Development edition · 16 November 2026. Commands using repository paths, Make targets or sample projects require access to the Quinzee source checkout. No public installer is available. Examples use disposable local environments.
8. Testing and verification
Choose checks by the changed behavior. Build and CTest entry points are in AGENTS.md (source checkout required); the completion requirements remain mandatory. Documentation-only changes do not require rebuilding the application, but runnable examples must be checked against their actual SDK signatures and setup instructions.
| Check | Coverage |
|---|---|
make test |
HTTP/TLS, host/modules, services, deployment expressions/runtime contracts, queues, DNS, monitoring, telemetry, QHQ and dependency lifecycle |
make PRESET=release test |
The same contracts with hardened optimized binaries |
make PRESET=asan test |
Memory/undefined behavior, on a compatible host or isolated builder |
make docker-tsan TSAN_ALLOW_ASLR_CHANGE=1 |
Race checks with the explicitly authorized test-container policy described below |
make docker-lint |
C++ formatting and first-party Clang-Tidy diagnostics |
make docker-test |
Runtime images, HTTPS, constrained CAs, DNS/Azure names, persistence, JSON logs and official SDK calls |
make engines-test |
Real isolated native engines, TLS, lifecycle, persistence and owned deletion |
make service-bus-sdk-test |
Official .NET AMQP/TLS client, compiled Bicep, settlement, fan-out, SAS/bearer and unsupported operations |
make functions-test |
Real Functions host, HTTP key enforcement, Service Bus trigger, settings/lifecycle, helper recovery and deletion; requires the local lint runner image |
make manuals-check, make pdf |
Book assembly, links, assets, complete code inclusion and printable output |
make application-check |
Locked dependencies, .NET/React builds/style, Bicep, Playwright and automated accessibility |
make application-test |
Official .NET service clients, both SQL engines, Redis, queue recovery and cleanup |
make application-observability-test |
Real telemetry backends, cross-service correlation, source redaction and database/collector failures |
Tests use private temporary state, ephemeral listeners or isolated Compose projects. Do not point destructive cleanup tests at a persistent developer instance. Application failures retain private logs/results; avoid publishing those directories without redaction.
Use ctest --preset debug -R 'monitoring|telemetry' for selected targets, or
build/debug/tests/monitoring_unit_test retention_and_durable_failure for a
single harness case. The exact registered targets are in
tests/CMakeLists.txt (source checkout required).
ASan/UBSan and TSan are verified in containers where host-injected libraries interfere with sanitizer runtime startup. The TSan runner permits one exact personality argument inside its test container only, after explicit approval; it does not change host sysctls or use an unconfined filter. Read the TSan policy (source checkout required) before opting in.
The observability verification record (source checkout required) records the 23-target debug/release/sanitizer suites and SDK/backend checks. The application record (source checkout required) covers the browser and workflow contracts. These are dated evidence, not guarantees about untested architectures or every upstream SDK version. Manual assistive-technology checks remain distinct from automated axe/keyboard coverage.
The manual verification record (source checkout required) describes the C# tutorial checks, SDK-container setup and exporter sampling configuration.