- Planned
- Assigned
- In Progress
- Blocked (reason: ...)
- Complete
medium
subagent (general)
- Test oracle design:
research/zig-cr/10-test-oracle.md - Feature matrix:
research/zig-cr/90-feature-matrix.md - Gap backlog:
research/zig-cr/92-gap-backlog.md - Existing test harnesses:
zig/harness/ - C oracle tests:
core/c/
Design and implement additional tests to systematically expose any gaps in the Zig implementation that aren't caught by existing test suites.
Focus areas:
- Edge cases and error conditions: Test boundary conditions, error handling, and failure modes
- Multi-table scenarios: Complex interactions between multiple CRR tables
- Merge conflict resolution: All tie-breaker paths (cl/col_version/value/site_id)
- Clock semantics: dbVersion/pendingDbVersion/seq invariants across transactions
- Serialization formats: PK blob packing/unpacking edge cases
- Virtual table operations: xUpdate paths, transaction boundaries, best-index scenarios
- Hook chaining: Commit/rollback hook interactions with other extensions
- Schema evolution: as_crr/as_table with different table configurations
- Performance characteristics: Large UNION query scaling, statement cache effectiveness
Each test should:
- Be deterministic and reproducible
- Have clear pass/fail criteria
- Document what functionality it's validating
- Run quickly (avoid long-running stress tests in CI)
zig/harness/(new test scripts or extensions to existing ones)zig/src/(add unit tests in relevant modules)research/zig-cr/92-gap-backlog.md(document test coverage improvements)
- Identify at least 5-10 additional test scenarios not covered by existing tests
- Document the test design in the
Progress Logwith rationale - Implement the new tests
- Tests successfully expose at least one previously undetected gap (if any exist)
- All new tests pass (or are marked as expected failures with task cards created for fixes)
- Test documentation explains what each test validates and why it's important
- Tests are integrated into existing test harness (make test-* targets)
- Task created to improve test coverage and expose implementation gaps
- Current test status: 154/154 MVP tests passing per gap backlog
- C oracle harness: 19/20 tests pass (fract_as_ordered stub causes 1 failure)
- Need to design tests that go beyond existing coverage
[fill in when done]