This makes the `bit::vec::bench::bench_bit_vec_big_union` benchmark go
from `774 ns/iter (+/- 190)` to `602 ns/iter (+/- 5)`.
(There's room for more work here too: if one can guarantee 128-bit
alignment for the vector, the compiler actually optimises `union`,
`intersection` etc. to SIMD instructions, which end up being ~5x faster
that the original version, and 4x faster than the optimised version in
this patch.)
This was one last spot where directories were being leaked through with
arguments of the form `\\?\` which neither `ld.exe` nor `gcc.exe` does
understands so the prefix needed to be stripped.
Closes#25072
This was one last spot where directories were being leaked through with
arguments of the form `\\?\` which neither `ld.exe` nor `gcc.exe` does
understands so the prefix needed to be stripped.
Closes#25072
This test has deadlocked on Windows once or twice now and we've had lots of
problems in the past of threads panicking when the process is being shut down.
One of the two threads in this test is guaranteed to panic because of the
`.unwrap()` on the `send` calls, so just call `recv` on both receivers after the
test executes to ensure that both threads are dying/dead.
I'm uncertain whether the 3 implementations in `net2` should unwrap the socket address values. Without unwrapping it looks like this:
```
UdpSocket { addr: Ok(V4(127.0.0.1:34354)), inner: 3 }
TcpListener { addr: Ok(V4(127.0.0.1:9123)), inner: 4 }
TcpStream { addr: Ok(V4(127.0.0.1:9123)), peer: Ok(V4(127.0.0.1:58360)), inner: 5 }
```
One issue is that you can create, e.g. `UdpSocket`s with bad addresses, which means you can't just unwrap in the implementation:
```
#![feature(from_raw_os)]
use std::net::UdpSocket;
use std::os::unix::io::FromRawFd;
let sock: UdpSocket = unsafe { FromRawFd::from_raw_fd(-1) };
println!("{:?}", sock); // prints "UdpSocket { addr: Err(Error { repr: Os(9) }), inner: -1 }"
```
Fixes#23134.
This makes the `bit::vec::bench::bench_bit_vec_big_union` benchmark go
from `774 ns/iter (+/- 190)` to `602 ns/iter (+/- 5)`.
(There's room for more work here too: if one can guarantee 128-bit
alignment for the vector, the compiler actually optimises `union`,
`intersection` etc. to SIMD instructions, which end up being ~5x faster
that the original version, and 4x faster than the optimised version in
this patch.)
- I found n error in the book, before contributing the patch to fix it, I had to find where they were hosted
- It took me quite look to find where within the rust-lang *organisation* it was! ... and this should make it easier for the next person in the same position
Without the inline annotation this:
str::from_utf8_unchecked( slice::from_raw_parts( ptr, len ) )
doesn't get inlined which can be pretty brutal performance-wise
when used in an inner loop of a low level string manipulation method.
This script used to be used to extract the grammar sections from the
reference, but there is now a separate src/doc/grammar.md where the
grammar sections that used to be in the reference live, so there is
no longer a need to extract the grammar from the reference.
Since the hashmap and its hasher are implemented in different crates, we
currently can't benefit from inlining, which means that especially for
small, fixed size keys, there is a huge overhead in hash calculations,
because the compiler can't apply optimizations that only apply for these
keys.
Fixes the brainfuck benchmark in #24014.
E.g. if `foo.rs` looks like
#![feature(test)]
extern crate test;
#[bench]
fn bar(b: &mut test::Bencher) {
b.iter(|| {
1
})
}
#[test]
fn baz() {}
#[bench]
fn qux(b: &mut test::Bencher) {
b.iter(|| {
panic!()
})
}
Then
$ rustc --test foo.rs
$ ./foo
running 3 tests
test baz ... ok
test qux ... FAILED
test bar ... ok
failures:
---- qux stdout ----
thread 'qux' panicked at 'explicit panic', bench.rs:17
failures:
qux
test result: FAILED. 2 passed; 1 failed; 0 ignored; 0 measured
$ ./foo --bench ba
running 2 tests
test baz ... ignored
test bar ... bench: 97 ns/iter (+/- 74)
test result: ok. 0 passed; 0 failed; 1 ignored; 1 measured
In particular, the two benchmark are being run as tests in the default
mode.
This helps for the main distribution, since benchmarks are only run with
`PLEASE_BENCH=1`, which is rarely set (and never set on the test bots),
and helps for code-coverage tools: benchmarks are run and so don't count
as dead code.
Fixes#15842.
Since the hashmap and its hasher are implemented in different crates, we
currently can't benefit from inlining, which means that especially for
small, fixed size keys, there is a huge overhead in hash calculations,
because the compiler can't apply optimizations that only apply for these
keys.
Fixes the brainfuck benchmark in #24014.
- `FIle::open` is for opening a file in read-only mode
- `FIle::create` is for opening a file in write-only mode, which is what we want instead for this example to make sense
Without the inline annotation this:
str::from_utf8_unchecked( slice::from_raw_parts( ptr, len ) )
doesn't get inlined which can be pretty brutal performance-wise
when used in an inner loop of a low level string manipulation method.
