rustc_metadata: Privatize more things and a couple of other refactorings
This PR continues https://github.com/rust-lang/rust/pull/66496 and hits the point of diminishing returns.
All fields of `CrateRoot` and `CrateMetadata` are privatized.
For read-only fields this certainly makes sense, but for a few fields updateable from outside of `rmeta.rs` (mostly `creader.rs`) it was done mostly for consistency, I can make them `pub(crate)` again if requested.
`cstore.rs` (which became small after #66496) was merged into `creader.rs`.
A few things noticed while making the privacy changes were addressed in the remaining refactoring commits.
Fixes https://github.com/rust-lang/rust/issues/66550
r? @eddyb @Mark-Simulacrum
Apply proper commit from PR #63934
While working on PR #63934, I accidentally reverted to an older version
of the PR while working on a rebase. The PR was then merged, not with
the later, approved changes, but with earlier, unapproved changes.
This PR applies the changes that were *suppoesd* to be mereged in
PR #63934. All of the proper tests appear to have been merged
in PR #63934, so this PR adds no new tests
Fallback to .init_array when no arguments are available on glibc Linux
Linux is one of the only platforms where `std::env::args` doesn't work in a cdylib.
Previously:
error: invalid format string: invalid argument name `_x`
--> src/main.rs:2:16
|
2 | println!("{_x}", a=0);
| ^^ invalid argument name in format string
|
= note: argument names cannot start with an underscore
Not supporting identifiers starting with underscore appears to have been
an arbitrary limitation from 2013 in code that was most likely never
reviewed:
https://github.com/rust-lang/rust/pull/8245/files#diff-0347868ef389c805e97636623e4a4ea6R277
The error message was dutifully improved in #50610 but is there any
reason that leading underscore would be a special case?
This commit updates the format_args parser to accept identifiers with
leading underscores.
While working on PR #63934, I accidentally reverted to an older version
of the PR while working on a rebase. The PR was then merged, not with
the later, approved changes, but with earlier, unapproved changes.
This PR applies the changes that were *suppoesd* to be mereged in
PR #63934. All of the proper tests appear to have been merged
in PR #63934, so this PR adds no new tests
Fixes#66580
Namely, `update_extern_crate`.
Also, stop tracking visited crates in `update_extern_crate`, the rank check does the same thing (prevents visiting dependencies if the rank didn't change), but more precisely.
Create promoted MIR fragments for `const` and `static`s
Resolves#65732.
The previous strategy of removing `Drop` and `StorageDead` for promoted locals only worked for rvalue lifetime extension and only if no `loop`s were present. This PR applies the approach currently used for `fn` and `const fn`s to `const` and `statics`.
This may have some performance impacts.
r? @eddyb
Simplify memory categorization
With AST borrowck gone, mem_categorization can be simplified, a lot.
* `cmt_` is now called `Place`. Most local variable names have been updated to reflect this, but the `cat_*` methods retain their names.
* `MemCategorizationContext` no longer needs a `ScopeTree` and always needs an `InferCtxt`.
* `Place` now uses a similar representation to `mir::Place` with a `Vec` of projections.
* `Upvar` places don't include the implicit environment and capture derefs. These are now handled by `regionck` when needed.
* Various types, methods and variants only used by AST borrowck have been removed.
* `ExprUseVisitor` now lives in `rustc_typeck::expr_use_visitor`.
* `MemCategorizationContext` and `Place` live in `rustc_typeck::mem_categorization`.
* `Place` is re-exported in `rustc_typeck::expr_use_visitor` so that Clippy can access it.
The loss of an error in `issue-4335.rs` is due to a change in capture inference in ill-formed programs. If any projection from a variable is moved from then we capture that variable by move, whether or not the place being moved from allows this.
Closes#66270
Add memoization for const function evaluations
When a const function is being evaluated, as long as all its arguments are zero-sized-types (or it has no arguments) then we can trivially memoize the evaluation result using the existing query mechanism.
With thanks to @oli-obk for mentoring me through this at RustFest Barcelona.
r? @oli-obk
Implement Debug for MaybeUninit
Precedent: `UnsafeCell` implements `Debug` even though it can't actually display the value. I noticed this omission while writing the following:
```
#[derive(Debug)]
pub struct SliceInitializer<'a, T> {
marker: PhantomData<&'a mut T>,
uninit: &'a mut [MaybeUninit<T>],
written: usize,
}
```
...which currently unergonomically fails to compile.
`UnsafeCell` does require `T: Debug`. Because of things like the above I think it'd be better to leave that requirement off. In fact, I'd also suggest removing that requirement for `UnsafeCell` too, which again I noticed in some low-level real world code.
