const-eval: make misalignment a hard error
It's been a future-incompat error (showing up in cargo's reports) since https://github.com/rust-lang/rust/pull/104616, Rust 1.68, released in March. That should be long enough.
The question for the lang team is simply -- should we move ahead with this, making const-eval alignment failures a hard error? (It turns out some of them accidentally already were hard errors since #104616. But not all so this is still a breaking change. Crater found no regression.)
Detect ruby-style closure in parser
When parsing a closure without a body that is surrounded by a block, suggest moving the opening brace after the closure head.
Fix#116608.
On type error involving closure, avoid ICE
When we encounter a type error involving a closure, we try to typeck prior closure invocations to see if they influenced the current expected type. When trying to do so, ensure that the closure was defined in our current scope.
Fix#116658.
Improve check-cfg diagnostics
This PR tries to improve some of the diagnostics of check-cfg.
The main changes is the unexpected name or value being added to the main diagnostic:
```diff
- warning: unexpected `cfg` condition name
+ warning: unexpected `cfg` condition name: `widnows`
```
It also cherry-pick the better sensible logic for when we print the list of expected values when we have a matching value for a very similar name.
Address https://github.com/rust-lang/rust/pull/111072#discussion_r1356818100
r? `@petrochenkov`
Streamline `rustc_driver_impl` pretty-printing.
This PR simplifies a lot of unnecessary structure in
`rustc_driver_impl/src/pretty.rs`. It removes some traits and functions,
simplifies some structs, renames some things for increased consistency, and
eliminates some boilerplate code. Overall it cuts more than 150 lines of code.
r? `@compiler-errors`
Remove cgu_reuse_tracker from Session
This removes a bit of global mutable state.
It will now miss post-lto cgu reuse when ThinLTO determines that a cgu doesn't get changed, but there weren't any tests for this anyway and a test for it would be fragile to the exact implementation of ThinLTO in LLVM.
When we encounter a type error involving a closure, we try to typeck
prior closure invocations to see if they influenced the current expected
type. When trying to do so, ensure that the closure was defined in our
current scope.
Fix#116658.
exhaustiveness: Rework constructor splitting
`SplitWildcard` was pretty opaque. I replaced it with a more legible abstraction: `ConstructorSet` represents the set of constructors for patterns of a given type. This clarifies responsibilities: `ConstructorSet` handles one clear task, and diagnostic-related shenanigans can be done separately.
I'm quite excited, I had has this in mind for years but could never quite introduce it. This opens up possibilities, including type-specific optimisations (like using a `FxHashSet` to collect enum variants, which had been [hackily attempted some years ago](https://github.com/rust-lang/rust/pull/76918)), my one-pass rewrite (https://github.com/rust-lang/rust/pull/116042), and future librarification.
`NoAnn` and `IdentifiedAnnotation` impl both `pprust_ast::PpAnn` and
`pprust_hir::PpAnn`, which is a bit confusing, because the optional
`tcx` is only needed for the HIR cases. (Currently the `tcx` is
unnecessarily provided in the `expanded` AST cases.)
This commit splits each one into `Ast` and `Hir` versions, which makes
things clear about where the `tcx` is needed. The commit also renames
all the traits so they consistently end with `Ann`.
`call_with_pp_support_ast` and `call_with_pp_support_hir` how each have
a single call site. This commit inlines and removes them, which also
removes the need for all the supporting traits: `Sess`,
`AstPrinterSupport`, and `HirPrinterSupport`. The `sess` member is also
removed from several structs.
The handling of the `PpMode` variants is currently spread across three
functions: `print_after_parsing`, `print_after_hir_lowering`, and
`print_with_analysis`. Each one handles some of the variants. This split
is primarily because `print_after_parsing` has slightly different
arguments to the other two.
This commit changes the structure. It merges the three functions into a
single `print` function, and encapsulates the different arguments in a
new enum `PrintExtra`.
Benefits:
- The code is a little shorter.
- All the `PpMode` variants are handled in a single `match`, with no
need for `unreachable!` arms.
- It enables the trait removal in the subsequent commit by reducing
the number of `call_with_pp_support_ast` call sites from two to one.
First, both `AstPrinterSupport` and `HirPrinterSupport` have a `sess`
method. This commit introduces a `Sess` trait and makes the support
traits be subtraits of `Sess`, to avoid some duplication.
Second, both support traits have a `pp_ann` method that isn't needed if
we enable `trait_upcasting`. This commit removes those methods.
(Both of these traits will be removed in a subsequent commit, as will
the `trait_upcasting` use.)
`phase_3_run_analysis_passes` no longer exists, and AFAICT this code has
been refactored so much since this comment was written that it no longer
has any useful meaning.
Copy 1-element arrays as scalars, not vectors
For `[T; 1]` it's silly to copy as `<1 x T>` when we can just copy as `T`.
Inspired by https://github.com/rust-lang/rust/issues/101210#issuecomment-1732470941, which pointed out that `Option<[u8; 1]>` was codegenning worse than `Option<u8>`.
(I'm not sure *why* LLVM doesn't optimize out `<1 x u8>`, but might as well just not emit it in the first place in this codepath.)
---
I think I bit off too much in #116479; let me try just the scalar case first.
r? `@ghost`
coverage: Clarify loop-edge detection and graph traversal
This is a collection of improvements to two semi-related pieces of code:
- The code in `counters` that detects which graph edges don't exit a loop, and would therefore be good candidates to have their coverage computed as an expression rather than having a physical counter.
- The code in `graph` that traverses the coverage BCB graph in a particular order, and tracks loops and loop edges along the way (which is relevant to the above).
I was originally only planning to make the `graph` changes, but there was going to be a lot of indentation churn in `counters` anyway, and once I started looking I noticed a lot of opportunities for simplification.
---
`@rustbot` label +A-code-coverage
Handle several `#[diagnostic::on_unimplemented]` attributes correctly
This PR fixes an issues where rustc would ignore subsequent `#[diagnostic::on_unimplemented]` attributes. The [corresponding RFC](https://rust-lang.github.io/rfcs/3368-diagnostic-attribute-namespace.html) specifies that the first matching instance of each option is used. Invalid attributes are linted and otherwise ignored.
Having to keep passing in a graph reference was a holdover from when the graph
was partly mutated during traversal. As of #114354 that is no longer necessary,
so we can simplify the traversal code by storing a graph reference as a field
in `TraverseCoverageGraphWithLoops`.
The previous code was storing the worklist in a vector, and then selectively
adding items to the start or end of the vector. That's a perfect use-case for a
double-ended queue.
This change also reveals that the existing code was a bit confused about which
end of the worklist is the front or back. For now, items are always removed
from the front of the queue (instead of the back), and code that adds items to
the queue has been flipped, to preserve the existing behaviour.
Fix duplicate note on internal feature gates with associated issues
Fixes#116293
Note sure if I should add tests because the issue occurs only for feature gates having associated issues and that set of feature gates will change unpredictably leading to an unnecessary churn in tests.
Use structured suggestion for #113174
When encountering a for loop that is rejected by the borrow checker because it is being advanced within its body, provide a structured suggestion for `while let Some(pat) = iter.next()`.
The BuiltinInternalFeatures gate already has a struct level #[note]
attribute. The additional note field in it caused a duplicate to be
displayed when it was set to Some(...) which happened when the
feature had an associated issue
This PR fixes an issues where rustc would ignore subsequent
`#[diagnostic::on_unimplemented]` attributes. The [corresponding
RFC](https://rust-lang.github.io/rfcs/3368-diagnostic-attribute-namespace.html)
specifies that the first matching instance of each option is used.
Invalid attributes are linted and otherwise ignored.
Rollup of 5 pull requests
Successful merges:
- #116219 (Relate alias ty with variance)
- #116315 (Do not check for impossible predicates in const-prop lint.)
- #116436 (Structurally normalize for closure)
- #116597 (Prevent showing methods from blanket impls of not available foreign traits to show up in the search results)
- #116627 (small cleanup)
r? `@ghost`
`@rustbot` modify labels: rollup
When encountering a for loop that is rejected by the borrow checker
because it is being advanced within its body, provide a structured
suggestion for `while let Some(pat) = iter.next()`.
Do not check for impossible predicates in const-prop lint.
The enclosing query already checks for them, and replaces the body with a single `unreachable` if they are indeed impossible.
Relate alias ty with variance
In the new solver, turns out that the subst-relate branch of the alias-relate predicate was relating args invariantly even for opaques, which have variance 💀.
This change is a bit more invasive, but I'd rather not special-case it [here](aeaa5c30e5/compiler/rustc_trait_selection/src/solve/alias_relate.rs (L171-L190)) and then have it break elsewhere. I'm doing a perf run to see if the extra call to `def_kind` is that expensive, if it is, I'll reconsider.
r? ``@lcnr``
Compute NLL loan scopes using the polonius model
For a *location-insensitive* analysis (that is, without expressiveness improvements for users yet), this PR implements loans going out of scope using reachability and liveness, rather than checking if the issuing region's values contain a given CFG point. This is equivalent to NLL scopes and computes the same data.
r? `@matthewjasper`
A couple of notes:
- there are some assumptions about SCC representatives, placeholders, free regions, and member constraints that I believe hold, and they're documented in the code
- this passes all the UI tests with `-Zpolonius=next` -- the perf is [not terrible](https://github.com/rust-lang/rust/pull/112432#issuecomment-1749685862) and there are a bunch of ways to improve it in the future.
- there's a fixme left, hopefully Matthew you know a clean way to get the information it mentions.
