FP single_component_path_imports

[djc]

Lint name:

single_component_path_imports

I tried this code:

#[cfg(feature = "log")]
use log;

mod foo {
    use crate::log::warn;

    pub fn bar() {
        warn!("bar");
    }
}

pub fn bar() {
    foo::bar()
}

(I think it doesn't matter whether the feature log is turned on or off.)

I expected to see this happen: no warning, since the log import is used in an inner module.

Instead, this happened:

warning: this import is redundant
 --> src/lib.rs:3:1
  |
3 | use log;
  | ^^^^^^^^ help: remove it entirely
  |
  = note: `#[warn(clippy::single_component_path_imports)]` on by default
  = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#single_component_path_imports

Meta

• cargo clippy -V: clippy 0.1.53 (42816d6 2021-04-24)
• rustc -Vv:

  rustc 1.53.0-nightly (42816d61e 2021-04-24)
  binary: rustc
  commit-hash: 42816d61ead7e46d462df997958ccfd514f8c21c
  commit-date: 2021-04-24
  host: x86_64-apple-darwin
  release: 1.53.0-nightly
  LLVM version: 12.0.0
  

cc @ThibsG (re #6905 (comment))

[camsteffen]

So you want the log crate to be accessible at crate::log?

[djc]

It is accessible as crate::log, and my code relies on that.

[camsteffen]

Yes, obviously the suggestion is not okay as is. I'm just asking from a code style viewpoint. Because this questions the assertion that "single component imports are always useless/bad style" which seems to be the idea behind the lint.

[djc]

From a code style viewpoint I think it's a great lint, I've applied many suggestions due to this lint. In this case the use log is conditional on a feature and gets switched out for a mock implementation of the log warnings, so it's definitely more of a special case. In general if the identifier gets used as a crate-rooted item, or maybe even in general through a path, the lint should probably be suppressed.

[camsteffen]

In general if the identifier gets used as a crate-rooted item, or maybe even in general through a path, the lint should probably be suppressed.

Yep that sounds reasonable. Hopefully it isn't too gnarly to implement. Scanning usages can be problematic. But I think this might be a relatively simple case.

[camsteffen]

We should also detect usages like super::log which means it doesn't need to be crate-rooted.