`empty_line_after_outer_attr` doesn't seem to work for rustdoc comments

[smoelius]

Summary

Pretty self-explanatory (I think).

Lint Name

empty_line_after_outer_attr

Reproducer

I tried this code (playground link):

#![warn(clippy::empty_line_after_outer_attr)]

/// Main function

fn main() {}

I expected to see this happen:
A warning.

Instead, this happened:
No warning.

Version

No response

[jdswensen]

I took a first pass at this, but wanted to make sure I'm on the right path before opening a PR. Two questions:

1. Is it desired that this lint encapsulate both outer attribute kinds? By that I mean, is it better as I've initially implemented it or would it be better to break the lint into two discrete configurations? The reason I ask is because of the next question.
2. The tests currently fail on:

/// This doc comment should not produce a warning

/** This is also a doc comment and should not produce a warning
 */

// This should not produce a warning
#[allow(non_camel_case_types)]
#[allow(missing_docs)]
#[allow(missing_docs)]
fn three_attributes() { assert!(true) }

By checking for AttrKind::DocComment in the outer attributes, clippy will end up catching cases like this. Is this new behavior acceptable? Or is there some other intended outcome? Making the change without introducing a different lint rule may fail code that currently has comments in this style.

It doesn't look too difficult to break up. My plan was to have a helper function that we could pass AttrKind and lint_msg into as both lints could use the same existing logic. The new rule could the be something like EMPTY_LINE_AFTER_OUTER_DOC_ATTR.

Either way, let me know! Thanks!

[smoelius]

The lines you cited are from this commit: a64724f

Apparently, the behavior I am complaining about was deliberate.

IMHO, lines like these should produce a warning:

/// This doc comment should not produce a warning

If they appear near the top of the module, it begs whether the author meant to use an inner doc comment, for example:

//! This doc comment should not produce a warning

I would like to know what the Clippy team thinks, though.

Thanks very much for looking into this issue, @jdswensen.

[smoelius]

@flip1995 Sorry for the ping, but I think @jdswensen could use some direction here.

[flip1995]

I would not include this into the empty_line_after_outer_attr lint. I wouldn't say that the average Rust user considers doc comments as outer/inner attributes.

So an additional lint (that shares code with empty_line_after_outer_attr) would be better IMO. Maybe empty_line_after_doc_comments?

[jdswensen]

That works for me! It's also the direction I was leaning after getting more familiar with the lint and test cases.