Region naming refactoring [6/N]
Followup to #67474
EDIT: this PR is probably best read commit-by-commit...
The major changes in this PR include:
- moving many functions around to modules that better suit them. In particular, a lot of methods were moved from `borrow_check::diagnostics::region_errors` to `borrow_check::region_infer`, and `report_region_errors` was moved from `borrow_check` to `borrow_check::diagnostics::region_errors`.
- `borrow_check::diagnostics::{region_errors, region_name}` are now most comprised of methods on `MirBorrowckCtxt` instead of `RegionInferenceContext`, allowing us to get rid of the annoying `pub(in crate::borrow_check)` on most of the fields of the latter, along with a number of method arguments on many methods.
- I renamed `MirBorrowckCtxt.nonlexical_regioncx` to just `regioncx` because their is no lexical lifetimes any more, and the old name was annoyingly verbose, causing many lines to wrap unnecessarily.
- I got rid of `ErrorRegionNamingContext`. Region naming is implemented as inherent methods on `MirBorrowckCtxt`, so we just move the naming stuff into that struct.
The PR is rather large, but the commits are fairly self-contained (though they don't all compile). There was one minor output change to one test with `compare-mode=nll`, which I think is acceptable.
Between this PR and the last one, a net of 200 lines are removed, most of which was function parameters and context structs 🎉
Some samples:
```diff
- self.nonlexical_regioncx.free_region_constraint_info(
- &self.body,
- &self.local_names,
- &self.upvars,
- self.mir_def_id,
- self.infcx,
- borrow_region_vid,
- region,
- );
+ self.free_region_constraint_info(borrow_region_vid, region);
```
```diff
- .or_else(|| {
- self.give_name_if_anonymous_region_appears_in_yield_ty(
- infcx,
- body,
- *mir_def_id,
- fr,
- renctx,
- )
- });
+ .or_else(|| self.give_name_if_anonymous_region_appears_in_arguments(fr))
```
r? @matthewjasper
cc @eddyb
Rollup of 6 pull requests
Successful merges:
- #67956 (Detail transitive containment in E0588 diagnostic)
- #68153 (resolve: Point at the private item definitions in privacy errors)
- #68195 (Account for common `impl Trait`/`dyn Trait` return type errors)
- #68288 (Fix some of the rustfmt fallout in Miri)
- #68292 (don't clone types that are copy)
- #68301 (Don't propagate __RUST_TEST_INVOKE to subprocess)
Failed merges:
r? @ghost
Don't propagate __RUST_TEST_INVOKE to subprocess
When -Z panic_abort_tests is enabled, we use an environment variable to
tell the subprocess which test to invoke. If that subprocess then
invokes another Rust test binary, chaos ensues.
r? @alexcrichton
Account for common `impl Trait`/`dyn Trait` return type errors
- When all return paths have the same type, suggest `impl Trait`.
- When all return paths implement the expected `trait`, suggest `Box<dyn Trait>` and mention using an `enum`.
- When multiple different types are returned and `impl Trait` is expected, extend the explanation.
- When return type is `impl Trait` and the return paths do not implement `Trait`, point at the returned values.
- Split `src/librustc/traits/error_reporting.rs` into multiple files to keep size under control.
Fix#68110, cc #66523.
When -Z panic_abort_tests is enabled, we use an environment variable to
tell the subprocess which test to invoke. If that subprocess then
invokes another Rust test binary, chaos ensues.
When a type error involves a `dyn Trait` as the return type, do not emit
the type error, as the "return type is not `Sized`" error will provide
enough information to the user.
Enable leak sanitizer test case
* Use `black_box` to avoid memory leak removal during optimization.
* Leak multiple objects to make test case more robust.
Clean up some diagnostics by making them more consistent
In general:
- Diagnostic should start with a lowercase letter.
- Diagnostics should not end with a full stop.
- Ellipses contain three dots.
- Backticks should encode Rust code.
I also reworded a couple of messages to make them read more clearly.
It might be sensible to create a style guide for diagnostics, so these informal conventions are written down somewhere, after which we could audit the existing diagnostics.
r? @Centril
Add some regression tests
Closes#64848 (fixed by #67631)
Closes#65918 (ICE is hidden by #67000, no longer ICE)
Closes#66473 (fixed by #68084)
Closes#67550 (set mir-opt-level to 3)
r? @Centril
Promoteds can contain raw pointers, but these must still only point to immutable allocations
fixes#67601
r? @RalfJung
cc @wesleywiser in order to not change behaviour in this PR, const prop uses the constant rules for interning, but at least there's an explicit mode for it now that we can think about this in the future
Don't run const propagation on items with inconsistent bounds
Fixes#67696
Using `#![feature(trivial_bounds)]`, it's possible to write functions
with unsatisfiable 'where' clauses, making them uncallable. However, the
user can act as if these 'where' clauses are true inside the body of the
function, leading to code that would normally be impossible to write.
Since const propgation can run even without any user-written calls to a
function, we need to explcitly check for these uncallable functions.
Add unreachable propagation mir optimization pass
@oli-obk suggested we create a MIR pass that optimizes away basic blocks that lead only to basic blocks with terminator kind **unreachable**. This is a first take on this, which we started with @gilescope at RustFest Impl Days.
The test currently fails when the compiled program runs (undefined behaviour). Is there a way to avoid running the compiled program?
perf: Eagerly convert literals to consts
Previousely even literal constants were being converted to an `Unevaluted` constant for evaluation later. This seems unecessary as no more information is needed to be able to convert the literal to a mir constant.
Hopefully this will also minimise the performance impact of #67717, as far less constant evaluations are needed.
