`unused_parens` doesn't fire for trailing return expression

[varkor]

fn foo() -> u32 {
    let _ = (5); //~ LINT unnecessary parentheses around assigned value
    (5) // no lint
}

[Tyg13]

I'd be interested in handling this.

Looking at librucst_lint/unused.rs, it looks fairly straightforward -- however it does look like there could be some ambiguity with return values in blocks (e.g. let _ = { (5) };) Would we want to disambiguate lints, or would it be fine to use the same lint diagnostic for both cases?

[varkor]

@Tyg13: what sort of ambiguity do you mean? I'd expect a single lint on (5) in your example (note that we don't yet lint about unused {}).

[Tyg13]

@varkor: To use your original example,

fn foo() -> u32 {
    let a = {
        let b = 5;
        (b)
    };
    (a)
}

Both (a) and (b) are StmtKind::Expr, since they represent the implicit return at the end of a block. I was wondering if we'd care about the (b) case, and if we did, if it made sense to treat it the same as (a). If you naively add a match arm to the Stmt visitor, you get the same diagnostic for both.

[varkor]

Yes, both (a) and (b) should be linted here.