Honor tracker conflict tiebreak in deterministic status policy

[gmickel]

Honor tracker conflict tiebreak in deterministic status policy

GitHub #268 exposed a contract gap: tracker.conflictTiebreak was configured and documented, but the deterministic status policy always returned status-deadlock. This change wires the existing setting into the shared provider-neutral decision path.

Spec: fn-146-honor-tracker-conflict-tiebreak-in
Branch: fn-146-honor-tracker-conflict-tiebreak-in → origin/main
Tasks: 1 completed
R-ID coverage: 7/7 satisfied

TL;DR

• Honors always-ask, flow-wins, and tracker-wins in the shared deterministic status policy.
• Rejects malformed persisted values before claims or sequence work, with zero side effects.
• Preserves merge-evidence gating, provider-neutral writes, local terminal folds, and candidate-bearing conflicts.
• Adds direct-status, facade, receipt, provider-parity, and distribution regression coverage.
• Updates canonical, Codex-mirror, bundled, and public documentation surfaces.

R-ID coverage

R-ID	Acceptance criterion	Task	Evidence
R1	always-ask preserves candidate-bearing conflicts for both terminal×active orientations with no mutation.	fn-146-honor-tracker-conflict-tiebreak-in.1	30341e0, c6db2eb
R2	flow-wins applies Flow state through the provider-neutral write path with merge-evidence gating.	fn-146-honor-tracker-conflict-tiebreak-in.1	30341e0, c6db2eb
R3	A terminal tracker uses the existing local fold and pulled/aggregate receipt path.	fn-146-honor-tracker-conflict-tiebreak-in.1	30341e0, c6db2eb
R4	An active tracker against terminal Flow returns a typed, candidate-bearing unrepresentable conflict.	fn-146-honor-tracker-conflict-tiebreak-in.1	30341e0, c6db2eb
R5	Missing config defaults safely; malformed runtime and CLI values fail with strict validation before mutation.	fn-146-honor-tracker-conflict-tiebreak-in.1	30341e0, c6db2eb
R6	Regression coverage spans both orientations, all policies, direct/facade paths, side effects, receipts, and providers.	fn-146-honor-tracker-conflict-tiebreak-in.1	30341e0, c6db2eb
R7	Runtime copies, manifests, mirrors, docs, changelog, focused/full tests, and Ruff remain aligned.	fn-146-honor-tracker-conflict-tiebreak-in.1	30341e0, c6db2eb

Verification

fn-146-honor-tracker-conflict-tiebreak-in.1 — cd plugins/flow-next/tests && python3 -m unittest test_tracker_status test_tracker_facade test_tracker_config -q · cd plugins/flow-next/tests && python3 -m unittest test_tracker_status test_tracker_facade test_tracker_config test_tracker_distribution -q · python3 scripts/run_tests_parallel.py · uvx ruff@0.16.0 check plugins/flow-next/scripts/flowctl.py plugins/flow-next/scripts/flowctl_tracker/status/policy.py plugins/flow-next/scripts/flowctl_tracker/status/verb.py plugins/flow-next/scripts/flowctl_tracker/facade/ops.py plugins/flow-next/tests/test_tracker_status.py plugins/flow-next/tests/test_tracker_facade.py plugins/flow-next/tests/test_tracker_config.py · uvx ruff@0.16.0 check . · ./scripts/sync-codex.sh (twice; idempotent and validation guards green) · cd /Users/gordon/work/flow-next.dev && pnpm build

Critical changes

• High-churn: plugins/flow-next/tests/test_tracker_status.py (+205/-1 lines)
• High-churn: plugins/flow-next/tests/test_tracker_facade.py (+88/-0 lines)
• High-churn: plugins/flow-next/scripts/flowctl_tracker/status/policy.py (+67/-1 lines)
• High-churn: .flow/specs/fn-146-honor-tracker-conflict-tiebreak-in.json (+120/-0 lines)
• High-churn: .flow/specs/fn-146-honor-tracker-conflict-tiebreak-in.md (+91/-0 lines)

How to review this PR

The pipeline verified all 7 R-IDs, 194 focused tests, the 3,286-test full suite, pinned Ruff, distribution parity, two idempotent Codex sync passes, and the public-doc build. Cross-model implementation and spec-completion reviews both returned SHIP. Human judgment should focus on authority selection at the deadlock boundary, fail-fast placement, and whether the unrepresentable mirror remains safely non-mutating.

Review plan

Must review (~13%)

• plugins/flow-next/scripts/flowctl_tracker/status/policy.py — high-churn authority-selection logic. Do all three policies preserve deadlock-first ordering and the unrepresentable conflict boundary in decide()?
• plugins/flow-next/scripts/flowctl_tracker/status/verb.py — runtime status path. Does status() validate before its claim and preserve existing apply/apply-local persistence semantics?
• plugins/flow-next/scripts/flowctl_tracker/facade/ops.py — lifecycle entry paths. Do op_push(), op_pull(), and op_reconcile() reject malformed config before sequence work?
• plugins/flow-next/scripts/flowctl.py — CLI validation. Does cmd_config_set() retain the strict enum error contract?

Spot-check

• plugins/flow-next/tests/test_tracker_status.py — both collision orientations, all policies, provider parity, receipts, and no-mutation assertions.
• plugins/flow-next/tests/test_tracker_facade.py — facade consumption and fail-fast side-effect matrix.
• plugins/flow-next/tests/test_tracker_config.py — strict CLI enum cases.
• plugins/flow-next/docs/flowctl.md, plugins/flow-next/docs/tracker-sync.md, plugins/flow-next/skills/flow-next-tracker-sync/references/status-sync.md, and CHANGELOG.md — runtime/documentation agreement.
• plugins/flow-next/scripts/flowctl_tracker/MANIFEST.json — distribution hashes.

Safe to skim

• .flow/bin/flowctl.py, .flow/bin/flowctl_tracker/facade/ops.py, .flow/bin/flowctl_tracker/status/policy.py, and .flow/bin/flowctl_tracker/status/verb.py — propagated runtime copies, parity-tested.
• .flow/bin/flowctl_tracker/MANIFEST.json — generated distribution state.
• plugins/flow-next/codex/skills/flow-next-tracker-sync/references/status-sync.md — regenerated by sync-codex.sh, guard-verified.
• .flow/specs/fn-146-honor-tracker-conflict-tiebreak-in.json, .flow/specs/fn-146-honor-tracker-conflict-tiebreak-in.md, .flow/tasks/fn-146-honor-tracker-conflict-tiebreak-in.1.json, and .flow/tasks/fn-146-honor-tracker-conflict-tiebreak-in.1.md — Flow task/spec state.

Memory left behind

Bugs captured during this spec:

• bug/integration/caller-facade-guards-must-cover-retro-2026-07-29 — The primary Capture lifecycle touchpoint used the fn-140 facade, but its bounded MISSING-receipt retro-fire still instructed the removed skill-dispatch grammar.
• bug/integration/caller-fakes-must-enforce-lifecycle-2026-07-29 — The tracker caller execution harness accepted incomplete lifecycle facade calls and returned no output, so active-path tests could pass while omitting required operation inputs or ignoring a JSON stream leak.
• bug/integration/caller-oracle-must-preserve-historical-2026-07-29 — The first caller oracle normalized desired future event tags into the historical baseline and stored active observations as prose.
• bug/integration/tracker-ownership-rewrites-require-2026-07-29 — The deterministic tracker ownership rewrite passed focused tests but review found stale adjacent contracts.

Glossary / strategy notes

• Spec-driven team patterns** — makes the documented tracker projection contract executable and auditable.
• Ralph autonomous mode** — preserves always-ask defer semantics while allowing explicitly configured deterministic recovery.
• Cross-platform parity** — fixes the shared provider-neutral path and propagates every bundled copy.
• Self-improving through normal work** — converts a live release-recovery finding into a regression-backed deterministic contract.

Closes #268

Generated by /flow-next:make-pr from fn-146-honor-tracker-conflict-tiebreak-in against origin/main on 2026-07-29.

[gmickel]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Bravo.

Reviewed commit: 86d5c4a33c

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".