All other arg properties use the `prop = value` style, which makes it
slightly annoying to use the `prop(value)` style for `global`. Change to
`prop = value` also for `global` for consistency.
thread_local: be excruciatingly explicit in dtor code
Use raw pointers to accomplish internal mutability, and clearly split references where applicable. This reduces the likelihood that any of these parts are misunderstood, either by humans or the compiler's optimizations.
Fixes#124317
r? ``@joboet``
ast: Generalize item kind visiting
And avoid duplicating logic for visiting `Item`s with different kinds (regular, associated, foreign).
The diff is better viewed with whitespace ignored.
Update cargo
9 commits in c9392675917adc2edab269eea27c222b5359c637..b60a1555155111e962018007a6d0ef85207db463
2024-04-23 19:35:19 +0000 to 2024-04-26 16:37:29 +0000
- fix(toml): Remove underscore field support in 2024 (rust-lang/cargo#13804)
- fix: emit 1.77 syntax error only when msrv is incompatible (rust-lang/cargo#13808)
- docs(ref): Index differences between virtual / real manifests (rust-lang/cargo#13794)
- refactor(toml): extract dependency-to-source-id to function (rust-lang/cargo#13802)
- Add where lint was set (rust-lang/cargo#13801)
- fix(toml): Don't double-warn when underscore is used in workspace dep (rust-lang/cargo#13800)
- fix(toml): Be more forceful with underscore/dash redundancy (rust-lang/cargo#13798)
- Fix warning suppression for config.toml vs config compat symlinks (rust-lang/cargo#13793)
- Cleanup linting system (rust-lang/cargo#13797)
r? ghost
PathBuf: replace transmuting by accessor functions
The existing `repr(transparent)` was anyway insufficient as `OsString` was not `repr(transparent)`. And furthermore, on Windows it was blatantly wrong as `OsString` wraps `Wtf8Buf` which is a `repr(Rust)` type with 2 fields:
51a7396ad3/library/std/src/sys_common/wtf8.rs (L131-L146)
So let's just be honest about what happens and add accessor methods that make this abstraction-breaking act of PathBuf visible on the APIs that it pierces through.
Fixes https://github.com/rust-lang/rust/issues/124409
Port run-make `--print=native-static-libs` to rmake.rs
This PR port the run-make `--print=native-static-libs` test to rmake.rs
The dedup was really awful in the `Makefile`, I'm glad to finally have a proper dedup detection for this.
Related to https://github.com/rust-lang/rust/issues/121876
r? `@jieyouxu`
resolve: Remove two cases of misleading macro call visiting
Macro calls are ephemeral, they should not add anything to the definition tree, even if their AST could contains something with identity.
Thankfully, macro call AST cannot contain anything like that, so these walks are just noops.
In majority of other places in def_collector / build_reduced_graph they are already not visited.
(Also, a minor match reformatting is included.)
bootstrap: keep all cargo test files in dist rustc-src
Cargo tests use some files that we would otherwise exclude, especially
the `cargo init` tests that are meant to deal with pre-existing `.git`
and `.hg` repos and their ignore files. Keeping these in our dist
tarball doesn't take much space, and allows distro builds to run these
tests successfully.
env: split up Windows and Unix environment variable handling
On Windows, manage them entirely outside the AM state; this also means we no longer report any data races for environment variable memory.
Fixes https://github.com/rust-lang/rust/issues/124411
`obligations_for_self_ty`: use `ProofTreeVisitor` for nested goals
As always, dealing with proof trees continues to be a hacked together mess. After this PR and #124380 the only remaining blocker for core is https://github.com/rust-lang/trait-system-refactor-initiative/issues/90. There is also a `ProofTreeVisitor` issue causing an ICE when compiling `alloc` which I will handle in a separate PR. This issue likely affects coherence diagnostics more generally.
The core idea is to extend the proof tree visitor to support visiting nested candidates without using a `probe`. We then simply recurse into nested candidates if they are the only potentially applicable candidate for a given goal and check whether the self type matches the expected one.
For that to work, we need to improve `CanonicalState` to also handle unconstrained inference variables created inside of the trait solver. This is done by extending the `var_values` of `CanoncalState` with each fresh inference variables. Furthermore, we also store the state of all inference variables at the end of each probe. When recursing into `InspectCandidates` we then unify the values of all these states.
r? `@compiler-errors`
Add support for run-make-support unit tests to be run with bootstrap
The `run-make-support` library needs to run its unit tests to ensure it is correct.
Close#124267
Convert some iter macros to normal functions
With all the MIR optimization changes that have happened since these were written, let's see if they still actually matter.
\*perf comes back\*
Well, it looks like it's not longer relevant for instruction, cycle, nor wall-time perf. Looks like a bunch of things are maybe 10kb bigger in debug, but some are also 50k *smaller* in debug.
So I think they should switch to being normal functions as the "greatly improves performance" justification for them being macros seems to no longer be true -- probably thanks to us always building `core` with `-Z inline-mir` so the difference is negligible.
