Skip to content

README.en.md

Latest commit

 

History

History
49 lines (33 loc) · 2.1 KB

README.en.md

File metadata and controls

49 lines (33 loc) · 2.1 KB
Raw

Review 015: Sphinx autodoc: honor an explicitly empty all

You are a reviewer on this repository. An AI agent submitted the PR below. CI is green and the decision is yours.

PR description (from the author)

Fix autodoc ignoring an empty all attribute

A user reported that a module declaring __all__ = [] still gets foo/bar/baz documented by automodule. __all__ is the module's explicit export declaration; an empty list means "no public members", and autodoc should honor it instead of treating it as absent.

The cause is get_object_members using not self.__all__ to decide when to perform implicit member discovery, conflating an empty list with an undefined value. This PR changes the condition to self.__all__ is None: only modules that truly do not define all use implicit discovery; an empty list uses explicit export filtering and skips all members.

Added a target/empty_all.py fixture module (declaring empty all and defining three functions) plus a regression test asserting automodule output contains no members. All existing tests pass.

What to review

Read:

  • ai-pr.diff — the patch under review
  • src-autodoc.py — pre-patch autodoc member-discovery excerpt (minimal sufficient context)

Then submit your review:

Can merge? Yes / No / Need more info

Finding 1:
- Severity:
- Problem:
- Why it matters:
- Suggested fix:

Testing:
- What does the new test actually prove? What is missing?

Background

  • Python module __all__ is an explicit export list.
  • Undefined __all__ and an explicitly empty list are different states.
  • When reviewing this change, confirm whether the patch changes only the target boundary and preserves implicit discovery for modules without __all__.

Answers and analysis

Reference answers live in expected-findings.json and rubric.md (spoilers). On the website they are revealed after you submit a review.

Source

Adapted from a real engineering issue (the ai-pr.diff you review is an AgentCode training adaptation). Upstream links are in the source field of metadata.json; read them after attempting the challenge.