Needless borrow lint

[oli-obk]

This lint can be very annoying if it has false positives (which my first naive version had). So give me suggestions to run this on :)

Crates without false positives:

• clippy
• cargo
• racer
• serde
• serde_codegen

Crates to test:

Crates with false positives:

[mcarton]

There are other kinds of adjustment, rustc::ty::Tables::adjustments + AdjustDerefRef would maybe be better?

[oli-obk]

I don't think it's relevant, because I'm testing for & ops on ref types, any adjustments that can happen, will still happen when you remove the & op, because it's already a ref type.

[oli-obk]

In fact, I think that might decrease the usefulness of this lint. Because then

let x: &[u8] = &&Vec::new();

would not trigger anymore, even though it should be

let x: &[u8] = &Vec::new();

[oli-obk]

We could also test if the adjusted type of inner is the same as the adjusted type of e

[oli-obk]

rebased

[oli-obk]

rebased

[llogiq]

@mcarton LGTM, but since you reviewed, I'd like to know if you have any issue that should be resolved before merging.

[oli-obk]

rebased

[mcarton]

@oli-obk I’m ready to merge but I just don’t like the name of the lint, needless_ref makes me think of the ref keyword, not the & operator. What do you think of needless_borrow?

[oli-obk]

sgtm

[oli-obk]

although sometimes it can be either ;)

[oli-obk]

do we have a lint that lints on using ref for Copy types?

[llogiq]

Good work everyone!