Skip to main content

test

Run Polygraphy tests with the correct environment setup. Activates when asked to run tests, run the test suite, or run a specific test file. Also use this proactively after making code changes — e.g. after a refactor, bug fix, or any edit that could affect runtime behaviour — to verify nothing is broken before committing.

소스 정보

저장소
NVIDIA/TensorRT
최근 소스 활동
2026년 9월 22일 19:37
감지된 SKILL.md 언어
영어
스타
13,438
포크
2,421

설치 방법

기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.

소스 파일 검토

설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.

SKILL.md 표시 중

SKILL.md
소스 지침 · 읽기 전용 미리보기
name
test
description
Run Polygraphy tests with the correct environment setup. Activates when asked to run tests, run the test suite, or run a specific test file. Also use this proactively after making code changes — e.g. after a refactor, bug fix, or any edit that could affect runtime behaviour — to verify nothing is broken before committing.
# Run Polygraphy Tests ## Required environment variables Set these before running any pytest command (from the Makefile `test` target): - `PYTHONPATH` must include the repo root - `POLYGRAPHY_INTERNAL_CORRECTNESS_CHECKS=1` - `CUDA_MODULE_LOADING=LAZY` ## Running a specific test file or subset ```bash PYTHONPATH="$(pwd):${PYTHONPATH}" \ POLYGRAPHY_INTERNAL_CORRECTNESS_CHECKS=1 \ CUDA_MODULE_LOADING=LAZY \ python3 -m pytest <path/to/test_file.py> -v --durations=15 --failed-first --new-first --script-launch-mode=subprocess ``` ## Running the full test suite Serial tests must run before parallel ones: ```bash PYTHONPATH="$(pwd):${PYTHONPATH}" \ POLYGRAPHY_INTERNAL_CORRECTNESS_CHECKS=1 \ CUDA_MODULE_LOADING=LAZY \ python3 -m pytest tests/ -m "serial and not slow" -v --durations=15 --failed-first --new-first --script-launch-mode=subprocess && \ python3 -m pytest tests/ -n 8 --dist=loadscope -m "not serial and not slow" -v --durations=15 --failed-first --new-first --script-launch-mode=subprocess ``` ## Notes - Always pass `--script-launch-mode=subprocess` to avoid in-process runner issues with stdout/stderr capture. - Skip `-n 8 --dist=loadscope` when running a small subset. - `slow` tests are excluded by default; drop `and not slow` to include them. - **Use this skill proactively** after any code change (refactor, bug fix, new feature) that could affect runtime behaviour — run the relevant test file before committing, without waiting for the user to ask.
GitHub에서 보기