#[allow(non_snake_case)] ignored

[sivizius]

The linter ignores #[allow(non_snake_case)] and warns about »… should have snake_case name, e.g. `…`«. In fact, this lint is redundant to the rustc-lint.

[lnicola]

I think this wasn't completely implemented in #6421, CC @popzxc

Sample code:

#![allow(non_snake_case)]

fn main() {
    #[allow(non_snake_case)]
    let FOO_BAR = 42;
    drop(FOO_BAR);
}

It's also shown twice, not sure why.

In fact, this lint is redundant to the rustc-lint.

It's not. This one works as you type, but for cargo check you need to save the file you're editing. In the long run, all of them will be supported directly by rust-analyzer.

[popzxc]

Yep, the support of this attribute is very basic, thus it is kind of expected.
Proper solution will require collecting attributes applied to the item up to the crate root, which is not a trivial task (given the fact that currently there is no API to do so).

[lnicola]

Shouldn't #[allow(non_snake_case)] on the declaration work?

[popzxc]

In theory, yes. However I could have skipped checking for attribute somewhere.

Anyway, unfortunately it doesn't seem that I'd be free from main work in the following couple of weeks to fix this :(

[lowr]

There have been bug reports on this: #13521, #14080. We currently don't (or, more precisely, can't) honor #[allow()]s on let statements, because we don't have a means to analyze attributes on statements and expressions at all iiuc.

[HKalbasi]

The #[allow] on expressions has been implemented in #14990, but the #![allow] is still not implemented.

[dladeira]

Hey, I'm not sure if this is a problem on my end, but when adding #[allow(non_camel_case_types)] to a struct, the warning still shows. It's only when at the top of the file I declare #![allow(non_snake_case)], does it go away.

Code for reference:

#[allow(non_camel_case_types)]
struct ResponseData {
    deviceName: String,
    _id: String,
    focus: i128
}

Am I doing something wrong? Is this intended?

[alibektas]

Can it be that you confuse what both lints are responsible for or did I misunderstand you? See what both non_* do on the following example

#![allow(dead_code)]
#![allow(non_snake_case)]

#[allow(non_camel_case_types)]
struct responseData {
    deviceName: String,
    _id: String,
    focus: i128,
}

The problem with your snippet is non_snake_case and not non_camel_case_types so it makes sense that warning still shows when there is no #[allow(no_snake_case)]

[dladeira]

My bad, I got confused. It works now, thank you for clarifying.