Begin experimental support for pin reborrowing
This commit adds basic support for reborrowing `Pin` types in argument position. At the moment it only supports reborrowing `Pin<&mut T>` as `Pin<&mut T>` by inserting a call to `Pin::as_mut()`, and only in argument position (not as the receiver in a method call).
This PR makes the following example compile:
```rust
#![feature(pin_ergonomics)]
fn foo(_: Pin<&mut Foo>) {
}
fn bar(mut x: Pin<&mut Foo>) {
foo(x);
foo(x);
}
```
Previously, you would have had to write `bar` as:
```rust
fn bar(mut x: Pin<&mut Foo>) {
foo(x.as_mut());
foo(x);
}
```
Tracking:
- #130494
r? `@compiler-errors`
Remove macOS 10.10 dynamic linker bug workaround
Rust's current minimum macOS version is 10.12, so the hack can be removed. This PR also updates the `remove_dir_all` docs to reflect that all supported macOS versions are protected against TOCTOU race conditions (the fallback implementation was already removed in #127683).
try-job: dist-x86_64-apple
try-job: dist-aarch64-apple
try-job: dist-apple-various
try-job: aarch64-apple
try-job: x86_64-apple-1
The issue-112505-overflow test just extended a case of transmute-fail.rs
so simply put them in the same file.
Then we normalize away other cases of this.
Rollup of 6 pull requests
Successful merges:
- #129542 (Add regression test for #129541)
- #129755 (test: cross-edition metavar fragment specifiers)
- #130566 (Break up compiletest `runtest.rs` into smaller helper modules)
- #130585 (Add tidy check for rustdoc templates to ensure the whitespace characters are all stripped)
- #130605 (Fix feature name in test)
- #130607 ([Clippy] Remove final std paths for diagnostic item)
r? `@ghost`
`@rustbot` modify labels: rollup
[Clippy] Remove final std paths for diagnostic item
Removes the paths to SeekFrom::Start/Current that were left in #130553.
This was split off as it involves introducing a utility to check for enum ctors, as both:
- enum variants cannot be diagnostic items
- even if they could, that wouldn't help because we need to get the enum variant ctor
While adding the `is_enum_variant_ctor`, I removed both `is_diagnostic_ctor` and `is_res_diagnostic_ctor` as they are unused and never worked due to the above bullet points.
Fix feature name in test
This is meant to test that the `box_patterns` feature isn't active due to the `cfg(FALSE)`, but uses the removed `box_syntax` feature. Fix this so it's testing what it should be.
Add tidy check for rustdoc templates to ensure the whitespace characters are all stripped
Fixes https://github.com/rust-lang/rust/issues/130559.
I'm planning to send a follow-up in case a tag at the end of a line isn't needed (if the next line starts with a jinja tag for example).
r? `@notriddle`
Break up compiletest `runtest.rs` into smaller helper modules
Previously compiletest's `runtest.rs` was a massive 4700 lines file that made reading and navigation very awkward. This PR breaks the `runtest.rs` file up into smaller helper modules, one for each test suite/mode.
> [!NOTE]
> This PR should not contain functional changes, it is intended to be mostly code motion to breakup `runtest.rs` into smaller helper modules to make it easier to digest.
>
> This PR intentionally does not neatly reorganize where all the methods on `TestCx` goes, that is intended for a follow-up PR. Some methods on `TestCx` do not need to be on `TestCx`. It also does not address a weirdness in valgrind, that is intended for a follow-up PR as well.
Part of a series of compiletest cleanups #130565.
Fixes#89475.
r? `@ghost` (I need to do a self-review pass first)
Get rid of niche selection's dependence on fields's order
Fixes#125630.
Use the optimal niche selection decided in `univariant()` rather than picking niche field manually.
r? `@the8472`
Rollup of 3 pull requests
Successful merges:
- #130485 (Do not expect infer/bound/placeholder/error in v0 symbol mangling)
- #130567 (Register tool docs for compiletest)
- #130582 (rustdoc: use the correct span for doctests)
r? `@ghost`
`@rustbot` modify labels: rollup
Register tool docs for compiletest
This PR registers tool docs for `src/tools/compiletest`, meaning that
```
$ ./x doc src/tools/compiletest
```
or
```
$ ./x doc compiletest
```
will now generate docs, like for `run-make-support`.
Fixes#130564.
Do not expect infer/bound/placeholder/error in v0 symbol mangling
Infer/bound/placeholder/error are not encounterable during codegen. Let's make sure v0 symbol mangling doesn't "accidentally" handle them.
As for aliases (namely: projections and uv consts) these may still be encounterable because of the way that we render the def paths of items. Specifically, when we have something like:
```
struct W<T>(T);
impl<T> W<T> {
fn x() {
fn y() {}
}
}
```
The path of `y` is rendered like `crate_name::W<T>:❌:y`. Specifically, since `y` doesn't inherit the generics of the impl, we use the *identity* substitutions for that impl. If the impl has any aliases, they will remain unnormalized if they're rigid.
r? `@BoxyUwU`
[perf] skip normalizing param env if it is already normalized
If the param env is already normalized after elaboration, then we can skip a bunch of expensive operations.
> [!note]
> This makes it so that outlives predicates are no longer sorted after non-outlives predicates. Surely this won't make a semantic difference.
r? ghost
The comment here about 48 bit addresses being enough was written in 2016
but was made incorrect in 2019 by 5-level paging, and then persisted for
another 5 years before being noticed and corrected.
Generating a call to `as_mut()` let to more restrictive borrows than
what reborrowing usually gives us. Instead, we change the desugaring to
reborrow the pin internals directly which makes things more expressive.
Rollup of 5 pull requests
Successful merges:
- #128001 (Improve documentation for <integer>::from_str_radix)
- #130553 ([Clippy] Get rid of most `std` `match_def_path` usage, swap to diagnostic items.)
- #130554 (`pal::unsupported::process::ExitCode`: use an `u8` instead of a `bool`)
- #130556 (Mark the `link_cfg` feature as internal)
- #130558 (Support 128-bit atomics on s390x)
r? `@ghost`
`@rustbot` modify labels: rollup
Mark the `link_cfg` feature as internal
This PR marks the `link_cfg` feature as internal because it's a perme-unstable feature, only used by `core`/`std`and `unwind`.
`pal::unsupported::process::ExitCode`: use an `u8` instead of a `bool`
`ExitCode` should “represents the status code the current process can return to its parent under normal termination”, but is currently represented as a `bool` on unsupported platforms, making the `impl From<u8> for ExitCode` lossy.
Fixes#130532.
History: [IRLO thread](https://internals.rust-lang.org/t/mini-pre-rfc-redesigning-process-exitstatus/5426) (`ExitCode` as a `main` return), #48618 (initial impl), #93445 (`From<u8>` impl).
