Skip to content

Handle Numba cleanup crash in GitHub Actions workflow #28

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
deepentropy merged 2 commits into from
Dec 2, 2025
Merged

Handle Numba cleanup crash in GitHub Actions workflow #28

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d5428b6
Initial plan
Copilot Dec 2, 2025
acacf1c
Fix GitHub Actions workflow cleanup crash after tests pass
Copilot Dec 2, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 34 additions & 2 deletions .github/workflows/test.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,13 +37,45 @@ jobs:

- name: Run unit tests
run: |
pytest -v --ignore=tests/accuracy --ignore=tests/benchmark -x
# Run pytest and capture output, allowing the command to fail
set +e
pytest -v --ignore=tests/accuracy --ignore=tests/benchmark -x 2>&1 | tee pytest_output.txt
PYTEST_EXIT_CODE=$?
set -e

# Check if all tests passed by looking at the pytest summary line
if grep -qE "^=+ .* passed" pytest_output.txt; then
Copy link

Copilot AI Dec 2, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The grep pattern ^=+ .* passed will match pytest summary lines that contain both "passed" and "failed" (e.g., "5 passed, 10 failed"). This could cause the workflow to incorrectly treat partial failures as success when a cleanup crash occurs.

Consider using a more precise pattern that ensures ALL tests passed:

if grep -qE "^=+ [0-9]+ passed in " pytest_output.txt && ! grep -qE " failed" pytest_output.txt; then

Or check for the absence of failures:

if grep -qE "^=+ .* passed" pytest_output.txt && ! grep -qE "^=+ .* failed" pytest_output.txt; then

Copilot uses AI. Check for mistakes.
# If tests passed but exit code is non-zero (cleanup crash), treat as success
if [ "$PYTEST_EXIT_CODE" -eq 134 ] || [ "$PYTEST_EXIT_CODE" -eq 139 ]; then
Copy link

Copilot AI Dec 2, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[nitpick] Consider adding a comment explaining why specific exit codes (134 = SIGABRT, 139 = SIGSEGV) are being handled. This would improve maintainability and help future developers understand the workaround.

Example:

# Exit codes: 134 = SIGABRT (free() corruption), 139 = SIGSEGV (segfault)
# These indicate Numba cleanup crashes that occur after all tests pass
if [ "$PYTEST_EXIT_CODE" -eq 134 ] || [ "$PYTEST_EXIT_CODE" -eq 139 ]; then

Copilot uses AI. Check for mistakes.
echo "::warning::Pytest cleanup crashed (exit code $PYTEST_EXIT_CODE) but all tests passed. This is a known Numba issue."
exit 0
fi
fi

# If we got here and exit code is non-zero, it's a real failure
if [ "$PYTEST_EXIT_CODE" -ne 0 ]; then
echo "Tests failed with exit code $PYTEST_EXIT_CODE"
cat pytest_output.txt
exit $PYTEST_EXIT_CODE
fi

- name: Run tests with coverage
if: matrix.python-version == '3.12'
run: |
pip install pytest-cov
pytest --cov=numta --cov-report=xml --ignore=tests/accuracy --ignore=tests/benchmark
set +e
pytest --cov=numta --cov-report=xml --ignore=tests/accuracy --ignore=tests/benchmark 2>&1 | tee pytest_cov_output.txt
PYTEST_EXIT_CODE=$?
set -e

if grep -qE "^=+ .* passed" pytest_cov_output.txt; then
Copy link

Copilot AI Dec 2, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[nitpick] Consider adding a comment explaining why specific exit codes (134 = SIGABRT, 139 = SIGSEGV) are being handled. This would improve maintainability and help future developers understand the workaround.

Example:

# Exit codes: 134 = SIGABRT (free() corruption), 139 = SIGSEGV (segfault)
# These indicate Numba cleanup crashes that occur after all tests pass
if [ "$PYTEST_EXIT_CODE" -eq 134 ] || [ "$PYTEST_EXIT_CODE" -eq 139 ]; then
Suggested change
if grep -qE "^=+ .* passed" pytest_cov_output.txt; then
if grep -qE "^=+ .* passed" pytest_cov_output.txt; then
# Exit codes: 134 = SIGABRT (free() corruption), 139 = SIGSEGV (segfault)
# These indicate Numba cleanup crashes that occur after all tests pass.
# If all tests passed, we ignore these cleanup crashes and mark the run as successful.

Copilot uses AI. Check for mistakes.
Copy link

Copilot AI Dec 2, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The grep pattern ^=+ .* passed will match pytest summary lines that contain both "passed" and "failed" (e.g., "5 passed, 10 failed"). This could cause the workflow to incorrectly treat partial failures as success when a cleanup crash occurs.

Consider using a more precise pattern that ensures ALL tests passed:

if grep -qE "^=+ [0-9]+ passed in " pytest_cov_output.txt && ! grep -qE " failed" pytest_cov_output.txt; then

Or check for the absence of failures:

if grep -qE "^=+ .* passed" pytest_cov_output.txt && ! grep -qE "^=+ .* failed" pytest_cov_output.txt; then
Suggested change
if grep -qE "^=+ .* passed" pytest_cov_output.txt; then
if grep -qE "^=+ [0-9]+ passed in " pytest_cov_output.txt && ! grep -qE " failed" pytest_cov_output.txt; then

Copilot uses AI. Check for mistakes.
if [ "$PYTEST_EXIT_CODE" -eq 134 ] || [ "$PYTEST_EXIT_CODE" -eq 139 ]; then
echo "::warning::Pytest cleanup crashed but all tests passed."
exit 0
fi
fi

exit $PYTEST_EXIT_CODE

- name: Upload coverage reports
if: matrix.python-version == '3.12'
Expand Down
Loading