Wave-3 decomposition: SQLite engine tasks (open/opts, seam+tx, mechanisms, integration, review gate) + core value-constructor pre-work; plan synced for waves-1-2 review + ADR-023 follow-through

2026-10-08 10:57:29 +00:00
1 parent 44637eea5b
commit 1269246faf
@@ -0,0 +1,74 @@
+---
+id: review-wave-3
+name: Review gate — wave 3 (SQLite engine)
+status: pending
+depends_on: [sqlite-engine-integration]
+scope: moderate
+risk: low
+impact: phase
+level: review
+tags: [wave-3, review]
+---
+
+## Description
+
+Review the SQLite engine before wave 4 (or 5) decomposes. Primary
+lens per the plan's review gates: **engine-vs-contract conformance** —
+the engine crate now carries the first real consumer surface, so this
+is the gate that checks the code against the pinned contract text
+line by line (the wave-1 gate's discipline, now against an
+implementation that must *honor* the surface, not just declare it).
+
+Check:
+
+- **Contract conformance**: every trait method's behavior against
+  core-contract.md's per-mechanism text — validation at entry points
+  (auto-commit AND tx paths), error arms (`Database`-only receivers,
+  `Codec` at payload seams, `LeadershipLost` vs clean stop), value
+  shapes (`Job`/`StreamEvent` field-for-field), the no-work-is-a-value
+  vocabulary (`None`/`false`/empty-`Vec` where pinned, never errors).
+- **ADR-023 follow-through**: the extent guards actually sit at the
+  trait-impl entry (not substrate-side patches); duration guards on
+  both lock call sites; plain-path open holds; `poll_interval` wired;
+  `encode_payload` consumed fallibly at every call site.
+- **Seam integrity**: drop = rollback under panic; writer-slot lease
+  released on every path (commit, rollback, drop, error);
+  `spawn_blocking` used consistently (no blocking call inside async
+  context, no connection moved across await).
+- **Substrate boundary**: still contract-blind (no `alkstore::` under
+  `src/substrate/`); the lint removal didn't leave dead ported code
+  behind; any new divergence registered in PROVENANCE.md.
+- **Backoff curve**: the engine-side formula matches ADR-010 §3's
+  pinned range and cap exactly (wave 4 re-derives it; wave 5 pins
+  equivalence — this gate checks the one implementation against the
+  text).
+- **Backlog column**: the engine's suite rows present, stamped, green.
+- **Family standards**: no panics, no unwrap/expect outside tests, no
+  inline comments (doc comments fine), module-per-file.
+- **Gates**: full workspace build/test/clippy/fmt.
+
+## Acceptance Criteria
+
+- [ ] Contract conformance spot-checked against the spec text (not
+      just the tests)
+- [ ] ADR-023's five backlog rows verified implemented as decided
+- [ ] Seam/lease/rollback paths verified (code-read, not test-name-trust)
+- [ ] Substrate boundary + provenance register still clean
+- [ ] All gates green
+- [ ] Findings recorded; wave 4/5 decomposition may proceed
+
+## References
+
+- docs/plans/implementation.md (Review gates)
+- docs/architecture/core-contract.md
+- docs/architecture/engine-sqlite.md
+- docs/architecture/decisions/023-fourth-review-round.md
+- docs/reviews/2026-10-08-waves-1-2-general-review.md (the posture this gate inherits)
+
+## Notes
+
+> To be filled by implementation agent
+
+## Summary
+
+> To be filled on completion