codontrace 0.3.0a2__tar.gz → 0.3.0b2__tar.gz

{codontrace-0.3.0a2 → codontrace-0.3.0b2}/.github/workflows/ci.yml

@@ -1,6 +1,6 @@
 name: CI
 
-on:
+"on":
   push:
   pull_request:
 
@@ -12,17 +12,18 @@ env:
   PIP_DISABLE_PIP_VERSION_CHECK: "1"
 
 jobs:
-  test:
-    name: Python ${{ matrix.python-version }}
-    runs-on: ubuntu-latest
-    timeout-minutes: 25
+  core-smoke-cross-os:
+    name: Core smoke (${{ matrix.os }}, Python ${{ matrix.python-version }})
+    runs-on: ${{ matrix.os }}
+    timeout-minutes: 30
     strategy:
       fail-fast: false
       matrix:
+        os: [ubuntu-latest, windows-latest, macos-latest]
         python-version: ["3.11", "3.12", "3.13", "3.14"]
     steps:
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
 
       - name: Set up Python
         uses: actions/setup-python@v6
@@ -38,15 +39,8 @@ jobs:
       - name: Compile package, tests, examples, and tools
         run: python -m compileall -q src tests examples tools
 
+      - name: Enforce core boundary with AST import guard
+        run: python tools/check_core_boundary.py
+
       - name: Run release-critical tests
-        run: |
-          python -m pytest \
-            tests/test_export_import.py \
-            tests/test_api_signature.py \
-            tests/test_genesis_causal_mechanism_hardening.py \
-            tests/test_genesis_causal_mechanism_runtime_wiring_integrity.py \
-            tests/test_genesis_integration_public_api_unification.py \
-            tests/test_genesis_integration_runtime_wiring.py \
-            tests/release \
-            tests/genesis_gates \
-            --disable-plugin-autoload
+        run: python -m pytest tests/test_export_import.py tests/test_api_signature.py tests/test_genesis_causal_mechanism_hardening.py tests/test_genesis_causal_mechanism_runtime_wiring_integrity.py tests/test_genesis_integration_public_api_unification.py tests/test_genesis_integration_runtime_wiring.py tests/release tests/genesis_gates --disable-plugin-autoload

{codontrace-0.3.0a2 → codontrace-0.3.0b2}/.github/workflows/publish-pypi.yml

@@ -1,6 +1,6 @@
 name: Publish to PyPI
 
-on:
+"on":
   release:
     types: [published]
 
@@ -17,7 +17,7 @@ jobs:
     timeout-minutes: 20
     steps:
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
 
       - name: Set up Python
         uses: actions/setup-python@v6
@@ -35,7 +35,7 @@ jobs:
         run: python -m twine check dist/*
 
       - name: Upload artifacts
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@v7
         with:
           name: python-distributions
           path: dist/
@@ -49,7 +49,7 @@ jobs:
       contents: read
     steps:
       - name: Download distributions
-        uses: actions/download-artifact@v4
+        uses: actions/download-artifact@v8
         with:
           name: python-distributions
           path: dist/

codontrace-0.3.0b2/CHANGELOG.md

@@ -0,0 +1,191 @@
+# Changelog
+
+## Unreleased
+
+Subsequent library work after `0.3.0b2` belongs here.
+
+## 0.3.0b2 — Science-gate hardening and Phase A–E substrates (2026-09-11)
+
+Public beta identity for science-gate hardening plus the Phase A–E library surfaces that landed on `main` after `0.3.0b1`. Package version remains `0.3.0b2` (not bumped to b3). This is not an Avida replacement and does not prove OEE, instinct evolution, AGI, collective intelligence, evolved plasticity, or phenotypic plasticity.
+
+### Release identity
+
+- Kept package version `0.3.0b2` (`pyproject.toml`, `codontrace.__version__`, `CITATION.cff`).
+- Aligned `RELEASE_LABEL` / `RELEASE_ARTIFACT_NAME` (and aliases), README / CLAIMS / RELEASE_EVIDENCE / BibTeX pins, and citation `date-released` with that same identity.
+
+### Notes (claim policy)
+
+- Phase A–C and Phase E are **opt-in substrates** (life-loop ecology; sexual recombination; dynamic/fluctuating environment; capsule/memory/role/deme). Research defaults stay unchanged unless a caller selects those presets.
+- Phase D is an **opt-in measurement** layer (`MultiGenerationEvidencePack`). Metric deltas are runtime observations. `instinct_improved` is ClaimGate-gated; OEE stays `oee_measurement_only` unless existing research-grade thresholds are met.
+- ClaimGate keeps `collective_intelligence`, `evolved_plasticity`, and `avida_replacement` blocked. Do not claim OEE proof, instinct/intelligence proof, or AGI.
+- No loosening of scientific claims from the science-gate hardening below. GENESIS remains research-beta software.
+
+### Phase E — capsule/memory/role/collective substrate
+
+#### Added
+
+- Opt-in `PhaseESubstrateConfig` (omitted from default serialization so Phase A–D presets and pinned digests stay stable).
+- Organism-local / lineage capsule memory with **real** subsequent effects: action substitution, ATP bonus, and COPY_SELF task gating under a matching sensory cue (Am Nat 2020 / odometry mapping). Ablation with capsules disabled yields a different replay digest.
+- Role / differentiation tags (`soma` / `germline` / `messenger`) that can gate reproduction or messaging (Avida GermlineReplication / GECCO 2008 analog).
+- Deme/group containers, Goldsby-style messaging buffer (`send_message`, `retrieve_message`, `broadcast_message`, `block_propagation`), and a minimal mean-fitness deme-replication trigger with recorded events (`DEMES_*` analog).
+- `PlasticityProtocolSpec` plus `read_environment_cue()` over Phase C regimes/local patches, including the Ghalambor/Clune four-condition experimental-design checklist (not evolved-plasticity proof).
+- OntoAvida-inspired `PhenotypeTranscriptomeEvidence` export and `AvidaParityProtocolSpec` head-to-head recipe (software capability only).
+- `GenesisRuntimeProfile.phase_e_substrate_world()`, `build_phase_e_evidence_pack`, `examples/genesis_phase_e_substrate.py`, `tests/test_genesis_phase_e_substrate.py`, and `docs/PHASE_E_LITERATURE.md`.
+- ClaimGate blocked aliases for `proved_collective_intelligence`, `evolved_plasticity`, `avida_replacement`, and `associative_learning_proved`. Ceiling remains `runtime_observation`.
+
+#### Notes
+
+Phase E is an opt-in **substrate**. Default life-loop, sexual, dynamic-environment, and Phase D measurement APIs keep their pinned digests. Phase D MODES/Bedau packs can observe Phase E when it is enabled because action/ATP/birth records actually change. This is not proof of learning, collective intelligence, evolved plasticity, OEE, or Avida-replacement status.
+
+### Phase D — multi-generation instinct/behavior evidence
+
+#### Added
+
+- First-class `MultiGenerationEvidencePack` (JSON + digest) measuring multi-generation fitness, survival, births, ATP, and resource intake from existing life-loop records.
+- Descriptive instinct/behavior trajectories (genome→action phenotype summaries) plus same-seed ancestor vs descendant cohort comparison and optional mutation-off ablation/control hooks.
+- Persistence-filtered MODES-inspired axes (change, novelty, complexity, ecological potential; Dolson et al. 2019) and Bedau-style novelty/diversity/activity surfaces as library objects.
+- ClaimGate labels `runtime_observation` / `instinct_improved` (downgrades without multi-seed + ablation) and blocked `open_ended_intelligence` / instinct-evolution-proved aliases. OEE remains `oee_measurement_only` unless existing research-grade thresholds are met.
+- `examples/genesis_multi_generation_evidence.py`, `tests/test_genesis_phase_d_multigen_evidence.py`, and `docs/PHASE_D_LITERATURE.md`.
+
+#### Notes
+
+Phase D is an opt-in **measurement** layer. Phase A/B/C default presets, research defaults, and pinned spec/run digests stay stable (`phase_d_instinct_claim_metrics` metadata remains `hooks_only_not_implemented` on those specs). Metric deltas are runtime observations. This is not proof of OEE, AGI, collective intelligence, or instinct evolution, and not an Avida replacement.
+
+### Phase C — dynamic / fluctuating environment
+
+#### Added
+
+- Opt-in `EnvironmentConfig` / `EnvironmentSchedule` / `ResourceSpec` (omitted from default serialization so asexual and sexual research digests stay stable).
+- Avida `RESOURCE` chemostat parity: per-resource `initial`, `inflow` per tick, `outflow` fraction of unused resource (Ofria & Wilke 2004; Cooper/Ofria ~1% outflow).
+- Periodic / seasonal schedules (Avida-ED periodic; period < 1 is non-periodic) and seeded regime-switch ticks.
+- Two selectable fluctuating regimes (high-food vs low-food, or two niche reward maps) as a substrate for later plasticity studies — not a claim that plasticity evolved.
+- Spatial options: global pool, local patches, or both, with deterministic von Neumann diffusion/decay hooks aligned with ElementGrid neighbor geometry.
+- Digest-backed `EnvironmentState`, per-tick `EnvironmentSnapshot`, `EnvironmentEvent` log, and `WorldEvent` records (`environment_inflow` / `outflow` / `regime_switch` / `periodic_toggle` / `hazard_changed`).
+- `EnvironmentSchedule` object API plus `verify_environment_trajectory_replay()` for env trajectory digests.
+- `GenesisRuntimeProfile.dynamic_environment_world()` and `life_loop_world(environment=...)`; organisms still eat/starve/reproduce while food availability follows the schedule.
+- `tests/test_genesis_phase_c_dynamic_environment.py` and `examples/genesis_dynamic_environment.py`.
+
+#### Deferred
+
+- Evolved phenotypic plasticity measurement (environment substrate only).
+- Avida reaction/task-resource coupling (logic tasks as metabolic reactions).
+
+#### Notes
+
+Phase C is a time-varying environment *substrate*, not an Avida replacement and not a claim of evolved plasticity or intelligence. Claim language stays software capability / runtime observation only. Grounded in Ofria & Wilke 2004, Cooper/Ofria limited-resource ecosystems, Avida-ED resource modes, and Avida fluctuating-environment / plasticity literature used only as experimental design context.
+
+### Phase B — sexual recombination substrate
+
+#### Added
+
+- Optional `ReproductionMode.SEXUAL_CROSSOVER` on `ReproductionConfig` (omitted from default serialization so asexual research digests stay stable).
+- `SexualRecombinationConfig` mirroring Avida `avida.cfg` `RECOMBINATION_GROUP`: `recombination_prob`, `same_length_only`, `two_fold_cost_sex`, `max_birth_wait_ticks`, `chamber_capacity`, timeout policy, plus stubs for mating types / lekking.
+- Birth-chamber pairing queue (`BirthChamberState` / `IncipientOffspring`) with capacity and wait-timeout (`fail` or `asexual_fallback`). Empty chambers are omitted from population snapshots so asexual digests stay stable.
+- Avida-style positional continuous corresponding segment exchange (`apply_positional_segment_exchange`, `recombine_positional_segment_pair`) with a dedicated deterministic RNG fork for chamber pairing (`genesis/population/birth_chamber`) and crossover breakpoints (`recombination/<slot>/<slot>`).
+- Dual recombinant products by default; `two_fold_cost_sex` places only one. Mutation still runs after recombination.
+- Two-parent lineage / birth / child-genome fields (`second_parent_id`, `parent_ids`) that appear only on sexual births.
+- Direct `reproduce(..., mate=...)` remains a one-child library API; population sexual mode uses the chamber.
+- `GenesisRuntimeProfile.life_loop_world(reproduction_mode=...)` opt-in; default life-loop remains asexual COPY_SELF → mutate → child.
+- Sexual path reuses Phase A AliveGate, ATP, capacity, and placement gates.
+- `tests/test_genesis_phase_b_sexual_recombination.py` and `examples/genesis_sexual_recombination.py`.
+
+#### Deferred
+
+- Diploid meiosis / selfing (Aevol Eukaryote) as Phase B.1.
+- Full `MATING_TYPES` / `LEKKING` (config stubs only).
+- Modular random-region swap when `CONT_REC_REGS=0` (Phase B ships continuous corresponding regions).
+
+#### Notes
+
+Phase B is a recombination *substrate*, not an Avida replacement and not a claim of sexual selection or intelligence. Claim language stays software capability / runtime observation only. Grounded in Misevic, Ofria, Lenski 2006 Proc B and `devosoft/avida` `avida.cfg` `RECOMBINATION_GROUP`.
+
+### Phase A — Darwinian life-loop
+
+#### Added
+
+- `GenesisRuntimeProfile.life_loop_world()`: explicit ecology / life-loop preset with limited depletable food, partial deterministic respawn, basal metabolism, starvation death records, adjacent offspring placement, AliveGate + ATP reproduction gates, fitness-proportional capacity selection, and asexual parent→mutate→child inheritance.
+- Opt-in `MetabolicConfig` basal ATP drain (default off so existing research presets/digests stay stable).
+- `DeathMonitoringConfig` starvation floor + consecutive-tick threshold with explicit `starvation` removal reason (omitted from default config serialization).
+- `OffspringPlacementPolicy.REPLACE_OCCUPIED`: optional Avida-like overwrite-neighbor policy; not the research or life-loop default.
+- `LifeLoopObservation` / `summarize_life_loop_observation()`: Phase D hook for runtime counts only, including eater/waiter births, starvation deaths, and remaining resource cells (not instinct/intelligence claims).
+- `examples/genesis_life_loop.py` and `tests/test_genesis_phase_a_life_loop.py` for the eat → survive → reproduce loop, depletable resources, spatial capacity, differential reproductive success, and fixed-seed replay digest stability.
+- Restored `codontrace.genesis.engine` as a re-export of the core engine implementation so the public Genesis import path works.
+
+#### Notes
+
+Phase A is literature-complete as a Darwinian life-loop *substrate* (Avida limited resources + energy-budget ALife). It is not an Avida replacement. Phase B (sexual crossover) and Phase C (fluctuating/seasonal environments) are opt-in substrates. Phase D (multi-generation evidence metrics) is an opt-in post-hoc measurement API that does not change these presets. Research defaults such as `ReproductionConfig.offspring_placement=SAME_CELL` and `ResourceConfig.density=0` are unchanged unless a caller selects the life-loop or dynamic-environment preset. Claim language remains software capability / runtime observation only.
+
+### Scientific evidence-gate hardening (capsule / memory / generalization)
+
+#### Added
+
+- `codontrace.genesis.capsule_utility`: pure outcome-based capsule utility evaluator (`capsule_outcome_utility_v2`). Utility is measured selection-fitness delta only; synthetic fixed rewards (e.g. `task_delta = 1.0`) are forbidden. `claim_eligible` requires adoption + measured positive delta + trusted source status (`measured` / `last_known`).
+- `codontrace.genesis.memory_evidence`: pure delayed-reward evidence classifier. Write→later reward without read is `temporal_correlation` only. Causal claim requires read-linked evidence plus ablation `control_digest`.
+- Science unit tests:
+  - `tests/test_capsule_utility_science.py`
+  - `tests/test_memory_delayed_evidence_science.py`
+  - `tests/test_generalization_protocol_science.py`
+
+#### Changed
+
+- `engine.capsule_utility_records` now delegates to the pure evaluator (single source of truth).
+- `engine.memory_use_records` classifies delayed-reward chains via the evidence ladder (`observed_write` → `temporal_correlation` → `read_linked` → `causal_support`).
+- `engine.generalization_records` no longer emits first/last-tick digest proxies. Without a real heldout protocol, status is `protocol_not_run` with digests `not_run:*` and `claim_eligible=False`.
+- `SignalActionLink` / `MemoryUseEvidence` schema → v2 evidence fields (`evidence_status`, `causal_status`, `control_digest`, `claim_eligible`).
+- `GeneralizationResult` schema → v2 with hard gate against `protocol_not_run` and identical train/heldout digests.
+
+#### Claim policy
+
+No loosening of scientific claims. Changes tighten evidence surfaces so ClaimGate cannot treat correlation or synthetic rewards as causal success. GENESIS remains research-beta software; it does not claim AGI, consciousness, proven collective intelligence, or peer-reviewed superiority.
+
+## 0.3.0b1 — Studio-readiness beta release
+
+### Changed
+
+- Promoted current package identity from `0.3.0a2` alpha to `0.3.0b1` beta.
+- Updated package metadata, runtime `codontrace.__version__`, citation metadata, release evidence, README install pins, and current release artifact identity.
+- Kept historical `added_in`/compatibility provenance fields intact where they describe APIs introduced during the alpha line.
+
+### Added
+
+- Added `docs/STUDIO_PHASE1_EXECUTION_SPEC.html` as the repo-ready Phase 1 Studio execution handoff.
+- Added `docs/STUDIO_BOUNDARY.md` to lock the library/UI boundary before Studio work begins.
+- Added `docs/PERFORMANCE_PHASE1.md` as a safe profiling and optimization plan for live execution without changing scientific semantics.
+
+### Notes
+
+This beta promotion does not make CodonTrace a UI product and does not loosen the claim boundary. The core remains a dependency-free research library; Studio/API/Desktop work belongs in a separate consumer repository.
+
+## 0.3.0a2 — AGPL metadata correction alpha release
+
+### Changed
+
+- Updated release identity from `0.3.0a1` to `0.3.0a2`.
+- Updated package, citation, and runtime version metadata for the AGPL public alpha line.
+- Removed deprecated license classifier from `pyproject.toml` to comply with modern Python packaging license-expression behavior.
+- Kept `AGPL-3.0-or-later` as the package license expression.
+
+### Notes
+
+This release is a metadata/legal-packaging correction release. It does not change the scientific claim boundary: CodonTrace Genesis remains alpha research software and does not claim final peer-reviewed benchmark results, AGI, consciousness, or proven collective intelligence.
+
+## 0.3.0a1 — Alpha research release
+
+This release prepares CodonTrace Genesis for public alpha distribution as a deterministic research-software library.
+
+### Added
+
+- Causal mechanism surfaces for capsule ablation, capsule outcome windows, signal-memory-action links, skill-compression ablation, and child-outcome audits.
+- Role, territory, collective-task, role-ablation, heldout-partner, multi-agent contribution, source-reputation, counterfactual-replay, and extended OEE schema surfaces.
+- Runtime-wiring manifest digests for the new causal mechanism policies.
+- Integration audit helpers for public API, replay-policy coverage, evidence consistency, package hygiene, and reference hygiene.
+- CI and PyPI Trusted Publishing workflow templates.
+
+### Changed
+
+- Release identity is aligned to `0.3.0a1`.
+- Public documentation was rewritten for release-facing clarity and claim control.
+
+### Claim policy
+
+This is an alpha research-software release. It exposes deterministic primitives and evidence structures; it does not claim AGI, consciousness, artificial life, benchmark superiority, causal certainty, or autonomous open-ended discovery.

{codontrace-0.3.0a2 → codontrace-0.3.0b2}/CITATION.cff

@@ -1,13 +1,15 @@
 cff-version: 1.2.0
 message: "If you use CodonTrace Genesis, please cite the versioned software release."
 title: "CodonTrace Genesis"
-version: "0.3.0a2"
+version: "0.3.0b2"
 authors:
   - family-names: "Jamei"
     given-names: "Parvaz"
 license: "AGPL-3.0-or-later"
 repository-code: "https://github.com/Parvaz-Jamei/codontrace-genesis"
-abstract: "CodonTrace Genesis is a deterministic, library-first research-alpha toolkit for digital-evolution experiments, causal mechanism auditing, capsule-mediated transfer, skill-compression inheritance, role and collective-behavior instrumentation, and replayable evidence artifacts."
+doi: "10.5281/zenodo.20337435"
+date-released: "2026-09-11"
+abstract: "CodonTrace Genesis is a deterministic, library-first research-beta toolkit for digital-evolution experiments, causal mechanism auditing, capsule-mediated transfer, skill-compression inheritance, role and collective-behavior instrumentation, and replayable evidence artifacts."
 keywords:
   - digital evolution
   - causal mechanism auditing