rustdoc: don't ICE on `TyKind::Typeof`
Fixes#102986
I'm not sure why rustdoc started seeing `TyKind::Typeof` all of a sudden (the code being editted was last touched 3 months ago), probably something to do with error recovery? idk.
Add suggestion to the "missing native library" error
If we fail to locate a native library that we are linking with, it could be the case the user entered a complete file name like `foo.lib` or `libfoo.a` when we expect them to simply provide `foo`.
In this situation, we now detect that case and suggest the user only provide the library name itself.
The two items it was really intended to target were the buttons, and those
both need to have the style set directly on them anyway because the buttons
are both child elements of wrappers.
Rollup of 6 pull requests
Successful merges:
- #102765 (Suggest `==` to the first expr which has `ExprKind::Assign` kind)
- #102854 (openbsd: don't reallocate a guard page on the stack.)
- #102904 (Print return-position `impl Trait` in trait verbosely if `-Zverbose`)
- #102947 (Sort elaborated existential predicates in `object_ty_for_trait`)
- #102956 (Use `full_res` instead of `expect_full_res`)
- #102999 (Delay `is_intrinsic` query until after we've determined the callee is a function)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Sort elaborated existential predicates in `object_ty_for_trait`
r? `@cjgillot`
I think that #102845 caused #102933. Depending on the order that we elaborate these existential projection predicates, there's no guarantee that they'll be sorted by def id, which is what is failing the assertion in the issue.
Fixes#102933Fixes#102973
openbsd: don't reallocate a guard page on the stack.
the kernel currently enforce that a stack is immutable. calling mmap(2) or mprotect(2) to change it will result in EPERM, which generate a panic!().
so just do like for Linux, and trust the kernel to do the right thing.
If we fail to locate a native library that we are linking with, it could
be the case the user entered a complete file name like `foo.lib` or
`libfoo.a` when we expect them to simply provide `foo`.
In this situation, we now detect that case and suggest the user only
provide the library name itself.
Optimize TLS on Windows
This implements the suggestion in the current TLS code to embed the linked list of destructors in the `StaticKey` structure to save allocations. Additionally, locking is avoided when no destructor needs to be run. By using one Windows-provided `Once` per key instead of a global lock, locking is more finely-grained (this unblocks #100579).
rustdoc: merge separate `.item-info` CSS
Rough timeline:
* The longer `.content .item-info` selector originated in 110e7270ab7b0700ce714b8b1c7e509195dea2c4. No reason seems to be given in the PR why it needed the `.content` part, but it was probably added because of <110e7270ab/src/librustdoc/html/static/rustdoc.css (L476-L478)>. That selector with the margin-bottom was removed when CSS containment was added in 8846c0853d8687fda0e5f23f6687b03b243980ee.
* `.stability` was renamed `.item-info` in caf6c5790a858893c1d32ed2054c9577d12e7493.
* The selector without the `.content` was added in d48a39a5e24ab08f727d1c919dc2af98c333ad14.
tidy: error if a lang feature is already present
If a lang feature gets declared twice, like for example as a result of a mistake during stabilization, emit an error in tidy. Library features already have this logic.
Inspired by a mistake done during `half_open_range_patterns` stabilization: https://github.com/rust-lang/rust/pull/102275/files#r991292215
The PR requires #102883 to be merged before CI turns green because the check is doing its job.
For reviewers, I suggest [turning off whitespace changes](https://github.com/rust-lang/rust/pull/102971/files?w=1) in the diff by adding `?w=1` to the url, as a large part of the diff is just about removing one level of indentation.
Support casting boxes to dyn*
Boxes have a pointer type at codegen time which LLVM does not allow to be transparently converted to an integer. Work around this by inserting a `ptrtoint` instruction if the argument is a pointer.
r? ``@compiler-errors``
Fixes#102427
