Add inlining annotations in `dec2flt`.
Currently, the combination of `dec2flt` being generic and the `FromStr` implementaions
containing inline anttributes causes massive amounts of assembly to be generated whenever
these implementation are used. In addition, the assembly has calls to function which ought to
be inlined, but they are not (even when using lto).
This Pr fixes this.
Stabilize `nonzero_min_max`
## Overall
Stabilizes `nonzero_min_max` to allow the "infallible" construction of ordinary minimum and maximum `NonZero*` instances.
The feature is fairly straightforward and already matured for some time in stable toolchains.
```rust
let _ = NonZeroU8::MIN;
let _ = NonZeroI32::MAX;
```
## History
* On 2022-01-25, implementation was [created](https://github.com/rust-lang/rust/pull/93293).
## Considerations
* This report is fruit of the inanition observed after two unsuccessful attempts at getting feedback.
* Other constant variants discussed at https://github.com/rust-lang/rust/issues/89065#issuecomment-923238190 are orthogonal to this feature.
Fixes https://github.com/rust-lang/rust/issues/89065
The point of these is to be seen lexically in the docs, so they should always be passed as the correct literal, not as an expression.
(Otherwise we could just compute `Min`/`Max` from `BITS`, for example.)
Avoid copy-pasting the `ilog` panic string in a bunch of places
I also ended up changing the implementations to `if let` because it doesn't work to
```rust
self.checked_ilog2().unwrap_or_else(panic_for_nonpositive_argument)
```
due to the `!`. But as a bonus that meant I could remove the `rustc_allow_const_fn_unstable` too.
Implement `signum` with `Ord`
Rather than needing to do things like #105840 for `signum` too, might as well just implement that method using `Ord`, since it's doing the same "I need `-1`/`0`/`+1`" behaviour that `cmp` is already doing.
This also seems to slightly improve the assembly: <https://rust.godbolt.org/z/5oEEqbxK1>
doc: rewrite doc for signed int::{carrying_add,borrowing_sub}
Reword the documentation for bigint helper methods, signed `int::{carrying_add,borrowing_sub}` (#85532).
This change is a follow-up to #101889, which was for the unsigned methods.
Fix doc example for `wrapping_abs`
The `max` variable is unused. This change introduces the `min_plus` variable, to make the example similar to the one from `saturating_abs`. An alternative would be to remove the unused variable.
Fix inconsistent rounding of 0.5 when formatted to 0 decimal places
As described in #70336, when displaying values to zero decimal places the value of 0.5 is rounded to 1, which is inconsistent with the display of other half-integer values which round to even.
From testing the flt2dec implementation, it looks like this comes down to the condition in the fixed-width Dragon implementation where an empty buffer is treated as a case to apply rounding up. I believe the change below fixes it and updates only the relevant tests.
Nevertheless I am aware this is very much a core piece of functionality, so please take a very careful look to make sure I haven't missed anything. I hope this change does not break anything in the wider ecosystem as having a consistent rounding behaviour in floating point formatting is in my opinion a useful feature to have.
Resolves#70336
The `max` variable is unused. This change introduces the `min_plus`
variable, to make the example similar to the one from `saturating_abs`.
An alternative would be to remove the unused variable.
Improve performance of `rem_euclid()` for signed integers
such code is copy from
https://github.com/rust-lang/rust/blob/master/library/std/src/f32.rs and
https://github.com/rust-lang/rust/blob/master/library/std/src/f64.rs
using `r+rhs.abs()` is faster than calc it with an if clause. Bench result:
```
$ cargo bench
Compiling div-euclid v0.1.0 (/me/div-euclid)
Finished bench [optimized] target(s) in 1.01s
Running unittests src/lib.rs (target/release/deps/div_euclid-7a4530ca7817d1ef)
running 7 tests
test tests::it_works ... ignored
test tests::bench_aaabs ... bench: 10,498,793 ns/iter (+/- 104,360)
test tests::bench_aadefault ... bench: 11,061,862 ns/iter (+/- 94,107)
test tests::bench_abs ... bench: 10,477,193 ns/iter (+/- 81,942)
test tests::bench_default ... bench: 10,622,983 ns/iter (+/- 25,119)
test tests::bench_zzabs ... bench: 10,481,971 ns/iter (+/- 43,787)
test tests::bench_zzdefault ... bench: 11,074,976 ns/iter (+/- 29,633)
test result: ok. 0 passed; 0 failed; 1 ignored; 6 measured; 0 filtered out; finished in 19.35s
```
It seems that, default `rem_euclid` triggered a branch prediction, thus `bench_default` is faster than `bench_aadefault` and `bench_aadefault`, which shuffles the order of calculations. but all of them slower than what it was in `f64`'s and `f32`'s `rem_euclid`, thus I submit this PR.
bench code:
```rust
#![feature(test)]
extern crate test;
fn rem_euclid(a:i32,rhs:i32)->i32{
let r = a % rhs;
if r < 0 { r + rhs.abs() } else { r }
}
#[cfg(test)]
mod tests {
use super::*;
use test::Bencher;
use rand::prelude::*;
use rand::rngs::SmallRng;
const N:i32=1000;
#[test]
fn it_works() {
let a: i32 = 7; // or any other integer type
let b = 4;
let d:Vec<i32>=(-N..=N).collect();
let n:Vec<i32>=(-N..0).chain(1..=N).collect();
for i in &d {
for j in &n {
assert_eq!(i.rem_euclid(*j),rem_euclid(*i,*j));
}
}
assert_eq!(rem_euclid(a,b), 3);
assert_eq!(rem_euclid(-a,b), 1);
assert_eq!(rem_euclid(a,-b), 3);
assert_eq!(rem_euclid(-a,-b), 1);
}
#[bench]
fn bench_aaabs(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=rem_euclid(*i,*j);
}
}
res
});
}
#[bench]
fn bench_aadefault(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=i.rem_euclid(*j);
}
}
res
});
}
#[bench]
fn bench_abs(b: &mut Bencher) {
let d:Vec<i32>=(-N..=N).collect();
let n:Vec<i32>=(-N..0).chain(1..=N).collect();
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=rem_euclid(*i,*j);
}
}
res
});
}
#[bench]
fn bench_default(b: &mut Bencher) {
let d:Vec<i32>=(-N..=N).collect();
let n:Vec<i32>=(-N..0).chain(1..=N).collect();
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=i.rem_euclid(*j);
}
}
res
});
}
#[bench]
fn bench_zzabs(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=rem_euclid(*i,*j);
}
}
res
});
}
#[bench]
fn bench_zzdefault(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=i.rem_euclid(*j);
}
}
res
});
}
}
```
benchmark result:
```
$ cargo bench
Compiling div-euclid v0.1.0 (/me/div-euclid)
Finished bench [optimized] target(s) in 1.01s
Running unittests src/lib.rs (target/release/deps/div_euclid-7a4530ca7817d1ef)
running 7 tests
test tests::it_works ... ignored
test tests::bench_aaabs ... bench: 10,498,793 ns/iter (+/- 104,360)
test tests::bench_aadefault ... bench: 11,061,862 ns/iter (+/- 94,107)
test tests::bench_abs ... bench: 10,477,193 ns/iter (+/- 81,942)
test tests::bench_default ... bench: 10,622,983 ns/iter (+/- 25,119)
test tests::bench_zzabs ... bench: 10,481,971 ns/iter (+/- 43,787)
test tests::bench_zzdefault ... bench: 11,074,976 ns/iter (+/- 29,633)
test result: ok. 0 passed; 0 failed; 1 ignored; 6 measured; 0 filtered out; finished in 19.35s
```
benchmark code:
```rust
#![feature(test)]
extern crate test;
#[inline(always)]
fn rem_euclid(a:i32,rhs:i32)->i32{
let r = a % rhs;
if r < 0 {
// if rhs is `integer::MIN`, rhs.wrapping_abs() == rhs.wrapping_abs,
// thus r.wrapping_add(rhs.wrapping_abs()) == r.wrapping_add(rhs) == r - rhs,
// which suits our need.
// otherwise, rhs.wrapping_abs() == -rhs, which won't overflow since r is negative.
r.wrapping_add(rhs.wrapping_abs())
} else {
r
}
}
#[cfg(test)]
mod tests {
use super::*;
use test::Bencher;
use rand::prelude::*;
use rand::rngs::SmallRng;
const N:i32=1000;
#[test]
fn it_works() {
let a: i32 = 7; // or any other integer type
let b = 4;
let d:Vec<i32>=(-N..=N).collect();
let n:Vec<i32>=(-N..0).chain(1..=N).collect();
for i in &d {
for j in &n {
assert_eq!(i.rem_euclid(*j),rem_euclid(*i,*j));
}
}
assert_eq!(rem_euclid(a,b), 3);
assert_eq!(rem_euclid(-a,b), 1);
assert_eq!(rem_euclid(a,-b), 3);
assert_eq!(rem_euclid(-a,-b), 1);
}
#[bench]
fn bench_aaabs(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=rem_euclid(*i,*j);
}
}
res
});
}
#[bench]
fn bench_aadefault(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=i.rem_euclid(*j);
}
}
res
});
}
#[bench]
fn bench_abs(b: &mut Bencher) {
let d:Vec<i32>=(-N..=N).collect();
let n:Vec<i32>=(-N..0).chain(1..=N).collect();
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=rem_euclid(*i,*j);
}
}
res
});
}
#[bench]
fn bench_default(b: &mut Bencher) {
let d:Vec<i32>=(-N..=N).collect();
let n:Vec<i32>=(-N..0).chain(1..=N).collect();
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=i.rem_euclid(*j);
}
}
res
});
}
#[bench]
fn bench_zzabs(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=rem_euclid(*i,*j);
}
}
res
});
}
#[bench]
fn bench_zzdefault(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=i.rem_euclid(*j);
}
}
res
});
}
}
```
such code is copy from
https://github.com/rust-lang/rust/blob/master/library/std/src/f32.rs
and
https://github.com/rust-lang/rust/blob/master/library/std/src/f64.rs
using r+rhs.abs() is faster than calc it directly.
Bench result:
```
$ cargo bench
Compiling div-euclid v0.1.0 (/me/div-euclid)
Finished bench [optimized] target(s) in 1.01s
Running unittests src/lib.rs (target/release/deps/div_euclid-7a4530ca7817d1ef)
running 7 tests
test tests::it_works ... ignored
test tests::bench_aaabs ... bench: 10,498,793 ns/iter (+/- 104,360)
test tests::bench_aadefault ... bench: 11,061,862 ns/iter (+/- 94,107)
test tests::bench_abs ... bench: 10,477,193 ns/iter (+/- 81,942)
test tests::bench_default ... bench: 10,622,983 ns/iter (+/- 25,119)
test tests::bench_zzabs ... bench: 10,481,971 ns/iter (+/- 43,787)
test tests::bench_zzdefault ... bench: 11,074,976 ns/iter (+/- 29,633)
test result: ok. 0 passed; 0 failed; 1 ignored; 6 measured; 0 filtered out; finished in 19.35s
```
bench code:
```
#![feature(test)]
extern crate test;
fn rem_euclid(a:i32,rhs:i32)->i32{
let r = a % rhs;
if r < 0 { r + rhs.abs() } else { r }
}
#[cfg(test)]
mod tests {
use super::*;
use test::Bencher;
use rand::prelude::*;
use rand::rngs::SmallRng;
const N:i32=1000;
#[test]
fn it_works() {
let a: i32 = 7; // or any other integer type
let b = 4;
let d:Vec<i32>=(-N..=N).collect();
let n:Vec<i32>=(-N..0).chain(1..=N).collect();
for i in &d {
for j in &n {
assert_eq!(i.rem_euclid(*j),rem_euclid(*i,*j));
}
}
assert_eq!(rem_euclid(a,b), 3);
assert_eq!(rem_euclid(-a,b), 1);
assert_eq!(rem_euclid(a,-b), 3);
assert_eq!(rem_euclid(-a,-b), 1);
}
#[bench]
fn bench_aaabs(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=rem_euclid(*i,*j);
}
}
res
});
}
#[bench]
fn bench_aadefault(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=i.rem_euclid(*j);
}
}
res
});
}
#[bench]
fn bench_abs(b: &mut Bencher) {
let d:Vec<i32>=(-N..=N).collect();
let n:Vec<i32>=(-N..0).chain(1..=N).collect();
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=rem_euclid(*i,*j);
}
}
res
});
}
#[bench]
fn bench_default(b: &mut Bencher) {
let d:Vec<i32>=(-N..=N).collect();
let n:Vec<i32>=(-N..0).chain(1..=N).collect();
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=i.rem_euclid(*j);
}
}
res
});
}
#[bench]
fn bench_zzabs(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=rem_euclid(*i,*j);
}
}
res
});
}
#[bench]
fn bench_zzdefault(b: &mut Bencher) {
let mut d:Vec<i32>=(-N..=N).collect();
let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
d.shuffle(&mut rng);
n.shuffle(&mut rng);
d.shuffle(&mut rng);
b.iter(||{
let mut res=0;
for i in &d {
for j in &n {
res+=i.rem_euclid(*j);
}
}
res
});
}
}
```
