Kind of improvement
test
Affected module
not applicable (maintained C++ algorithms and error paths)
Affected area
area/testing
Summary
Continue the maintained-code coverage stage of #2884 using the repository-scoped measurement foundation merged in #2887. Keep algorithm/test changes separate from CI infrastructure.
Prioritize deterministic edge/error paths in active MCI recursion, RaBitQ/SIMQ variants, HGraph/IVF serialization compatibility and corruption handling, API/factory validation, lifecycle failures, and IO/cache failures. Audit MCI's unused legacy no-argument enumerator before deciding whether removal merits a separate algorithm change; do not execute dead code merely to increase coverage. Deprecated HNSW and DiskANN are explicit non-goals.
Acceptance:
Expected benefit
Improve regression protection and reach the parent's 90% objective through useful tests and reviewed dead-code decisions, independently of the pre-merge CI rollout.
Benchmark evidence
The historical gap analysis is in #2884. Post-foundation full reports were still pending when this stage was split; no new full-suite percentage is claimed.
References
Related issues
Duplicate search for coverage error paths returned #2884, #1949, #2447, #2582, and #2661. This issue tracks the coverage stage of #2884; dependency removal and new IO/analyzer features are outside scope.
Drafted with AI assistant: Codex:GPT-6
Kind of improvement
test
Affected module
not applicable (maintained C++ algorithms and error paths)
Affected area
area/testing
Summary
Continue the maintained-code coverage stage of #2884 using the repository-scoped measurement foundation merged in #2887. Keep algorithm/test changes separate from CI infrastructure.
Prioritize deterministic edge/error paths in active MCI recursion, RaBitQ/SIMQ variants, HGraph/IVF serialization compatibility and corruption handling, API/factory validation, lifecycle failures, and IO/cache failures. Audit MCI's unused legacy no-argument enumerator before deciding whether removal merits a separate algorithm change; do not execute dead code merely to increase coverage. Deprecated HNSW and DiskANN are explicit non-goals.
Acceptance:
Expected benefit
Improve regression protection and reach the parent's 90% objective through useful tests and reviewed dead-code decisions, independently of the pre-merge CI rollout.
Benchmark evidence
The historical gap analysis is in #2884. Post-foundation full reports were still pending when this stage was split; no new full-suite percentage is claimed.
References
.github/workflows/coverage.yml:29Related issues
Duplicate search for coverage error paths returned #2884, #1949, #2447, #2582, and #2661. This issue tracks the coverage stage of #2884; dependency removal and new IO/analyzer features are outside scope.
Drafted with AI assistant: Codex:GPT-6