You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
# Default collection: only tests/unit/. Running bare `pytest` from the repo
# root will NOT pick up tests/perf/ or eval/ -- those are opt-in only, since
# perf benchmarks need real hardware/models and eval scripts aren't pytest
# tests at all (see eval/README.md).
pythonpath = .
testpaths = tests/unit
markers =
perf: performance/latency benchmarks requiring real hardware (Jetson) and real models -- excluded from default test runs, invoke explicitly with `-m perf --benchmark-only` against tests/perf/
integration: real components wired together (contextvars + real SQLite + fake embedder) -- excluded from default test runs, invoke explicitly with `-m integration` against tests/integration/
stress: concurrency/load tests (real queue + threads + real timing) -- excluded from default test runs, invoke explicitly with `-m stress` against tests/stress/
# Fail loudly on unknown markers instead of silently ignoring typos.