Bug: The ESLint react-hooks/preserve-manual-memoization rule warns against optional chaining operator

[4eb0da]

eslint-plugin-react-hooks version: 7.0.1

Steps To Reproduce

const Test = ({ obj }) => {
    const onClick = useCallback(() => {
        if (obj?.id) {
            console.log(obj.id);
        }
    }, [obj?.id]);

    return (
        <button onClick={onClick}>
            Hello
        </button>
    );
};

The current behavior

Compilation Skipped: Existing memoization could not be preserved

React Compiler has skipped optimizing this component because the existing manual memoization could not be preserved. The inferred dependencies did not match the manually specified dependencies, which could cause the value to change more or less frequently than expected. The inferred dependency was `obj`, but the source dependencies were [obj?.id]. Inferred less specific property than source.

The expected behavior

No errors

This can be fixed in several trivial ways:

1. Use obj?.id in all places
2. Introduce the variable:

const onClick = useCallback(() => {
    const id = obj?.id;
    if (id) {
        console.log(id);
    }
}, [obj?.id]);

[SaxenaAarohi]

Accessing obj.id still counts as reading obj. Extracting id
avoids an implicit object dependency and prevents React Compiler
from skipping optimization.

[karissa-uka]

I'd like to help fix this issue. The problem stems from the mismatch between inferred dependencies and manually specified ones when using optional chaining.

Would it be appropriate to update the linter rule to either:

Accept obj?.id in the dependency array when the body safely guards its usage
Provide a more specific error message suggesting developers use consistent access patterns