syntax: Continue refactoring literals
A follow up to https://github.com/rust-lang/rust/pull/60679.
a2fd002bd5: Similarly to `EscapeError`, literal parsing now produces a `LitError`.
This way we can get rid of `diag: Option<(Span, &Handler)>` in interfaces while leaving attr/mod alone.
d9516d1120: Gathers all components of a literal token in a single struct.
librustc_errors: Remove unused annotation style `OldSchoolNoteText`
I could not find any references to it and the `snippet` module does not
seem to be exported publicly, so I think it can be safely removed.
This was originally removed in 17bd76a51 and I'm not sure why it is still there.
Revert "Add implementations of last in terms of next_back on a bunch of DoubleEndedIterators."
This changed observable behavior for several iterator types.
r? @alexcrichton
Solaris CI: Build with dilos2 stable
dilos2-testing has problems since the last repository update, so get the packages from dilos2 stable.
Fixes#61022.
Disable LLVM/debug assertions in gnu-full-bootstrap
This builder already is very close to the 2 hour mark and the debug
assertions aren't really buying us anything over what all the other
containers are enabling, so let's disable them for this slow builder.
Make -Zemit-artifact-notifications also emit the artifact type
This is easier for tooling to handle than trying to reverse-engineer the type from the filename extension. The field name and value is intended to reflect the `--emit` command-line option.
Related issues https://github.com/rust-lang/rust/issues/60988https://github.com/rust-lang/rust/issues/58465
cc @alexcrichton
Bump compiler-builtins to 0.1.15
This commit bumps the `compiler-builtins` dependency to 0.1.15 which
expects to have the source for `compiler-rt` provided externally if the
`c` feature is enabled. This then plumbs through the necessary support
in the build system to ensure that if the `llvm-project` directory is
checked out and present that we enable the `c` feature of
`compiler-builtins` and compile in all the C intrinsics.
Add match arm scopes and other scope fixes
* Add drop and lint scopes for match arms.
* Lint attributes are now respected on match arms.
* Make sure we emit a StorageDead if we diverge when initializing a temporary.
* Adjust MIR pretty printing of scopes for locals.
* Don't generate duplicate lint scopes for `let statements`.
* Add some previously missing fake borrows for matches.
closes#46525
cc @rust-lang/compiler
Simplify use of keyword symbols
They mirror non-keyword symbols now (see https://github.com/rust-lang/rust/pull/60630).
`keywords::MyKeyword.name()` -> `kw::MyKeyword`
`keywords::MyKeyword.ident()` -> `Ident::with_empty_ctxt(kw::MyKeyword)` (not common)
`keywords::Invalid.ident()` -> `Ident::invalid()` (more common)
Keywords are simply `Symbol` constants now, the `Keyword` struct is eliminated.
This means `kw::MyKeyword` can now be used in `match` in particular.
Update clippy submodule
r? @Manishearth
If anyone is wondering where the odd old commits are coming from, we merged all beta backport commits and so into master in order to make sure we don't need to keep those branches around.
rustc_metadata: parametrize schema::CrateRoot by 'tcx and rip out old unused incremental infra.
These are the first two commits of #59953, already reviewed and approved by @michaelwoerister.
r? @michaelwoerister
Simplify RefCell minimum_spanning_tree example
This simplifies the implementation of the `minimum_spanning_tree` example of `RefCell` in the `cell` module-level docs, avoiding an unnecessary recursive call. This also eliminates the need for a block to contain the scope of the borrow in this example. But since that use of a block served an important didactic purpose, we make up for this by instead introducing a block in the initial, simpler example of `RefCell`, where the point will hopefully be conveyed to the reader more easily.
Always try to project predicates when finding auto traits in rustdoc
Fixes#60726
Previous, AutoTraitFinder would only try to project predicates when the
predicate type contained an inference variable. When finding auto
traits, we only project to try to unify inference variables - we don't
otherwise learn any new information about the required bounds.
However, this lead to failing to properly generate a negative auto trait
impl (indicating that a type never implements a certain auto trait) in
the following unusual scenario:
In almost all cases, a type has an (implicit) negative impl of an auto
trait due some other type having an explicit *negative* impl of that
auto trait. For example:
struct MyType<T> {
field: *const T
}
has an implicit 'impl<T> !Send for MyType<T>', due to the explicit
negative impl (in libcore) 'impl<T: ?Sized> !Send for *const T'.
However, as exposed by the 'abi_stable' crate, this isn't always the
case. This minimzed example shows how a type can never implement
'Send', due to a projection error:
```
pub struct True;
pub struct False;
pub trait MyTrait {
type Project;
}
pub struct MyStruct<T> {
field: T
}
impl MyTrait for u8 {
type Project = False;
}
unsafe impl<T> Send for MyStruct<T>
where T: MyTrait<Project=True> {}
pub struct Wrapper {
inner: MyStruct<u8>
}
```
In this example, `<u8 as MyTrait>::Project == True'
must hold for 'MyStruct<u8>: Send' to hold.
However, '<u8 as MyTrait>::Project == False' holds instead
To properly account for this unusual case, we need to call
'poly_project_and_unify' on *all* predicates, not just those with
inference variables. This ensures that we catch the projection error
that occurs above, and don't incorrectly determine that 'Wrapper: Send'
holds.
Allow null-pointer-optimized enums in FFI if their underlying representation is FFI safe
I'm not sure if this requires an RFC. I attempted to start [a discussion on internals.rust-lang.org](https://internals.rust-lang.org/t/options-ffi-safety-and-guarantees-for-abi-compatibility-with-nonnull-optimizations/9784) and when no one really objected I figured I'd go ahead and try implementing this.
This allows types like `Option<NonZeroU8>` to be used in FFI without triggering the `improper_ctypes` lint. This works by changing the `is_repr_nullable_ptr` function to consider an enum `E` to be FFI-safe if:
- `E` has no explicit `#[repr(...)]`.
- It only has two variants.
- One of those variants is empty (meaning it has no fields).
- The other variant has only one field.
- That field is one of the following:
- `&T`
- `&mut T`
- `extern "C" fn`
- `core::num::NonZero*`
- `core::ptr::NonNull<T>`
- `#[repr(transparent)] struct` wrapper around one of the types in this list.
- The size of `E` and its field are both known and are both the same size (implying `E` is participating in the nonnull optimization).
This logic seems consistent with [the Rust nomicon](https://doc.rust-lang.org/nomicon/repr-rust.html).
This builder already is very close to the 2 hour mark and the debug
assertions aren't really buying us anything over what all the other
containers are enabling, so let's disable them for this slow builder.
This commit bumps the `compiler-builtins` dependency to 0.1.15 which
expects to have the source for `compiler-rt` provided externally if the
`c` feature is enabled. This then plumbs through the necessary support
in the build system to ensure that if the `llvm-project` directory is
checked out and present that we enable the `c` feature of
`compiler-builtins` and compile in all the C intrinsics.
This allows types like Option<NonZeroU8> to be used in FFI without triggering the improper_ctypes lint. This works by changing the is_repr_nullable_ptr function to consider an enum E to be FFI-safe if:
- E has no explicit #[repr(...)].
- It only has two variants.
- One of those variants is empty (meaning it has no fields).
- The other variant has only one field.
- That field is one of the following:
- &T
- &mut T
- extern "C" fn
- core::num::NonZero*
- core::ptr::NonNull<T>
- #[repr(transparent)] struct wrapper around one of the types in this list.
- The size of E and its field are both known and are both the same size (implying E is participating in the nonnull optimization).
* Clean up the `install-windows-build-deps.yml` file and add some more
comments where appropriate.
* Add some comments to `run.yml`
* Don't fast path the `rustfmt` submodule, but we'll take care of that
later if necessary.
Ban multi-trait objects via trait aliases
Obviously, multi-trait objects are not normally supported, so they should not be supported via trait aliases.
This has been factored out from the previous PR https://github.com/rust-lang/rust/pull/55994 (see point 1).
r? @Centril
CC @nikomatsakis
------------------
### RELNOTES:
We now allow `dyn Send + fmt::Debug` with equivalent semantics to `dyn fmt::Debug + Send`.
That is, the order of the mentioned traits does not matter wrt. principal/not-principal traits.
This is a small change that might deserve a mention in the blog post because it is a language change but most likely not.
See ce2ee305f9/src/test/ui/traits/wf-trait-object-reverse-order.rs.
// @Centril
Preserve local scopes in generator MIR
Part of #52924, depended upon by the generator layout optimization #60187.
This PR adds `StorageDead` statements in more places in generators, so we can see when non-`Drop` locals have gone out of scope and recover their storage.
The reason this is only done for generators is compiler performance. See https://github.com/rust-lang/rust/pull/60187#issuecomment-485637811 for what happens when we do this for all functions.
For `Drop` locals, we modify the `MaybeStorageLive` analysis to use `drop` to indicate that storage is no longer live for the local. Once `drop` returns or unwinds to our function, we implicitly assume that the local is `StorageDead`.
Instead of using `drop`, it is possible to emit more `StorageDead` statements in the MIR for `Drop` locals so we can handle all locals the same. I am fine with doing it that way, but this was the simplest approach for my purposes. It is also likely to be more performant.
r? @Zoxc (feel free to reassign)
cc @cramertj @eddyb @RalfJung @rust-lang/wg-async-await
