boxed_local false positive for ?Sized

[CAD97]

Playground

fn lol(x: Box<[u8]>) {
    unimplemented!();
}

warning: local variable doesn't need to be boxed here
 --> src/lib.rs:1:8
  |
1 | fn lol(x: Box<[u8]>) {
  |        ^
  |
  = note: `#[warn(clippy::boxed_local)]` on by default
  = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#boxed_local

This is kind of an interesting case; the suggestion isn't wrong per se, but it's definitely not optimal. The lint explainer as currently worded strongly suggests "an unboxed T would work fine", whereas with ?Sized values this obviously cannot be done. The owned value has to be boxed or otherwise behind an indirection.

[flip1995]

The lint triggering here is correct, we should suggest to turn this into &[u8] though. See #4351 (comment)