Prefer .cloned() over .map(|x| x.clone())

[wartmanm]

Implicit cloning by dereferencing Copy types is included as well. It seems to be almost as common, at least in the code I've written.

I feel like I'm misusing if_let_chain!(), I hope not too badly...

[llogiq]

Looks good, just one potential problem: The deref_only part may lead to false positives if the Deref contains an implicit type conversion (e.g. syntax::ptr::P<T> derefs to T, whereas .clone() will create another P<T>.

Solving this will require checking for auto-adjustments, I think.

[wartmanm]

You're right, I completely overlooked that. I figured the walk_ptrs_ty_depth check would avoid it, but it didn't work if you derefed a &P<T> twice.

[llogiq]

Do you need help with the check? I think we should have something similar in eta_reduction.rs.

[wartmanm]

Sorry, it's been done for a little while, but I force-pushed the change, which wasn't very helpful of me. I moved the check in eta_reduction.rs into util.rs, I hope that's okay.

[Manishearth]

Force push is fine with us

[llogiq]

Yes, I just overlooked it. I'll merge.