Deny associated type bindings within associated type bindings
Fixes#102335
This was made worse by #100865, which unified the way we generate substs for GATs and non-generic associated types. However, the issue was not _caused_ by #100865, evidenced by the test I added for GATs:
```rust
trait T {
type A: S<C<(), i32 = ()> = ()>;
//~^ ERROR associated type bindings are not allowed here
}
trait Q {}
trait S {
type C<T>: Q;
}
fn main() {}
```
^ which passes on beta (where GATs are stable) and presumably ever since GATs support was added to `create_substs_for_associated_item` in astconv.
Suggest unwrapping `???<T>` if a method cannot be found on it but is present on `T`.
This suggests various ways to get inside wrapper types if the method cannot be found on the wrapper type, but is present on the wrappee.
For this PR, those wrapper types include `Localkey`, `MaybeUninit`, `RefCell`, `RwLock` and `Mutex`.
Stabilize bench_black_box
This PR stabilize `feature(bench_black_box)`.
```rust
pub fn black_box<T>(dummy: T) -> T;
```
The FCP was completed in https://github.com/rust-lang/rust/issues/64102.
`@rustbot` label +T-libs-api -T-libs
- Rename `unescape_raw_str_or_raw_byte_str` as
`unescape_raw_str_or_byte_str`, which is more accurate.
- Remove the unused `Mode::in_single_quotes` method.
- Make some assertions more precise, and add a missing one to
`unescape_char_or_byte`.
- Change all the assertions to `debug_assert!`, because this code is
reasonably hot, and the assertions aren't required for memory safety,
and any violations are likely to be sufficiently obvious that normal
tests will trigger them.
rustdoc: remove redundant `#help-button` CSS
When the separate top and bottom styles were added in cd3f4da244, some of the CSS rules were needlessly duplicated.
The `text-align: initial` rule on `.side-by-side` was always redundant, since the rules that centered the text were set on children, not parents.
session: remove now-unnecessary lint `#[allow]`s
In #101230, the internal diagnostic migration lints - `diagnostic_outside_of_impl` and `untranslatable_diagnostic` - were modified so that they wouldn't trigger on functions annotated with `#[rustc_lint_diagnostics]`. However, this change has to make it into the bootstrap compiler before the `#[allow]` annotations that it aims to remove can be removed, which is possible now that #102051 has landed.
Avoid LLVM-deprecated `Optional::hasValue`
LLVM 15 added `Optional::has_value`, and LLVM `main` (16) has deprecated
`hasValue`. However, its `explicit operator bool` does the same thing,
and was added long ago, so we can use that across our full LLVM range of
compatibility.
rustdoc: remove no-op CSS `.srclink { font-weight; font-size }`
When this CSS was added in 34bd2b845b, source links were nested below headers.
34bd2b845b/src/librustdoc/html/render.rs (L4015-L4019)
Now, thanks to 458e7219bc, they are now siblings of headers, and thanks to 270d09dca9, they have the same font size that they would've had anyway.
rustdoc: use CSS containment to speed up render
https://developer.mozilla.org/en-US/docs/Web/CSS/CSS_Containment
This affected layout a little and required adjustments to the CSS to keep spacing the same. In particular, the margins of adjacent items usually overlap with each other. However, when an item has contain: layout, any margins of child nodes push out the size of the item itself. This was making spacing between items a little too big. To solve that, I removed margins in some places: in particular for certain classes that often occur at the end of a `details.rustdoc-toggle` block, I removed their bottom margin. Generally, the margins provided by the next item down are sufficient.
Also remove an unnecessary margin-top on .code-header.
In particular this helps with the problem that rustdoc in some situations can generate giant HTML pages, which can crash a Chrome tab on typical modern hardware, for instance: `https://docs.rs/iced-x86/1.16.0/iced_x86/code_asm/struct.CodeAssembler.html` (26MB, 409k DOM nodes). This doesn't, of course, universally solve the problem, but it pushes out the boundary of the largest page rustdoc can produce without crashing a browser tab.
Demos:
https://rustdoc.crud.net/jsha/css-contain/std/string/struct.String.html
(warning: giant page, _may_ crash a browser tab) https://rustdoc.crud.net/jsha/css-contain-icedx86/iced_x86/code_asm/struct.CodeAssembler.html
r? `@notriddle`
Stabilize `#![feature(mixed_integer_ops)]`
Tracked and FCP completed in #87840.
````@rustbot```` label +T-libs-api +S-waiting-on-review +relnotes
r? rust-lang/t-libs-api
When the separate top and bottom styles were added in
cd3f4da244, some of the CSS rules were
needlessly duplicated.
The `text-align: initial` rule on `.side-by-side` was always redundant, since
the rules that centered the text were set on children, not parents.
https://developer.mozilla.org/en-US/docs/Web/CSS/CSS_Containment
This affected layout a little and required adjustments to the CSS to
keep spacing the same. In particular, the margins of adjacent items
usually overlap with each other. However, when an item has contain:
layout, any margins of child nodes push out the size of the item itself.
This was making spacing between items a little too big. To solve that, I
removed margins in some places: in particular for certain classes that
often occur at the end of a `details.rustdoc-toggle` block, I removed
their bottom margin. Generally, the margins provided by the next item
down are sufficient.
Also remove an unnecessary margin-top on .code-header.
In #101230, the internal diagnostic migration lints -
`diagnostic_outside_of_impl` and `untranslatable_diagnostic` - were
modified so that they wouldn't trigger on functions annotated with
`#[rustc_lint_diagnostics]`. However, this change has to make it into
the bootstrap compiler before the `#[allow]` annotations that it aims to
remove can be removed, which is possible now that #102051 has landed.
Signed-off-by: David Wood <david.wood@huawei.com>
macros: diagnostic derive on enums
Part of #100717.
Extends `#[derive(Diagnostic)]` to work on enums too where each variant acts like a distinct diagnostic - being able to represent diagnostics this way can be quite a bit simpler for some parts of the compiler.
r? `@compiler-errors`
cc `@Xiretza`
LLVM 15 added `Optional::has_value`, and LLVM `main` (16) has deprecated
`hasValue`. However, its `explicit operator bool` does the same thing,
and was added long ago, so we can use that across our full LLVM range of
compatibility.
