904380e96a
Commits the ADR-0243 (Wave-Field Cognitive Lifecycle) proposal verbatim as Proposed, and the arc-governing implementation plan derived from it plus docs/analysis/core_cohesion_master_plan.md and the ADR-0241/0242 post-Accept close audit findings. The plan records that most of ADR-0243's organs already exist on main (sensorium feed, WaveManifold, energy classes, GoldTether, surprise/discovery plumbing, self-authorship, biography, packs/en readback) and that tests/test_third_door_cohesion.py already pins I-01..I-05, A-02/A-04, and core_ha absence from the cohesion master plan — so this arc is the §2.2 relaxation reasoner plus three wiring lanes, not a rebuild. Also pins three defects in the ADR's §3 reference prototype so it can never be ported verbatim: the egress gate's antisymmetric quadratic form is identically zero (rejects every state, valid or not); the unitary relaxation loop cannot dissipate toward a ground state (the honest relaxer is the imaginary-time semigroup); and the prototype's block-matrix "pseudoscalar proxy" is a different operator from the real Cl(4,1) pseudoscalar action despite sharing I^2 = -Id. [Verification]: uv run core test --suite smoke -q -> 176 passed (133.41s); uv run python -m pytest tests/test_adr_0243_sketch_defect_pins.py tests/test_third_door_cohesion.py -q -> 19 passed (0.93s)
94 lines
12 KiB
Markdown
94 lines
12 KiB
Markdown
# ADR-0241 / ADR-0242 Mastery-Close: Plan & Outcome
|
||
|
||
**Status:** CLOSED — both ADRs Accepted (ratified by Joshua Shay, 2026-07-15). This document is the durable, versioned status surface for the arc; it consolidates the original plan with what actually happened, phase by phase, including the parts that ran past the plan's original scope.
|
||
|
||
**Why this file exists:** the operational W0–W5 plan that drove this arc (`docs/plans/adr-0241-0242-mastery-close.md` in the authoring session's Claude Code plan store) was never committed — it lived only as a gitignored `.claude/plan/` artifact in a since-orphaned worktree. This doc is the fix: the plan-to-outcome record now lives in the repo, not in an assistant's session memory.
|
||
|
||
**Related, already-committed artifacts** (unchanged by this doc, cross-referenced below):
|
||
- `docs/plans/adr-0241-cohesion-mastery-plan-p0-p12.md` — Plan A (P0–P12), archived point-in-time
|
||
- `docs/plans/adr-0242-drive-complete-plan-d0-d10.md` — Plan B (D0–D10), archived point-in-time
|
||
- `docs/plans/adr-0241-0242-session-completion-summary-2026-07-15.md` — the 2026-07-15 completion summary
|
||
- `docs/audit/adr-0241-0242-acceptance-packet-2026-07-15.md` — the D10 acceptance packet + §8 ruling record
|
||
- `docs/audit/adr_0241_cohesion_acceptance_checklist.md` — per-box acceptance evidence
|
||
- `docs/adr/ADR-0241-wave-field-driven-hyperbolic-atlas-and-resonant-cognition.md`, `docs/adr/ADR-0242-atlas-packing-and-fibonacci.md` — both now **Accepted**
|
||
|
||
---
|
||
|
||
## Original goal (as authored 2026-07-15)
|
||
|
||
Complete the wave-field substrate program (ADR-0241/0242 + core_ha deprecation + Fibonacci vectors) to the mastery bar: merged green without shortcuts, adversarially verified, acceptance package ready for Joshua.
|
||
|
||
---
|
||
|
||
## Phase-by-phase: plan vs. outcome
|
||
|
||
### W0 — Unblock #38 ✅ COMPLETE
|
||
**Plan:** diagnose whether the lane-shas CI failure on PR #38 was a per-lane timeout or a genuine SHA mismatch; fix forward without re-pinning unless mismatch was proven.
|
||
**Outcome:** root cause was resource starvation on the (since-retired) Act host, not a timeout or a SHA mismatch — pins were byte-identical 9/9 locally. No re-pin occurred. A debug line that had leaked into the tip commit (`print(f"DEBUG ENV: {dict(os.environ)}")` in `evals/public_demo/runner.py:131`, a live secret-exposure vector under any CI that logs env) was caught and reverted (`4853a55c`) before merge, as the plan's W3 gate required.
|
||
**PR:** #38 (`9e1455b9`), #39 (`b80f262f`).
|
||
|
||
### W1 — Adversarial verification pass ✅ COMPLETE
|
||
**Plan:** five cross-authorship attacks against P8 chiral non-vacuity, quarantine transitivity, Fibonacci cert digest/budget exactness, P9 seal discipline, and the fidelity ledger — each attack executed by an operator who did not author the code under attack.
|
||
**Outcome:** all five reads executed (not read-only — attacks were run, not just reasoned about). Confirmed solid: chiral charge genuinely non-vacuous and conserved (Q=−2.5), Fibonacci cert budget-exact and digest-stable. **Caught a live, real defect**: `core/physics/__init__.py` eagerly loaded 5 Tier-2 (off-serving) modules into `chat.runtime` via its barrel import — a genuine A-04 transitive serve breach, not a false alarm. Fixed via PEP 562 lazy exports + a new process-level guard test (`tests/test_serve_quarantine_transitive.py`).
|
||
**PR:** #40 (`54228d45`).
|
||
|
||
### W2 — Commit the session artifacts ✅ COMPLETE
|
||
**Plan:** copy Grok's three plan/summary docs from an ephemeral worktree into `docs/plans/` so they survive worktree cleanup.
|
||
**Outcome:** done as specified; all three files present on `main` today (see "Related artifacts" above).
|
||
**PR:** #40 (bundled with W1).
|
||
|
||
### W3 — Merge + cleanup ✅ COMPLETE (original scope)
|
||
**Plan:** merge #38 with explicit authorization, delete arc branches/worktrees, confirm post-merge main is green.
|
||
**Outcome:** done. **Naming note:** the label "W3" was later reused for an unrelated, later phase — PR #46 ("isolate xdist polluters blocking `-n auto` default (W3)") is a *different* W3, part of the post-ratification hardening pass below, not this merge-and-cleanup gate. Treat them as distinct; the collision is cosmetic, not a scope conflict.
|
||
**PR:** #38, #39, #40.
|
||
|
||
### W4 — D10 acceptance package ✅ COMPLETE
|
||
**Plan:** one-page packet for Joshua — acceptance checklist walkthrough, fidelity ledger state, W1 findings, explicit deviations for ruling (P7 polar demotion, sensorium staged packets, progressive continuous-field integrals), R&D-memo caveats already filtered by implementation.
|
||
**Outcome:** packet written (`docs/audit/adr-0241-0242-acceptance-packet-2026-07-15.md`), including a §7 memo-slice addendum (impl-stronger-than-memo table) added after independent verification via the Google Drive connector against the private authority docs. **Joshua ruled "Ratify"** — both ADRs flipped to Accepted with a full ruling record (§8) and provenance-guarded governance test (`tests/test_adr_0241_governance_p12.py`).
|
||
**PR:** #41 (chiral sign-gate + packet), #42 (CRDT-vs-bit-exact dossier — one of the ruling's explicit deviations), #43 (ratification: status flip + memo-slice closure + core_ha Path-B correction).
|
||
|
||
### W5 — Post-Accept backlog — **PARTIALLY SUPERSEDED, see below**
|
||
**Plan (as originally written):** D9 Rust f64 parity; runner capacity (second Act runner / larger VM, later superseded by a GitHub-Actions-mirror idea); sensorium real compilers; V2 energy promotion (Fibonacci vs dyadic vs log, benchmark-gated); progressive continuous field.
|
||
**Outcome (as of mid-arc):** none of the W5 items had been started then; CI-infra under-forecast how bad the situation would get — see "Unplanned: CI collapse and local-first pivot" below. **Post-Accept follow-through (2026-07-16, this arc):** D9 f64 GP parity (Python SOT), sensorium real-compiler → ψ feed, V2 fib/dyadic/log fixed-replay evidence (no production promotion), `-n auto` fast-lane default, INV-21 allowlist dedup, and runner doctrine (local-first + `ubuntu-latest:host`, no Docker CI for merge) are closed. Progressive continuous field remains deferred research. See **Remaining backlog** below.
|
||
|
||
---
|
||
|
||
## Unplanned: post-ratification hardening + CI collapse and local-first pivot
|
||
|
||
This work was not anticipated by the original plan (W5 assumed a GitHub-Actions-mirror fix for CI capacity; what actually happened was messier and is worth recording honestly).
|
||
|
||
**Post-ratification test/governance hardening:**
|
||
- #45 (`8f67590d`) — cleared pre-existing full-suite reds; confirmed the "~31 pre-existing red tests" figure was stale (retired).
|
||
- #46 (`b0156406`) — isolated `-n auto` xdist polluters (the *second* "W3", unrelated to the merge-gate W3 above).
|
||
- #47 (`56e55e6e`) — evolved the anti-self-Accept governance guard (`P12`) for the post-ratification state; the guard correctly went red against the ratification itself and had to be updated to check *provenance* (ratified-by-whom) rather than merely "not yet Accepted."
|
||
|
||
**CI collapse:**
|
||
GitHub Actions became billing-locked (~$6, account-level) mid-arc, forcing a pivot away from the "GitHub-Actions-mirror" idea W5.2 had assumed would supersede the Act-runner ceiling. That idea is now dead; the actual fix was:
|
||
- #48 (`9d284d95`) — locked dependency installs (`uv sync --locked`) to stop lane-pin thrash from PyPI drift. Initial version was not self-validating: `uv.lock` was gitignored (existence ≠ tracked), so every locked CI run failed "Unable to find lockfile" until the lockfile itself was committed.
|
||
- #49 (`a7279a54`) — ADR-0242 §5 carry-seam items: κ-cert telemetry event + F5–F7 cross-band discovery gate (Tier-2, off-serving).
|
||
- CI doctrine rewrite: local-first validation is now the real gate (`uv run core test --suite smoke -q` in the worktree before every push, `[Verification]:` tag in PR descriptions), with a local Mac Act runner (`ubuntu-latest:host`) as secondary/observability CI, replacing the GCP e2-micro runner entirely (that box cannot survive an Actions job — 1GB RAM, OOM class). Doctrine landed in `AGENTS.md` §CI/CD Runner Architecture + a pointer in this repo's `CLAUDE.md`, plus the global `~/.claude/CLAUDE.md`.
|
||
- #50 (`31073d55`) — lane-shas streaming progress + configurable timeout, so a hang is visible in the log the moment it happens instead of being silently truncated (same class of failure that caused the original W0 diagnosis to take real forensic work).
|
||
- #51 (`207d9c6a`) — this CI doctrine committed to the repo (`AGENTS.md`, `CLAUDE.md`) — also the new Mac runner's first live dispatch test.
|
||
- #52 (`18d3c70d`) — assessment note on a rejected parallel-session re-implementation submission (its "three critical bugs" claim did not survive verification; one had a literal Python syntax error).
|
||
|
||
**Open technical note (infra, not doctrine):** some Act runner configs under `~/infra/runner-m1` still map `ubuntu-latest` → `docker://node:20-bookworm` and pull containers. That is a **runner misconfiguration**, not the merge gate. Doctrine: local-first in the worktree; optional secondary runner is `ubuntu-latest:host` (native, no Docker). Fix the label/platform map on the Mac runner when infra is next touched — do not re-author doctrine around Docker.
|
||
|
||
---
|
||
|
||
## Current state (2026-07-16)
|
||
|
||
- Both ADRs: **Accepted**, ratified, full audit trail.
|
||
- Post-Accept backlog **merged**: PR #55 (`d388fc9e` feature commit + status notes) landed on `main` @ `19819194`; all `[Verification]` numbers were independently reproduced by the 2026-07-16 audit (smoke 176; fast lane 11709 passed / 108 skipped; f64 parity 21/21 incl. Rust-path; invariants+demos+A-04 140). Three seam-vs-integration caveats carried into the ADR-0243 arc — see `docs/plans/adr-0243-implementation-plan.md` §1.
|
||
- CI: **local-first is the real merge gate** (no Docker CI for merge); `uv.lock` committed and locked installs enforced in workflows; optional Mac Act runner must be `ubuntu-latest:host` (native), not `docker://…`.
|
||
|
||
## Remaining backlog (post-Accept follow-through; 2026-07-16 mastery close)
|
||
|
||
Status of items that were open after ratification / CI pivot:
|
||
|
||
1. **D9** — ✅ **DONE (this arc):** frozen Python f64 Cl(4,1) GP parity suite (`tests/test_geometric_product_f64_parity.py`) is authority; backend proves f64 never silent-f32-truncates under `CORE_BACKEND=rust` (Rust f32 GP only; f64 stays Python SOT until a future `geometric_product_f64` PyO3 export is parity-gated). No scipy-as-truth.
|
||
2. **Sensorium real compilers → ψ (I-04)** — ✅ **DONE (this arc):** `packet_from_compiler_versor` / `packet_from_compilation_unit` lift live audio/vision compiler units into `ModalityPacket`; tests exercise compile→ψ→superpose→algebraic `phase_correlation` (not only `fake_deterministic_packet`). Cosine/ANN still banned; A-04 quarantine holds.
|
||
3. **V2 energy comparative evidence** — ✅ **DONE without production promotion (this arc):** Fibonacci vs dyadic vs log under fixed replay via `comparative_three_way` / `fixed_replay_compare_artifact` + `python -m evals.adr_0242_v2_energy_compare`. Artifact flags `joshua_gate_required=True`; `FieldEnergyOperator` / `energy.py` unchanged. **Promotion remains blocked** until Joshua’s explicit gate.
|
||
4. **Progressive continuous field** — ⏸ **deferred** — research track, non-blocking (out of scope for this close).
|
||
5. **Runner Docker-vs-host discrepancy** — ✅ **DONE (this arc, doctrine restored):** Global + repo doctrine is **local-first** (run locally, then merge). Optional Mac Act runner must be **`ubuntu-latest:host`** = native host execution — **not Docker job containers**. A misconfigured `~/infra/runner-m1` that mapped `ubuntu-latest` → `docker://node:20-bookworm` was the bug; doctrine was never “bless Docker.” `AGENTS.md` re-states host-native + local-first as the gate.
|
||
6. **`-n auto` fast-lane default flip** — ✅ **DONE (this arc):** `make test-fast` / `test-full` default to `-n auto`; serial escape hatches `test-fast-serial` / `test-full-serial`; `docs/testing-lanes.md` updated. #46 polluter isolation is the prerequisite.
|
||
7. **INV-21 allowlist dedup** — ✅ **DONE (this arc):** tuple-first pin with import-time uniqueness check; demos no longer re-list the full frozenset (single ownership in `TestINV21OneMutationPath`).
|