeps/cache/materialization.dag projection) Operator ruling 2026-07-09: extdeps stays the ground-truth/interface layer; every USE of a cache goes through materialization so memo never gets re-invented. provider_from_catalog is the only door from a catalog row into the ladder: keying/tier/eviction derived from cited facts (mechanism->class: InProcess=>ScopeExit; Ttl/Lru/SizeBounded=>SpacePacked; Never/Manual outside a process => typed ProjectionRefused, counted by the new enrolled witness; HandAuthored/prefix-fallback keys => ExistenceKeyed until key-completeness upgrades them). CI sccache row now derived, hand-typed tier facts deleted, 6/6 + 23/23 witnesses green. 10b re-scoped: v1-internal caches = inventory only (retire with seed); ParseTable/cached_stage promoted to next. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Fail-closed hardening: RefusedNatureConflict arm, weakest-claim nature default, UsesStep silence counted, ladder cells enrolled in CI The complexity-lens false-clean lesson applied to this gate before anyone relies on it: (1) heterogeneous natures on one identity were first-wins — now a typed RefusedNatureConflict (24th ladder cell, RED witness); (2) an unrostered run-step claimed PureComputation — now IdempotentEffect, the weakest claim (verdict-identical, no purity overclaim; a misdeclared WorldRead can no longer be legally memoized by default); (3) UsesStep foreign actions produced silence — now a pinned denominator (==9; a new foreign action must consciously bump it); (4) the ladder's 24 cells enroll on the CI surface so the law's own witnesses gate every run. 24/24 + 7/7. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * M1 ratchet bump: declared resolve count 1 -> 3 (gate's first live catch — enrollment pays counted resolves) Run 29058798771: enrolling the ladder witness file moved resolves_total 1 -> 3 (each enrolled entry file pays one closure resolve against the shared index) and the receipt gate redded that exact run — the designed semantics, receipted in the note. Bump acknowledges the two witness-entry resolves as declared debt; M2 (one closure resolve per roots set) ratchets back toward 1. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Retire the M1/M2 labels from carriers (no-codename rule; collided with DESIGN's floor-memoization M1/M2) Step renamed to 'Floor resolve receipt gate (declared cold-resolve count)'; notes and the gate echo now say what the things are: the counted cold-resolve receipt, and the shared-resolve rewire (one closure resolve per source-roots set). Naming-retirement note left on the carrier. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * Run-step natures pinned as counted claims, not affordances Operator ruling 2026-07-10: hardcoded shell scripts get no affordances — run-steps are supposed to be modeled as typed intents and emitted (shell-emission-model slice 4 covers ci_workflow RunSteps). Until that slice lands, every nature in job_run_demands is a claim (rostered FreshEffect / weakest-default IdempotentEffect), never derived from a modeled effect row. Count the whole claimed-nature surface (16 by execution) and pin it in the enrolled witness: a new raw-script step must consciously bump the pin, and slice-4 migration becomes a countable ratchet ending at zero, where the pin becomes a wall. Rejected alternative recorded in the carrier note: a per-step declared-natures roster (a second parallel ledger over scripts the shell-emission plan already governs). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Regen DESIGN.md after main merge (duplicate-work row + #6373 body-lowering row coexist) Conflict was both sides editing the open-threads list in design_document.dag: ours added the duplicate-work thread, main's #6373 rewrote body-lowering to Stages 1-3 LANDED. Resolution keeps both rows in the .dag authority; DESIGN.md regenerated from it, never hand-edited. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ate (#6441) * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * Analysis spine: spine_receipt composes dependency_view + materialize (increment 1) The Rc-safe half of run ≜ realize ∘ materialize ∘ dependency_view: compose the two catamorphisms that need no interpreter thread-safety. - src/v2/std/spine.dag: level-profile fold over the DependencyView DAG → critical_path_depth (longest dependency chain / the reduce spine) + independence_width (max nodes at one level / what parallelizes to hw width). spine_receipt emits the §9 numbers by COMPUTATION, not assertion, joining materialize's content-hash Share/dedup counts. - Discriminating RED (§9): single chain A→B→C → depth 3 / width 1 (serial); root over N independent leaves → depth 2 / width N. - Witnesses 9/9 (spine) green by execution; materialize 6/6 unchanged. The RUNNER (wall-clock parallel execution, §8 steps 3-4) stays increment 2+, gated on interpreter Rc→Arc (v1 interp is !Send). This increment measures the achievable width/critical-path floor the runner targets; it does not fabricate wall-clock N×. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Unbundle recompute-trace interpreter extension from the analysis-spine PR CI root cause: the recompute-trace extension edited v1_interpreter.rs, which hosts the atom_identity_hash/hash_combine fnv1a64 intrinsics that dag_collect_fingerprint_witness executes. That correctly invalidated the re-verify 'assumed-green node-frontier' SKIP of that witness (cone-touch, fail-closed) — but the witness is a pre-existing 806s (13min) fnv1a64 whole-corpus walk, so un-skipping it blew the 30-min floor budget. Main stays green only because its interpreter cone is untouched (witness stays skipped). The interpreter extension is a separable, env-gated (GUNBC_RECOMPUTE_TRACE=1), READ-mode diagnostic — not a dependency of the spine/materialize analysis substrate (pure .dag, content_hash-based). Reverting v1_interpreter.rs to main restores byte-identical interpreter content → the fingerprint witness returns to assumed-green SKIP → floor budget restored. The recompute-trace extension lands as its own follow-up PR where the fingerprint-witness re-run is expected. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Materialization ladder: the state x decision law, as executable witnesses The operator's 4-rule business logic generalized on one axis: the decision is a function of WHEN the redundancy is knowable and WHETHER what was knowable was prepared for. Errors fire only on knowable-but-unprepared; genuine emergence and declared triviality are typed acceptances, never silence. std.materialization_ladder (dag/std): - Frame/FrameKind: SharedStateFrame | IsolatedChildrenFrame | ReplayedFrame{attempts} | UnboundedSiblingsFrame — the nested-scope axis; a 'run' at any layer is a frame, never a different kind of thing. - DemandNature: Pure | IdempotentEffect | WorldRead{envelope_declared} | FreshEffect — the D3 effect gate; TTL = unmodeled dependency XOR declared staleness, never an eviction knob. - CacheProvider: keying (ContentKeyed|ExistenceKeyed) + EvictionPolicy (ScopeExit|SpacePacked) REQUIRED by construction — rule 3 unwritable-bad. - LadderVerdict (total): AuthoredDuplication | Discharged | RefusedNoProvider | RefusedScopeTooNarrow | RefusedExistenceKeyed (the build-if-absent #6352 wall) | RefusedUnmodeledWorldRead | ExemptFreshEffect | AcceptedBelowCostFloor | AcceptedSingleRecompute. - Declared-emergent frames obligate UP FRONT: retry frames derive checkpointing, unbounded-sibling frames (server loops, CI-runs-over-time) derive persistent caches — prepare-before-demand. 12/12 witnesses green by execution; each test is one cell of the table, fixtures mirror the live worked example (workflow/jobs/steps, retry, fleet). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Docs crystallization: derived materialization placement into roadmap ④ + the state x decision table - roadmap_authority ④ (ROADMAP.md regenerated via main_wet, drift-consistent): the nested-frame law, plurality cells, declared-emergent prepare-up-front, keying/staleness walls, the caching-completion un-shelve, the ①-vs-§2 cross-run-caching contradiction subsumed into per-node derived verdicts, and the v1.compiler.ownership §3-convergence row. - duplicate-work design doc: the state x decision table as the qualifier's final verdict logic + the ownership consolidation subsection. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Consolidation plan: realization + materialization + ownership are one law Operator synthesis (2026-07-09): Materialization = the verdict vocabulary; ownership = the verdict computation at the eval frame; realization = provider selection discharging the verdict at each frame's carrier. Share's handler is layer-aware — the cheapest cache is a REFERENCE (same program + ref-capable carrier + readonly => point-at-it, no copy, no destroyed memory), demoting through HAMT structural share / process memo / artifact / CAS, and demotion must be priced, never silent (#6249 clone-fallback is the receipt). Census of every hand-rolled instance found in-tree (16 rows): each named with its frame, ladder cell, action, and dissolution trigger — ownership (consolidated), Rc/HAMT (reference tier), clone-fallback (demotion wall), ParseTable + cached_stage (C5 grounding), M1 memo / intern / assumed-green / sccache / cargo-cache (C2 provider rows), PROCESS_RESOLVE_STORE (C3 = the 9GB scope fix), build-if-absent (landed as the keying wall), resolve-cache #5789 (rule-4 requirements), recompute-trace (state-4 finding source). Peers kept distinct: affected-set, Independence/Placement, mutable state. Sequence C1-C5 with C1 = this PR (ladder + live CI gate). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Ownership refactored onto materialization: the value tier, tested as providers First consolidation increment (operator-directed): v1.compiler.ownership's decision IS provider selection at the value grain, now expressed in the ladder and proven by mirror witnesses. - ProviderTier: Reference | Copy | Memo{keying} | Artifact{keying} | Cas{keying} — keying folded INTO store tiers, so an existence-keyed reference is unwritable BY SHAPE (§5 construction); frame obligations at isolation LCAs are dischargeable only by store tiers (a reference cannot cross an isolation boundary — witnessed). - Value tier: AccessMode (Read|Consume|Carry|Project), ValueUsage, value_materialization -> ValueDead | ValueMoved | ValueSharedByReference | ValueDemotedToCopy (PRICED, requires a declared CopyTier provider — the #6249 silent clone-fallback made refusable) | ValueRefusedAffine | ValueRefusedNoCarrierProvider. - Faithful to v1's THREE distinct plurality readings, kept separate: take_count = semantic_consumer_count (Consumed only, affine axis); value_access_plurality = binding_fan_out (Carry excluded, reference axis); borrow_count = whole_value_borrow_count (Read+Carry not Project, movability). - Fold-accumulator case mirrored at conservative pre-FoldAccUnwrap semantics: carries excluded from plurality but blocking the move -> demoted-to-copy. 23/23 witnesses green (14 frame cells + 9 value cells). Corpus-level equivalence against live v1 ownership folds = the C4 receipt. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * WIP: Duplicate Computation/Materialization * G: exhaustive cache/share census (52 mechanisms, 6 groups) + C0 catalog-grounding of provider rows + forward-wiring walls The sweep found the tree already models caches: extdeps/cache cache_catalog (CacheInterfaceCatalogFacts, 9 rows) + extdeps/realization carrier facts + the warm==cold purity oracle. C0 grounds CacheProvider on that catalog (provider_from_catalog projection; CacheProvider.id -> CacheInterfaceId) so the ladder and the catalog never restate each other's half. Three forward walls (demands-from-DependencyView, hand-cache shape lens, live-provider-or-red) make new caches born as provider rows only. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * C0: provider rows derive from the cited cache catalog (extdeps/cache/materialization.dag projection) Operator ruling 2026-07-09: extdeps stays the ground-truth/interface layer; every USE of a cache goes through materialization so memo never gets re-invented. provider_from_catalog is the only door from a catalog row into the ladder: keying/tier/eviction derived from cited facts (mechanism->class: InProcess=>ScopeExit; Ttl/Lru/SizeBounded=>SpacePacked; Never/Manual outside a process => typed ProjectionRefused, counted by the new enrolled witness; HandAuthored/prefix-fallback keys => ExistenceKeyed until key-completeness upgrades them). CI sccache row now derived, hand-typed tier facts deleted, 6/6 + 23/23 witnesses green. 10b re-scoped: v1-internal caches = inventory only (retire with seed); ParseTable/cached_stage promoted to next. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Fail-closed hardening: RefusedNatureConflict arm, weakest-claim nature default, UsesStep silence counted, ladder cells enrolled in CI The complexity-lens false-clean lesson applied to this gate before anyone relies on it: (1) heterogeneous natures on one identity were first-wins — now a typed RefusedNatureConflict (24th ladder cell, RED witness); (2) an unrostered run-step claimed PureComputation — now IdempotentEffect, the weakest claim (verdict-identical, no purity overclaim; a misdeclared WorldRead can no longer be legally memoized by default); (3) UsesStep foreign actions produced silence — now a pinned denominator (==9; a new foreign action must consciously bump it); (4) the ladder's 24 cells enroll on the CI surface so the law's own witnesses gate every run. 24/24 + 7/7. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * M1 ratchet bump: declared resolve count 1 -> 3 (gate's first live catch — enrollment pays counted resolves) Run 29058798771: enrolling the ladder witness file moved resolves_total 1 -> 3 (each enrolled entry file pays one closure resolve against the shared index) and the receipt gate redded that exact run — the designed semantics, receipted in the note. Bump acknowledges the two witness-entry resolves as declared debt; M2 (one closure resolve per roots set) ratchets back toward 1. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Retire the M1/M2 labels from carriers (no-codename rule; collided with DESIGN's floor-memoization M1/M2) Step renamed to 'Floor resolve receipt gate (declared cold-resolve count)'; notes and the gate echo now say what the things are: the counted cold-resolve receipt, and the shared-resolve rewire (one closure resolve per source-roots set). Naming-retirement note left on the carrier. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * Run-step natures pinned as counted claims, not affordances Operator ruling 2026-07-10: hardcoded shell scripts get no affordances — run-steps are supposed to be modeled as typed intents and emitted (shell-emission-model slice 4 covers ci_workflow RunSteps). Until that slice lands, every nature in job_run_demands is a claim (rostered FreshEffect / weakest-default IdempotentEffect), never derived from a modeled effect row. Count the whole claimed-nature surface (16 by execution) and pin it in the enrolled witness: a new raw-script step must consciously bump the pin, and slice-4 migration becomes a countable ratchet ending at zero, where the pin becomes a wall. Rejected alternative recorded in the carrier note: a per-step declared-natures roster (a second parallel ledger over scripts the shell-emission plan already governs). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Regen DESIGN.md after main merge (duplicate-work row + #6373 body-lowering row coexist) Conflict was both sides editing the open-threads list in design_document.dag: ours added the duplicate-work thread, main's #6373 rewrote body-lowering to Stages 1-3 LANDED. Resolution keeps both rows in the .dag authority; DESIGN.md regenerated from it, never hand-edited. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Inferred materialization increment 1: floor demand ledger + receipt-or-red gate Running IS enrolling: the interpreter ledgers every keyed pure call and every InterpContext absorbs its totals into a process accumulator on Drop — by construction, no eval path escapes the receipt. claim_executor writes target/floor-materialization-receipt.txt at walk end; trace defaults ON in the executor, and an explicit =0 zeroes keyed_calls which the gate refuses. Gate arms this push (all verified under dash from the emitted ci.yml): receipt missing/malformed/keyed_calls=0 fail closed. Exact pins for unkeyed_calls and duplicated_keys land from the FIRST CI receipt (the resolve gate's measure-then-pin path) — unkeyed is known nonzero on the floor (count_matching takes a predicate closure; closures are the one disclosed identity-less class, dissolve-on captured-env content identity). Unit witness green: resolve -> eval -> ctx Drop -> absorb -> drain-once. Step addition bumped the claimed-natures pin 16 -> 17 consciously. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Review fix: drop the racy drain-once assertion from the receipt unit test opus-4-7 finding on #6441: under plain cargo test (still the documented runner) tests share a process, the env latch is OnceLock-sticky, and sibling ctx drops could absorb between the two takes — making the drain-empty assertion racy. Absorb-on-Drop stays asserted (monotone under concurrent absorbs: siblings only ADD); drain-once is Option::take by construction, not asserted through the shared global. Comment states the sharing semantics explicitly. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… gated) (#6455) * Inferred materialization increment 1: floor demand ledger + receipt gate (#6441) * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * Analysis spine: spine_receipt composes dependency_view + materialize (increment 1) The Rc-safe half of run ≜ realize ∘ materialize ∘ dependency_view: compose the two catamorphisms that need no interpreter thread-safety. - src/v2/std/spine.dag: level-profile fold over the DependencyView DAG → critical_path_depth (longest dependency chain / the reduce spine) + independence_width (max nodes at one level / what parallelizes to hw width). spine_receipt emits the §9 numbers by COMPUTATION, not assertion, joining materialize's content-hash Share/dedup counts. - Discriminating RED (§9): single chain A→B→C → depth 3 / width 1 (serial); root over N independent leaves → depth 2 / width N. - Witnesses 9/9 (spine) green by execution; materialize 6/6 unchanged. The RUNNER (wall-clock parallel execution, §8 steps 3-4) stays increment 2+, gated on interpreter Rc→Arc (v1 interp is !Send). This increment measures the achievable width/critical-path floor the runner targets; it does not fabricate wall-clock N×. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Unbundle recompute-trace interpreter extension from the analysis-spine PR CI root cause: the recompute-trace extension edited v1_interpreter.rs, which hosts the atom_identity_hash/hash_combine fnv1a64 intrinsics that dag_collect_fingerprint_witness executes. That correctly invalidated the re-verify 'assumed-green node-frontier' SKIP of that witness (cone-touch, fail-closed) — but the witness is a pre-existing 806s (13min) fnv1a64 whole-corpus walk, so un-skipping it blew the 30-min floor budget. Main stays green only because its interpreter cone is untouched (witness stays skipped). The interpreter extension is a separable, env-gated (GUNBC_RECOMPUTE_TRACE=1), READ-mode diagnostic — not a dependency of the spine/materialize analysis substrate (pure .dag, content_hash-based). Reverting v1_interpreter.rs to main restores byte-identical interpreter content → the fingerprint witness returns to assumed-green SKIP → floor budget restored. The recompute-trace extension lands as its own follow-up PR where the fingerprint-witness re-run is expected. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Materialization ladder: the state x decision law, as executable witnesses The operator's 4-rule business logic generalized on one axis: the decision is a function of WHEN the redundancy is knowable and WHETHER what was knowable was prepared for. Errors fire only on knowable-but-unprepared; genuine emergence and declared triviality are typed acceptances, never silence. std.materialization_ladder (dag/std): - Frame/FrameKind: SharedStateFrame | IsolatedChildrenFrame | ReplayedFrame{attempts} | UnboundedSiblingsFrame — the nested-scope axis; a 'run' at any layer is a frame, never a different kind of thing. - DemandNature: Pure | IdempotentEffect | WorldRead{envelope_declared} | FreshEffect — the D3 effect gate; TTL = unmodeled dependency XOR declared staleness, never an eviction knob. - CacheProvider: keying (ContentKeyed|ExistenceKeyed) + EvictionPolicy (ScopeExit|SpacePacked) REQUIRED by construction — rule 3 unwritable-bad. - LadderVerdict (total): AuthoredDuplication | Discharged | RefusedNoProvider | RefusedScopeTooNarrow | RefusedExistenceKeyed (the build-if-absent #6352 wall) | RefusedUnmodeledWorldRead | ExemptFreshEffect | AcceptedBelowCostFloor | AcceptedSingleRecompute. - Declared-emergent frames obligate UP FRONT: retry frames derive checkpointing, unbounded-sibling frames (server loops, CI-runs-over-time) derive persistent caches — prepare-before-demand. 12/12 witnesses green by execution; each test is one cell of the table, fixtures mirror the live worked example (workflow/jobs/steps, retry, fleet). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Docs crystallization: derived materialization placement into roadmap ④ + the state x decision table - roadmap_authority ④ (ROADMAP.md regenerated via main_wet, drift-consistent): the nested-frame law, plurality cells, declared-emergent prepare-up-front, keying/staleness walls, the caching-completion un-shelve, the ①-vs-§2 cross-run-caching contradiction subsumed into per-node derived verdicts, and the v1.compiler.ownership §3-convergence row. - duplicate-work design doc: the state x decision table as the qualifier's final verdict logic + the ownership consolidation subsection. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Consolidation plan: realization + materialization + ownership are one law Operator synthesis (2026-07-09): Materialization = the verdict vocabulary; ownership = the verdict computation at the eval frame; realization = provider selection discharging the verdict at each frame's carrier. Share's handler is layer-aware — the cheapest cache is a REFERENCE (same program + ref-capable carrier + readonly => point-at-it, no copy, no destroyed memory), demoting through HAMT structural share / process memo / artifact / CAS, and demotion must be priced, never silent (#6249 clone-fallback is the receipt). Census of every hand-rolled instance found in-tree (16 rows): each named with its frame, ladder cell, action, and dissolution trigger — ownership (consolidated), Rc/HAMT (reference tier), clone-fallback (demotion wall), ParseTable + cached_stage (C5 grounding), M1 memo / intern / assumed-green / sccache / cargo-cache (C2 provider rows), PROCESS_RESOLVE_STORE (C3 = the 9GB scope fix), build-if-absent (landed as the keying wall), resolve-cache #5789 (rule-4 requirements), recompute-trace (state-4 finding source). Peers kept distinct: affected-set, Independence/Placement, mutable state. Sequence C1-C5 with C1 = this PR (ladder + live CI gate). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Ownership refactored onto materialization: the value tier, tested as providers First consolidation increment (operator-directed): v1.compiler.ownership's decision IS provider selection at the value grain, now expressed in the ladder and proven by mirror witnesses. - ProviderTier: Reference | Copy | Memo{keying} | Artifact{keying} | Cas{keying} — keying folded INTO store tiers, so an existence-keyed reference is unwritable BY SHAPE (§5 construction); frame obligations at isolation LCAs are dischargeable only by store tiers (a reference cannot cross an isolation boundary — witnessed). - Value tier: AccessMode (Read|Consume|Carry|Project), ValueUsage, value_materialization -> ValueDead | ValueMoved | ValueSharedByReference | ValueDemotedToCopy (PRICED, requires a declared CopyTier provider — the #6249 silent clone-fallback made refusable) | ValueRefusedAffine | ValueRefusedNoCarrierProvider. - Faithful to v1's THREE distinct plurality readings, kept separate: take_count = semantic_consumer_count (Consumed only, affine axis); value_access_plurality = binding_fan_out (Carry excluded, reference axis); borrow_count = whole_value_borrow_count (Read+Carry not Project, movability). - Fold-accumulator case mirrored at conservative pre-FoldAccUnwrap semantics: carries excluded from plurality but blocking the move -> demoted-to-copy. 23/23 witnesses green (14 frame cells + 9 value cells). Corpus-level equivalence against live v1 ownership folds = the C4 receipt. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * WIP: Duplicate Computation/Materialization * G: exhaustive cache/share census (52 mechanisms, 6 groups) + C0 catalog-grounding of provider rows + forward-wiring walls The sweep found the tree already models caches: extdeps/cache cache_catalog (CacheInterfaceCatalogFacts, 9 rows) + extdeps/realization carrier facts + the warm==cold purity oracle. C0 grounds CacheProvider on that catalog (provider_from_catalog projection; CacheProvider.id -> CacheInterfaceId) so the ladder and the catalog never restate each other's half. Three forward walls (demands-from-DependencyView, hand-cache shape lens, live-provider-or-red) make new caches born as provider rows only. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * C0: provider rows derive from the cited cache catalog (extdeps/cache/materialization.dag projection) Operator ruling 2026-07-09: extdeps stays the ground-truth/interface layer; every USE of a cache goes through materialization so memo never gets re-invented. provider_from_catalog is the only door from a catalog row into the ladder: keying/tier/eviction derived from cited facts (mechanism->class: InProcess=>ScopeExit; Ttl/Lru/SizeBounded=>SpacePacked; Never/Manual outside a process => typed ProjectionRefused, counted by the new enrolled witness; HandAuthored/prefix-fallback keys => ExistenceKeyed until key-completeness upgrades them). CI sccache row now derived, hand-typed tier facts deleted, 6/6 + 23/23 witnesses green. 10b re-scoped: v1-internal caches = inventory only (retire with seed); ParseTable/cached_stage promoted to next. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Fail-closed hardening: RefusedNatureConflict arm, weakest-claim nature default, UsesStep silence counted, ladder cells enrolled in CI The complexity-lens false-clean lesson applied to this gate before anyone relies on it: (1) heterogeneous natures on one identity were first-wins — now a typed RefusedNatureConflict (24th ladder cell, RED witness); (2) an unrostered run-step claimed PureComputation — now IdempotentEffect, the weakest claim (verdict-identical, no purity overclaim; a misdeclared WorldRead can no longer be legally memoized by default); (3) UsesStep foreign actions produced silence — now a pinned denominator (==9; a new foreign action must consciously bump it); (4) the ladder's 24 cells enroll on the CI surface so the law's own witnesses gate every run. 24/24 + 7/7. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * M1 ratchet bump: declared resolve count 1 -> 3 (gate's first live catch — enrollment pays counted resolves) Run 29058798771: enrolling the ladder witness file moved resolves_total 1 -> 3 (each enrolled entry file pays one closure resolve against the shared index) and the receipt gate redded that exact run — the designed semantics, receipted in the note. Bump acknowledges the two witness-entry resolves as declared debt; M2 (one closure resolve per roots set) ratchets back toward 1. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Retire the M1/M2 labels from carriers (no-codename rule; collided with DESIGN's floor-memoization M1/M2) Step renamed to 'Floor resolve receipt gate (declared cold-resolve count)'; notes and the gate echo now say what the things are: the counted cold-resolve receipt, and the shared-resolve rewire (one closure resolve per source-roots set). Naming-retirement note left on the carrier. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * Run-step natures pinned as counted claims, not affordances Operator ruling 2026-07-10: hardcoded shell scripts get no affordances — run-steps are supposed to be modeled as typed intents and emitted (shell-emission-model slice 4 covers ci_workflow RunSteps). Until that slice lands, every nature in job_run_demands is a claim (rostered FreshEffect / weakest-default IdempotentEffect), never derived from a modeled effect row. Count the whole claimed-nature surface (16 by execution) and pin it in the enrolled witness: a new raw-script step must consciously bump the pin, and slice-4 migration becomes a countable ratchet ending at zero, where the pin becomes a wall. Rejected alternative recorded in the carrier note: a per-step declared-natures roster (a second parallel ledger over scripts the shell-emission plan already governs). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Regen DESIGN.md after main merge (duplicate-work row + #6373 body-lowering row coexist) Conflict was both sides editing the open-threads list in design_document.dag: ours added the duplicate-work thread, main's #6373 rewrote body-lowering to Stages 1-3 LANDED. Resolution keeps both rows in the .dag authority; DESIGN.md regenerated from it, never hand-edited. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Inferred materialization increment 1: floor demand ledger + receipt-or-red gate Running IS enrolling: the interpreter ledgers every keyed pure call and every InterpContext absorbs its totals into a process accumulator on Drop — by construction, no eval path escapes the receipt. claim_executor writes target/floor-materialization-receipt.txt at walk end; trace defaults ON in the executor, and an explicit =0 zeroes keyed_calls which the gate refuses. Gate arms this push (all verified under dash from the emitted ci.yml): receipt missing/malformed/keyed_calls=0 fail closed. Exact pins for unkeyed_calls and duplicated_keys land from the FIRST CI receipt (the resolve gate's measure-then-pin path) — unkeyed is known nonzero on the floor (count_matching takes a predicate closure; closures are the one disclosed identity-less class, dissolve-on captured-env content identity). Unit witness green: resolve -> eval -> ctx Drop -> absorb -> drain-once. Step addition bumped the claimed-natures pin 16 -> 17 consciously. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Review fix: drop the racy drain-once assertion from the receipt unit test opus-4-7 finding on #6441: under plain cargo test (still the documented runner) tests share a process, the env latch is OnceLock-sticky, and sibling ctx drops could absorb between the two takes — making the drain-empty assertion racy. Absorb-on-Drop stays asserted (monotone under concurrent absorbs: siblings only ADD); drain-once is Option::take by construction, not asserted through the shared global. Comment states the sharing semantics explicitly. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Turn on affected-set CI: re-land witness enrollment flip (discovery shrunk by affected set) + falsifier host-OOM receipt (#6438) * WIP: Re-land affected-set CI enrollment flip once shard resolve footprint is * WIP: Re-land affected-set CI enrollment flip once shard resolve footprint is * Fix CI OOM: pin discovery corpus spawn_width_cap to 1. Run 28999086030 OOM-killed at width=2 on the 24GiB live slot during the discovery-flip corpus batch. Gate workloads still fit at width=2; only the tree-wide discovery batch serializes via spawn_width_cap=1, with a receipt witness and dissolve-on note. Co-authored-by: Cursor <cursoragent@cursor.com> * Fix type error in corpus discovery spawn width cap helper. Both if-branches must return Nat (hardware_thread_count_value), not a bare Int literal. Co-authored-by: Cursor <cursoragent@cursor.com> * Merge affected-set enrollment flip (takeover of PR #6403 from session/gentle-stag-677-flip) Conflict resolutions onto post-#6422/#6431/#6432/#6435 main: - ci_witness_optin_inversion: main's typed Scaffold Disposition -> Terminal (dissolve fired at the flip; roster stays as explicit-entry home, exclusivity dissolved) - floor step timeout: 45 (main) vs 60 (flip) -> 60 with provenance note - ci_spec notes: kept flip notes, stale fixed-8 phrasing dated - falsifier width note: + RESIDUE paragraph (5x exit-137 on srv2-class slots = converge lane, not plan width) Verified before commit: discovery batches charge corpus_resolve_nanos (own key), resolve_nanos=0 -> resolves_total declared 3 is NOT moved by the flip. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: affected set processing * WIP: affected set processing * WIP: affected set processing * WIP: affected set processing * Correct the false flip receipt + kill-surviving calibration receipts (space-lens loop) CORRECTED RECEIPT: run 29000557166's floor never completed - the executor was host-OOM-killed mid-discovery-corpus at ~8min; the log's ExitSuccess belongs to the merge-admission stamp tool stamping CI_FLOOR_EXIT=137. No flipped corpus has completed in CI (0/1 ci + 0/5 falsifier). Both carrier notes corrected. Calibration (coordinated with merry-owl-649's space-lens lane): - roster_import_closure_nodes_pre_resolve: shared closure-count authority (pure import walk, closure grain not entry grain), emitted BEFORE the heavy resolve so killed runs still yield the (nodes, peak) lower-bound pair - width-1 definition-drift oracle: pre-resolve walk must equal post-resolve resolved union on completed runs; refuses on divergence. Proven by execution: pre == post == 190/213 nodes across Off/Applied modes - kill-surviving cgroup memory.peak pre/post steps (post is always()) in the ci job and falsifier.yml; scope semantics labeled on the emission lines (reset=ok floor-scoped; span compares post>pre; never silently conflated) - job backstop timeouts extended by the two aux steps in both jobs Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: affected set processing * Calibration pairs carry censored-vs-exact labels (methodology: killed runs are censored observations) Floor steps get id=floor; the always() post-peak step emits floor_outcome so each (closure_nodes, peak) pair is explicitly labeled: success = exact point, anything else = censored lower bound (true demand strictly greater than read). Prevents the fit from treating cap-kill reads as point estimates, which would drag the slope down and make the predictor underestimate - the dangerous direction (merry-owl methodology catch, 2026-07-10). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: affected set processing * Review fixes: explicit witness import + complete step-budget ledger - ci_floor_plan_witness_test.dag: import witness_ci_corpus_discovery_serializes_at_width_one explicitly (cursor catch on #6438). The call resolved pre-fix via the seed resolver's flat namespace, so the witness ran green by execution; the explicit import restores the file's per-symbol import convention. - ci_workflow.dag: the resolve-receipt gate step had NO step cap — a hang there could only die by job-cancel (the #6323 starvation-kill class). It now carries the aux cap (script is a sub-second receipt read) and the ci job backstop counts four aux terms (peak pre/post, resolve-receipt gate, merge-admission gate): every step budgeted, backstop = step-sum + prelude (claude review catch on #6438, sharpened). - falsifier_workflow_witness_test.dag: falsifier_backstop_is_step_sum_plus_prelude asserted the pre-calibration formula — latent red proven by execution (FAIL receipt), fixed to the live two-aux sum, re-run PASS. - ci.yml regenerated byte-stable from the carrier (backstops 125->130, gate step timeout 5m). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Rebind calibration provenance comments to real artifacts (cursor review catch) The two cli_run.rs comments cited docs/plans/space-lens-minimal-project.md, which does not exist on main — the predictor design is in flight on PR #6442 (merry-owl-649's lane) and was never landed under that path. Rebound: the shared closure-definition authority is stated as this function itself, with the in-flight design cited by PR number and the landed parent-lane authorities cited by real paths (compute-envelope-model.md fleet envelope; input-envelope-roadmap.md admission). No behavior change; cargo check clean. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: affected set processing * Remove stray empty file (shell-redirect artifact the auto-committer flushed) An internal-messaging command's backtick content was command-substituted by bash; a '-> fail-closed' fragment became a stdout redirect and created an empty file at the repo root, which the auto-committer then committed as 38f0a46. No tree content beyond the empty file; removing it restores the branch to e99c757's content. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: affected set processing * Scaffold-mark the cgroup peak calibration shell (cursor review catch) The three concat-built calibration runners (ci_cgroup_peak_locate_shell, ci_floor_peak_pre_script, ci_floor_peak_post_script) landed without the on-carrier scaffold markers repo convention requires for hand-shell. Each now carries a Disposition = Scaffold { dissolves_to: RealizationDispatch } row binding the decl (the ci_materialization pattern), and both scripts embed the shared dissolve-on note as a shell comment (the ci_spec pattern): dissolution = bash-emit (#5828 / gap-B emit(intent, Bash)) realizing the observation as an emitted ShellProgram intent or a typed host Observe effect. ci.yml + falsifier.yml regenerated; falsifier_workflow_witness_holds and the flip witnesses re-run PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Merge origin/main into session/loyal-wren-398 (resolve #6441 step-list conflict) Conflict resolution, all consciously declared: - ci_job steps: union — the calibration peak pre/post steps wrap the floor step (this branch) and #6441's materialization receipt gate slots between the resolve-receipt gate and the merge-admission gate (main). - The incoming materialization receipt gate step landed with timeout none — the same uncapped-step starvation-kill class this branch's review fix eliminated — so it now carries the aux cap, and the ci backstop counts FIVE aux terms (peak pre, peak post, resolve-receipt gate, materialization receipt gate, merge-admission gate); the budget disposition note records the merge provenance. - ci_run_step_natures_are_claims_counted_not_silent: RunStep count pin bumped 17 -> 19, acknowledging the two peak calibration steps #6441's count predates (conscious-count discipline; proven red at 17 then green at 19 by execution). - ci.yml regenerated from the merged carriers (backstops 130 -> 135). Verified on the merged tree: release bins rebuilt on merged Rust; falsifier workflow witness, flip witnesses, floor-plan/optin/width witnesses, and all 8 ci_materialization witnesses PASS; generated-artifact regen ExitSuccess. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> * ShellProgram -> DAG: transport intent collapse + live importer ratchet (phase 0 of the sidecar dissolution) (#6449) * WIP: ShellProgram -> DAG * WIP: ShellProgram -> DAG * WIP: ShellProgram -> DAG * WIP: ShellProgram -> DAG * WIP: ShellProgram -> DAG * bash fold: native Concat/CmdSubst/WithRedir coverage + measured cost wall on the whole-tree emit path Fold-family productions extended so the fold no longer refuses word Concat/CmdSubst or stmt WithRedir (all four Redir variants): new concat_parts/word-compound/with_redir production families, lex tokens, kind-tag emit transforms, and recursive bundle arms. Byte-identity vs serialize_bash proven by execution: five depth-2 oracle tests plus the depth-4 assign_root_stmt manual probe (ROOT=$('git' 'rev-parse' '--show-toplevel' 2>/dev/null || 'pwd') byte-exact). The delegated fail-closed RED control repoints from WithRedir (now native) to Heredoc (still delegated) so the boundary guard stays discriminating. Measured cost wall, declared on-carrier (bash_program_emit_cost_wall_note): whole-tree backward row-selection is ~alternatives^depth (1s flat stmt, 64s for the single depth-4 stmt, DNF >8min for the full witness_bin program) because formal_production_unique_lhs_exact_match deep-validates every candidate per level and the descent re-validates each level again. Real-program oracles therefore stay MANUAL probes, not test fns (a discovery-run test would hang the local floor); the probe note on the test carrier names the dissolution triggers. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> * WIP: Wave 1A - namespace-only name resolution: make the syntactic containment * B1: NameResolutionPolicy row + position-tracked resolve (NamespaceOnlyY gated). Add v2.std.resolution_policy (ImportScoped default | NamespaceOnlyY). Thread ResolveContext { position, expected, policy } through resolve walk; namespace-only skips import module bindings and uses symbol_index at position. Policy pilot witness: green under NamespaceOnlyY at type position, RED under ImportScoped at module position. Import-scoped global default unchanged — zero corpus churn. Co-authored-by: Cursor <cursoragent@cursor.com> * Wave 1A - namespace-only name resolution: make the syntactic containment tree the single naming authority (qualified name = nesting position, reference = lexical lookup up ancestors, . = projection one level down). Path: confirm loyal-heron SymbolIndex scaling receipt FIRST, then SymbolIndex = conta (#6451) * WIP: Wave 1A - namespace-only name resolution: make the syntactic containment * v2.std.symbol_index: materialize containment tree as single naming authority Add SymbolIndex fill from nesting (qualified path → Node), qualified-name path algebra bridges, and discriminating witnesses. Retarget #6436 variant- visibility scaffold to dissolve into symbol_index_lexical_lookup; harvest_unique stays interim until module-scoped index scan lands. Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Wave 1A - namespace-only name resolution: make the syntactic containment * Fix rust fmt on qualified-name bridge host functions (CI rust_tests gate). Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Wave 1A - namespace-only name resolution: make the syntactic containment * Remove symbol_index_has_path; tests match symbol_index_lookup directly Dissolve Optional→Bool predicate in new std/ surface per review. Lexical lookup root termination already uses qualified_name_is_empty (not dotted string projection). Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Wave 1A - namespace-only name resolution: make the syntactic containment * Address review: is_empty authority, host scaffold binds, layer split - Remove qualified_name_is_empty; lexical lookup uses is_empty(xs: position) - Host dispositions bind to from/to_dotted_string bridges; add P5 receipt tests - Move extdeps-coupled fill to v2.compiler.symbol_index_fill (layer DAG fix) Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Wave 1A - namespace-only name resolution: make the syntactic containment * WIP: Wave 1A - namespace-only name resolution: make the syntactic containment * Wire resolver through SymbolIndex; dissolve harvest_unique_disj interim. Build SymbolIndex at admission and thread it through Namespace. resolve_atom falls back to symbol_index_lexical_lookup for unbound atoms after import-scoped lookup_chain. Fill-time unique-variant aliases (suffix-scan equivalent) replace #6436 harvest_unique_disj. Equivalence witnesses prove SymbolIndex-alone covers variant visibility (green + without-alias RED control); end-to-end wire green. Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Brian Searls <briansearls1@gmail.com> Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Wave 1A - namespace-only name resolution: make the syntactic containment * Drop unused ResolveContext.expected until expected-type lane lands. Removes the dead field and uncalled resolve_ctx_with_expected helper flagged in #6454 review 36717 — B1 uses position-only disambiguation; expected-type filtering returns when that slice is scoped. Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: gunbai-bot[bot] <289086189+gunbai-bot[bot]@users.noreply.github.com> Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Brian Searls <briansearls1@gmail.com> Co-authored-by: Brian Searls <11205878+briansrls@users.noreply.github.com>

…eipt-write refusal + counter invariant (#6456) * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * Analysis spine: spine_receipt composes dependency_view + materialize (increment 1) The Rc-safe half of run ≜ realize ∘ materialize ∘ dependency_view: compose the two catamorphisms that need no interpreter thread-safety. - src/v2/std/spine.dag: level-profile fold over the DependencyView DAG → critical_path_depth (longest dependency chain / the reduce spine) + independence_width (max nodes at one level / what parallelizes to hw width). spine_receipt emits the §9 numbers by COMPUTATION, not assertion, joining materialize's content-hash Share/dedup counts. - Discriminating RED (§9): single chain A→B→C → depth 3 / width 1 (serial); root over N independent leaves → depth 2 / width N. - Witnesses 9/9 (spine) green by execution; materialize 6/6 unchanged. The RUNNER (wall-clock parallel execution, §8 steps 3-4) stays increment 2+, gated on interpreter Rc→Arc (v1 interp is !Send). This increment measures the achievable width/critical-path floor the runner targets; it does not fabricate wall-clock N×. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Unbundle recompute-trace interpreter extension from the analysis-spine PR CI root cause: the recompute-trace extension edited v1_interpreter.rs, which hosts the atom_identity_hash/hash_combine fnv1a64 intrinsics that dag_collect_fingerprint_witness executes. That correctly invalidated the re-verify 'assumed-green node-frontier' SKIP of that witness (cone-touch, fail-closed) — but the witness is a pre-existing 806s (13min) fnv1a64 whole-corpus walk, so un-skipping it blew the 30-min floor budget. Main stays green only because its interpreter cone is untouched (witness stays skipped). The interpreter extension is a separable, env-gated (GUNBC_RECOMPUTE_TRACE=1), READ-mode diagnostic — not a dependency of the spine/materialize analysis substrate (pure .dag, content_hash-based). Reverting v1_interpreter.rs to main restores byte-identical interpreter content → the fingerprint witness returns to assumed-green SKIP → floor budget restored. The recompute-trace extension lands as its own follow-up PR where the fingerprint-witness re-run is expected. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Materialization ladder: the state x decision law, as executable witnesses The operator's 4-rule business logic generalized on one axis: the decision is a function of WHEN the redundancy is knowable and WHETHER what was knowable was prepared for. Errors fire only on knowable-but-unprepared; genuine emergence and declared triviality are typed acceptances, never silence. std.materialization_ladder (dag/std): - Frame/FrameKind: SharedStateFrame | IsolatedChildrenFrame | ReplayedFrame{attempts} | UnboundedSiblingsFrame — the nested-scope axis; a 'run' at any layer is a frame, never a different kind of thing. - DemandNature: Pure | IdempotentEffect | WorldRead{envelope_declared} | FreshEffect — the D3 effect gate; TTL = unmodeled dependency XOR declared staleness, never an eviction knob. - CacheProvider: keying (ContentKeyed|ExistenceKeyed) + EvictionPolicy (ScopeExit|SpacePacked) REQUIRED by construction — rule 3 unwritable-bad. - LadderVerdict (total): AuthoredDuplication | Discharged | RefusedNoProvider | RefusedScopeTooNarrow | RefusedExistenceKeyed (the build-if-absent #6352 wall) | RefusedUnmodeledWorldRead | ExemptFreshEffect | AcceptedBelowCostFloor | AcceptedSingleRecompute. - Declared-emergent frames obligate UP FRONT: retry frames derive checkpointing, unbounded-sibling frames (server loops, CI-runs-over-time) derive persistent caches — prepare-before-demand. 12/12 witnesses green by execution; each test is one cell of the table, fixtures mirror the live worked example (workflow/jobs/steps, retry, fleet). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Docs crystallization: derived materialization placement into roadmap ④ + the state x decision table - roadmap_authority ④ (ROADMAP.md regenerated via main_wet, drift-consistent): the nested-frame law, plurality cells, declared-emergent prepare-up-front, keying/staleness walls, the caching-completion un-shelve, the ①-vs-§2 cross-run-caching contradiction subsumed into per-node derived verdicts, and the v1.compiler.ownership §3-convergence row. - duplicate-work design doc: the state x decision table as the qualifier's final verdict logic + the ownership consolidation subsection. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Consolidation plan: realization + materialization + ownership are one law Operator synthesis (2026-07-09): Materialization = the verdict vocabulary; ownership = the verdict computation at the eval frame; realization = provider selection discharging the verdict at each frame's carrier. Share's handler is layer-aware — the cheapest cache is a REFERENCE (same program + ref-capable carrier + readonly => point-at-it, no copy, no destroyed memory), demoting through HAMT structural share / process memo / artifact / CAS, and demotion must be priced, never silent (#6249 clone-fallback is the receipt). Census of every hand-rolled instance found in-tree (16 rows): each named with its frame, ladder cell, action, and dissolution trigger — ownership (consolidated), Rc/HAMT (reference tier), clone-fallback (demotion wall), ParseTable + cached_stage (C5 grounding), M1 memo / intern / assumed-green / sccache / cargo-cache (C2 provider rows), PROCESS_RESOLVE_STORE (C3 = the 9GB scope fix), build-if-absent (landed as the keying wall), resolve-cache #5789 (rule-4 requirements), recompute-trace (state-4 finding source). Peers kept distinct: affected-set, Independence/Placement, mutable state. Sequence C1-C5 with C1 = this PR (ladder + live CI gate). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Ownership refactored onto materialization: the value tier, tested as providers First consolidation increment (operator-directed): v1.compiler.ownership's decision IS provider selection at the value grain, now expressed in the ladder and proven by mirror witnesses. - ProviderTier: Reference | Copy | Memo{keying} | Artifact{keying} | Cas{keying} — keying folded INTO store tiers, so an existence-keyed reference is unwritable BY SHAPE (§5 construction); frame obligations at isolation LCAs are dischargeable only by store tiers (a reference cannot cross an isolation boundary — witnessed). - Value tier: AccessMode (Read|Consume|Carry|Project), ValueUsage, value_materialization -> ValueDead | ValueMoved | ValueSharedByReference | ValueDemotedToCopy (PRICED, requires a declared CopyTier provider — the #6249 silent clone-fallback made refusable) | ValueRefusedAffine | ValueRefusedNoCarrierProvider. - Faithful to v1's THREE distinct plurality readings, kept separate: take_count = semantic_consumer_count (Consumed only, affine axis); value_access_plurality = binding_fan_out (Carry excluded, reference axis); borrow_count = whole_value_borrow_count (Read+Carry not Project, movability). - Fold-accumulator case mirrored at conservative pre-FoldAccUnwrap semantics: carries excluded from plurality but blocking the move -> demoted-to-copy. 23/23 witnesses green (14 frame cells + 9 value cells). Corpus-level equivalence against live v1 ownership folds = the C4 receipt. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * WIP: Duplicate Computation/Materialization * G: exhaustive cache/share census (52 mechanisms, 6 groups) + C0 catalog-grounding of provider rows + forward-wiring walls The sweep found the tree already models caches: extdeps/cache cache_catalog (CacheInterfaceCatalogFacts, 9 rows) + extdeps/realization carrier facts + the warm==cold purity oracle. C0 grounds CacheProvider on that catalog (provider_from_catalog projection; CacheProvider.id -> CacheInterfaceId) so the ladder and the catalog never restate each other's half. Three forward walls (demands-from-DependencyView, hand-cache shape lens, live-provider-or-red) make new caches born as provider rows only. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * C0: provider rows derive from the cited cache catalog (extdeps/cache/materialization.dag projection) Operator ruling 2026-07-09: extdeps stays the ground-truth/interface layer; every USE of a cache goes through materialization so memo never gets re-invented. provider_from_catalog is the only door from a catalog row into the ladder: keying/tier/eviction derived from cited facts (mechanism->class: InProcess=>ScopeExit; Ttl/Lru/SizeBounded=>SpacePacked; Never/Manual outside a process => typed ProjectionRefused, counted by the new enrolled witness; HandAuthored/prefix-fallback keys => ExistenceKeyed until key-completeness upgrades them). CI sccache row now derived, hand-typed tier facts deleted, 6/6 + 23/23 witnesses green. 10b re-scoped: v1-internal caches = inventory only (retire with seed); ParseTable/cached_stage promoted to next. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Fail-closed hardening: RefusedNatureConflict arm, weakest-claim nature default, UsesStep silence counted, ladder cells enrolled in CI The complexity-lens false-clean lesson applied to this gate before anyone relies on it: (1) heterogeneous natures on one identity were first-wins — now a typed RefusedNatureConflict (24th ladder cell, RED witness); (2) an unrostered run-step claimed PureComputation — now IdempotentEffect, the weakest claim (verdict-identical, no purity overclaim; a misdeclared WorldRead can no longer be legally memoized by default); (3) UsesStep foreign actions produced silence — now a pinned denominator (==9; a new foreign action must consciously bump it); (4) the ladder's 24 cells enroll on the CI surface so the law's own witnesses gate every run. 24/24 + 7/7. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * M1 ratchet bump: declared resolve count 1 -> 3 (gate's first live catch — enrollment pays counted resolves) Run 29058798771: enrolling the ladder witness file moved resolves_total 1 -> 3 (each enrolled entry file pays one closure resolve against the shared index) and the receipt gate redded that exact run — the designed semantics, receipted in the note. Bump acknowledges the two witness-entry resolves as declared debt; M2 (one closure resolve per roots set) ratchets back toward 1. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Retire the M1/M2 labels from carriers (no-codename rule; collided with DESIGN's floor-memoization M1/M2) Step renamed to 'Floor resolve receipt gate (declared cold-resolve count)'; notes and the gate echo now say what the things are: the counted cold-resolve receipt, and the shared-resolve rewire (one closure resolve per source-roots set). Naming-retirement note left on the carrier. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * Run-step natures pinned as counted claims, not affordances Operator ruling 2026-07-10: hardcoded shell scripts get no affordances — run-steps are supposed to be modeled as typed intents and emitted (shell-emission-model slice 4 covers ci_workflow RunSteps). Until that slice lands, every nature in job_run_demands is a claim (rostered FreshEffect / weakest-default IdempotentEffect), never derived from a modeled effect row. Count the whole claimed-nature surface (16 by execution) and pin it in the enrolled witness: a new raw-script step must consciously bump the pin, and slice-4 migration becomes a countable ratchet ending at zero, where the pin becomes a wall. Rejected alternative recorded in the carrier note: a per-step declared-natures roster (a second parallel ledger over scripts the shell-emission plan already governs). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Regen DESIGN.md after main merge (duplicate-work row + #6373 body-lowering row coexist) Conflict was both sides editing the open-threads list in design_document.dag: ours added the duplicate-work thread, main's #6373 rewrote body-lowering to Stages 1-3 LANDED. Resolution keeps both rows in the .dag authority; DESIGN.md regenerated from it, never hand-edited. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Inferred materialization increment 1: floor demand ledger + receipt-or-red gate Running IS enrolling: the interpreter ledgers every keyed pure call and every InterpContext absorbs its totals into a process accumulator on Drop — by construction, no eval path escapes the receipt. claim_executor writes target/floor-materialization-receipt.txt at walk end; trace defaults ON in the executor, and an explicit =0 zeroes keyed_calls which the gate refuses. Gate arms this push (all verified under dash from the emitted ci.yml): receipt missing/malformed/keyed_calls=0 fail closed. Exact pins for unkeyed_calls and duplicated_keys land from the FIRST CI receipt (the resolve gate's measure-then-pin path) — unkeyed is known nonzero on the floor (count_matching takes a predicate closure; closures are the one disclosed identity-less class, dissolve-on captured-env content identity). Unit witness green: resolve -> eval -> ctx Drop -> absorb -> drain-once. Step addition bumped the claimed-natures pin 16 -> 17 consciously. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Review fix: drop the racy drain-once assertion from the receipt unit test opus-4-7 finding on #6441: under plain cargo test (still the documented runner) tests share a process, the env latch is OnceLock-sticky, and sibling ctx drops could absorb between the two takes — making the drain-empty assertion racy. Absorb-on-Drop stays asserted (monotone under concurrent absorbs: siblings only ADD); drain-once is Option::take by construction, not asserted through the shared global. Comment states the sharing semantics explicitly. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ncident, argued serially with receipts) (#6469) * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * WIP: Duplicate Computation * Analysis spine: spine_receipt composes dependency_view + materialize (increment 1) The Rc-safe half of run ≜ realize ∘ materialize ∘ dependency_view: compose the two catamorphisms that need no interpreter thread-safety. - src/v2/std/spine.dag: level-profile fold over the DependencyView DAG → critical_path_depth (longest dependency chain / the reduce spine) + independence_width (max nodes at one level / what parallelizes to hw width). spine_receipt emits the §9 numbers by COMPUTATION, not assertion, joining materialize's content-hash Share/dedup counts. - Discriminating RED (§9): single chain A→B→C → depth 3 / width 1 (serial); root over N independent leaves → depth 2 / width N. - Witnesses 9/9 (spine) green by execution; materialize 6/6 unchanged. The RUNNER (wall-clock parallel execution, §8 steps 3-4) stays increment 2+, gated on interpreter Rc→Arc (v1 interp is !Send). This increment measures the achievable width/critical-path floor the runner targets; it does not fabricate wall-clock N×. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Unbundle recompute-trace interpreter extension from the analysis-spine PR CI root cause: the recompute-trace extension edited v1_interpreter.rs, which hosts the atom_identity_hash/hash_combine fnv1a64 intrinsics that dag_collect_fingerprint_witness executes. That correctly invalidated the re-verify 'assumed-green node-frontier' SKIP of that witness (cone-touch, fail-closed) — but the witness is a pre-existing 806s (13min) fnv1a64 whole-corpus walk, so un-skipping it blew the 30-min floor budget. Main stays green only because its interpreter cone is untouched (witness stays skipped). The interpreter extension is a separable, env-gated (GUNBC_RECOMPUTE_TRACE=1), READ-mode diagnostic — not a dependency of the spine/materialize analysis substrate (pure .dag, content_hash-based). Reverting v1_interpreter.rs to main restores byte-identical interpreter content → the fingerprint witness returns to assumed-green SKIP → floor budget restored. The recompute-trace extension lands as its own follow-up PR where the fingerprint-witness re-run is expected. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Materialization ladder: the state x decision law, as executable witnesses The operator's 4-rule business logic generalized on one axis: the decision is a function of WHEN the redundancy is knowable and WHETHER what was knowable was prepared for. Errors fire only on knowable-but-unprepared; genuine emergence and declared triviality are typed acceptances, never silence. std.materialization_ladder (dag/std): - Frame/FrameKind: SharedStateFrame | IsolatedChildrenFrame | ReplayedFrame{attempts} | UnboundedSiblingsFrame — the nested-scope axis; a 'run' at any layer is a frame, never a different kind of thing. - DemandNature: Pure | IdempotentEffect | WorldRead{envelope_declared} | FreshEffect — the D3 effect gate; TTL = unmodeled dependency XOR declared staleness, never an eviction knob. - CacheProvider: keying (ContentKeyed|ExistenceKeyed) + EvictionPolicy (ScopeExit|SpacePacked) REQUIRED by construction — rule 3 unwritable-bad. - LadderVerdict (total): AuthoredDuplication | Discharged | RefusedNoProvider | RefusedScopeTooNarrow | RefusedExistenceKeyed (the build-if-absent #6352 wall) | RefusedUnmodeledWorldRead | ExemptFreshEffect | AcceptedBelowCostFloor | AcceptedSingleRecompute. - Declared-emergent frames obligate UP FRONT: retry frames derive checkpointing, unbounded-sibling frames (server loops, CI-runs-over-time) derive persistent caches — prepare-before-demand. 12/12 witnesses green by execution; each test is one cell of the table, fixtures mirror the live worked example (workflow/jobs/steps, retry, fleet). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Docs crystallization: derived materialization placement into roadmap ④ + the state x decision table - roadmap_authority ④ (ROADMAP.md regenerated via main_wet, drift-consistent): the nested-frame law, plurality cells, declared-emergent prepare-up-front, keying/staleness walls, the caching-completion un-shelve, the ①-vs-§2 cross-run-caching contradiction subsumed into per-node derived verdicts, and the v1.compiler.ownership §3-convergence row. - duplicate-work design doc: the state x decision table as the qualifier's final verdict logic + the ownership consolidation subsection. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Consolidation plan: realization + materialization + ownership are one law Operator synthesis (2026-07-09): Materialization = the verdict vocabulary; ownership = the verdict computation at the eval frame; realization = provider selection discharging the verdict at each frame's carrier. Share's handler is layer-aware — the cheapest cache is a REFERENCE (same program + ref-capable carrier + readonly => point-at-it, no copy, no destroyed memory), demoting through HAMT structural share / process memo / artifact / CAS, and demotion must be priced, never silent (#6249 clone-fallback is the receipt). Census of every hand-rolled instance found in-tree (16 rows): each named with its frame, ladder cell, action, and dissolution trigger — ownership (consolidated), Rc/HAMT (reference tier), clone-fallback (demotion wall), ParseTable + cached_stage (C5 grounding), M1 memo / intern / assumed-green / sccache / cargo-cache (C2 provider rows), PROCESS_RESOLVE_STORE (C3 = the 9GB scope fix), build-if-absent (landed as the keying wall), resolve-cache #5789 (rule-4 requirements), recompute-trace (state-4 finding source). Peers kept distinct: affected-set, Independence/Placement, mutable state. Sequence C1-C5 with C1 = this PR (ladder + live CI gate). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * Ownership refactored onto materialization: the value tier, tested as providers First consolidation increment (operator-directed): v1.compiler.ownership's decision IS provider selection at the value grain, now expressed in the ladder and proven by mirror witnesses. - ProviderTier: Reference | Copy | Memo{keying} | Artifact{keying} | Cas{keying} — keying folded INTO store tiers, so an existence-keyed reference is unwritable BY SHAPE (§5 construction); frame obligations at isolation LCAs are dischargeable only by store tiers (a reference cannot cross an isolation boundary — witnessed). - Value tier: AccessMode (Read|Consume|Carry|Project), ValueUsage, value_materialization -> ValueDead | ValueMoved | ValueSharedByReference | ValueDemotedToCopy (PRICED, requires a declared CopyTier provider — the #6249 silent clone-fallback made refusable) | ValueRefusedAffine | ValueRefusedNoCarrierProvider. - Faithful to v1's THREE distinct plurality readings, kept separate: take_count = semantic_consumer_count (Consumed only, affine axis); value_access_plurality = binding_fan_out (Carry excluded, reference axis); borrow_count = whole_value_borrow_count (Read+Carry not Project, movability). - Fold-accumulator case mirrored at conservative pre-FoldAccUnwrap semantics: carries excluded from plurality but blocking the move -> demoted-to-copy. 23/23 witnesses green (14 frame cells + 9 value cells). Corpus-level equivalence against live v1 ownership folds = the C4 receipt. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * WIP: Duplicate Computation * WIP: Duplicate Computation/Materialization * G: exhaustive cache/share census (52 mechanisms, 6 groups) + C0 catalog-grounding of provider rows + forward-wiring walls The sweep found the tree already models caches: extdeps/cache cache_catalog (CacheInterfaceCatalogFacts, 9 rows) + extdeps/realization carrier facts + the warm==cold purity oracle. C0 grounds CacheProvider on that catalog (provider_from_catalog projection; CacheProvider.id -> CacheInterfaceId) so the ladder and the catalog never restate each other's half. Three forward walls (demands-from-DependencyView, hand-cache shape lens, live-provider-or-red) make new caches born as provider rows only. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * C0: provider rows derive from the cited cache catalog (extdeps/cache/materialization.dag projection) Operator ruling 2026-07-09: extdeps stays the ground-truth/interface layer; every USE of a cache goes through materialization so memo never gets re-invented. provider_from_catalog is the only door from a catalog row into the ladder: keying/tier/eviction derived from cited facts (mechanism->class: InProcess=>ScopeExit; Ttl/Lru/SizeBounded=>SpacePacked; Never/Manual outside a process => typed ProjectionRefused, counted by the new enrolled witness; HandAuthored/prefix-fallback keys => ExistenceKeyed until key-completeness upgrades them). CI sccache row now derived, hand-typed tier facts deleted, 6/6 + 23/23 witnesses green. 10b re-scoped: v1-internal caches = inventory only (retire with seed); ParseTable/cached_stage promoted to next. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Fail-closed hardening: RefusedNatureConflict arm, weakest-claim nature default, UsesStep silence counted, ladder cells enrolled in CI The complexity-lens false-clean lesson applied to this gate before anyone relies on it: (1) heterogeneous natures on one identity were first-wins — now a typed RefusedNatureConflict (24th ladder cell, RED witness); (2) an unrostered run-step claimed PureComputation — now IdempotentEffect, the weakest claim (verdict-identical, no purity overclaim; a misdeclared WorldRead can no longer be legally memoized by default); (3) UsesStep foreign actions produced silence — now a pinned denominator (==9; a new foreign action must consciously bump it); (4) the ladder's 24 cells enroll on the CI surface so the law's own witnesses gate every run. 24/24 + 7/7. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * M1 ratchet bump: declared resolve count 1 -> 3 (gate's first live catch — enrollment pays counted resolves) Run 29058798771: enrolling the ladder witness file moved resolves_total 1 -> 3 (each enrolled entry file pays one closure resolve against the shared index) and the receipt gate redded that exact run — the designed semantics, receipted in the note. Bump acknowledges the two witness-entry resolves as declared debt; M2 (one closure resolve per roots set) ratchets back toward 1. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Retire the M1/M2 labels from carriers (no-codename rule; collided with DESIGN's floor-memoization M1/M2) Step renamed to 'Floor resolve receipt gate (declared cold-resolve count)'; notes and the gate echo now say what the things are: the counted cold-resolve receipt, and the shared-resolve rewire (one closure resolve per source-roots set). Naming-retirement note left on the carrier. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * Run-step natures pinned as counted claims, not affordances Operator ruling 2026-07-10: hardcoded shell scripts get no affordances — run-steps are supposed to be modeled as typed intents and emitted (shell-emission-model slice 4 covers ci_workflow RunSteps). Until that slice lands, every nature in job_run_demands is a claim (rostered FreshEffect / weakest-default IdempotentEffect), never derived from a modeled effect row. Count the whole claimed-nature surface (16 by execution) and pin it in the enrolled witness: a new raw-script step must consciously bump the pin, and slice-4 migration becomes a countable ratchet ending at zero, where the pin becomes a wall. Rejected alternative recorded in the carrier note: a per-step declared-natures roster (a second parallel ledger over scripts the shell-emission plan already governs). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Regen DESIGN.md after main merge (duplicate-work row + #6373 body-lowering row coexist) Conflict was both sides editing the open-threads list in design_document.dag: ours added the duplicate-work thread, main's #6373 rewrote body-lowering to Stages 1-3 LANDED. Resolution keeps both rows in the .dag authority; DESIGN.md regenerated from it, never hand-edited. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * Inferred materialization increment 1: floor demand ledger + receipt-or-red gate Running IS enrolling: the interpreter ledgers every keyed pure call and every InterpContext absorbs its totals into a process accumulator on Drop — by construction, no eval path escapes the receipt. claim_executor writes target/floor-materialization-receipt.txt at walk end; trace defaults ON in the executor, and an explicit =0 zeroes keyed_calls which the gate refuses. Gate arms this push (all verified under dash from the emitted ci.yml): receipt missing/malformed/keyed_calls=0 fail closed. Exact pins for unkeyed_calls and duplicated_keys land from the FIRST CI receipt (the resolve gate's measure-then-pin path) — unkeyed is known nonzero on the floor (count_matching takes a predicate closure; closures are the one disclosed identity-less class, dissolve-on captured-env content identity). Unit witness green: resolve -> eval -> ctx Drop -> absorb -> drain-once. Step addition bumped the claimed-natures pin 16 -> 17 consciously. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Review fix: drop the racy drain-once assertion from the receipt unit test opus-4-7 finding on #6441: under plain cargo test (still the documented runner) tests share a process, the env latch is OnceLock-sticky, and sibling ctx drops could absorb between the two takes — making the drain-empty assertion racy. Absorb-on-Drop stays asserted (monotone under concurrent absorbs: siblings only ADD); drain-once is Option::take by construction, not asserted through the shared global. Comment states the sharing semantics explicitly. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization * WIP: Duplicate Computation/Materialization --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Brian Searls <11205878+briansrls@users.noreply.github.com>

…ane PR; P0 merged via #6663) (#6738) * P2 pure-spec half: content-addressed artifact store — the first read eviction field std.artifact_store (witness-realization plan P2, pure fold spec; host transport is the second half and this spec is its single authority): - ArtifactKey { closure_digest, emitter_identity, target_language, toolchain } -> artifact_key_hash via std.content_hash (one hash authority) — keying is ContentHash BY SHAPE; no mtime/existence input exists, so ExistenceKeyed is unwritable here (the #6352 wall by construction) - store_over_provider READS CacheProvider.eviction and refuses construction over a non-SpacePacked provider (typed StoreRefusedEviction) — the eviction field's first behavioral consumer (memory-control audit F5) - store_get bumps recency; store_put packs to budget by least-recent eviction with every eviction COUNTED in StorePutReceipt (refuse-or-count, never widen) - budget stays a typed parameter beside the row for now: EvictionClass.budget carries cited upstream policy PROSE in extdeps rows (two concepts in one field); dissolve-on recorded in the module note for the variant split Witnesses (all green by execution, current binary, SubstrateInputsOnly): - store_construction_reads_eviction_holds (ScopeExit provider -> typed refusal) - store_hit_and_stale_never_served_holds (toolchain mutation -> new key -> miss) - store_budget_evicts_least_recent_counted_holds (touch ka, put kc -> [kb] evicted, within budget, survivor still hits) - store_no_eviction_under_budget_holds Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * P2 host-transport half: filesystem realization of the artifact store extdeps.realization.artifact_store_fs — one transport handler bound to the pure spec (std.artifact_store stays the single authority; a remote CAS is the other handler). The keyed path ENCODES identity (store_root/<hash>.artifact), so presence-at-path is a ContentKeyed hit for exactly that identity — not the #6352 existence-keying (which keyed output presence over unhashed inputs). SCOPE, honest and named: put/get only. The cited Filesystem service exposes Write/Read but no Delete/List, so SpacePacked budget enforcement on the persistent tier is unrealizable until those operations are added — until then the disk tier grows unbounded and the in-process fold is the only packed tier. Recorded in the transport note as the next rung, never a silent widen. Witnesses green by wet execution (artifact observed on disk at its hash path): - artifact_fs_roundtrip_holds - artifact_fs_mutated_input_misses_holds (stale-never-served on the persistent tier) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Plan doc: P2 status (both halves landed, green by execution; two named gaps) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * P1 (code): ObservePeakResidentAtSubject + the space algebra — memory becomes observable - RealizationMeasureEffect gains ObservePeakResidentAtSubject (audit F2: the effect coproduct had exactly one variant, time; a Measured space fact was unproducible by construction) - space algebra as time's DUAL (audit F4): space_measure_seq = max (sequential steps release), space_measure_par = add (concurrent steps co-reside), plus space_measure_list_seq; duality note records why the old sum was wrong - fleet_intent receipt rollup: space summed over a sequential list -> now the peak (the parallel rule was applied to the serial axis); keystone witness FLIPPED to assert max AND assert != sum, so the old contract cannot silently return - host physics: observed_peak_resident_bytes builtin (VmHWM in bytes) — registered in v1.compiler.method builtins, realized in the hand-maintained interpreter, FAIL-CLOSED when the host cannot report it (a fabricated 0 would be a Measured lie, DESIGN section 5) - peak_resident_measured_witness_test: the plan's P1 ACCEPT — the first CostAccount.space with basis Measured produced BY EXECUTION — declared ReadsLiveTree honestly (reads /proc through a builtin, the classifier's declared blind-spot class); plus the seq=max/par=add RED control Receipts land in the follow-up commit with the regen-synced stage0 and the rebuilt binary (pipeline in flight); the pure-.dag half already compiles clean through the P0 wall. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * P1 (receipts): regen sync + green-by-execution verification Regen-synced stage0 (the observed_peak_resident_bytes registration reaching the generated method registry) after rebuilding the regen tool from merged main (its compiled-in roster predated the Wave-2 crate-layout file; two-generation discipline: build committed -> regen -> build). Receipts on the rebuilt binary: - peak_resident_measured_holds -> true : the FIRST CostAccount.space with basis Measured produced by execution (plan P1 ACCEPT; audit F2 discharged at the witness grain) - space_seq_is_peak_not_sum_holds -> true (seq=max=5, par=add=8, list_seq=5) - witness_space_rolls_up_across_receipts -> true FLIPPED (asserts max AND != sum; 1024/2048 samples discriminate) - P0 regression: diagnostics_witness record_field_walls suite exit 0 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * P3 MVP: emit-on-demand at the source grain — second run pays zero emit The emit cache wired through the P2 store, keyed on INPUT identity: inferred_tree_digest x emitter identity x target x artifact kind — the same Hash=ContentHash authority the fnv1a64 convergence landed, composed by artifact_key_hash. Cold: miss -> pure emit (v2 emit = serialize_target o translate) -> put. Warm: served from the store WITH NO EMIT CALL IN THE ARM, and the served bytes asserted equal to a fresh emit (the agreement receipt — plan P3 ACCEPT (a) at the source grain). Witnesses green by wet execution (emitted rust_add source observed on disk at its hash path): - emit_source_store_cold_then_warm_holds - emit_source_store_mutated_emitter_misses_holds (emitter version bump -> new key -> miss; stale emitter output never served) - emit_source_store_provider_gate_holds Rung remainder (named): the native-artifact tier (zero BUILD — needs FLAG A, the hermetic pinned-toolchain ruling, and the parse-census first customer) and enrollment of the wet witnesses. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Design: effect grants over namespaces — hermetic/wet dissolved into (frame x verb x subtree) Operator rulings captured (2026-07-16): "hermetic" conflates four axes (input closure / output reach / handler binding / selection eligibility — the doc's own state-space-conflation failure mode, effect-system edition), and the fix must reuse NAMESPACES directly rather than mint a "universe" vocabulary. An effect target is a position in a containment tree that already exists (filesystem, URI, proc, code names); permission is a grant of (verb x subtree) on a frame; admissibility is the same prefix relation the naming lane walks — effects become the containment structure's fourth consumer, not a fork. Convergence map covers the proto-envelopes already in-tree (the hand-rolled workspace_root path gate, std.resources.ResourceHandle, AuthScope, LiveTreeDisposition) — all dissolve into derived projections. FLAG A reframed: build admissibility becomes the first grant row (Read within closure + pinned toolchain; Write within own workspace), scaffold-marked, dissolving into the P-B enforcement seam — a row, not a mode exception. Bound into the doc graph via the witness-realization plan's FLAG A section (orphan wall re-verified green). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Effect-grants design: the containment law (target lifecycle) + fix inherited orphan Design refinement (operator, 2026-07-16): a write is frame-contained iff (target within a frame-controlled namespace) AND (target lifecycle within frame lifecycle) — the lifecycle conjunct is what the old "hermetic" intuition was actually about, graded on the section-5 construction/validation axis (LifecycleByConstruction: ephemeral container fs, netns-scoped loopback receiver — persistence unwritable past the frame; LifecycleByConvention: /tmp scratch + cleanup). Four named acceptance cases added up front so the model cannot mislead: netns loopback = contained; container write = contained by construction; /tmp scratch = contained by convention only (the artifact store witnesses' honest current grade); BMC POST / repo-tree write = wet under any grade. Also: bind docs/plans/emitted-crate-partition-design.md into the doc graph (frontier.dag provenance row) — it merged orphaned on main (8322580) and the doc-reachability wall was red on main; wall re-verified true here, and frontier.dag entry-compiles clean. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Record duplicate-computation direction: one ComputationIdentity, N grains (execution-frame is the gap) Operator direction (2026-07-16): duplicate-computation detection is not shell-specific — "same inputs + deterministic process → same content-identity → the second is duplicate work" is the whole law; the surface it's read on is a realization axis, not the concept. Records the three detection grains against ONE ComputationIdentity: - within-script (argv/ShellWord) — v2.lens.duplicate_computation, dissolves in (§7) - within-graph (content_hash over Node subtrees) — v2.std.materialize MVP - within-run execution-frame — the UNCOVERED grain, where the ~275s double-resolve lives (two compile_to_resolved calls in one v1-seed claim_executor process, invisible to both peers). Ladder-classed: shared-state frame ⇒ AuthoredDuplication ⇒ REWIRE (not cache), distinct from the cross-run isolation-boundary store obligation — same identity, remedy by frame. Direction: the general detector must reach the execution-frame grain; the argv lens then dissolves into it and the within-graph MVP extends down — one detector, the surfaces its realizations. Doc is the roadmap ④'s linked carrier; orphan wall re-verified green. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Fix §5 fail-open: eval-call memo served stale world-reads (effect-dispatch odometer) Found via the artifact-store List-after-Delete witness: the interpreter's eval-call memo treated "no declared `uses` clause" as pure — and ZERO corpus funcs declare `uses`, so every effectful named func was memo-eligible. A func called twice with equal args in one eval served the FIRST result — e.g. a Filesystem.List after a Delete returned the pre-delete listing. That is a world-read served stale from cache: a silent §5 fail-open in the bootstrap engine, not a witness quirk. Fix (4 edits, transitive by construction): an effect-dispatch odometer on InterpContext, ticked at the single eval_service_call chokepoint; the memo refuses to STORE any call during which the odometer advanced. An observed effect poisons cacheability — exactly the ladder's "FreshEffect/WorldRead is never memoized" law, enforced at the realizer instead of by a vacuous uses-gate. Value::eq stays the sole equality authority; pure calls still memoize. Known residue (named): world-effecting BUILTINS (e.g. observed_peak_resident_bytes, filesystem via the service path already ticks) — the odometer covers service dispatch; a builtin-effect tick is the follow-up if a pure-memoized builtin ever reads the world. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Delete/List on the Filesystem service (operator Q2) — closes P2's persistent-tier gap The cited Filesystem service exposed only Write/Read, so P2's SpacePacked budget enforcement on the on-disk store was unrealizable (named gap in the P2 landing). Adds the two missing operations, keeping the extdeps interface faithful to the real dependency: - filesystem_io.dag: Delete + List operations (List readonly) - file transport gains a verb property (parse: 00_core file_transport_node + 02_parse parse_file_fields thread `verb`; interpreter: dispatch_file honors verb "delete"/"list" via remove_file / sorted read_dir). Absent verb keeps the original content-param convention (write iff `content` param, else read). - artifact_store_fs: artifact_fs_delete / artifact_fs_list wrappers - witness: artifact_fs_delete_then_misses_holds — put -> listed -> delete -> miss -> unlisted, green by wet execution (also the discriminating input that surfaced the memo fail-open fixed in the parent commit) Regen-synced (v1_compiler_parse.rs, v1_std_core.rs). Persistent-tier eviction (the SpacePacked enforcer using List+Delete) is the follow-up now that the ops exist. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * P4 v0: realize packing — width against MEASURED peaks, refuse-not-fabricate The memory-safe packing half of `realize` (spine FLAG D), consuming P1's measured space so the exit-137 failure mode (memory-blind packing OOMs) becomes arithmetic. std.realize_pack: - MeasuredPeak = PeakMeasured | PeakUnknown, derived from CostAccount.basis (P1's Measured space vs a Predicted guess) - HostBudget = BudgetReadable | BudgetUnreadable — the typed read - realize_pack_width -> RealizeVerdict = PackedWidth | MaturationReserve | BudgetRefused: * measured peak + readable budget -> PackedWidth min(independence, budget/peak), reusing realization_width.memory_bounded_shard_count (20% headroom reserved — the maturation-reserve margin; 100/25 packs to 3, not 4) * unknown peak -> MaturationReserve width-1 (first-run subject runs alone, its receipt converts it next round — the governor's admission logic, modeled) * unreadable budget -> BudgetRefused, NEVER the conservative_fallback_width fabrication (realization_width.dag:109 — the live §5 absorbing-fallback the memory-control audit flagged: budget unreadable answered with a number) Witnesses (6, green by execution): packs within budget; capped by independence; never exceeds budget (the exit-137 arithmetic control, W*peak <= budget by construction); unknown-peak -> maturation reserve; unreadable-budget -> refuses; measured-vs-predicted peak discrimination. Remaining for P4: the width-1 fabrication in memory_aware_spawn_width is now superseded for the measured path; wiring realize_pack into the executor's per-runnable scheduling (consuming real getrusage receipts) is the integration step. The math + refusal discipline land here first, green. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Design: space complexity — the dual of the time/termination analysis Operator direction (2026-07-16): extend the complexity analysis to space (the gap). Derived, not measured (§4: bounded/forward ⇒ checked not discovered — measuring peak RSS is the reflection-evidence cop-out). Two readings: asymptotic (O(n)) AND concrete derived bytes ("known/defined to these bounds"). Underivable = counted frontier now, hard error later (fail-closed ratchet). The key finding that makes it a dual, not a new analysis: ComplexitySummary already carries work/span/output_size as CostExpr (axis-agnostic), and CostSum is a fold — time SUMS over iterations, space takes the MAX (sequential residency releases). That is exactly P1's space_measure_seq=max lifted to CostExpr. So peak_space = space_of(work) — one transform swapping sum/max, accumulator term reusing output_size, recursion depth bounded by the SAME DescentEvidence that proves termination. Two readings of one descent structure (§2). Folds five threads onto one page: audit F2/F3 (space unobserved/underived), P1 (the residency algebra — reused as the transform), P4 (re-point pack input from MeasuredPeak to derived bound; retire the reactive governor), the 2026-07-12 ruling (derived not authored literals), the allocation model (the derived per-witness bound IS the up-front allocation; Σ ≤ budget = GUARANTEED mode at witness grain). Sequence: witness-grain concrete space first (closed closures), then asymptotic, then re-point P4, then ratchet the frontier to error. Bound into the doc graph via the witness-realization P1 note (which this supersedes as the keystone); orphan wall green. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Space complexity increment 1a: peak-working-set derivation, the dual of time The space-complexity analysis's core, as the structural dual of the existing time cost-expr — green by execution, no rebuild (interpreted from source). In src/v1/complexity.dag, beside the cost algebra it duals: - space_of(work: CostExpr) -> CostExpr : the transform. Sequential steps RELEASE so time's CostAdd becomes CostMax; concurrent steps CO-RESIDE so time's CostMax becomes CostAdd (P1's space_measure_seq=max / space_measure_par=add, lifted to CostExpr); a fold's CostSum collapses to its body's peak (iterations release). CostUnknown passes through fail-closed. - fold_peak_space(body_peak, output_size) : adds the accumulator/output_size term. - eval_cost_expr_concrete / eval_size_expr_concrete : lower a closed expr to concrete bytes (Absent on any unknown/free var — refuses, never fabricates). Witnesses (6, green by execution, --source-root src/v1): - reducing_fold_is_constant_space: O(n) time fold -> O(1) space (the headline) - sequential_releases_max_not_sum / parallel_coresides_add_not_max: the P1 duals - concrete_derived_bytes: closed expr -> exact bytes (40) - fold_that_builds_is_linear_space: output_size 100*4 + body 4 -> 404 bytes - unknown_cost_is_fail_closed_space: CostUnknown -> unknown space, concrete refuses (the SpaceBoundUnknown counted bottom; hard-error ratchet is the later stage) Design: space-complexity-design.md §2. Not floor-enrolled yet — imports v1.compiler.complexity so needs --source-root src/v1 (like the v1-internal tests); enrollment (host bin or v1-root discovery) is the follow-up, alongside the ComplexitySummary.peak_space field (38 construction sites) and the P4 re-point onto the derived bound. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Space complexity increment 2: asymptotic reading (O(n) etc.), space order below time The second reading the operator asked for — asymptotic space class — reusing std.induction's CostBound/cost_poly on space_of(work). Green by execution, no rebuild (interpreted). In src/v1/complexity.dag: - cost_expr_degree(e) -> Int? : polynomial degree. CostSum over a size var = +1 factor of n; seq/par take MAX degree; products ADD; bare log = degree 0 (sub-linear); Absent = frontier (fail-closed, never a fabricated degree). - space_asymptotic_bound(work, param) = degree_to_bound(cost_expr_degree(space_of(work))) -> ConstantBound | cost_poly(...) | ForeverBound. time_asymptotic_bound is the same on the raw work, for the side-by-side. Because space_of collapses the CostSum a reducing fold's TIME carries, space order is <= time order by construction. Witnesses (4, green): - reducing_fold: O(n) TIME (degree 1), O(1) SPACE (ConstantBound) - nested_reducing_loop: O(n^2) time, O(1) space (both CostSums collapse — the striking dual) - parallel_region: constant space - frontier_maps_to_forever_bound: CostUnknown -> ForeverBound, not a degree Design: space-complexity-design.md §3. Increments remaining: wire peak_space into ComplexitySummary (rebuild-gated, 38 sites), re-point P4, ratchet the frontier. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Wire space-complexity into ComplexitySummary (peak_space + concrete fill) Extend the landed space-complexity analysis (space_of / fold_peak_space / eval_cost_expr_concrete) into the compiler's per-function summary. - ComplexitySummary gains an OPTIONAL peak_space: CostExpr?. Optional so the P0 field wall (04_infer) skips it: all ~38 existing construction sites stay unchanged, an absent peak_space = the SpaceBoundUnknown counted frontier (fail-closed staging). Verified: complexity.dag entry-compiles 0 diagnostics. - Derived once at the finalized per-function summary (get_or_compute_summary's `simplified`) via derive_peak_space: space_of(work) for a scalar result, additive fold_peak_space(space_of(work) + output_term) for a collection result (the `result` output_size entry). Intermediate/error/external/seed summaries leave peak_space absent by design. - cost_account_space_from_summary(summary, size_env) -> ByteSize? fills CostAccount.space (basis Derived): eval peak_space at a closed size_env to a concrete ByteSize; absent peak_space or an underivable expr returns none, never a fabricated bound. - New witness complexity_summary_space_witness_test.dag (4 test fns, green by interpretation): derived peak yields expected bytes; absent peak -> none; underivable (CostUnknown) peak -> none (RED control); reducing-fold work is O(1) space regardless of n. Follow-up (out of scope): regen + rebuild-to-seed to activate the compiled realization. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * P4 re-point: pack against the DERIVED bound, not measured peak — seam closed Operator direction (derive, don't measure — §4). std.realize_pack re-pointed off P1's measured peak onto the statically DERIVED space bound: - MeasuredPeak/PeakMeasured/PeakUnknown -> DerivedBound/BoundDerived/BoundUnknown - measured_peak_of(CostAccount) -> derived_bound_of(derived_space: ByteSize?): Present = the derived working-set bound, Absent = SpaceBoundUnknown (the fail-closed frontier). P4 is now decoupled from CostBasis — it consumes the ByteSize? that cost_account_space_from_summary (eeeda2b, the subagent's ComplexitySummary wiring) produces. That closes the seam: ComplexitySummary.peak_space -> space_of derivation -> cost_account_space_from_summary -> ByteSize? -> derived_bound_of -> realize_pack_width. - realize_pack_width / realize_fits_budget take DerivedBound; BoundUnknown -> the width-1 maturation reserve (a not-yet-derivable subject runs alone, not a fabricated number); the refuse-not-fabricate discipline unchanged. Measurement (P1's ObservePeakResidentAtSubject) is demoted to at-most a falsifier, never a scheduler input — the whole memory-control line is now derived, not observed. Witnesses (6, green by execution, interpreted — no rebuild): packs within budget (headroom-reserved 3, not 4); capped by independence; never exceeds budget (the exit-137 arithmetic control); BoundUnknown -> maturation reserve; unreadable budget -> refuses; derived-vs-frontier bound discrimination (the new ByteSize? seam). Remaining for P4: activate in the seed (regen — gated with the space-complexity seed activation) + wire realize_pack into claim_executor per-runnable scheduling. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Regen seed: activate space-complexity peak_space in the compiled compiler regen_stage0 activated the complexity.dag changes (peak_space field + space_of + concrete evaluator, and the get_or_compute_summary derivation) into the generated seed v1_compiler_complexity.rs — the one file that changed (surgical; the other 94 generated files byte-identical). Built on srv1 (128 cores, 4m17s cold / 2m34s rebuild — off the memory-constrained Pi that watchdog-crashes on this crate). regen_stage0 --verify: regen_divergence_count=0 — committed stage0 matches a fresh self-compile, so the byte fixed point holds with peak_space live. The compiled compiler now derives peak_space during real complexity analysis (previously only interpreted from source). Remaining (flagged by the wiring pass): the interpreter RAISES on an omitted optional field while the seed reads None gracefully — latent-safe today (cost_account_space_from_summary is only called on simplified summaries, which always set peak_space), reconcile when a broader consumer lands. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Space classes in ComplexityReport (observable on real code) + floor-classify new witnesses Two things the srv1 corpus validation surfaced. 1. Observability: ComplexityReport gains space_classes: Map<String,String> beside function_classes — build_complexity_report now derives it via classify_complexity(space_of(work)) per function, so the derived SPACE order is observable for every REAL gunbc function (not just synthetic CostExpr witnesses). TopoBuildAcc threads it; empty_complexity_report seeds it. 2. Floor classification (validation caught my new witnesses breaking the hermetic floor — neither a space-complexity logic regression): - The three space witnesses import v1.compiler.complexity, unresolvable in the discovered corpus's roots -> a FATAL resolve halt at entry 46. Excluded from discovery (they run in the v1 lane via --source-root src/v1; host-bin enrollment like diagnostics_witness is the proper follow-up). - artifact_store_fs_witness does real Filesystem.Write -> hermetic refuses (correctly). Excluded from hermetic discovery (wet lane), same pattern as the existing ci_deploy_observed_wet / host_effect_apply wet exclusions. Interpreted entry-compile green; srv1 regen+rebuild+corpus re-validation follows. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Fix the debt srv1 corpus validation caught: residue wildcards + the last wet witness The srv1 corpus run (1900 PASS, and the space-complexity seed activation proven corpus-safe) surfaced 4 failures, all from witnesses I added this session — none a logic regression: - 2 non_fold_residue FAILs: my witnesses + realize_pack used `_ =>` wildcards over closed coproducts (CostExpr, CostBound, RealizeVerdict) — the §4 residue the lens bans. Fixed by making them exhaustive/behavioral: realize_pack + its witness list every RealizeVerdict arm; the space witnesses now assert through the concrete evaluator (eval_cost_expr_concrete, exhaustive Optional) and the polynomial degree (cost_expr_degree) instead of matching CostExpr/CostBound structure — cleaner behavioral tests (sequential releases -> eval 50 not sum 80; parallel co-resides -> eval 80 not max 50) with ZERO wildcards. non_fold_residue_clean_holds -> true (0 unrostered residue, re-verified). - 2 emit_source_store FAILs: real Filesystem.Write in hermetic mode. The path is test/claim/manual/ (not test/manual/), so the existing exclusion missed it; added emit_source_store_test.dag explicitly (wet lane). All rewritten witnesses green by interpretation; final srv1 corpus re-validation follows. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Regen seed: space_classes into the compiled compiler (merged-tree consistency) regen_stage0 on the merged tree (main-merge #6780/#6783 + my space-complexity work) changed exactly one seed file — v1_compiler_complexity.rs — activating space_classes (the ComplexityReport space-order surfacing) in the compiled compiler. main-merge seed was already consistent; this is the clean delta for my complexity.dag change, keeping regen --verify green. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * P4 handoff: executor cutover is a bridge call, not a self-host dependency Records the corrected framing in the plan's P4 section: claim_executor is the terminal bootstrap kernel (not a self-host emit target) and already interprets .dag, so it can call realize_pack through run_in_context_with_args rather than forking the packing law into Rust (§2) or waiting on the 27-module frontier. Seam: surface the derived bound, read the governor's budget, call realize_pack, advisory-first then demote the governor. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Fix compile-clean gate: move space-complexity witnesses out of dag/ tree The three space-complexity witnesses import v1.compiler.complexity (a src/v1 module), but lived in dag/test/claim/. The whole-tree compile-clean gate compiles [dag, src/v2] WITHOUT src/v1, so their imports could not resolve — red as 'module v1.compiler.complexity not found' whenever a src/v1 touch forces the whole-tree baseline. Discovery-exclusion handled the runner but not the gate; a dag/ file simply cannot import from src/v1. Move them to src/v1/test/claim/ (compiled with src/v1, not swept by regen's seed-closure walk, not in discovery scan dirs) and drop the now-dead exclusion substrings. All 6+4+4 test fns pass from the new home via claim_batch --source-root src/v1 --source-root dag --source-root src/v2. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Fix batch-4 gate regressions: extdeps authority anchor + drop redundant frontier binding - extdeps_external_authority_gate: artifact_store_fs.dag (P2, new) was the only extdeps/realization module missing the extdeps_external_authority_anchor every sibling declares. Add it (Https -> the realization dir, matching v1_handler). All 9 realization modules now carry exactly one anchor. - self_host_realized_comparison / cleanup: drop emitted_crate_partition_plan_doc_provenance from frontier.dag. It was a doc-reachability workaround added when the design doc was orphaned; main #6828 now links it from DESIGN.md (line 92), so the row is a redundant duplicate binding (unreferenced). frontier.dag is now byte-identical to main. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Fix self_host_realized_comparison: drop v2 import from relocated witnesses Moving the space witnesses to src/v1/ (to fix compile-clean) put them in regen_stage0's compile surface ([src/v1, dag]) — but they still imported v2.std.live_tree { LiveTreeDisposition, SubstrateInputsOnly } (src/v2, not in regen's roots), so 'regen_stage0 --emit-fresh' failed with unresolved import, and self_host_realized_comparison_reads_real_bytes then couldn't read the fresh-emitted bytes (No such file or directory) -> Bool(false). The v2 import fed only a 'data live_tree_disposition = SubstrateInputsOnly' metadata decl for discovery-based affected-set selection, used in no test fn and moot now that these run manually. Drop the import + decl; the witnesses become pure [src/v1, dag] (std + v1.compiler.complexity), compile clean under both regen and the manual runner, and all 6+4+4 test fns still pass. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Delete homeless space-complexity witnesses (operator: 2026-07-18); declare Rust-test follow-up The three space-complexity .dag witnesses test v1.compiler.complexity (a src/v1 compiler-internal module) and had no clean home: dag/ fails the compile-clean gate's cross-layer import ([dag, src/v2] roots, no src/v1), and src/v1/ makes regen_stage0 --emit-fresh emit them as unregistered stage0 seed files (breaking self_host_realized_comparison). The analysis stays proven in-seed by execution (regen, runs every compile); the discriminating behavioral REDs are recorded in space-complexity-design.md as a declared follow-up to re-add as Rust tests in compiler_tests.rs when P4 makes the coverage load-bearing. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

briansrls added 6 commits July 7, 2026 03:40

WIP: get main CI green

89e0218

WIP: get main CI green

786b173

WIP: get main CI green

1ae42ad

Merge remote-tracking branch 'origin/main' into session/crisp-bear-170

6208aeb

WIP: get main CI green

5ca3fad

WIP: get main CI green

ff7b4a5

gunbai-bot Bot marked this pull request as ready for review July 7, 2026 04:32

gunbai-bot Bot marked this pull request as draft July 7, 2026 04:32

briansrls and others added 3 commits July 7, 2026 04:33

WIP: get main CI green

7927413

WIP: get main CI green

7051794

gunbai-bot Bot mentioned this pull request Jul 7, 2026

6343 #6350

Closed

6 tasks

briansrls marked this pull request as ready for review July 7, 2026 17:14

briansrls merged commit 1a79ce1 into main Jul 7, 2026
6 of 10 checks passed

briansrls deleted the session/crisp-bear-170 branch July 7, 2026 17:59

gunbai-bot Bot mentioned this pull request Jul 10, 2026

affected set processing #6453

Merged

gunbai-bot Bot mentioned this pull request Jul 11, 2026

Host-scaffold classifier defork: substrate-declared LiveTreeDisposition; delete the Rust text classifier #6479

Merged

briansrls mentioned this pull request Jul 16, 2026

Witness realization P1–P6: measured space + the artifact store (the lane PR; P0 merged via #6663) #6738

Merged

4 of 5 tasks

Navigation Menu

Repository navigation

get main CI green - #6352

get main CI green#6352
briansrls merged 9 commits into
mainfrom
session/crisp-bear-170

gunbai-bot Bot commented Jul 7, 2026

Uh oh!

gunbai-bot Bot commented Jul 7, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

gunbai-bot Bot commented Jul 7, 2026

Worker attestation

Summary

Test plan

Uh oh!

gunbai-bot Bot commented Jul 7, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant