Pytest plugin to skip tests marked as slow by default

Hello pytest team, it seems that pytest_configure is not compatible with --pyargs. Can you confirm ? here is my function in conftest.py (I confirm that conftest.py is in the root of the package): d...

今回はpytestを使用したテスト方法を紹介します 名称バージョンPython3.12.3pytest8.4.2 pytestでテストするための準備 pytestには実施するためにいくつか決まり事があ

pytest 9.2.0 phát hành ngày 27/10/2026, với thay đổi lớn nhất ở @pytest.mark.parametrize: ba tham số indirect, ids và scope giờ chỉ nhận theo từ khóa — truyền theo vị trí sẽ lỗi TypeError ngay lúc thu thập test. Bản này cũng phân loại lỗi import plugin rõ ràng hơn. Nếu dự án của bạn dùng parametrize ở hàng trăm chỗ, hãy đọc kỹ trước khi nâng cấp: thay đổi tuy nhỏ nhưng đủ làm đỏ pipeline CI nếu code cũ truyền tham số theo vị trí — đúng kiểu lỗi mà pytest 9.2 muốn dẹp bỏ.

The maker’s benchmark log includes a useful complication: an earlier accelerated pytest run failed tests because it reused the wrong application module.

코딩 에이전트를 쓰다 보면 어제 정한 걸 오늘 세션에서 다시 설명하게 됩니다. "테스트는 pytest로", "이 레포는 마이그레이션 손으로 작성", "그 방식은 지난주에 해 봤는데 안 됐음" 같은 것들이요. 에이전트를 바꾸면(Claude Code → Codex) 처음부터 다시입니다. Heirmos는 이걸 자동으로 이...

The maker’s benchmark log includes a useful complication: an earlier accelerated pytest run failed tests because it reused the wrong application module.

This matches my experience fine-tuning a YOLO26n-cls classifier: abstention is where most evals break. What worked for me: * Hold out whole sessions, not random rows. Adjacent frames leak; random splits overstate accuracy by 20-30 points. * Per-class metrics: overall accuracy hid that my minority class was at 40% while the majority class hit 100%. * A CI-safe eval harness: recorded case files replayed in pytest, no API calls during tests. Abstention is worth measuring separately from accuracy. What rate does the model defer versus make something up? That number moved my decisions more than accuracy did.

This matches my experience training my first classifier: * Temporal validation, holding out whole sessions rather than random frames, dropped my measured accuracy by more than 20 points. * Per-class metrics matter: a single overall accuracy figure hid a minority class doing badly. * I keep evaluation separate from training by replaying recorded case files in pytest, so tests never call the model live. A compass gate sounds like it’s doing something similar for self-updating models: keeping the evaluation gate frozen while the weights change. I’d be curious how you handle the eval set drifting as the model improves.

Some date bugs only show up on specific days: Feb 29, the last day of a month, New Year's Eve, a DST change. You can't wait for those dates to test them, so the bugs ship. pytest-calendar-chaos runs a marked test once for each of those moments, with the clock frozen using time-machine: @pytest.mark.calendar_chaos( cases=["leap_day", "month_end", "dst"], zones=["UTC", "America/New_York"] ) def test_subscription_renewal(): ... The presets cover leap day, month end, year end, midnight, ISO week 53, the 2038 limit and DST changes. When a case fails, the report shows the exact mocked time and a command that reruns only that scenario. It's opt-in, so unmarked tests don't change. pip install pytest-calendar-chaos It needs Python 3.10+ and pytest 8+. It's MIT licensed and at version 0.3.0 on PyPI. It's a small project, so bug reports are the most useful thing you can send me. Repo: https://github.com/nsijwali/pytest-calendar-chaos

Problem & Context Currently, the RunRecap card summarizes touched files (write_file, edit_file, delete_file) and git diff stats at task completion. However, users on mobile often want to see which ...