Remove space after negative sign in Literal to_string
Negative proc macro literal tokens used to be printed with a space between the minus sign and the magnitude. That's because `impl ToString for Literal` used to convert the Literal into a TokenStream, which splits the minus sign into a separate Punct token.
```rust
Literal::isize_unsuffixed(-10).to_string() // "- 10"
```
This PR updates the ToString impl to directly use `rustc_ast::token::Lit`'s ToString, which matches the way Rust negative numbers are idiomatically written without a space.
```rust
Literal::isize_unsuffixed(-10).to_string() // "-10"
```
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()?;
```
Places are usually shallow and quick to visit. By contrast, computing
`is_freeze` can be much costlier, involving inference and trait
solving. Making sure to call `is_freeze` only when necessary should be
beneficial for performance in most cases.
Remove invalid suggestion involving `Fn` trait bound
This pull request closes#85735. The actual issue is a duplicate of #21974, but #85735 contains a further problem, which is an invalid suggestion if `Fn`/`FnMut`/`FnOnce` trait bounds are involved: The suggestion code checks whether the trait bound ends with `>` to determine whether it has any generic arguments, but the `Fn*` traits have a special syntax for generic arguments that doesn't involve angle brackets. The example given in #85735:
```rust
trait Foo {}
impl<'a, 'b, T> Foo for T
where
T: FnMut(&'a ()),
T: FnMut(&'b ()), {
}
```
currently produces:
```
error[E0283]: type annotations needed
--> src/lib.rs:4:8
|
4 | T: FnMut(&'a ()),
| ^^^^^^^^^^^^^ cannot infer type for type parameter `T`
|
= note: cannot satisfy `T: FnMut<(&'a (),)>`
help: consider specifying the type arguments in the function call
|
4 | T: FnMut(&'a ())::<Self, Args>,
| ^^^^^^^^^^^^^^
error: aborting due to previous error
```
which is incorrect, because there is no function call, and applying the suggestion would lead to a parse error. With my changes, I get:
```
error[E0283]: type annotations needed
--> test.rs:4:8
|
4 | T: FnMut(&'a ()),
| ^^^^^^^^^^^^^ cannot infer type for type parameter `T`
|
::: [...]/library/core/src/ops/function.rs:147:1
|
147 | pub trait FnMut<Args>: FnOnce<Args> {
| ----------------------------------- required by this bound in `FnMut`
|
= note: cannot satisfy `T: FnMut<(&'a (),)>`
error: aborting due to previous error
```
i.e. I have added a check to prevent the invalid suggestion from being issued for `Fn*` bounds, while the underlying issue #21974 remains for now.
Trait upcasting coercion (part2)
This is the second part of trait upcasting coercion implementation.
Currently this is blocked on #86264 .
The third part might be implemented using unsafety checking
r? `@bjorn3`
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.
This way, we can show the plus and minus buttons on screens, while voice
control will read off actual words "Collapse" and "Expand" instead of reading
"open brace minus close brace" and "open brace plus close brace".
Part of #87059
Rollup of 8 pull requests
Successful merges:
- #87645 (Properly find owner of closure in THIR unsafeck)
- #87646 (Fix a parser ICE on invalid `fn` body)
- #87652 (Validate that naked functions are never inlined)
- #87685 (Write docs for SyncOnceCell From and Default impl)
- #87693 (Add `aarch64-apple-ios-sim` as a possible target to the manifest)
- #87708 (Add convenience method for handling ipv4-mapped addresses by canonicalizing them)
- #87711 (Correct typo)
- #87716 (Allow generic SIMD array element type)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Add convenience method for handling ipv4-mapped addresses by canonicalizing them
This simplifies checking common properties in an address-family-agnostic
way since #86335 commits to not checking IPv4 semantics
of IPv4-mapped addresses in the `Ipv6Addr` property methods.
Add `aarch64-apple-ios-sim` as a possible target to the manifest
This should allow rustup and similar to actually make use of this new
target now.
r? ```@Mark-Simulacrum```
Followup to #85782
Fix a parser ICE on invalid `fn` body
Fixes#87635
A better fix would add a check for `fn` body on `expected_one_of_not_found` but I haven't come up with a graceful way. Any idea?
r? ```@oli-obk``` ```@estebank```
Properly find owner of closure in THIR unsafeck
Previously, when encountering a closure in a constant, the THIR unsafeck gets invoked on the owner of the constant instead of the constant itself, producing cycles.
Supersedes #87492. ```@FabianWolff``` thanks for your work on that PR, I copied your test file and added you as a co-author.
Fixes#87414.
r? ```@oli-obk```
Support negative numbers in Literal::from_str
proc_macro::Literal has allowed negative numbers in a single literal token ever since Rust 1.29, using https://doc.rust-lang.org/stable/proc_macro/struct.Literal.html#method.isize_unsuffixed and similar constructors.
```rust
let lit = proc_macro::Literal::isize_unsuffixed(-10);
```
However, the suite of constructors on Literal is not sufficient for all use cases, for example arbitrary precision floats, or custom suffixes in FFI macros.
```rust
let lit = proc_macro::Literal::f64_unsuffixed(0.101001000100001000001000000100000001); // :(
let lit = proc_macro::Literal::i???_suffixed(10ulong); // :(
```
For those, macros construct the literal using from_str instead, which preserves arbitrary precision, custom suffixes, base, and digit grouping.
```rust
let lit = "0.101001000100001000001000000100000001".parse::<Literal>().unwrap();
let lit = "10ulong".parse::<Literal>().unwrap();
let lit = "0b1000_0100_0010_0001".parse::<Literal>().unwrap();
```
However, until this PR it was not possible to construct a literal token that is **both** negative **and** preserving of arbitrary precision etc.
This PR fixes `Literal::from_str` to recognize negative integer and float literals.
Commit to not supporting IPv4-in-IPv6 addresses
Stabilization of the `ip` feature has for a long time been blocked on the question of whether Rust should support handling "IPv4-in-IPv6" addresses: should the various `Ipv6Address` property methods take IPv4-mapped or IPv4-compatible addresses into account. See also the IPv4-in-IPv6 Address Support issue #85609 and #69772 which originally asked the question.
# Overview
In the recent PR #85655 I proposed changing `is_loopback` to take IPv4-mapped addresses into account, so `::ffff:127.0.0.1` would be recognized as a looback address. However, due to the points that came up in that PR, I alternatively propose the following: Keeping the current behaviour and commit to not assigning any special meaning for IPv4-in-IPv6 addresses, other than what the standards prescribe. This would apply to the stable method `is_loopback`, but also to currently unstable methods like `is_global` and `is_documentation` and any future methods. This is implemented in this PR as a change in documentation, specifically the following section:
> Both types of addresses are not assigned any special meaning by this implementation, other than what the relevant standards prescribe. This means that an address like `::ffff:127.0.0.1`, while representing an IPv4 loopback address, is not itself an IPv6 loopback address; only `::1` is. To handle these so called "IPv4-in-IPv6" addresses, they have to first be converted to their canonical IPv4 address.
# Discussion
In the discussion for or against supporting IPv4-in-IPv6 addresses the question what would be least surprising for users of other languages has come up several times. At first it seemed most big other languages supported IPv4-in-IPv6 addresses (or at least considered `::ffff:127.0.0.1` a loopback address). However after further investigation it appears that supporting IPv4-in-IPv6 addresses comes down to how a language represents addresses. .Net and Go do not have a separate type for IPv4 or IPv6 addresses, and do consider `::ffff:127.0.0.1` a loopback address. Java and Python, which do have separate types, do not consider `::ffff:127.0.0.1` a loopback address. Seeing as Rust has the separate `Ipv6Addr` type, it would make sense to also not support IPv4-in-IPv6 addresses. Note that this focuses on IPv4-mapped addresses, no other language handles IPv4-compatible addresses.
Another issue that was raised is how useful supporting these IPv4-in-IPv6 addresses would be in practice. Again with the example of `::ffff:127.0.0.1`, considering it a loopback address isn't too useful as to use it with most of the socket APIs it has to be converted to an IPv4 address anyway. From that perspective it would be better to instead provide better ways for doing this conversion like stabilizing `to_ipv4_mapped` or introducing a `to_canonical` method.
A point in favour of not supporting IPv4-in-IPv6 addresses is that that is the behaviour Rust has always had, and that supporting it would require changing already stable functions like `is_loopback`. This also keeps the documentation of these functions simpler, as we only have to refer to the relevant definitions in the IPv6 specification.
# Decision
To make progress on the `ip` feature, a decision needs to be made on whether or not to support IPv4-in-IPv6 addresses.
There are several options:
- Keep the current implementation and commit to never supporting IPv4-in-IPv6 addresses (accept this PR).
- Support IPv4-in-IPv6 addresses in some/all `IPv6Addr` methods (accept PR #85655).
- Keep the current implementation and but not commit to anything yet (reject both this PR and PR #85655), this entire issue will however come up again in the stabilization of several methods under the `ip` feature.
There are more options, like supporting IPv4-in-IPv6 addresses in `IpAddr` methods instead, but to my knowledge those haven't been seriously argued for by anyone.
There is currently an FCP ongoing on PR #85655. I would ask the libs team for an alternative FCP on this PR as well, which if completed means the rejection of PR #85655, and the decision to commit to not supporting IPv4-in-IPv6 addresses.
If anyone feels there is not enough evidence yet to make the decision for or against supporting IPv4-in-IPv6 addresses, let me know and I'll do whatever I can to resolve it.
This simplifies checking common properties in an address-family-agnostic
way since since #86335 commits to not checking IPv4 semantics
of IPv4-mapped addresses in the `Ipv6Addr` property methods.
