The float modules in `std` are currently top-level but for `core`, they
are nested within the `num` directory and referenced by `#[path = ...]`.
For consistency, adjust `std` to use the same structure as `core`.
Also change the `f16` and `f128` gates from outer attributes to inner
attributes like `core` has.
Split `autodiff` into `autodiff_forward` and `autodiff_reverse`
This PR splits `#[autodiff]` macro so `#[autodiff(df, Reverse, args)]` would become `#[autodiff_reverse(df, args)]` and `#[autodiff(df, Forward, args)]` would become `#[autodiff_forwad(df, args)]`.
Remove #![feature(let_chains)] from library and src/librustdoc
PR https://github.com/rust-lang/rust/pull/132833 has stabilized the `let_chains` feature. This PR removes the last occurences from the library, the compiler, and librustdoc (also because #140887 missed the conditional in one of the crates as it was behind the "rustc" feature).
We keep `core` as exercise for the future as updating it is non-trivial (see PR thread).
Since [1], `compiler-builtins` makes a certain set of math symbols
weakly available on all platforms. This means we can begin exposing some
of the related functions in `core`, so begin this process here.
It is not possible to provide inherent methods in both `core` and `std`
while giving them different stability gates, so standalone functions are
added instead. This provides a way to experiment with the functionality
while unstable; once it is time to stabilize, they can be converted to
inherent.
For `f16` and `f128`, everything is unstable so we can move the inherent
methods.
The following are included to start:
* floor
* ceil
* round
* round_ties_even
* trunc
* fract
* mul_add
* div_euclid
* rem_euclid
* powi
* sqrt
* abs_sub
* cbrt
These mirror the set of functions that we have in `compiler-builtins`
since [1].
Tracking issue: https://github.com/rust-lang/rust/issues/137578
[1]: https://github.com/rust-lang/compiler-builtins/pull/763
Delegate from `std::env::ArgsOs` to the methods of the inner
platform-specific iterators, when it would be more efficient than just
using the default methods of its own impl. Most platforms use
`vec::IntoIter` as the inner type, so prioritize delegating to the
methods it provides.
`std::env::Args` is implemented atop `std::env::ArgsOs` and performs
UTF-8 validation with a panic for invalid data. This is a visible effect
which users certainly rely on, so we can't skip any arguments. Any
further iterator methods would skip some elements, so no change is
needed for that type.
Add `#[inline]` for any methods which simply wrap the inner iterator.
Create `Atomic<T>` type alias (rebase)
Rebase of #130543.
Additional changes:
- Switch from `allow` to `expect` for `private_bounds` on `AtomicPrimitive`
- Unhide `AtomicPrimitive::AtomicInner` from docs, because rustdoc shows the definition `pub type Atomic<T> = <T as AtomicPrimitive>::AtomicInner;` and generated links for it.
- `NonZero` did not have this issue, because they kept the new alias private before the direction was changed.
- Use `Atomic<_>` in more places, including inside `Once`'s `Futex`. This is possible thanks to https://github.com/rust-lang/rust-clippy/pull/14125
The rest will either get moved back to #130543 or #130543 will be closed in favor of this instead.
---
* ACP: https://github.com/rust-lang/libs-team/issues/443#event-14293381061
* Tracking issue: #130539
docs: fix incorrect stability markers on `std::{todo, matches}`
This regression appeared in 916cfbcd3e. The change is behaving as expected (a non-glob re-export uses the stability marker on the `use` item, not the original one), but this part of the standard library didn't follow it.
Fixes https://github.com/rust-lang/rust/issues/140344
Stabilise `std::ffi::c_str`
This finished FCP in #112134 but never actually got a stabilisation PR. Since the FCP in #120048 recently passed to add the `os_str` module, it would be nice to also merge this too, to ensure that both get added in the next version.
Note: The added stability attributes which *somehow* were able to be omitted before (rustc bug?) were added based on the fact that they were added in 302551388b, which ended up in 1.85.0.
Closes: https://github.com/rust-lang/rust/issues/112134
r? libs-api
This regression appeared in 916cfbcd3e.
The change is behaving as expected (a non-glob re-export uses the
stability marker on the `use` item, not the original one), but
this part of the standard library didn't follow it.
`concat_idents` has been around unstably for a long time, but there is
now a better (but still unstable) way to join identifiers using
`${concat(...)}` syntax with `macro_metavar_expr_concat`. This resolves
a lot of the problems with `concat_idents` and is on a better track
toward stabilization, so there is no need to keep both versions around.
`concat_idents!` still has a lot of use in the ecosystem so deprecate it
before removing, as discussed in [1].
Link: https://github.com/rust-lang/rust/issues/124225
[1]: https://rust-lang.zulipchat.com/#narrow/channel/219381-t-libs/topic/Removing.20.60concat_idents.60
Expose algebraic floating point intrinsics
# Problem
A stable Rust implementation of a simple dot product is 8x slower than C++ on modern x86-64 CPUs. The root cause is an inability to let the compiler reorder floating point operations for better vectorization.
See https://github.com/calder/dot-bench for benchmarks. Measurements below were performed on a i7-10875H.
### C++: 10us ✅
With Clang 18.1.3 and `-O2 -march=haswell`:
<table>
<tr>
<th>C++</th>
<th>Assembly</th>
</tr>
<tr>
<td>
<pre lang="cc">
float dot(float *a, float *b, size_t len) {
#pragma clang fp reassociate(on)
float sum = 0.0;
for (size_t i = 0; i < len; ++i) {
sum += a[i] * b[i];
}
return sum;
}
</pre>
</td>
<td>
<img src="https://github.com/user-attachments/assets/739573c0-380a-4d84-9fd9-141343ce7e68" />
</td>
</tr>
</table>
### Nightly Rust: 10us ✅
With rustc 1.86.0-nightly (8239a37f9) and `-C opt-level=3 -C target-feature=+avx2,+fma`:
<table>
<tr>
<th>Rust</th>
<th>Assembly</th>
</tr>
<tr>
<td>
<pre lang="rust">
fn dot(a: &[f32], b: &[f32]) -> f32 {
let mut sum = 0.0;
for i in 0..a.len() {
sum = fadd_algebraic(sum, fmul_algebraic(a[i], b[i]));
}
sum
}
</pre>
</td>
<td>
<img src="https://github.com/user-attachments/assets/9dcf953a-2cd7-42f3-bc34-7117de4c5fb9" />
</td>
</tr>
</table>
### Stable Rust: 84us ❌
With rustc 1.84.1 (e71f9a9a9) and `-C opt-level=3 -C target-feature=+avx2,+fma`:
<table>
<tr>
<th>Rust</th>
<th>Assembly</th>
</tr>
<tr>
<td>
<pre lang="rust">
fn dot(a: &[f32], b: &[f32]) -> f32 {
let mut sum = 0.0;
for i in 0..a.len() {
sum += a[i] * b[i];
}
sum
}
</pre>
</td>
<td>
<img src="https://github.com/user-attachments/assets/936a1f7e-33e4-4ff8-a732-c3cdfe068dca" />
</td>
</tr>
</table>
# Proposed Change
Add `core::intrinsics::f*_algebraic` wrappers to `f16`, `f32`, `f64`, and `f128` gated on a new `float_algebraic` feature.
# Alternatives Considered
https://github.com/rust-lang/rust/issues/21690 has a lot of good discussion of various options for supporting fast math in Rust, but is still open a decade later because any choice that opts in more than individual operations is ultimately contrary to Rust's design principles.
In the mean time, processors have evolved and we're leaving major performance on the table by not supporting vectorization. We shouldn't make users choose between an unstable compiler and an 8x performance hit.
# References
* https://github.com/rust-lang/rust/issues/21690
* https://github.com/rust-lang/libs-team/issues/532
* https://github.com/rust-lang/rust/issues/136469
* https://github.com/calder/dot-bench
* https://www.felixcloutier.com/x86/vfmadd132ps:vfmadd213ps:vfmadd231ps
try-job: x86_64-gnu-nopt
try-job: x86_64-gnu-aux
By marking `__errno_location` as `#[ffi_const]` and `std::sys::os::errno` as `#[inline]`, this PR allows merging multiple calls to `io::Error::last_os_error()` into one.
Add `MAX_LEN_UTF8` and `MAX_LEN_UTF16` Constants
This pull request adds the `MAX_LEN_UTF8` and `MAX_LEN_UTF16` constants as per #45795, gated behind the `char_max_len` feature.
The constants are currently applied in the `alloc`, `core` and `std` libraries.
* Renames the methods:
* `get_many_mut` -> `get_disjoint_mut`
* `get_many_unchecked_mut` -> `get_disjoint_unchecked_mut`
* Does not rename the feature flag: `get_many_mut`
* Marks the feature as stable
* Renames some helper stuff:
* `GetManyMutError` -> `GetDisjointMutError`
* `GetManyMutIndex` -> `GetDisjointMutIndex`
* `get_many_mut_helpers` -> `get_disjoint_mut_helpers`
* `get_many_check_valid` -> `get_disjoint_check_valid`
This only touches slice methods.
HashMap's methods and feature gates are not renamed here
(nor are they stabilized).
Implement `ByteStr` and `ByteString` types
Approved ACP: https://github.com/rust-lang/libs-team/issues/502
Tracking issue: https://github.com/rust-lang/rust/issues/134915
These types represent human-readable strings that are conventionally,
but not always, UTF-8. The `Debug` impl prints non-UTF-8 bytes using
escape sequences, and the `Display` impl uses the Unicode replacement
character.
This is a minimal implementation of these types and associated trait
impls. It does not add any helper methods to other types such as `[u8]`
or `Vec<u8>`.
I've omitted a few implementations of `AsRef`, `AsMut`, and `Borrow`,
when those would be the second implementation for a type (counting the
`T` impl), to avoid potential inference failures. We can attempt to add
more impls later in standalone commits, and run them through crater.
In addition to the `bstr` feature, I've added a `bstr_internals` feature
for APIs provided by `core` for use by `alloc` but not currently
intended for stabilization.
This API and its implementation are based *heavily* on the `bstr` crate
by Andrew Gallant (`@BurntSushi).`
r? `@BurntSushi`
Use `NonNull::without_provenance` within the standard library
This API removes the need for several `unsafe` blocks, and leads to clearer code. It uses feature `nonnull_provenance` (#135243).
Close#135343
Approved ACP: https://github.com/rust-lang/libs-team/issues/502
Tracking issue: https://github.com/rust-lang/rust/issues/134915
These types represent human-readable strings that are conventionally,
but not always, UTF-8. The `Debug` impl prints non-UTF-8 bytes using
escape sequences, and the `Display` impl uses the Unicode replacement
character.
This is a minimal implementation of these types and associated trait
impls. It does not add any helper methods to other types such as `[u8]`
or `Vec<u8>`.
I've omitted a few implementations of `AsRef`, `AsMut`, `Borrow`,
`From`, and `PartialOrd`, when those would be the second implementation
for a type (counting the `T` impl) or otherwise may cause inference
failures. These impls are important, but we can attempt to add them
later in standalone commits, and run them through crater.
In addition to the `bstr` feature, I've added a `bstr_internals` feature
for APIs provided by `core` for use by `alloc` but not currently
intended for stabilization.
This API and its implementation are based *heavily* on the `bstr` crate
by Andrew Gallant (@BurntSushi).
This greatly reduces the number of places that actually use the `rustc_layout_scalar_valid_range_*` attributes down to just 3:
```
library/core\src\ptr\non_null.rs
68:#[rustc_layout_scalar_valid_range_start(1)]
library/core\src\num\niche_types.rs
19: #[rustc_layout_scalar_valid_range_start($low)]
20: #[rustc_layout_scalar_valid_range_end($high)]
```
Everything else -- PAL Nanoseconds, alloc's `Cap`, niched FDs, etc -- all just wrap those `niche_types` types.
`CheckAttrVisitor::check_doc_keyword` checks `#[doc(keyword = "..")]`
attributes to ensure they are on an empty module, and that the value is
a non-empty identifier.
The `rustc::existing_doc_keyword` lint checks these attributes to ensure
that the value is the name of a keyword.
It's silly to have two different checking mechanisms for these
attributes. This commit does the following.
- Changes `check_doc_keyword` to check that the value is the name of a
keyword (avoiding the need for the identifier check, which removes a
dependency on `rustc_lexer`).
- Removes the lint.
- Updates tests accordingly.
There is one hack: the `SelfTy` FIXME case used to used to be handled by
disabling the lint, but now is handled with a special case in
`is_doc_keyword`. That hack will go away if/when the FIXME is fixed.
Co-Authored-By: Guillaume Gomez <guillaume1.gomez@gmail.com>
`UniqueRc` trait impls
UniqueRc tracking Issue: #112566
Stable traits: (i.e. impls behind only the `unique_rc_arc` feature gate)
* Support the same formatting as `Rc`:
* `fmt::Debug` and `fmt::Display` delegate to the pointee.
* `fmt::Pointer` prints the address of the pointee.
* Add explicit `!Send` and `!Sync` impls, to mirror `Rc`.
* Borrowing traits: `Borrow`, `BorrowMut`, `AsRef`, `AsMut`
* `Rc` does not implement `BorrowMut` and `AsMut`, but `UniqueRc` can.
* Unconditional `Unpin`, like other heap-allocated types.
* Comparison traits `(Partial)Ord` and `(Partial)Eq` delegate to the pointees.
* `PartialEq for UniqueRc` does not do `Rc`'s specialization shortcut for pointer equality when `T: Eq`, since by definition two `UniqueRc`s cannot share an allocation.
* `Hash` delegates to the pointee.
* `AsRawFd`, `AsFd`, `AsHandle`, `AsSocket` delegate to the pointee like `Rc`.
* Sidenote: The bounds on `T` for the existing `Pointer<T>` impls for specifically `AsRawFd` and `AsSocket` do not allow `T: ?Sized`. For the added `UniqueRc` impls I allowed `T: ?Sized` for all four traits, but I did not change the existing (stable) impls.
Unstable traits:
* `DispatchFromDyn`, allows using `UniqueRc<Self>` as a method receiver under `feature(arbitrary_self_types)`.
* Existing `PinCoerceUnsized for UniqueRc` is generalized to allow non-`Global` allocators, like `Rc`.
* `DerefPure`, allows using `UniqueRc` in deref-patterns under `feature(deref_patterns)`, like `Rc`.
For documentation, `Rc` only has documentation on the comparison traits' methods, so I copied/adapted the documentation for those, and left the rest without impl-specific docs.
~~Edit: Marked as draft while I figure out `UnwindSafe`.~~
Edit: Ignoring `UnwindSafe` for this PR
