rustc: Fill out remaining parts of C-unwind ABI
This commit intends to fill out some of the remaining pieces of the
C-unwind ABI. This has a number of other changes with it though to move
this design space forward a bit. Notably contained within here is:
* On `panic=unwind`, the `extern "C"` ABI is now considered as "may
unwind". This fixes a longstanding soundness issue where if you
`panic!()` in an `extern "C"` function defined in Rust that's actually
UB because the LLVM representation for the function has the `nounwind`
attribute, but then you unwind.
* Whether or not a function unwinds now mainly considers the ABI of the
function instead of first checking the panic strategy. This fixes a
miscompile of `extern "C-unwind"` with `panic=abort` because that ABI
can still unwind.
* The aborting stub for non-unwinding ABIs with `panic=unwind` has been
reimplemented. Previously this was done as a small tweak during MIR
generation, but this has been moved to a separate and dedicated MIR
pass. This new pass will, for appropriate functions and function
calls, insert a `cleanup` landing pad for any function call that may
unwind within a function that is itself not allowed to unwind. Note
that this subtly changes some behavior from before where previously on
an unwind which was caught-to-abort it would run active destructors in
the function, and now it simply immediately aborts the process.
* The `#[unwind]` attribute has been removed and all users in tests and
such are now using `C-unwind` and `#![feature(c_unwind)]`.
I think this is largely the last piece of the RFC to implement.
Unfortunately I believe this is still not stabilizable as-is because
activating the feature gate changes the behavior of the existing `extern
"C"` ABI in a way that has no replacement. My thinking for how to enable
this is that we add support for the `C-unwind` ABI on stable Rust first,
and then after it hits stable we change the behavior of the `C` ABI.
That way anyone straddling stable/beta/nightly can switch to `C-unwind`
safely.
#[inline] slice::Iter::advance_by
https://github.com/rust-lang/rust/pull/87387#issuecomment-891942661 was marked as a regression. One of the methods in the PR was missing an inline annotation unlike all the other methods on slice iterators.
Let's see if that makes a difference.
Make wrapping_neg() use wrapping_sub(), #[inline(always)]
This is a follow-up change to the fix for #75598. It simplifies the implementation of wrapping_neg() for all integer types by just calling 0.wrapping_sub(self) and always inlines it. This leads to much less assembly code being emitted for opt-level≤1 and thus much better performance for debug-compiled code.
Background is [this discussion on the internals forum](https://internals.rust-lang.org/t/why-does-rust-generate-10x-as-much-unoptimized-assembly-as-gcc/14930).
Add `core::stream::from_iter`
_Tracking issue: https://github.com/rust-lang/rust/issues/81798_
This_ PR implements `std::stream::from_iter`, as outlined in the _"Converting an Iterator to a Stream"_ section of the [Stream RFC](https://github.com/nellshamrell/rfcs/blob/add-async-stream-rfc/text/0000-async-stream.md#converting-an-iterator-to-a-stream). This function enables converting an `Iterator` to a `Stream` by wrapping each item in the iterator with a `Poll::Ready` instance.
r? `@tmandry`
cc/ `@rust-lang/libs` `@rust-lang/wg-async-foundations`
## Example
Being able to convert from an iterator into a stream is useful when refactoring from iterative loops into a more functional adapter-based style. This is fairly common when using more complex `filter` / `map` / `find` chains. In its basic form this conversion looks like this:
**before**
```rust
let mut output = vec![];
for item in my_vec {
let out = do_io(item).await?;
output.push(out);
}
```
**after**
```rust
use std::stream;
let output = stream::from_iter(my_vec.iter())
.map(async |item| do_io(item).await)
.collect()?;
```
Having a way to convert an `Iterator` to a `Stream` is essential in enabling this flow.
## Implementation Notes
This PR makes use of `unsafe {}` to pin an item. Currently we're having conversations on the libs stream in Zulip how to bring `pin-project` in as a dependency to `core` so we can omit the `unsafe {}`.
This PR also includes a documentation block which references `Stream::next` which currently doesn't exist in the stdlib (originally included in the RFC and PR, but later omitted because of an unresolved issue). `stream::from_iter` can't stabilize before `Stream` does, and there's still a chance we may stabilize `Stream` with a `next` method. So this PR includes documentation referencing that method, which we can remove as part of stabilization if by any chance we don't have `Stream::next`.
## Alternatives Considered
### `impl IntoStream for T: IntoIterator`
An obvious question would be whether we could make it so every iterator can automatically be converted into a stream by calling `into_stream` on it. The answer is: "perhaps, but it could cause type issues". Types like `std::collections` may want to opt to create manual implementations for `IntoStream` and `IntoIter`, which wouldn't be possible if it was implemented through a catch-all trait.
Possibly an alternative such as `impl IntoStream for T: Iterator` could work, but it feels somewhat restrictive. In the end, converting an iterator to a stream is likely to be a bit of a niche case. And even then, **adding a standalone function to convert an `Iterator` into a `Stream` would not be mutually exclusive with a blanket implementation**.
### Naming
The exact name can be debated in the period before stabilization. But I've chosen `stream::from_iter` rather than `stream::iter` because we are _creating a stream from an iterator_ rather than _iterating a stream_. We also expect to add a stream counterpart to `iter::from_fn` later on (blocked on async closures), and having `stream::from_fn` and `stream::from_iter` would feel like a consistent pair. It also has prior art in `async_std::stream::from_iter`.
## Future Directions
### Stream conversions for collections
This is a building block towards implementing `stream/stream_mut/into_stream` methods for `std::collections`, `std::vec`, and more. This would allow even quicker refactorings from using loops to using iterator adapters by omitting the import altogether:
**before**
```rust
use std::stream;
let output = stream::from_iter(my_vec.iter())
.map(async |item| do_io(item).await)
.collect()?;
```
**after**
```rust
let output = my_vec
.stream()
.map(async |item| do_io(item).await)
.collect()?;
```
This commit intends to fill out some of the remaining pieces of the
C-unwind ABI. This has a number of other changes with it though to move
this design space forward a bit. Notably contained within here is:
* On `panic=unwind`, the `extern "C"` ABI is now considered as "may
unwind". This fixes a longstanding soundness issue where if you
`panic!()` in an `extern "C"` function defined in Rust that's actually
UB because the LLVM representation for the function has the `nounwind`
attribute, but then you unwind.
* Whether or not a function unwinds now mainly considers the ABI of the
function instead of first checking the panic strategy. This fixes a
miscompile of `extern "C-unwind"` with `panic=abort` because that ABI
can still unwind.
* The aborting stub for non-unwinding ABIs with `panic=unwind` has been
reimplemented. Previously this was done as a small tweak during MIR
generation, but this has been moved to a separate and dedicated MIR
pass. This new pass will, for appropriate functions and function
calls, insert a `cleanup` landing pad for any function call that may
unwind within a function that is itself not allowed to unwind. Note
that this subtly changes some behavior from before where previously on
an unwind which was caught-to-abort it would run active destructors in
the function, and now it simply immediately aborts the process.
* The `#[unwind]` attribute has been removed and all users in tests and
such are now using `C-unwind` and `#![feature(c_unwind)]`.
I think this is largely the last piece of the RFC to implement.
Unfortunately I believe this is still not stabilizable as-is because
activating the feature gate changes the behavior of the existing `extern
"C"` ABI in a way that has no replacement. My thinking for how to enable
this is that we add support for the `C-unwind` ABI on stable Rust first,
and then after it hits stable we change the behavior of the `C` ABI.
That way anyone straddling stable/beta/nightly can switch to `C-unwind`
safely.
Add missing "allocated object" doc link to `<*mut T>::add`
The portion of the documentation expecting the link was already there, but it was rendered as "[allocated object]". The added reference is just copied from the documentation for `<*const T>::add`.
rfc3052 followup: Remove authors field from Cargo manifests
Since RFC 3052 soft deprecated the authors field, hiding it from
crates.io, docs.rs, and making Cargo not add it by default, and it is
not generally up to date/useful information for contributors, we may as well
remove it from crates in this repo.
Partially stabilize `const_slice_first_last`
This stabilizes the non-`mut` methods of `const_slice_first_last` as `const`. These methods are trivial to implement and have no blockers that I am aware of.
`@rustbot` label +A-const-fn +S-waiting-on-review +T-libs-api
Move UnwindSafe, RefUnwindSafe, AssertUnwindSafe to core
They were previously only available in std::panic, not core::panic.
- https://doc.rust-lang.org/1.51.0/std/panic/trait.UnwindSafe.html
- https://doc.rust-lang.org/1.51.0/std/panic/trait.RefUnwindSafe.html
- https://doc.rust-lang.org/1.51.0/std/panic/struct.AssertUnwindSafe.html
Where this is relevant: trait objects! Inside a `#![no_std]` library it's otherwise impossible to have a struct holding a trait object, and at the same time can be used from downstream std crates in a way that doesn't interfere with catch_unwind.
```rust
// common library
#![no_std]
pub struct Thing {
pub(crate) x: &'static (dyn SomeTrait + Send + Sync),
}
pub(crate) trait SomeTrait {...}
```
```rust
// downstream application
fn main() {
let thing: library::Thing = ...;
let _ = std::panic::catch_unwind(|| { let _ = thing; }); // does not work :(
}
```
See a4131708e2/src/gradient.rs (L7-L15) for a real life example of needing to work around this problem. In particular that workaround would not even be viable if implementors of the trait were provided externally by a caller, as the `feature = "std"` would become non-additive in that case.
What happens without the UnwindSafe constraints:
```rust
fn main() {
let gradient = colorous::VIRIDIS;
let _ = std::panic::catch_unwind(|| { let _ = gradient; });
}
```
```console
error[E0277]: the type `(dyn colorous::gradient::EvalGradient + Send + Sync + 'static)` may contain interior mutability and a reference may not be safely transferrable across a catch_unwind boundary
--> src/main.rs:3:13
|
3 | let _ = std::panic::catch_unwind(|| { let _ = gradient; });
| ^^^^^^^^^^^^^^^^^^^^^^^^ `(dyn colorous::gradient::EvalGradient + Send + Sync + 'static)` may contain interior mutability and a reference may not be safely transferrable across a catch_unwind boundary
|
::: .rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/std/src/panic.rs:430:40
|
430 | pub fn catch_unwind<F: FnOnce() -> R + UnwindSafe, R>(f: F) -> Result<R> {
| ---------- required by this bound in `catch_unwind`
|
= help: within `Gradient`, the trait `RefUnwindSafe` is not implemented for `(dyn colorous::gradient::EvalGradient + Send + Sync + 'static)`
= note: required because it appears within the type `&'static (dyn colorous::gradient::EvalGradient + Send + Sync + 'static)`
= note: required because it appears within the type `Gradient`
= note: required because of the requirements on the impl of `UnwindSafe` for `&Gradient`
= note: required because it appears within the type `[closure@src/main.rs:3:38: 3:62]`
```
Implement advance_by, advance_back_by for slice::{Iter, IterMut}
Part of #77404.
Picking up where #77633 was closed.
I have addressed https://github.com/rust-lang/rust/pull/77633#issuecomment-771842599 by restoring `nth` and `nth_back`. So according to that comment this should already be r=m-ou-se, but it has been sitting for a while.
Add docs about performance and `Iterator::map` to `[T; N]::map`
This suboptimal code gen for some usages of array::map got a bit of
attention by multiple people throughout the community. Some cases:
- https://github.com/rust-lang/rust/issues/75243#issuecomment-866051086
- https://github.com/rust-lang/rust/issues/75243#issuecomment-874732134
- https://www.reddit.com/r/rust/comments/oeqqf7/unexpected_high_stack_usage/
My *guess* is that this gets the attention it gets because in JavaScript
(and potentially other languages), a `map` function on arrays is very
commonly used since in those languages, arrays basically take the role
of Rust's iterator. I considered explicitly naming JavaScript in the
first paragraph I added, but I couldn't find precedence of mentioning
other languages in standard library doc, so I didn't add it.
When array::map was stabilized, we still wanted to add docs, but that
somehow did not happen in time. So here we are. Not sure if this sounds
crazy but maybe it is worth considering beta backporting this? Only if
it's not a lot of work, of course! But yeah, stabilized array::map is
already in beta and if this problem is really as big as it sometimes seems,
might be worth having the docs in place when 1.55 is released.
CC ``@CryZe``
r? ``@m-ou-se`` (since you were involved in that discussion and the stabilization)
Optimize fmt::PadAdapter::wrap
After adding the first `write!` usage to my project and printing the result to the console, I noticed, that my binary contains the strings "called `Option::unwrap()` on a `None` value`" and more importantly "C:\Users\Patrick Fischer\.rustup\toolchains\nightly-x86_64-pc-windows-msvc\lib\rustlib\src\rust\library\core\src\fmt\builders.rs", with my release build being configured as follows:
```
[profile.release]
panic = "abort"
codegen-units = 1
strip = "symbols" # the important bit
lto = true
```
I am in a no_std environment and my custom panic handler is a simple `loop {}`. I did not expect the above information to be preserved. I heavily suspect the edited function to be the culprit. It contains the only direct use of `Option::unwrap` in the entire file and I tracked the symbols in the assembly to be used from the section `_ZN68_$LT$core..fmt..builders..PadAdapter$u20$as$u20$core..fmt..Write$GT$9write_str17ha1d5e5efe167202aE`.
Aside from me suspecting this function to be the culprit, the replaced code performs the same operation as `Option::insert`, but without the `unreachable_unchecked` optimization `Option::insert` provides. Therefore, it makes sense to me to use the more optimized version, instead.
As I don't change any semantics, I hope a simple pull request suffices.
Fix may not to appropriate might not or must not
I went through and changed occurrences of `may not` to be more explicit with `might not` and `must not`.
Since RFC 3052 soft deprecated the authors field anyway, hiding it from
crates.io, docs.rs, and making Cargo not add it by default, and it is
not generally up to date/useful information, we should remove it from
crates in this repo.
Add #[track_caller] for some function in core::mem.
These functions can panic for some types. This makes the panic point to the code that calls e.g. mem::uninitialized(), instead of inside the definition of mem::uninitialized.
Make const panic!("..") work in Rust 2021.
During const eval, this replaces calls to core::panicking::panic_fmt and std::panicking::being_panic_fmt with a call to a new const fn: core::panicking::const_panic_fmt. That function uses fmt::Arguments::as_str() to get the str and calls panic_str with that instead.
panic!() invocations with formatting arguments are still not accepted, as the creation of such a fmt::Arguments cannot be done in constant functions right now.
r? `@RalfJung`
Remove unsound TrustedRandomAccess implementations
Removes the implementations that depend on the user-definable trait `Copy`.
Fixes#85873 in the most straightforward way.
<hr>
_Edit:_ This PR now contains additional trait infrastructure to avoid performance regressions around in-place collect, see the discussion in this thread starting from the codegen test failure at https://github.com/rust-lang/rust/pull/85874#issuecomment-872327577.
With this PR, `TrustedRandomAccess` gains additional documentation that specifically allows for and specifies the safety conditions around subtype coercions – those coercions can happen in safe Rust code with the `Zip` API’s usage of `TrustedRandomAccess`. This PR introduces a new supertrait of `TrustedRandomAccess`(currently named `TrustedRandomAccessNoCoerce`) that _doesn’t allow_ such coercions, which means it can be still be useful for optimizing cases such as in-place collect where no iterator is handed out to a user (who could do coercions) after a `get_unchecked` call; the benefit of the supertrait is that it doesn’t come with the additional safety conditions around supertraits either, so it can be implemented for more types than `TrustedRandomAccess`.
The `TrustedRandomAccess` implementations for `vec::IntoIter`, `vec_deque::IntoIter`, and `array::IntoIter` are removed as they don’t conform with the newly documented safety conditions, this way unsoundness is removed. But this PR in turn (re-)adds a `TrustedRandomAccessNoCoerce` implementation for `vec::IntoIter` to avoid performance regressions from stable in a case of in-place collecting of `Vec`s [the above-mentioned codegen test failure]. Re-introducing the (currently nightly+beta-only) impls for `VecDeque`’s and `[T; N]`’s iterators is technically possible, but goes beyond the scope of this PR (i.e. it can happen in a future PR).
Remove P: Unpin bound on impl Future for Pin
We can safely produce a `Pin<&mut P::Target>` without moving out of the `Pin` by using `Pin::as_mut` directly.
The `Unpin` bound was originally added in #56939 following the recommendation of ``@withoutboats`` in https://github.com/rust-lang/rust/issues/55766#issue-378417538
That comment does not give explicit justification for why the bound should be added. The relevant context was:
> [ ] Remove `impl<P> Unpin for Pin<P>`
>
> This impl is not justified by our standard justification for unpin impls: there is no pointer direction between `Pin<P>` and `P`. Its usefulness is covered by the impls for pointers themselves.
>
> This futures impl (link to the impl changed in this PR) will need to change to add a `P: Unpin` bound.
The decision to remove the unconditional impl of `Unpin for Pin` is sound (these days there is just an auto-impl for when `P: Unpin`). But, I think the decision to also add the `Unpin` bound for `impl Future` may have been unnecessary. Or if that's not the case, I'd be very interested to have the argument for why written down somewhere. The bound _appears_ to not be needed, as demonstrated by the change requiring no unsafe code and by the existence of `Pin::as_mut`.
Stabilize core::task::ready!
_Tracking issue: https://github.com/rust-lang/rust/issues/70922_
This PR stabilizes the `task::ready!` macro. Similar to https://github.com/rust-lang/rust/pull/80886, this PR was waiting on https://github.com/rust-lang/rust/issues/74355 to be fixed.
The `task::ready!` API has existed in the futures ecosystem for several years, and was added on nightly last year in https://github.com/rust-lang/rust/pull/70817. The motivation for this macro is the same as it was back then: virtually every single manual future implementation makes use of this; so much so that it's one of the few things included in the [futures-core](https://docs.rs/futures-core/0.3.12/futures_core) library.
r? ``@tmandry``
cc/ ``@rust-lang/wg-async-foundations`` ``@rust-lang/libs``
## Example
```rust
use core::task::{Context, Poll};
use core::future::Future;
use core::pin::Pin;
async fn get_num() -> usize {
42
}
pub fn do_poll(cx: &mut Context<'_>) -> Poll<()> {
let mut f = get_num();
let f = unsafe { Pin::new_unchecked(&mut f) };
let num = ready!(f.poll(cx));
// ... use num
Poll::Ready(())
}
```
During const eval, this replaces calls to core::panicking::panic_fmt and
std::panicking::being_panic_fmt with a call to a new const fn:
core::panicking::const_panic_fmt. That function uses
fmt::Arguments::as_str() to get the str and calls panic_str with that
instead.
panic!() invocations with formatting arguments are still not accepted,
as the creation of such a fmt::Arguments cannot be done in constant
functions right now.
These functions can panic for some types. This makes the panic point to
the code that calls e.g. mem::uninitialized(), instead of inside the
definition of mem::uninitialized.
Include new details regarding coercions to a subtype.
These conditions also explain why the previously removed implementations
for {array, vec, vec_deque}::IntoIter<T> were unsound, because they introduced
an extra `T: Clone` for the TrustedRandomAccess impl, even though their parameter T
is covariant.
Document math behind MIN/MAX consts on integers
Currently the documentation for `[integer]::{MIN, MAX}` doesn't explain where the constants come from. This documents how the values of those constants are related to powers of 2.
Implement RFC 3107: `#[derive(Default)]` on enums with a `#[default]` attribute
This PR implements RFC 3107, which permits `#[derive(Default)]` on enums where a unit variant has a `#[default]` attribute. See comments for current status.
implement fold() on array::IntoIter to improve flatten().collect() perf
With #87168 flattening `array::IntoIter`s is now `TrustedLen`, the `FromIterator` implementation for `Vec` has a specialization for `TrustedLen` iterators which uses internal iteration. This implements one of the main internal iteration methods on `array::Into` to optimize the combination of those two features.
This should address the main issue in #87411
```
# old
test vec::bench_flat_map_collect ... bench: 2,244,024 ns/iter (+/- 18,903)
# new
test vec::bench_flat_map_collect ... bench: 172,863 ns/iter (+/- 2,141)
```
Make StrSearcher behave correctly on empty needle
Fix#85462.
This will not affect ABI since the other variant of the enum is bigger.
It may break some code, but that would be very strange: usually people
don't continue after the first `Done` (or `None` for a normal iterator).
`@rustbot` label T-libs A-str A-patterns
DOC: remove unnecessary feature crate attribute from example code
I'm not sure whether I fully understand the stabilization process (I most likely don't), but I think this attribute isn't necessary here, right?
This was recently stabilized in #86344.
DOC: fix hypothetical Rust code in `step_by()` docstring
I don't know how important that is, but if I'm not mistaken, the hypothetical code in the docstring of `step_by()` (see https://doc.rust-lang.org/std/iter/trait.Iterator.html#method.step_by) isn't correct.
I guess writing `next()` instead of `self.next()` isn't a biggie, but this would also imply that `advance_n_and_return_first()` is a method, which AFAICT it isn't.
I've also done some re-formatting in a separate commit and a parameter renaming in yet another commit.
Feel free to take or leave any combination of those commits.
Regression fix to avoid further beta backports: Remove unsound TrustedRandomAccess implementations
Removes the implementations that depend on the user-definable trait `Copy`.
Only fix regressions to ensure merge in 1.55: Does not modify `vec::IntoIter`.
<hr>
This PR applies the beta-`1.53` backport #86222 (merged as part of #86225), a reduced version of #85874 that only fixes regressions, to `master` in order to avoid the need for further backports from `1.55` onwards. Beta-`1.54` backport already happened with #87136. In case that #85874 gets merged quickly (within a week), this PR would be unnecessary.
r? `@cuviper`
docs: GlobalAlloc: completely replace example with one that works
Since this is an example, this could really do with some review from someone familiar with unsafe stuff!
I made the example no longer `no_run` since it works for me.
Fixes#81847
Removes the implementations that depend on the user-definable trait `Copy`.
Only fix regressions to ensure merge in 1.55: Does not modify `vec::IntoIter`.
implement TrustedLen for Flatten/FlatMap if the U: IntoIterator == [T; N]
This only works if arrays are passed directly instead of array iterators
because we need to be sure that they have not been advanced before
Flatten does its size calculation.
resolves#87094
Since this is an example, this could really do with some review from
someone familiar with unsafe stuff !
I made the example no longer `no_run` since it works for me.
Fixes#81847
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
Co-authored-by: Amanieu d'Antras <amanieu@gmail.com>
Move asm! and global_asm! to core::arch
Follow-up to https://github.com/rust-lang/stdarch/pull/1183 .
Implements the libs-api team decision from rust-lang/rust#84019 (comment) .
In order to not break nightly users, this PR also adds the newly-moved items to the prelude. However, a decision will need to be made before stabilization as to whether these items should remain in the prelude. I will file an issue for this separately.
Fixes#84019 .
r? `@Amanieu`
Mark `Option::insert` as must_use
Some people seems misled by the function name and use it in case where a simple assignment just works.
If the return value is not used, `option = Some(value);` should be preferred instead of `option.insert(value);`
Add diagnostic items for Clippy
This adds a bunch of diagnostic items to `std`/`core`/`alloc` functions, structs and traits used in Clippy. The actual refactorings in Clippy to use these items will be done in a different PR in Clippy after the next sync.
This PR doesn't include all paths Clippy uses, I've only gone through the first 85 lines of Clippy's [`paths.rs`](ecf85f4bdc/clippy_utils/src/paths.rs) (after rust-lang/rust-clippy#7466) to get some feedback early on. I've also decided against adding diagnostic items to methods, as it would be nicer and more scalable to access them in a nicer fashion, like adding a `is_diagnostic_assoc_item(did, sym::Iterator, sym::map)` function or something similar (Suggested by `@camsteffen` [on Zulip](https://rust-lang.zulipchat.com/#narrow/stream/147480-t-compiler.2Fwg-diagnostics/topic/Diagnostic.20Item.20Naming.20Convention.3F/near/225024603))
There seems to be some different naming conventions when it comes to diagnostic items, some use UpperCamelCase (`BinaryHeap`) and some snake_case (`hashmap_type`). This PR uses UpperCamelCase for structs and traits and snake_case with the module name as a prefix for functions. Any feedback on is this welcome.
cc: rust-lang/rust-clippy#5393
r? `@Manishearth`
Update Rust Float-Parsing Algorithms to use the Eisel-Lemire algorithm.
# Summary
Rust, although it implements a correct float parser, has major performance issues in float parsing. Even for common floats, the performance can be 3-10x [slower](https://arxiv.org/pdf/2101.11408.pdf) than external libraries such as [lexical](https://github.com/Alexhuszagh/rust-lexical) and [fast-float-rust](https://github.com/aldanor/fast-float-rust).
Recently, major advances in float-parsing algorithms have been developed by Daniel Lemire, along with others, and implement a fast, performant, and correct float parser, with speeds up to 1200 MiB/s on Apple's M1 architecture for the [canada](0e2b5d163d/data/canada.txt) dataset, 10x faster than Rust's 130 MiB/s.
In addition, [edge-cases](https://github.com/rust-lang/rust/issues/85234) in Rust's [dec2flt](868c702d0c/library/core/src/num/dec2flt) algorithm can lead to over a 1600x slowdown relative to efficient algorithms. This is due to the use of Clinger's correct, but slow [AlgorithmM and Bellepheron](http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.45.4152&rep=rep1&type=pdf), which have been improved by faster big-integer algorithms and the Eisel-Lemire algorithm, respectively.
Finally, this algorithm provides substantial improvements in the number of floats the Rust core library can parse. Denormal floats with a large number of digits cannot be parsed, due to use of the `Big32x40`, which simply does not have enough digits to round a float correctly. Using a custom decimal class, with much simpler logic, we can parse all valid decimal strings of any digit count.
```rust
// Issue in Rust's dec2fly.
"2.47032822920623272088284396434110686182e-324".parse::<f64>(); // Err(ParseFloatError { kind: Invalid })
```
# Solution
This pull request implements the Eisel-Lemire algorithm, modified from [fast-float-rust](https://github.com/aldanor/fast-float-rust) (which is licensed under Apache 2.0/MIT), along with numerous modifications to make it more amenable to inclusion in the Rust core library. The following describes both features in fast-float-rust and improvements in fast-float-rust for inclusion in core.
**Documentation**
Extensive documentation has been added to ensure the code base may be maintained by others, which explains the algorithms as well as various associated constants and routines. For example, two seemingly magical constants include documentation to describe how they were derived as follows:
```rust
// Round-to-even only happens for negative values of q
// when q ≥ −4 in the 64-bit case and when q ≥ −17 in
// the 32-bitcase.
//
// When q ≥ 0,we have that 5^q ≤ 2m+1. In the 64-bit case,we
// have 5^q ≤ 2m+1 ≤ 2^54 or q ≤ 23. In the 32-bit case,we have
// 5^q ≤ 2m+1 ≤ 2^25 or q ≤ 10.
//
// When q < 0, we have w ≥ (2m+1)×5^−q. We must have that w < 2^64
// so (2m+1)×5^−q < 2^64. We have that 2m+1 > 2^53 (64-bit case)
// or 2m+1 > 2^24 (32-bit case). Hence,we must have 2^53×5^−q < 2^64
// (64-bit) and 2^24×5^−q < 2^64 (32-bit). Hence we have 5^−q < 2^11
// or q ≥ −4 (64-bit case) and 5^−q < 2^40 or q ≥ −17 (32-bitcase).
//
// Thus we have that we only need to round ties to even when
// we have that q ∈ [−4,23](in the 64-bit case) or q∈[−17,10]
// (in the 32-bit case). In both cases,the power of five(5^|q|)
// fits in a 64-bit word.
const MIN_EXPONENT_ROUND_TO_EVEN: i32;
const MAX_EXPONENT_ROUND_TO_EVEN: i32;
```
This ensures maintainability of the code base.
**Improvements for Disguised Fast-Path Cases**
The fast path in float parsing algorithms attempts to use native, machine floats to represent both the significant digits and the exponent, which is only possible if both can be exactly represented without rounding. In practice, this means that the significant digits must be 53-bits or less and the then exponent must be in the range `[-22, 22]` (for an f64). This is similar to the existing dec2flt implementation.
However, disguised fast-path cases exist, where there are few significant digits and an exponent above the valid range, such as `1.23e25`. In this case, powers-of-10 may be shifted from the exponent to the significant digits, discussed at length in https://github.com/rust-lang/rust/issues/85198.
**Digit Parsing Improvements**
Typically, integers are parsed from string 1-at-a-time, requiring unnecessary multiplications which can slow down parsing. An approach to parse 8 digits at a time using only 3 multiplications is described in length [here](https://johnnylee-sde.github.io/Fast-numeric-string-to-int/). This leads to significant performance improvements, and is implemented for both big and little-endian systems.
**Unsafe Changes**
Relative to fast-float-rust, this library makes less use of unsafe functionality and clearly documents it. This includes the refactoring and documentation of numerous unsafe methods undesirably marked as safe. The original code would look something like this, which is deceptively marked as safe for unsafe functionality.
```rust
impl AsciiStr {
#[inline]
pub fn step_by(&mut self, n: usize) -> &mut Self {
unsafe { self.ptr = self.ptr.add(n) };
self
}
}
...
#[inline]
fn parse_scientific(s: &mut AsciiStr<'_>) -> i64 {
// the first character is 'e'/'E' and scientific mode is enabled
let start = *s;
s.step();
...
}
```
The new code clearly documents safety concerns, and does not mark unsafe functionality as safe, leading to better safety guarantees.
```rust
impl AsciiStr {
/// Advance the view by n, advancing it in-place to (n..).
pub unsafe fn step_by(&mut self, n: usize) -> &mut Self {
// SAFETY: same as step_by, safe as long n is less than the buffer length
self.ptr = unsafe { self.ptr.add(n) };
self
}
}
...
/// Parse the scientific notation component of a float.
fn parse_scientific(s: &mut AsciiStr<'_>) -> i64 {
let start = *s;
// SAFETY: the first character is 'e'/'E' and scientific mode is enabled
unsafe {
s.step();
}
...
}
```
This allows us to trivially demonstrate the new implementation of dec2flt is safe.
**Inline Annotations Have Been Removed**
In the previous implementation of dec2flt, inline annotations exist practically nowhere in the entire module. Therefore, these annotations have been removed, which mostly does not impact [performance](https://github.com/aldanor/fast-float-rust/issues/15#issuecomment-864485157).
**Fixed Correctness Tests**
Numerous compile errors in `src/etc/test-float-parse` were present, due to deprecation of `time.clock()`, as well as the crate dependencies with `rand`. The tests have therefore been reworked as a [crate](https://github.com/Alexhuszagh/rust/tree/master/src/etc/test-float-parse), and any errors in `runtests.py` have been patched.
**Undefined Behavior**
An implementation of `check_len` which relied on undefined behavior (in fast-float-rust) has been refactored, to ensure that the behavior is well-defined. The original code is as follows:
```rust
#[inline]
pub fn check_len(&self, n: usize) -> bool {
unsafe { self.ptr.add(n) <= self.end }
}
```
And the new implementation is as follows:
```rust
/// Check if the slice at least `n` length.
fn check_len(&self, n: usize) -> bool {
n <= self.as_ref().len()
}
```
Note that this has since been fixed in [fast-float-rust](https://github.com/aldanor/fast-float-rust/pull/29).
**Inferring Binary Exponents**
Rather than explicitly store binary exponents, this new implementation infers them from the decimal exponent, reducing the amount of static storage required. This removes the requirement to store [611 i16s](868c702d0c/library/core/src/num/dec2flt/table.rs (L8)).
# Code Size
The code size, for all optimizations, does not considerably change relative to before for stripped builds, however it is **significantly** smaller prior to stripping the resulting binaries. These binary sizes were calculated on x86_64-unknown-linux-gnu.
**new**
Using rustc version 1.55.0-dev.
opt-level|size|size(stripped)
|:-:|:-:|:-:|
0|400k|300K
1|396k|292K
2|392k|292K
3|392k|296K
s|396k|292K
z|396k|292K
**old**
Using rustc version 1.53.0-nightly.
opt-level|size|size(stripped)
|:-:|:-:|:-:|
0|3.2M|304K
1|3.2M|292K
2|3.1M|284K
3|3.1M|284K
s|3.1M|284K
z|3.1M|284K
# Correctness
The dec2flt implementation passes all of Rust's unittests and comprehensive float parsing tests, along with numerous other tests such as Nigel Toa's comprehensive float [tests](https://github.com/nigeltao/parse-number-fxx-test-data) and Hrvoje Abraham [strtod_tests](https://github.com/ahrvoje/numerics/blob/master/strtod/strtod_tests.toml). Therefore, it is unlikely that this algorithm will incorrectly round parsed floats.
# Issues Addressed
This will fix and close the following issues:
- resolves#85198
- resolves#85214
- resolves#85234
- fixes#31407
- fixes#31109
- fixes#53015
- resolves#68396
- closes https://github.com/aldanor/fast-float-rust/issues/15
Implementation is based off fast-float-rust, with a few notable changes.
- Some unsafe methods have been removed.
- Safe methods with inherently unsafe functionality have been removed.
- All unsafe functionality is documented and provably safe.
- Extensive documentation has been added for simpler maintenance.
- Inline annotations on internal routines has been removed.
- Fixed Python errors in src/etc/test-float-parse/runtests.py.
- Updated test-float-parse to be a library, to avoid missing rand dependency.
- Added regression tests for #31109 and #31407 in core tests.
- Added regression tests for #31109 and #31407 in ui tests.
- Use the existing slice primitive to simplify shared dec2flt methods
- Remove Miri ignores from dec2flt, due to faster parsing times.
- resolves#85198
- resolves#85214
- resolves#85234
- fixes#31407
- fixes#31109
- fixes#53015
- resolves#68396
- closes https://github.com/aldanor/fast-float-rust/issues/15
Due to #20400 the corresponding TrustedLen impls need a helper trait
instead of directly adding `Item = &[T;N]` bounds.
Since TrustedLen is a public trait this in turn means
the helper trait needs to be public. Since it's just a workaround
for a compiler deficit it's marked hidden, unstable and unsafe.
Correct invariant documentation for `steps_between`
Given that the previous example involves stepping forward from A to B, the equivalent example on this line would make most sense as stepping backward from B to A.
I should probably add a caveat here that I’m fairly new to Rust, and this is my first contribution to this repo, so it’s very possible that I’ve misunderstood how this is supposed to work (either on a technical level or a social one). If this is the case, please do let me know.
This only works if arrays are passed directly instead of array iterators
because we need to be sure that they have not been advanced before
Flatten does its size calculation.
This is a follow-up change to the fix for #75598. It simplifies the implementation of wrapping_neg() for all integer types by just calling 0.wrapping_sub(self) and always inlines it. This leads to much less assembly code being emitted for opt-level≤1.
Make the specialized Fuse still deal with None
Fixes#85863 by removing the assumption that we'll never see a cleared iterator in the `I: FusedIterator` specialization. Now all `Fuse` methods check for the possibility that `self.iter` is `None`, and the specialization only avoids _setting_ that to `None` in `&mut self` methods.
Given that the previous example involves stepping forward from A to B,
the equivalent example on this line would make most sense as stepping
backward from B to A.
expand: Support helper attributes for built-in derive macros
This is needed for https://github.com/rust-lang/rust/pull/86735 (derive macro `Default` should have a helper attribute `default`).
With this PR we can specify helper attributes for built-in derives using syntax `#[rustc_builtin_macro(MacroName, attributes(attr1, attr2, ...))]` which mirrors equivalent syntax for proc macros `#[proc_macro_derive(MacroName, attributes(attr1, attr2, ...))]`.
Otherwise expansion infra was already ready for this.
The attribute parsing code is shared between proc macro derives and built-in macros (`fn parse_macro_name_and_helper_attrs`).
create method overview docs for core::option and core::result
The `Option` and `Result` types have large lists of methods. They each could use an overview page of methods grouped by category. These proposed overviews include "truth tables" for the underappreciated boolean operators/combinators of these types. The methods are already somewhat categorized in the source, but some logical groupings are broken up by the necessities of putting related methods in different `impl` blocks, for example.
This is based on #86209, but those are small changes and unlikely to conflict.
Split MaybeUninit::write into new feature gate and stabilize it
This splits off the `MaybeUninit::write` function from the `maybe_uninit_extra` feature gate into a new `maybe_uninit_write` feature gate and stabilizes it.
Earlier work to improve the documentation of the write function: #86220
Tracking issue: #63567
This will not affect ABI since the other variant of the enum is bigger.
It may break some code, but that would be very strange: usually people
don't continue after the first `Done` (or `None` for a normal iterator).
- Add `:Sized` assertion in interpreter impl
- Use `Scalar::from_bool` instead of `ScalarInt: From<bool>`
- Remove unneeded comparison in intrinsic typeck
- Make this UB to call with undef, not just return undef in that case
special case for integer log10
Now that #80918 has been merged, this PR provides a faster version of `log10`.
The PR also adds some tests for values close to all powers of 10.
