Resolve lifetimes independently for each item-like.
Now that the heavy-lifting is done on the AST and during lowering, we do not need to perform HIR lifetime resolution on a full item at once. Instead, we can treat each item-like independently, and look at `generics_of` the parent exceptionally for associated items.
Remove lock wrappers in `sys_common`
This moves the lazy allocation to `sys` (SGX and UNIX). While this leads to a bit more verbosity, it will simplify future improvements by making room in `sys_common` for platform-independent implementations.
This also removes the condvar check on SGX as it is not necessary for soundness and will be removed anyway once mutex has been made movable.
For simplicity's sake, `libunwind` also uses lazy allocation now on SGX. This will require an update to the C definitions before merging this (CC `@raoulstrackx).`
r? `@m-ou-se`
Retry failed macro matching for diagnostics
When a declarative macro fails to match, retry the matching to collect diagnostic info instead of collecting it on the fly in the hot path. Split out of #103439.
You made a bunch of changes to declarative macro matching, so
r? `@nnethercote`
This change should produce a few small perf wins: https://github.com/rust-lang/rust/pull/103439#issuecomment-1294249602
Recover wrong-cased keywords that start items
(_this pr was inspired by [this tweet](https://twitter.com/Azumanga/status/1552982326409367561)_)
r? `@estebank`
We've talked a bit about this recovery, but I just wanted to make sure that this is the right approach :)
For now I've only added the case insensitive recovery to `use`s, since most other items like `impl` blocks, modules, functions can start with multiple keywords which complicates the matter.
Use 64 bits for incremental cache in-file positions
We currently use a 32-bit integer to encode byte positions into the incremental cache.
This is not enough when the query chache file is >4GB.
As the overflow check was a `debug_assert`, it was removed in released compilers, making compilation succeed silently.
At the next compilation, cache decoding would try to read unrelated data because of garbled file position, triggering an ICE.
Fixes https://github.com/rust-lang/rust/issues/79786
(I'm closing that bug since it the original report and the subsequent questions are probably different instances. A new bug should be opened for new instances of that ICE.)
Rollup of 9 pull requests
Successful merges:
- #102763 (Some diagnostic-related nits)
- #103443 (Parser: Recover from using colon as path separator in imports)
- #103675 (remove redundent "<>" for ty::Slice with reference type)
- #104046 (bootstrap: add support for running Miri on a file)
- #104115 (Migrate crate-search element to CSS variables)
- #104190 (Ignore "Change InferCtxtBuilder from enter to build" in git blame)
- #104201 (Add check in GUI test for file loading failure)
- #104211 (⬆️ rust-analyzer)
- #104231 (Update mailmap)
Failed merges:
- #104169 (Migrate `:target` rules to use CSS variables)
r? `@ghost`
`@rustbot` modify labels: rollup
Add check in GUI test for file loading failure
Since https://github.com/rust-lang/rust/pull/101702, some resources location need to be updated in case their content changed because then their hash will change too. This will prevent errors like https://github.com/rust-lang/rust/pull/104114 to happen again.
The second commit is to prevent CORS errors: when a file is linked from a file itself imported, the web browser considers they come from a different domain and therefore triggers the error. The option tells the web browser to ignore this case.
cc ```@jsha```
r? ```@notriddle```
Ignore "Change InferCtxtBuilder from enter to build" in git blame
Because it changed the indentation of many things, this commit caused a lot of diff with no functional changes, so we should ignore it.
r? ```@compiler-errors``` as you've complained about this before
The relevant commit: 283abbf0e7
bootstrap: add support for running Miri on a file
This enables:
```
./x.py run src/tools/miri --stage 0 --args src/tools/miri/tests/pass/hello.rs
```
That can be super helpful for debugging.
Also avoid sharing the Miri sysroot dir with a system-wide (rustup-managed) installation of Miri.
Fixes https://github.com/rust-lang/rust/issues/76666
Parser: Recover from using colon as path separator in imports
I don't know if this is the right approach, any feedback is welcome.
r? ```@compiler-errors```
Fixes#103269
Some diagnostic-related nits
1. Use `&mut Diagnostic` instead of `&mut DiagnosticBuilder<'_, T>`
2. Make `diag.span_suggestions` take an `IntoIterator` instead of `Iterator`, just to remove some `.into_iter` calls on the caller.
idk if I should add a lint to make sure people use `&mut Diagnostic` instead of `&mut DiagnosticBuilder<'_, T>` in cases where we're just, e.g., adding subdiagnostics to the diagnostic... maybe a followup.
Const Compare for Tuples
Makes the impls for Tuples of ~const `PartialEq` types also `PartialEq`, impls for Tuples of ~const `PartialOrd` types also `PartialOrd`, for Tuples of ~const `Ord` types also `Ord`.
behind the `#![feature(const_cmp)]` gate.
~~Do not merge before #104113 is merged because I want to use this feature to clean up the new test that I added there.~~
r? ``@fee1-dead``
Don't intra linkcheck reference
This removes the reference from the intra-doc link checks. This causes problems if any of the reference content needs to change, it causes the linkchecker to break. The reference has its own broken link check (https://github.com/rust-lang/reference/tree/master/style-check) which uses pulldown-cmark on the source to find actual broken links (instead of false-positives like this regex does).
I think the intra-doc link check could potentially be removed completely, since I think rustdoc is now checking for them well enough. However, it may serve as a decent regression check.
Refactor build-manifest to minimize the number of changes needed to add a new component
- Add all components to `PkgType`
- Automate functionality wherever possible, so functions often don't have to be manually edited
- Where that's not possible, use exhaustive matches on `PkgType` instead of adding individual strings.
- Add documentation for how to add a component. Improve the existing documentation for how to test changes.
I tested locally that this generates an identical manifest before and after my change, as follows:
```sh
git checkout d44e14225ab00e164aa9ea9e8d9e1bee40f96b3e
cargo +nightly run --manifest-path src/tools/build-manifest/Cargo.toml build/dist build/manifest-before 1970-01-01 http://example.com nightly
git checkout refactor-build-manifest
cargo +nightly run --manifest-path src/tools/build-manifest/Cargo.toml build/dist build/manifest-before 1970-01-01 http://example.com nightly
sort -u build/manifest-before/channel-rust-nightly.toml | diff - <(sort -u build/manifest-after/channel-rust-nightly.toml)
```
I then verified by hand that the differences before sorting are inconsequential (mostly targets being slightly reordered).
The only change in behavior is that `llvm-tools` is now properly renamed to `llvm-tools-preview`:
```
; sort -u build/manifest-before/channel-rust-nightly.toml | diff - <(sort -u build/manifest-after/channel-rust-nightly.toml)
784a785
> [renames.llvm-tools]
894a896
> to = "llvm-tools-preview"
```
This is based on https://github.com/rust-lang/rust/pull/102241 and should not be merged before.
Remove #![allow(rustc::potential_query_instability)] from rustc_infer
Related to #84447
This PR probably needs to be benchmarked to check for regressions.
internal: Migrate `ide_assists::utils` and `ide_assists::handlers` to use format arg captures (part 1)
This not only serves as making future migration to mutable syntax trees easier, it also finds out what needs to be migrated in the first place.
~~Aside from the first commit, subsequent commits are structured to only deal with one file/handler at a time.~~
This is the first of 3 PRs, migrating:
Utils:
- `gen_trait_fn_body`
- `render_snippet`
- `ReferenceConversion`
- `convert_type`
- `getter`
Handlers:
- `add_explicit_type`
- `add_return_type`
- `add_turbo_fish`
- `apply_demorgan`
- `auto_import`
- `convert_comment_block`
- `convert_integer_literal`
- `convert_into_to_from`
- `convert_iter_for_each_to_for`
- `convert_let_else_to_match`
- `convert_tuple_struct_to_named_struct`
- `convert_two_arm_bool_match_to_matches_macro`
- `destructure_tuple_binding`
- `extract_function`
- `extract_module`
- `extract_struct_from_enum_variant`
- `extract_type_alias`
- `extract_variable`
- `fix_visibility`
[debuginfo] Make cpp-like debuginfo type names for slices and str consistent.
Before this PR, the compiler would emit the debuginfo name `slice$<T>` for all kinds of slices, regardless of whether they are behind a reference or not and regardless of the kind of reference. As a consequence, the types `Foo<&[T]>`, `Foo<[T]>`, and `Foo<&mut [T]>` would end up with the same type name `Foo<slice$<T> >` in debuginfo, making it impossible to disambiguate between them by name. Similarly, `&str` would get the name `str` in debuginfo, so the debuginfo name for `Foo<str>` and `Foo<&str>` would be the same. In contrast, `*const [bool]` and `*mut [bool]` would be `ptr_const$<slice$<bool> >` and `ptr_mut$<slice$<bool> >`, i.e. the encoding does not lose information about the type.
This PR removes all special handling for slices and `str`. The types `&[bool]`, `&mut [bool]`, and `&str` thus get the names `ref$<slice2$<bool> >`, `ref_mut$<slice2$<bool> >`, and `ref$<str$>` respectively -- as one would expect.
The new special name for slices is `slice2$` to differentiate it from the previous name `slice$`, which has different semantics. The same is true for `str` and `str$`. This kind of versioning already has a precedent with the case of `enum$` and `enum2$` and hopefully will make it easier to transition existing consumers of these names.
cc `@rust-lang/wg-debugging` `@vadimcn`
r? `@wesleywiser`
UPDATE: Here is a table to clarify the changes
| Rust type | DWARF name | C++-like name (before) | C++-like name (after) |
|-----------|------------|------------------------|------------------------|
| `[T]` | `[T]` | `slice$<T>` | `slice2$<T>` |
| `&[T]` | `&[T]` | `slice$<T>` | `ref$<slice2$<T> >` |
| `&mut [T]` | `&mut [T]` | `slice$<T>` | `ref_mut$<slice2$<T> >`|
| `str` | `str` | `str` | `str$` |
| `&str` | `&str` | `str` | `ref$<str$>` |
| `&mut str` | `&mut str` | `str` | `ref_mut$<str$>`|
| `*const [T]` | `*const [T]` | `ptr_const$<slice$<T> >` | `ptr_const$<slice2$<T> >` |
| `*mut [T]` | `*mut [T]` | `ptr_mut$<slice$<T> >` | `ptr_mut$<slice2$<T> >` |
As you can see, before the PR many types would end up with the same name, making it impossible to distinguish between them in NatVis or other places where types are matched or looked up by name. The DWARF version of names is not changed.
Ensure that compile-flags arguments are the last in UI tests
Before this PR, compiletest would add `-L path/to/aux` at the end of the rustc flags, even after the custom ones set with the compile-flags header comment. This made it impossible to check how rustc would behave when a flag requiring an argument was passed without the argument, because the argument would become `-L`.
This PR fixes that by adding the `-L path/to/aux` before the arguments defined in compile-flags, at least for UI tests. Other test suites might either be fixed as well by this change, or still present the old behavior (`-L` is now always passed before, but other tests suites might add additional flags after the custom ones).
