[dor-sr]

Summary

Closes #28989

SLF001 already skips instances created with super().__new__(cls), but not the same pattern written with object:

class A:
    def __new__(cls):
        obj = object().__new__(cls)
        obj._private_attr = "bar"  # SLF001 before this change

This extends the special case to object().__new__(...) and object.__new__(...), as suggested in the issue. object has to resolve to the builtin, and object() has to be called without arguments.

Test Plan

Added cases to SLF001_1.py: the two object forms with cls, object.__new__(Amet) inside a classmethod, and two that should still be flagged (object(1) and a shadowed object). The snapshot only gains those two diagnostics.

Also checked the reproducer from the issue with a local build: flagged before, clean after.

[astral-sh-bot]

ruff-ecosystem results

Linter (stable)

ℹ️ ecosystem check detected linter changes. (+1 -2 violations, +0 -0 fixes in 2 projects; 56 projects unchanged)

apache/airflow (+0 -2 violations, +0 -0 fixes)

ruff check --no-cache --exit-zero --no-fix --output-format concise --no-preview --select ALL

- task-sdk/src/airflow/sdk/io/path.py:139:9: SLF001 Private member accessed: `_conn_id`
- task-sdk/src/airflow/sdk/io/path.py:145:13: SLF001 Private member accessed: `_inject_authenticated_fs`

home-assistant/core (+1 -0 violations, +0 -0 fixes)

+ homeassistant/config_entries.py:177:41: RUF100 [*] Unused `noqa` directive (unused: `SLF001`)

Changes by rule (2 rules affected)

code	total	+ violation	- violation	+ fix	- fix
SLF001	2	0	2	0	0
RUF100	1	1	0	0	0

Linter (preview)

ℹ️ ecosystem check detected linter changes. (+1 -3 violations, +0 -0 fixes in 2 projects; 56 projects unchanged)

apache/airflow (+0 -2 violations, +0 -0 fixes)

ruff check --no-cache --exit-zero --no-fix --output-format concise --preview --select ALL

- task-sdk/src/airflow/sdk/io/path.py:139:9: private-member-access Private member accessed: `_conn_id`
- task-sdk/src/airflow/sdk/io/path.py:145:13: private-member-access Private member accessed: `_inject_authenticated_fs`

home-assistant/core (+1 -1 violations, +0 -0 fixes)

ruff check --no-cache --exit-zero --no-fix --output-format concise --preview

- homeassistant/config_entries.py:177:41: noqa-comments [*] `noqa` comment used instead of `ruff: ignore`
+ homeassistant/config_entries.py:177:41: unused-noqa [*] Unused `noqa` directive (unused: `SLF001`)

Changes by rule (3 rules affected)

code	total	+ violation	- violation	+ fix	- fix
private-member-access	2	0	2	0	0
unused-noqa	1	1	0	0	0
noqa-comments	1	0	1	0	0

[ntBre]

Thanks, this looks good to me overall, just one question.

[ntBre]

This same validation could be applied to the super version right? I think I'd either apply it to both or omit this additional change from the PR, and I'd probably lean toward the latter since it's unrelated to the issue.

[dor-sr]

Makes sense, I dropped it. super().__new__ and object.__new__ are handled the same way now, with no check on the argument, and I removed the other-class test case.

[ntBre]

Thank you!