Use allow(unused_imports) instead of cfg(doc) for imports used only for intra-doc links

[jyn514]

This prevents links from breaking when items are re-exported in a
different crate and the original isn't being documented.

Spotted in #75832 (comment) (thanks ollie!)

r? @ollie27

[ollie27]

Yeah, there are currently broken links on https://doc.rust-lang.org/nightly/std/intrinsics/fn.atomic_xchg.html which this should fix.

@bors r+ rollup

[bors]

📌 Commit 602dd14 has been approved by ollie27

[jyn514]

I'm a little confused how this slipped through CI ... the place where this breaks is when documenting a re-export in std, right? So it seems strange it wasn't caught by x.py doc library/std :/

[ollie27]

We're explicitly ignoring warnings from other crates:

rust/src/librustdoc/passes/collect_intra_doc_links.rs

Lines 952 to 954 in 5180f3d

	// If non-local, no need to check anything.
	info!("ignoring warning from parent crate: {}", msg);
	return;

[jyn514]

Yup, that would do it. We should probably close #56922 as fixed, then. I'll make a PR with a test case.

[JohnTitor]

Failed in rollup: #75862 (comment)
@bors r-

[jyn514]

@JohnTitor I pushed a commit which I hope will fix it, but I'm not sure how to test thumb targets locally ... x.py doc library/std --target thumbv8m.main-none-eabihf panics at couldn't find required command: "arm-none-eabi-gcc". Do you have any suggestions?

[jyn514]

well, I'll ask bors to test in the meantime and if it fails, it fails.

@bors r=ollie27 rollup=iffy

[bors]

📌 Commit 5aac94a has been approved by ollie27

[JohnTitor]

@jyn514 I don't familiar with that target so let's ask bors :)

[bors]

⌛ Testing commit 5aac94a with merge d795eb4...

[bors]

☀️ Test successful - checks-actions, checks-azure
Approved by: ollie27
Pushing d795eb4 to master...