nightshift: test-gap

[nightshift-micr]

Test Gap Analysis: Microck/boilify

nightshift automated analysis — test-gap task

Summary

Analyzed test coverage for Boilify — an OpenFX video effect plugin (C++, 401 lines in Boilify.cpp). The project has zero tests — no test files, no test framework, no CI test step.

Repository Structure

File	Lines	Purpose
Boilify.cpp	401	Single source: OpenFX plugin with Perlin noise distortion
CMakeLists.txt	85	Build config (no test targets)
tuner/app.js	~50	Web-based parameter tuner UI
tuner/index.html	~30	Tuner HTML shell
.github/workflows/	3 files	Release builds for Linux/macOS/Windows

Test Gap Assessment

Critical Gaps (No Coverage)

Component	Risk	Recommended Test Type	Priority
fastFloor()	🟡 Medium	Unit test — boundary values (negative, zero, positive floats)	High
mixBits()	🟡 Medium	Unit test — known input/output pairs for hash function	High
hashFast()	🟡 Medium	Unit test — distribution, determinism, seed isolation	High
perlinFast()	🟡 Medium	Unit test — output range [0,1], continuity, seed variation	High
applyNoiseType()	🟢 Low	Unit test — smooth passthrough, ridged transform	Medium
fbm()	🟡 Medium	Unit test — octaves, amplitude normalization, noise types	High
renderSlice()	🔴 High	Integration test — pixel output with controlled inputs	Critical
onLoad()	🔴 High	Integration test — null host, missing suites	High
describe()	🟡 Medium	Integration test — property setup correctness	Medium
describeInContext()	🟡 Medium	Integration test — parameter ranges and defaults	Medium
render()	🔴 High	Integration test — end-to-end with mock OpenFX host	Critical

Test Infrastructure Gap

Need	Status	Recommendation
Test framework	❌ None	Add a simple C++ test runner (Catch2, doctest, or Google Test)
Mock OpenFX host	❌ None	Create minimal mock OfxHost with stubbed suites for unit testing
CI test step	❌ None	Add test execution to GitHub Actions workflows
Coverage reporting	❌ None	Add gcov/lcov for coverage metrics

Specific Test Cases to Write

Unit Tests (no OpenFX dependency needed)

1. fastFloor edge cases: fastFloor(-0.5) → -1, fastFloor(0.0) → 0, fastFloor(1.9) → 1, fastFloor(-1.0) → -1
2. hashFast determinism: Same inputs → same output; different seed → different output
3. hashFast distribution: 10k samples should have approximately uniform distribution in [0, 1)
4. mixBits avalanche: Single bit change in input should change ~50% of output bits
5. perlinFast range: Output should always be in [0, 1) for any input
6. fbm normalization: With 3 octaves, output should be normalized to approximately [0, 1]
7. fbm octaves: More octaves should produce different (more detailed) output
8. applyNoiseType smooth: Should pass value through unchanged
9. applyNoiseType ridged: Should transform to 1.0 - |2*v - 1| pattern

Integration Tests (require mock OpenFX host)

10. render passthrough: With strength=0, output should equal input
11. render bounds clamping: Displaced pixels should never read outside source rect
12. render seed stability: Same seed + time → identical output (deterministic)
13. render pixel depth: Should handle byte, short, and float pixel depths correctly
14. render multithreading: Output should be identical for 1 thread vs N threads
15. onLoad missing host: Should return kOfxStatErrMissingHostFeature when gHost is NULL
16. describeInContext parameter ranges: Verify strength [0,100], size [1,600], seed [0,9999]

Build Test Gaps

17. Build verification: CI workflows build but don't verify the .ofx plugin loads correctly
18. Cross-platform consistency: No test that the tuner web app matches plugin parameter ranges

Quantitative Assessment

Metric	Value
Source files	1 (Boilify.cpp)
Test files	0
Lines of code	401
Lines of test code	0
Test coverage	0%
Functions/methods	12
Tested functions	0
Test-to-code ratio	0:1

Recommendations

1. Priority 1: Create a test harness with a mock OpenFX host for renderSlice testing
2. Priority 2: Add unit tests for pure functions (fastFloor, hashFast, perlinFast, fbm, applyNoiseType) — these require no mocking
3. Priority 3: Add Catch2 or doctest as a header-only test framework (minimal build impact)
4. Priority 4: Add CMake test target and integrate with CI
5. Priority 5: Test the tuner app's parameter ranges match the plugin's defined ranges

Notes

• The pure math functions (fastFloor, hashFast, perlinFast, fbm) are highly testable in isolation — no OpenFX dependency needed
• The renderSlice function is the most critical to test — it handles pixel memory directly and has bounds-checking logic that could silently fail
• The OpenFX plugin architecture makes full integration testing challenging without a host application (DaVinci Resolve, Nuke, etc.), but mock-based testing covers the logic layer