gh-105080: Fixed inconsistent signature on derived classes

[gaogaotiantian]

classB: def__new__(self, *args, **kwargs): returnsuper().__new__(self) def__init__(self, value): self.value=valueclassD1(B): def__init__(self, value): super().__init__(value) classD2(D1): pass

For the code above, inspect.signature(D2) will give a different result than inspect.signature(D1), which is counter-intuitive. This also affects help(D2) vs help(D1).

We should keep the consistency for the signatures for derived classes (of course when the signature is not changed).

This is due to a corner case in inspect.signature() where only the direct defined __init__ and __new__ method on the class takes priority, then __new__ is always used if defined on any base classes. This fix searches __new__ and __init__ based on MRO and prioritize __new__ if both are defined on a base class.

• Issue: The help function shows incorrect signature for subclass #105080

[miss-islington]

Thanks @gaogaotiantian for the PR, and @carljm for merging it 🌮🎉.. I'm working now to backport this PR to: 3.12.
🐍🍒⛏🤖

[bedevere-bot]

GH-105257 is a backport of this pull request to the 3.12 branch.

[miss-islington]

Thanks @gaogaotiantian for the PR, and @carljm for merging it 🌮🎉.. I'm working now to backport this PR to: 3.11.
🐍🍒⛏🤖

[miss-islington]

Sorry, @gaogaotiantian and @carljm, I could not cleanly backport this to 3.11 due to a conflict.
Please backport using cherry_picker on command line.
cherry_picker 9ad199ba36791711f596393ca9a20dbf118ef858 3.11

[bedevere-bot]

GH-105274 is a backport of this pull request to the 3.11 branch.