Library test coverage — OutputManager integration, cyclic deps, serialization, filekit fallback

[djdarcy]

Library test coverage — OutputManager integration, cyclic deps, serialization, filekit fallback

Problem

core_lib and plan_lib shipped with solid unit tests (33 + 50 = 83 tests), but six areas identified during implementation remain untested:

1. OutputManager + renderer integration — THAC0 level filtering through the plan renderer is untested. render_plan() calls print_*() functions but no test verifies that verbosity levels actually filter plan output.

2. Cyclic dependency detection — Action.depends_on supports dependency chains, and _resolve_execution_order() handles them, but no test exercises a cycle (A→B→A). The handler exists but is unverified.

3. Serialization round-trip — Action and Plan are dataclasses with .asdict() support, but no test verifies asdict() → JSON → reconstruct for the "saved plans" promise.

4. dazzle_filekit fallback — plan_lib's file_ops.py has a conditional import for dazzle_filekit with a stdlib fallback. No test verifies the fallback path works when dazzle_filekit isn't installed.

5. Large plan stress — No test exercises plans with 50+ actions to verify performance of dependency resolution and rendering at scale.

6. ANSI rendering correctness — The renderer uses ANSI escape codes for terminal output, but no test verifies the output matches expected formatting (could use snapshot testing).

These gaps were identified in the 2026-03-02__18-33-23 gap analysis during the core_lib/plan_lib implementation session.

Proposed solution

A focused test module for each gap area:

# tests/test_lib_integration.py

class TestOutputManagerRendererIntegration:
    """Verify THAC0 level 0 shows summary, level 2 shows full details"""

class TestCyclicDependency:
    """Verify A→B→A raises CyclicDependencyError with clear message"""

class TestSerializationRoundTrip:
    """Verify Action/Plan survive asdict() → JSON → reconstruct"""

class TestFileKitFallback:
    """Verify plan_lib works without dazzle_filekit installed"""

class TestLargePlanPerformance:
    """Verify 100-action plan resolves and renders in <1 second"""

class TestANSIRendering:
    """Snapshot test for renderer output format"""

Implementation approach

These can be built incrementally — each test class is independent:

• Phase A (high value): Cyclic deps + serialization round-trip — these catch real bugs
• Phase B (medium value): OutputManager + renderer integration + filekit fallback — these catch integration gaps
• Phase C (low priority): Large plan stress + ANSI rendering — these catch edge cases

Acceptance criteria

• Cyclic dependency test exists and verifies error handling
• Serialization round-trip test: Action → asdict() → JSON string → dict → verify all fields preserved
• OutputManager integration test: renderer respects THAC0 verbosity levels
• dazzle_filekit fallback test: file_ops.py works when import fails
• All new tests pass alongside existing 83 tests
• At least one test per gap area from the list above

Related issues

• Refs Plan-Execute architecture — trustworthy --dry-run across all subcommands #53 — Plan-Execute epic (strengthens confidence before closing)
• Refs Plan-Execute Phase 0: core infrastructure — plan.py dataclasses and display #54 — Phase 0 core infrastructure (these are the Phase 0 test gaps)

Analysis

See 2026-03-02__20-16-20__full-postmortem_core-lib-plan-lib-and-session-discoveries.md for the gap identification during implementation.
See 2026-03-02__21-48-45__dev-workflow-process_missing-issues-and-plan-gaps.md for gap analysis (Gap 4).