Repository navigation
fix: resolve IterableList names independently of list attributes - #2283
Byron merged 1 commit into
Conversation
Sorry, something went wrong.
Normal attribute lookup makes string indexing return list methods for item names such as `count`, and reports absent `append` identifiers as present. Use `__getattr__` explicitly for identifier lookup and remove its fallback to list attributes. Ordinary attribute access still exposes list methods. Add twelve present/absent collision cases with and without prefixes. The full suite passes 1701 tests, 89 conditional skips, one existing xfail, and 64 subtests; pinned hooks, type checks, docs and packaging also pass.
There was a problem hiding this comment.
Warning
Copilot couldn't run its full agentic review because it didn't start before the timeout. Make sure your repository has a runner available, or add a copilot-code-review.yml file specifying one with the runs-on attribute. See the docs for more details.
Copilot review overview
2 open findings__getitem__ duplicates the 'No item found...' message formatting that __getattr__ already owns,… · New
The new AttributeError message interpolates the identifier without quoting ({attr}), which can… · New
Updates IterableList name-based lookup so string indexing and membership checks resolve items by identifier rather than colliding with built-in list attributes/methods (e.g., append, sort), and adds regression tests for these edge cases.
Changes:
- Route __contains__ and string __getitem__ through IterableList.__getattr__ instead of Python attribute lookup.
- Change __getattr__ fallback from list attribute access to raising AttributeError when no matching item exists.
- Add parametrized regression tests covering list-method names, dunder/internal names, prefixes, and deletion.
| File | Description |
|---|---|
| git/util.py | Adjusts IterableList lookup semantics to avoid list attribute collisions and to raise on missing identifiers. |
| test/test_util.py | Adds regression tests ensuring identifier lookup works even for names that match list methods/dunders and after deletion. |
🧠 Review effort: Lite
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Sorry, something went wrong.
This PR was prepared by an OpenAI Codex agent.
IterableList string indexing currently uses normal attribute lookup. Consequently, names such as count, sort, or append return list methods rather than matching items, and an absent append is reported as present. This also affects named references and submodules, as described in #1091 (currently closed as not planned).
Route string indexing and identifier membership through the existing identifier lookup, and raise AttributeError when no matching item exists instead of falling back to list attributes. Ordinary attribute access to list methods remains available.
Twelve regression cases cover list-method names, internal/dunder names, ordinary names, and prefixed identifiers, including absence after deletion. Ten fail against pristine source; the two ordinary-name controls pass.
Validation on CPython 3.12.14 / Linux / Git 2.52.0:
The unchanged skips are platform/filesystem-specific, and the existing xfail concerns shallow submodules. One unchanged path-environment deprecation warning remains. Other operating systems and Python versions were not run locally; hosted results are reported separately.