cargo clippy --all misses crates that are not part of the workspace

[gnzlbg]

Currently cargo clippy --all applies to sub-crates that are part of a workpace. However, if the root crate is not part of a workspace, sub-crates in top-level directories are not considered. This behavior differs from cargo fmt --all, which also formats crates in the root directory that are not part of a workspace.

[sanmai-NL]

Isn't the behavior of cargo fmt wrong, in your view?

[nk9]

I would agree that clippy should lint what cargo would build. Less sure which way cargo fmt --all should behave, although at first blush I agree with @sanmai-NL. Regardless, I don't think clippy should feel bound by their behavior.

@rustbot label S-needs-discussion A-ui

[samueltardieu]

How does cargo check behave?

[nk9]

It ignores the subcrate.

$ find . -name "main.rs"
./subcrate/src/main.rs
./src/main.rs

$ cat subcrate/src/main.rs
fn main() {
    let a = 42;
    println!("Hello, world!");
}

$ cargo check
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.03s

$ cd subcrate

$ cargo check
warning: unused variable: `a`
 --> src/main.rs:2:9
  |
2 |     let a = 42;
  |         ^ help: if this is intentional, prefix it with an underscore: `_a`
  |
  = note: `#[warn(unused_variables)]` on by default

warning: `subcrate` (bin "subcrate") generated 1 warning
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.03s

[samueltardieu]

Sorry, I meant with --all: if cargo clippy is consistent with cargo check, then this is not a Clippy bug.

[nk9]

Ah, makes sense. Yes, it ignores the subcrate in that instance as well.

$ ls
Cargo.lock Cargo.toml src        subcrate   target

$ cargo check --all
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.03s

[samueltardieu]

So let's ignore this in Clippy.