Hushwheel Quality Instrumentation¶
- Audit date:
2026-03-18(Asia/Tbilisi) - Repository root:
/home/standard/dspy_rag_in_repo_docs_and_impl1 - Working branch during verification:
codex/hushwheel-quality-instrumentation-20260318
Scope¶
This audit captures the hushwheel follow-up work that expands the fixture-local verification stack with harder build flags, sanitizer reruns, profiling logs, richer persisted reports, and the matching repository/CI surface updates.
The scope includes:
- new hushwheel
Makefiletargets for static analysis, complexity, coverage, reports, and a combinedqualitygate - new hushwheel
Makefiletargets forhardening,sanitizers, andprofiling - ELF hardening audits that persist
file,size,readelf, andnmoutput for the hardened hushwheel binary - AddressSanitizer plus UndefinedBehaviorSanitizer reruns with persisted unit, integration, and BDD logs
- repeated CLI workload profiling with raw timing tables, Markdown summaries, and sample command logs
- report-aware integration and BDD entrypoints that accept
HUSHWHEEL_BIN - stronger CLI coverage assertions for spoke entries, no-match cases, missing arguments, and unknown commands
- a dedicated
.github/workflows/hushwheel-quality.ymlworkflow that runs only when hushwheel surfaces change, now with explicitbinutilsinstallation for the binary-audit commands - updated hushwheel README and testing docs plus repository workflow tests that pin the new CI surface
Executed Commands¶
Executed successfully in this turn:
make -C tests/fixtures/hushwheel_lexiconarium hardeningmake -C tests/fixtures/hushwheel_lexiconarium sanitizersmake -C tests/fixtures/hushwheel_lexiconarium profilingmake -C tests/fixtures/hushwheel_lexiconarium qualityuv run python -m compileall src testsuv run pytest tests/test_utilities.py tests/test_repository_rag_bdd.pyuv run repo-rag smoke-testcargo build --manifest-path rust-cli/Cargo.tomlmake verify-surfacesuv run pytest tests/test_hushwheel_fixture.py tests/test_hushwheel_program_surface.py tests/test_project_surfaces.pyuv run ruff format tests/test_hushwheel_program_surface.py tests/test_project_surfaces.py tests/fixtures/hushwheel_lexiconarium/tools/lint_hushwheel.pyuv run ruff format tests/test_hushwheel_fixture.pymake qualitymake hooks-install
Results¶
make -C tests/fixtures/hushwheel_lexiconarium hardening: passed- hardened build flags now include
_FORTIFY_SOURCE=3, stack protector, stack-clash protection, PIE, RELRO,BIND_NOW, non-executable stack, frame pointers, and-Werror=format-security - persisted ELF audit under
tests/fixtures/hushwheel_lexiconarium/build/reports/hardening/reports:PIE:passRELRO segment:passBIND_NOW:passNon-executable stack:passStack protector symbol:not-detected
make -C tests/fixtures/hushwheel_lexiconarium sanitizers: passed- AddressSanitizer plus UndefinedBehaviorSanitizer reran the unit, integration, and BDD surfaces without findings
- persisted logs under
tests/fixtures/hushwheel_lexiconarium/build/reports/sanitizers/ make -C tests/fixtures/hushwheel_lexiconarium profiling: passed- persisted raw timing data at
tests/fixtures/hushwheel_lexiconarium/build/reports/profiling/runtime-profile.tsv - profiled
25iterations each ofabout,stats, canonical and generatedlookup, canonical and generatedprefix, andcategory storm-index - measured means from the current run:
stats:10.137 msabout:11.021 msprefix-canonical:11.065 mslookup-generated:12.657 msprefix-generated:12.914 mscategory-storm:13.314 mslookup-canonical:16.907 ms
make -C tests/fixtures/hushwheel_lexiconarium quality: passedcppcheck: passed with XML plus GCC-style text reports undertests/fixtures/hushwheel_lexiconarium/build/reports/cppcheck/lizard: passed with text, CSV, and checkstyle XML reports undertests/fixtures/hushwheel_lexiconarium/build/reports/complexity/- hardened binary audit, sanitizer logs, and runtime profile reports were added under:
tests/fixtures/hushwheel_lexiconarium/build/reports/hardening/tests/fixtures/hushwheel_lexiconarium/build/reports/sanitizers/tests/fixtures/hushwheel_lexiconarium/build/reports/profiling/
gcovr: passed with persisted text, Cobertura XML, HTML, JSON, and Markdown summary outputs undertests/fixtures/hushwheel_lexiconarium/build/reports/coverage/- executable coordinator coverage reached:
line_percent: 100.0function_percent: 100.0branch_percent: 98.6
uv run python -m compileall src tests: passeduv run pytest tests/test_utilities.py tests/test_repository_rag_bdd.py: passed,13 passeduv run repo-rag smoke-test: passed with:answer_contains_repository: truemcp_candidate_count: 1manifest_path: artifacts/azure/repo-rag-smoke.jsoncargo build --manifest-path rust-cli/Cargo.toml: passedmake verify-surfaces: passed withissue_count: 0- changed-surface pytest slice: passed,
26 passed make quality: passed,121 passedand repository coverage87.56%- the first repo-wide
make qualityrerun exposed two real blockers in hushwheel-adjacent Python surfaces:tests/test_hushwheel_fixture.pyneeded repository-standardruff formattests/fixtures/hushwheel_lexiconarium/tools/regenerate_hushwheel_fixture.pycarried a staleF541suppression
- both were fixed in this turn and the subsequent
make qualitypass succeeded make hooks-install: passed and refreshed managedpre-commitpluspre-pushhooks- post-push GitHub Actions evidence:
- run
23238384378(Hushwheel Quality, head108754e5d1a05e76d15e94e7efa0d1ee15925daa) failed in47s- fixture quality suite, report snapshot, quality-summary publish, and artifact upload passed
- repository surface step failed because the local hushwheel generator/corpus sync in
docs/catalog.mdplus the eightsrc/hushwheel_spoke_*.cfiles had not been included in the first scoped commit
- follow-up run
23238475251(Hushwheel Quality, headbfe74cb96bcfb79f8076eabcaaac53af1c16ecc3) succeeded in44s- every workflow step passed, including artifact upload
- the workflow still emits the existing Node.js 20 deprecation annotation for
actions/checkout@v4,actions/upload-artifact@v4, andastral-sh/setup-uv@v6
Current Verification Status¶
Configured and exercised in this turn:
- hushwheel fixture lint, unit, integration, BDD, static analysis, complexity, and coverage gates
- hushwheel hardened-build verification with persisted binary-analysis artifacts
- hushwheel sanitizer reruns with persisted command logs
- hushwheel runtime profiling with raw timing tables and sample CLI logs
- persisted hushwheel report generation for local and CI artifact upload
- hushwheel workflow parsing and path-filter coverage in repository tests
- compile checks
- repository surface verification
- focused repository pytest verification for hushwheel fixture, hushwheel program surfaces, and project workflow surfaces
- full repository quality gate with lint, type checks, tests, and coverage
- Rust wrapper build
Not exercised in this turn:
- live Azure endpoint probes against remote services
- a clean-tree file-summary regeneration, because no new tracked files were added during this pass
Notes¶
- The hushwheel coverage report intentionally excludes
src/hushwheel_spoke_*.candsrc/hushwheel_spokes.c, because those generated tables are data-heavy catalog surfaces rather than executable control-flow logic. - The repository-wide
make qualityrun now passes, but getting there required one formatted test refresh and removal of one stale hushwheel-generator lint suppression that the tighter pass surfaced. - The first post-push hushwheel workflow run failed for a repository-surface consistency reason: the generated hushwheel catalog and spoke files were already updated in the local worktree, but they were omitted from the initial scoped commit. The follow-up commit synced those generated surfaces and the rerun passed.
- The fixture-local report tree is intentionally regenerated after the repository pytest slice, because the existing packaging/docs tests clean the hushwheel build directory as part of their normal fixture contract.
- The repository worktree contained unrelated pending changes before this task continued. Those changes were not reverted during this audit.