Skip to content

TESTING.md

Latest commit

 

History

History
81 lines (64 loc) · 2.93 KB

TESTING.md

File metadata and controls

81 lines (64 loc) · 2.93 KB
Raw

Testing ArrayBridge

This is the practical test guide. The workflow files under .github/workflows are the authority for the current CI matrix.

Local core checks

python -m pip install -e ".[dev,docs]"
ruff check src tests
black --check src tests
mypy src stubs
pytest
python -m sphinx -E -W --keep-going -b html docs/source /tmp/arraybridge-docs

The default suite uses real optional frameworks when installed and otherwise exercises declared unavailable paths and focused fakes. Markers such as torch, tensorflow, jax, cupy, pyclesperanto, and gpu can select or exclude framework-specific tests.

Real GPU lattice check

On a GPU host, validate more than a single NumPy round trip:

  • import all installed frameworks in one fresh process;
  • enumerate each library's devices through ArrayLibrary.all() (available_devices());
  • convert a small exact array through every available source-target pair;
  • stack and unstack one, two, and several planes in every framework;
  • verify same-framework movement across multiple local devices when present;
  • confirm absent frameworks are not imported by discovery or cleanup;
  • exercise DLPack success and explicit CPU fallback paths;
  • target OOM cleanup to the device that owned execution.

Use a fresh process and unset TensorFlow/JAX allocator variables when checking that declaration-owned import defaults work. Framework warnings about shared CUDA plugin registration should be recorded separately from value, device, or shape failures.

Commands for a GPU machine

0.5.0's GPU run predates three later fixes, and 0.5.1 was not run on a GPU. On a machine with CUDA 12 and about 15 GB of free disk:

git clone https://github.com/OpenHCSDev/ArrayBridge.git && cd ArrayBridge
git checkout v0.5.1
uv venv -p 3.12 .venv-gpu
uv pip install -p .venv-gpu/bin/python -e ".[dev,gpu]"
.venv-gpu/bin/python -m pytest -q --no-cov -ra
.venv-gpu/bin/python -m pytest -q --no-cov -ra -m "gpu or cupy or torch or tensorflow or jax or pyclesperanto"
.venv-gpu/bin/python -m pytest -q --no-cov -ra tests/test_operation_soundness.py
.venv-gpu/bin/python host_call_check.py

tests/test_operation_soundness.py calls every array operation of every installed library on its first device; no skip should name an installed library. host_call_check.py is:

import torch
from arraybridge import torch as on_torch

@on_torch
def add_one(image):
    return image + 1

add_one(torch.ones(2, 3, 3))
assert not torch.cuda.is_initialized(), "a host call initialised CUDA"
if torch.cuda.is_available():
    assert add_one(torch.ones(2, 3, 3, device="cuda:0")).is_cuda
print("ok")

Interpreting CI

The main CI workflow runs the test suite across its declared Python and OS matrix. The manual GPU-named workflow installs CPU-capable Torch and JAX on a standard runner; it does not prove CUDA behavior. Real multi-framework and multi-device results therefore remain a required release check on suitable hardware.