`use_debug` lint should not trigger on `debug!()` log macro calls

[hcpl]

Example use-case:

#![cfg_attr(feature = "cargo-clippy", warn(use_debug))]

#[macro_use]
extern crate log;

pub fn euclid_steps(mut x: u64, mut y: u64) -> Vec<u64> {
    debug!("x = {}, y = {}", x, y);

    let mut steps = vec![];

    loop {
        let tmp = x % y;

        if tmp == 0 {
            debug!("Steps performed in Euclidean algorithm: {:?}", steps);
            return steps;
        }

        steps.push(tmp);
        x = y;
        y = tmp;
    }
}

Placing #![cfg_attr(feature = "cargo-clippy", allow(use_debug))] before each debug!() usage seems tedious to me when there are several dozens of them; and turning it off (even if locally) defeats the purpose of the lint.

[sanmai-NL]

There’s also trace! and various other logging crates. I propose that these macros themselves are patched to contain an expression-level allow on the format string.

[oli-obk]

should be very easy with the new system. basically we only want this lint on explicit calls to print!, println!, write! and writeln!

Basically pass in the current macro name as an argument to https://github.com/rust-lang-nursery/rust-clippy/blob/master/clippy_lints/src/write.rs#L220 and then use that to lint conditionally in https://github.com/rust-lang-nursery/rust-clippy/blob/master/clippy_lints/src/write.rs#L247

[sanmai-NL]

What about format! and then reusing the resulting string in some inappropriate way, e.g. printing the string later on?

[oli-obk]

that, too :D good catch

[sanmai-NL]

I’m trying to say, when the programmer uses format!, it’s unclear what happens with the output and also whether that’s inappropriate.

[oli-obk]

oh... seems to me that this would be a case for "moderate sprinkling of allow"?

[sanmai-NL]

So deny on format! by default? Seems fine to me.