clang-tidy's readability-redundant-member-init has false positives for aggregates

[foonathan]

The following code generates a clang-tidy warning for readability-redundant-member-init.

struct foo {
    std::string a;
    std::string b  = {};
};

However, the initializer for b is not redundant. Consider:

void expect_foo(const foo& f);

expect_foo({"hello"});

With the initializer for b present, the code is fine. Without the initializer, the code generates a missing-field-initializer warning.

[chrchr-github]

Seems like this is a problem with missing-field-initializer then. That check should probably only warn for types that require initialization.
https://godbolt.org/z/Toscx3rsP

[chrchr-github]

Related to/duplicate of #90235

[foonathan]

That check should probably only warn for types that require initialization.

I'd disagree. I rely on that warning to force me to update all call sites when I add a new member.

[5chmidti]

I'm closing this as a duplicate of #90235.

Here's what I think the resolution of this should be: #90235 (comment)

I think that the check and the warning are correct in what they diagnose. The check detects two kinds of issues. Unnecessary initializers in constructors, and unnecessary initializers at the member-declaration. I think it would make sense to allow to configure what the check should detect. For the provided example, you'd disable the checking for redundant field initializations, while keeping the redundant constructor field initialization diagnostics.

[Denzeli]

missing-field-initializers does not complain about member-initialized fields of built-in types but warns about fields of class types with user-provided default constructors which is counter-intuitive. Had to disable this otherwise useful warning