use char instead of &str for single char patterns

[matthiaskrgr]

No description provided.

[ecstatic-morse]

r=me after removing unnecessary escapes.

[matthiaskrgr]

diff: https://github.com/rust-lang/rust/compare/b974a814870359c5957c554be18291b71f7758f2..8ceb9096ac72bab8b5566f22645c16e84634603f

r? @ecstatic-morse

[ecstatic-morse]

@bors r+ rollup=always

[bors]

📌 Commit 8ceb9096ac72bab8b5566f22645c16e84634603f has been approved by ecstatic-morse

[Dylan-DPC-zz]

failed in rollup
@bors r-

[petrochenkov]

#41993 was fixed, but I still think this replacement is ideologically wrong.
All these places don't care about unicode code points, they do substring search, it's just the substring happens to be pretty short.
char is a pretty rare data type that shouldn't be used unless you want unicode code points specifically.

[matthiaskrgr]

When running these benchmarks that I found via the discussion you linked

#![feature(test)]

extern crate test;
use crate::test::black_box;
use crate::test::Bencher;

fn main() {
    println!("Hello, world!");
}

#[bench]
fn starts_with_char(b: &mut Bencher) {
    let text = black_box("kdjsfhlakfhlsghlkvcnljknfqiunvcijqenwodind");
    b.iter(|| {
        for _ in 0..1024 {
            black_box(text.starts_with('k'));
        }
    })
}

#[bench]
fn starts_with_str(b: &mut Bencher) {
    let text = black_box("kdjsfhlakfhlsghlkvcnljknfqiunvcijqenwodind");
    b.iter(|| {
        for _ in 0..1024 {
            black_box(text.starts_with("k"));
        }
    })
}

#[bench]
fn ends_with_char(b: &mut Bencher) {
    let text = black_box("kdjsfhlakfhlsghlkvcnljknfqiunvcijqenwodind");
    b.iter(|| {
        for _ in 0..1024 {
            black_box(text.ends_with('k'));
        }
    })
}

#[bench]
fn ends_with_str(b: &mut Bencher) {
    let text = black_box("kdjsfhlakfhlsghlkvcnljknfqiunvcijqenwodind");
    b.iter(|| {
        for _ in 0..1024 {
            black_box(text.ends_with("k"));
        }
    })
}

, char is consistently faster for me with default optimization levels:

running 4 tests
test ends_with_char   ... bench:         465 ns/iter (+/- 31)
test ends_with_str    ... bench:         544 ns/iter (+/- 79)
test starts_with_char ... bench:         356 ns/iter (+/- 38)
test starts_with_str  ... bench:         691 ns/iter (+/- 10)

running 4 tests
test ends_with_char   ... bench:         464 ns/iter (+/- 10)
test ends_with_str    ... bench:         544 ns/iter (+/- 15)
test starts_with_char ... bench:         355 ns/iter (+/- 11)
test starts_with_str  ... bench:         692 ns/iter (+/- 11)

running 4 tests
test ends_with_char   ... bench:         448 ns/iter (+/- 11)
test ends_with_str    ... bench:         527 ns/iter (+/- 2)
test starts_with_char ... bench:         340 ns/iter (+/- 33)
test starts_with_str  ... bench:         670 ns/iter (+/- 8)

[petrochenkov]

If this is a perf improvement now, rather than just a stylistic change, then it should be ok.

[bors]

☔ The latest upstream changes (presumably #69507) made this pull request unmergeable. Please resolve the merge conflicts.

[matthiaskrgr]

rebased

[matthiaskrgr]

r? @petrochenkov or @ecstatic-morse

[Dylan-DPC-zz]

@bors r=ecstatic-morse

[bors]

📌 Commit 7c84ba1 has been approved by ecstatic-morse