run-pass/extern-pass-empty is probably a bogus thing to test

[pnkfelix]

The following code hits the lint that warns about trying to put #[repr(C)] on an empty struct.

rust/src/test/run-pass/extern-pass-empty.rs

Lines 33 to 34 in c2afca3

	#[repr(C)]
	struct Empty;

(This seems like a bogus test that was caught by working on fixing #53764.)

[pnkfelix]

If I understand correctly from the history of the test, #12762 explicitly added support for empty structs ... by ignoring them ... (?)

I'm not 100% sure what our policy should be here.

[pnkfelix]

nominating for discussion at compiler team meeting. (Though it might need to be discussed more broadly than that.)

[pnkfelix]

unsafe code guidelines group actually discussed this very topic, see here: rust-lang/unsafe-code-guidelines#31 (comment)

[nikomatsakis]

@pnkfelix ultimately I think we decided that our current behavior makes sense and effectively documented it here in this section; it matches the extensions provided by gcc and other C compilers (though it does not match the C++ behavior). So it does seem like a good thing to test to me.

[nikomatsakis]

This does not imply that our particular test is well setup.

[nagisa]

Discussed during the compiler meeting. It seems that we cannot really make a decision here before UCG becomes an official document.