unused allow annotation

[laralove143]

What it does

we add #[allow(lint)] to a function to suppress a lint, when that function's body changes and no longer raises that lint, we have the old #[allow(lint)] annotation silently staying there

we might also add #[allow(lint)] to a function because we're sure it will raise that lint, and we turn out to be wrong we dont realize we added that for no reason

Lint Name

unused_allow

Category

style

Advantage

• keeps code clean
• doesnt give the idea that the function raises the lint
• gives no hint to the history of a function

Drawbacks

we might wanna allow a lint to express that we dont care if this happens in the future, in which case we can also allow this lint to express that more explicitly

Example

say we think ignoring an error is fine

#[allow(unused_must_use)]
fn foo() {
  let _: Result<_, _> = fallible();
}

then we realize we do need to handle the error so we rewrite the function to

#[allow(unused_must_use)]
fn foo() {
  fallible().handle();
}

but here #[allow(unused_must_use)] stays even tho it doesnt need to

[xFrednet]

Hey @laralove143, this is a feature which would need to be implemented in the compiler and not Clippy. There was already an RFC to add a feature like this, see: rust-lang/rfcs#2383. If you're interested, you can ready the tracking issue and the linked Zulip conversation. The stabilization is currently held up on a discussion how the new attribute should exactly work. If you have any thoughts on the matter or additional use cases, I'd highly appreciate your input :)

• Tracking issue for RFC 2383, "Lint Reasons RFC" rust#54503
• https://rust-lang.zulipchat.com/#narrow/stream/213817-t-lang/topic/Expect.20attribute.20mental.20model/near/340449104

[xFrednet]

I'm closing this issue, as the discussion moved into the tracking issue. If such a lint would be added, it would need to be implemented in rustc anyways :)