fix: async trait method for `unnecessary_async` by koka831 · Pull Request #13508...

Contributor

We can simplify this a bit via

// Check if this node is contained in a trait, second ancestors: ast::Fn -> ast::AssocItemList -> ast::Trait
if let Some(_) = function.syntax().ancestors().nth(2).and_then(ast::Trait::cast) {
    return None
}

We basically check the ancestor nodes of the function, take the second ancestor which is the Trait it is contained in (if it is a trait function), and if that is indeed the case we are done here.

@Veykril Thank you for review! I see, it make sense.

I tried as you suggested but found ast::Trait only matches trait definitions.
(the ancestor of a method in a trait implementation is ast::Impl, which is indistinguishable from impl for struct.)

So I think we would need at least the following procedure, right?

if let Some(impl_) = ctx.find_node_at_offset::<ast::Impl>() {
  if let Some(_) = resolve_target_trait(&ctx.sema, &impl_) {
    return None;
  }
}

Sorry, we can do it with the following:

    if let Some(impl_) = function.syntax().ancestors().nth(2).and_then(ast::Impl::cast) {
        if let Some(_) = impl_.trait_() {
            return None;
        }
    }