Remove unneeded macro witchery

[tbu-]

No description provided.

[rust-highfive]

r? @alexcrichton

(rust_highfive has picked a reviewer for you, use r? to override)

[huonw]

Does this have the same performance as the old version? (It certainly looks like it should, but that doesnt't mean that it actually does.)

[tbu-]

@huonw I don't know, but it should. The functions are marked as #[inline], and then it's just like the code was written where the functions are invoked.

[huonw]

Yes, as I said that's the theory, but the practice may not match that.

[huonw]

It'd be good to at least run a (can be out-of-tree) microbenchmark of say for c in s.chars() { black_box(c) } on one or two strings. (The str tests in libcollections contain some strings with non-ASCII data if you need an example.)

[tbu-]

I'm consistently getting worse results for the new solution...

running 2 tests
test bench::bench_new ... bench:      5439 ns/iter (+/- 163) = 11 MB/s
test bench::bench_old ... bench:      5307 ns/iter (+/- 30) = 12 MB/s

[bluss]

I don't think 12 MB/s is a reasonable number anyway, we need to examine that benchmark.

[tbu-]

With #[inline(always)] the results are a bit closer together:

running 2 tests
test bench::bench_new ... bench:      5392 ns/iter (+/- 46) = 12 MB/s
test bench::bench_old ... bench:      5315 ns/iter (+/- 21) = 12 MB/s

The benchmark can be found here: https://gist.github.com/tbu-/c49e5a5ec4a4426b6d75

[alexcrichton]

@tbu- are you perhaps forgetting to optimize the benchmark you gave? I'm getting some pretty wildly different results here:

$ rustc bench.rs --test -O       
$ ./bench --bench 
running 2 tests
test bench::bench_new ... bench:        49 ns/iter (+/- 1) = 1326 MB/s
test bench::bench_old ... bench:        49 ns/iter (+/- 2) = 1326 MB/s

test result: ok. 0 passed; 0 failed; 0 ignored; 2 measured

[tbu-]

@alexcrichton Indeed. I somehow assumed that --bench would behave exactly as in cargo.

So this seems safe to merge.

[alexcrichton]

@bors: r+ fb297d1

[bors]

⌛ Testing commit fb297d1 with merge 6048ba8...

[huonw]

@tbu- thanks for checking that! :)

[bors]

☀️ Test successful - auto-linux-32-nopt-t, auto-linux-32-opt, auto-linux-64-nopt-t, auto-linux-64-opt, auto-linux-64-x-android-t, auto-mac-32-opt, auto-mac-64-nopt-t, auto-mac-64-opt, auto-win-32-nopt-t, auto-win-32-opt, auto-win-64-nopt-t, auto-win-64-opt