Reduce false positives number in rustdoc html diff

[GuillaumeGomez]

cc @rust-lang/dev-tools
r? @nrc

Very simple trick but should lighten html diff a bit

[QuietMisdreavus]

small note: .find(...).is_some() is the same as .any(...)

[GuillaumeGomez]

Ah good point!

[ollie27]

What false positives does this avoid that

rust/src/librustdoc/html/render.rs

Lines 1760 to 1770 in 2f1ef9e

	let differences = differences.into_iter()
	.filter(|s| {
	match *s {
	html_diff::Difference::NodeText { ref elem_text,
	ref opposite_elem_text,
	.. }
	if match_non_whitespace(elem_text, opposite_elem_text) => false,
	_ => true,
	}
	})
	.collect::<Vec<_>>();

(which is cleaned up a bit in #44329) won't?

EDIT: actually I noticed that this just reduces the output a bit for already found differences, so it would be good to see some before and afters.

[GuillaumeGomez]

@ollie27: Yes, it works on already found differences (like almost all upcoming rustdoc PRs will most likely).

[QuietMisdreavus]

[00:31:59] error[E0308]: mismatched types
[00:31:59] --> /checkout/src/librustdoc/html/render.rs:665:31
[00:31:59] |
[00:31:59] 665 | .any(|&(a, b)| a.trim() != b.trim()) {
[00:31:59] | ^^^^^^^ expected tuple, found reference
[00:31:59] |
[00:31:59] = note: expected type (&str, &str)
[00:31:59] found type &_

oh hey, .any() gives the real item, not a reference

[GuillaumeGomez]

Damn!

[QuietMisdreavus]

I'm not entirely sure which specific entries this will help, but if you say it helps, i'm willing to land this anyway. However, since #44350 changed the signature and layout of render_difference, this will need to be rebased on top of that one when it's landed. r=me once all that happens.

[GuillaumeGomez]

I think we should merge this one first if you don't mind. The second one will be delayed.

[QuietMisdreavus]

Yeah, since that other one got delayed, i'll go ahead and push this one forward.

@bors r+

[bors]

📌 Commit 502e707 has been approved by QuietMisdreavus

[frewsxcv]

@bors rollup