13 Commits

Include renames

Author	SHA1	Message	Date
Nick	da36073a0c	feat(warm_router): strangler exit — _py_base_level_from_uses retired, Rust authoritative ... 28-line ACT-R oracle deleted to delegation stub. Frozen oracle (cold/single/ multi/bad-rows/future/Z-suffix + non-string edge). The kernel handles the AttributeError edge internally with identical observable (-2.0). Tests migrated to public delegation. 50 regression + ruff clean.	2026-10-10 20:01:01 +01:00
Nick	3239080c9b	feat(memory_store): strangler exit — 10 _py_ oracles retired (~250 lines), Rust authoritative ... The largest single-file exit: all ten retrieval kernels deleted to delegation stubs (score/fusion/map/position/fts/chunk/norm/merge/fuse/ project). Frozen oracle covers every canary section (abstention edges, 24-row matrix, fusion/floor/gap/improvement/assoc/map/pos/fts/chunk/norm/ fuse). Canary 75/75 vs frozen. The oracle-paths shims test rewritten for post-exit fail-loud semantics (stubs RAISE under the python pin, no recursion). recall()/SQLite/cache/uplift orchestration stays Python reactive by design. One pre-existing ffi-smoke failure (fails on stashed tree too — fault-watch scope, not this batch).	2026-10-10 18:28:54 +01:00
Nick	4f31741ac6	fix(ci): four unit/lint gates repaired — spine store staleness, date time bomb, lint ledger, digest ... 1. spine store rebuilt (d8ce21b-era classification + cc_loader/memory_consolidation relationships) — rebuild-neutrality test now passes against the CURRENT map. 2. DATE-TIME BOMB (test_warm_base_level_math_and_gather): base_level_activation used the real clock against fixed fixture rows; three days of decay crossed the -2.0 clamp and broke the test with no code change. Fix: optional now_iso seam (no-clock law, backward-compatible default), test pins NOW_ISO. Discovered via live probe: the oracle returned exactly -2.0 under the real clock. 3. LINT LEDGER (+1 BLE001 from 3c79a0a consolidation kernel): _core() probe except annotated per the compliant-handler spec (kernel-unavailable degrades to the Python oracle). Baseline 345 restored without re-baselining. 4. AGENTS.md digest synced (henxels sync).	2026-10-07 11:06:24 +01:00
Nick	98b5a73ed1	feat(cognition): R2 W3.3 — identity/growth/pins kernels + full coverage (72%) ... - identity_anchors: check_drift delegates (pattern table stays the module's data authority via _pattern_triples); _py_check_drift oracle covers the four prefix families; drift parity proven on the full table (31 patterns) + the full-text multi-class case - growth_delta: _near_dup/growth_profile delegate; oracle accepts delta dicts AND raw strings; near-dup kernel corrected to the REAL semantics: prof-key = strip-punct lowercase words >4 chars, split at |A∩B|/max(|A|,|B|) >= 0.25 (NOT a jaccard-min heuristic) - pins: recall kernel projects via _project parity (details-blob parse, defaults, 'reason' field); empty-context abstention frozen - probe branches exercised honestly: pinned-fallback + ImportError via the from-import resolution path (the from-import goes through the package attribute, not sys.modules — the PEP resolution law frozen in the battery) - reflect malformed-JSON + [:3] cap + apply budget cut frozen - suite 2295/0; diff-cover 72% (floor 70); canaries 9/9+18/18+38/38 x3; henxels hold; ruff/type clean	2026-10-05 12:50:33 +01:00
Nick	538bbeb821	feat(cognition): R2 W3.2 — persona_gate decision kernels (cognition_core) ... Contract BUILDER-2026-10-05T07:19:27Z-cli (Wave 3 slice 2, #2461). - cognition_core/broker.rs gains the persona-decision kernel: persona_lex (distinctive content words: punct strip, light plural stem, acronym preservation — 55-word stoplist), persona_overlap (intersection over min-size), persona_decide (engagement matrix: amplifier set, semantic gates 0.45/0.30/0.68, user-rule 0.4 bar, permissive/legacy-restrictive verdicts, _cite parity '[topic:pN] text') - persona_gate.py: _lex/_overlap/decide delegate; semantic cosines gathered Python-side (embedder calls are IO — the membrane's discipline) and passed into the kernel; _py_ oracles are the load-bearing fallback - FFI: cc_persona_lex/overlap/decide; loader extended - canary_persona.py (canary-first red on every seam): 18/18 x3 — lex matrix incl. acronyms/3-char exclusions, decide engagement paths, citation format parity - shim battery 35/35; suite 2294/0; ruff clean; type ratchet PASS; henxels all-hold	2026-10-05 09:00:43 +01:00
Nick	290d264f31	feat(cognition): R2 W3.1 — context_broker kernel + the surrogate-pair parser fix across all nine crates ... Contract BUILDER-2026-10-05T07:19:27Z-cli, closed: module:context_broker correctly_predicted; cognition_core expected_new_observed; cc_loader surfaces honestly new (#2463 wave epic). - cognition_core crate (zero-dep, no-clock): context_broker's four deterministic kernels — default_estimator (chars//4; utf16-length corruption caught — the canary's astral cases), default_budget, default_memory_budget (+cap), memory_disabled (env gate value-parsed) - THE PARSER BUG (nine pyjson copies repaired): the surrogate-pair branch tested self.b[i+1..] for the second escape's backslash — but hex4 had already advanced i past the first escape's 'u+digits', so the correct guard window is [i..] not [i+1..]; and after the guard, i advanced +2 leaving hex4 to read a shifted window ('bad \\u digits' on \\ud83c\\udfb8). Fixed: guard at [i..], pre-hex4 advance +1. Astral chars (emoji) now parse as ONE code point — char-count parity restored everywhere (estimator/chunk/norm) - context_broker: four functions delegate (load-bearing _py_ oracles); the env gate is cross-checked Python-side (Python verdict stays authority) - cc_loader: ctypes shim, CC_RUST_LIB file-form; battery-covered - CI: crate builds in both jobs, CC_RUST_LIB envs, canary registered Gates: canary 38/38 x3 + full seven-crate wall green, suite 2293/0, type ratchet PASS, ruff clean, henxels all-hold, diff-cover 75%.	2026-10-05 08:43:19 +01:00
Nick	1283b0bf94	fix(memory): W2.1 coverage 62->84 + pair-list contract harmonized ... - chunk gather + projection now speak the W2.1 pair-list contract consistently on BOTH engines (fusion['ranks'] is [[rowid, rrf], ...]; dict-form still accepted for backward callers) - oracle-path tests: all three modes, abstention edges, malformed timestamps, hybrid accumulation, chunk SQL gather through a real fixture (vec-table absent by design; embed stubbed — vector execution stays Python per the W2.0 ruling) - battery 32->32 tests extended; suite 2291/0; canary 75/75 x3; diff-cover 84% (floor 70)	2026-10-05 07:51:40 +01:00
Nick	1ee35d76c9	fix(memory): W2.1 coverage + stopword parity + chunk projection kernel ... - battery: delegation + projection coverage for the kernel consumers (29-test shim battery; CI's 45% was the pre-extension tree) - chunk projection (chunk_recall tail) moved into the kernel: RRF-desc order, text[:400] chars, budget cap, page/room/default semantics — canary 75/75 x3 - stopword list AST-extracted exactly (51 words incl. needs/wants/likes) - memory_store: recall/fused/association/chunk kernels delegate with _py_ oracles as load-bearing fallback Gates: canary 75/75 x3, shim battery 29/29, suite 2286/0, ruff clean, type ratchet PASS, henxels all-hold. PR 2460 updated with the propagation evidence line.	2026-10-05 07:20:38 +01:00
Nick	19976c310c	feat(memory): R2 W2.1 — deterministic Rust memory-retrieval kernel (#2458) ... Contract BUILDER-2026-10-04T20:10:25Z-cli, closed: memory_store and memory_selector correctly_predicted; kernel expected_new_observed; shim surfaces honestly new. Stacked on W2.0 (merged de54453). - memory_retrieval crate (zero-dep, no-clock, rlib+cdylib+bin): the retrieval transform family in ONE kernel surface — map_reduce, position_aware, fts_query_conjunctive (51-word stoplist extracted via AST, no transcription drift), chunk_text (byte-vs-char law honored), norm_key, plus the extracted kernels: rank_fusion (RRF/lexical normalization/mode selection), score_recall_rows (metadata blend + mode-specific QPP abstention), merge_association_rows (association projection), fuse_results (fact/research projection + per-kind min-max scale alignment) - PARITY-PLUS-IMPROVEMENT proven: recall's N+1 (per-candidate entry reads + repeated DF probes) ELIMINATED — rows batch through the W2.0 read boundary once and the kernel runs per retrieval stage - Parity trap frozen: naive timestamps take Python's designed TypeError->recency=1.0 fallback (parse_iso now refuses naive, exactly like aware-vs-naive subtraction) — the 34-case matrix caught this - UTF-8 transport fix: ensure_ascii=True surrogate pairs corrupt astral chars (emoji) in char-count semantics — direct UTF-8 transport, frozen - both _py_ oracles and delegating publics; RUNE_RETRIEVAL_CORE=python pins the fallback - CI: crate builds in both jobs, MR_RUST_LIB envs, canary registered Gates: canary 72/72 x3, suite 2286/0, type ratchet PASS, ruff clean, henxels all-hold. Store: 218 entities, rust-crate:memory_retrieval modeled.	2026-10-05 06:38:57 +01:00
Nick	0816c5997d	feat(memory): R2 W2.0 — retain SQLite behind a zero-dep Rust read boundary (#2456) ... Contract BUILDER-2026-10-04T15:54:17Z-cli closed: module:memory_store correctly_predicted; memory_storage crate, mems_loader, and canary honestly new. - memory_storage crate: zero Cargo dependencies; system libsqlite3 FFI; sqlite3_open_v2(SQLITE_OPEN_READONLY) makes writes impossible at the connection level; no write-capable mems_* exports - typed read parity: INTEGER/FLOAT/TEXT/NULL/BLOB, bound string/numeric/bool/NULL params, aggregates, empty results, unicode, and honest SQLite errors - mutation SQL is rejected; fixture SHA-256 is identical before/after all reads and the rejected mutation - memory_store gains an explicit backend seam: stdlib SQLite remains the default; Rust read path is opt-in via RUNE_MEMORY_STORAGE=rust; connection-based transactional callers are untouched - stats_for_path is the proving consumer; both backends return identical result shape - canary-first: crate-absent red proven, now 19/19 three consecutive - CI builds the crate in both jobs, pins MEMS_RUST_LIB, and runs the crate canary; pytest shim battery covers loader/seam lines Evidence: full suite 2283/0, diff-cover 71% >= 70, type ratchet PASS, ruff clean, henxels all-hold. SQLite is retained exactly as Nick directed; this slice exports no Rust write path.	2026-10-04 21:07:43 +01:00
Nick	8404cddb1d	feat(scaffolding): R2 Wave 1 — the scaffolding engine reborn (Go start / Rust execution) ... Contract BUILDER-2026-10-04T15:25:33Z-cli, closed: scaffold, construction_decision, go-surface:rune-entry correctly_predicted; crate + shim + canary honestly new. The engine governed its own rebuild. - scaffold_core crate (zero-dep, no-clock, rlib+cdylib+bin): lifecycle.rs (the LEGAL_TRANSITIONS state machine with exact error strings — the single authority, terminals derived), close.rs (the closure classification — all buckets incl. the dict-literal overwrite quirk and legacy aliases), language.rs (Nick's protocol addition: the placement verdict — rune dichotomy rule, other-project fit assessment) - THE PARITY PROOF: all 43 production contracts in the ledger replay to IDENTICAL closure verdicts — the new engine reproduces the V1-accepted system's real history byte-for-byte - Go surface: rune scaffold subcommand — pure ops direct to the Rust binary over stdio NDJSON (the tranche's first contracted membrane junction); store-backed ops proxy to the engine (strangler membrane); 4 new Go unit tests (stub-binary junction + failure surfacing) - scaffold.py: bucket math factored to _py_close_quality, consumed by close_contract and the replay battery (one computation, two engines) - construction_decision.py: language_placement in the classification path - Canary-first: canary_scaffold_core.py red-proven, now 152/152 x3 (legality matrix 81 pairs, ledger replay, determinism, Go membrane, placement matrix incl. delegation agreement) - CI: crate builds in both jobs, SC_RUST_LIB envs, canary registered Gates: pytest green (one bridge-flake rerun), canary 152/152 x3, type ratchet PASS, ruff clean, henxels all-hold (shim battery extended). Store: 214 entities — rust-crate:scaffold_core modeled.	2026-10-04 16:52:13 +01:00
Nick	171f3fb250	fix(tests): close the diff-cover hole — 63% -> 76% on the changed-lines floor ... Runs 18367/18369 failed unit-tests on diff-cover (63% < 70%, 338 missing lines) with pytest itself green — the changed-lines floor counts PYTEST-traced lines only, and three slice surfaces had canary-only coverage. Battery extension (the shim-battery law applied to my own slices): - warm_router: gate branches, finish stack, public score_neuron wrapper (designed-fallback determinism), base-level math + gather (hermetic via monkeypatched table ensure — STORE_DB resolution is suite-order dependent, the order-dependent failure the coverage run exposed) - tui_transport: InProcess four-method drive, V1 against an in-process core, factory modes incl. the bogus-mode ValueError - token_optimizer: optimize_prompt/get_status/get_cache_stats/flush CI-exact verification: full coverage suite 2278/0, diff-cover 76% (floor 70). ruff clean.	2026-10-04 13:09:53 +01:00
Nick	9e6215e605	fix(ci): henxels holds — intentional mirrors registered + shim-layer pytest battery ... Contract BUILDER-2026-10-04T10:29:39Z-cli (successor to 10:20:24Z-cli, same red evidence run 18322 lint job 51891). - henxels.yaml: the deliberate-duplicate registry — shared/** (deployed- vintage mirror tree), the named src/rune strangler shims, per-crate pyjson copies (self-contained zero-dep design), tiny zero-dep lockfiles, store/scaffold data, issue-template + completion_file mirrors, the interop/rebase-proof fixture - tests/test_rust_core_shims.py (new, 11/11): the thin pytest battery the diff-cover law owed the shim layer — drives the DELEGATION lines (loader resolves, envelope unpack, public-vs-fallback agreement) for spine_loader, ms_loader, cs_loader, tc_loader, and core_server's in-process shapes; skips cleanly on unbuilt crates Local: henxels all-hold green, battery 11/11.	2026-10-04 11:30:19 +01:00