We already handle this case this way on the coherence side, and it matches the new solver's behaviour. While there is some breakage around type-alias-impl-trait (see new "type annotations needed" in tests/ui/type-alias-impl-trait/issue-84660-unsoundness.rs), no stable code breaks, and no new stable code is accepted.
Don't fatal when calling `expect_one_of` when recovering arg in `parse_seq`
In `parse_seq`, when parsing a sequence of token-separated items, if we don't see a separator, we try to parse another item eagerly in order to give a good diagnostic and recover from a missing separator:
d1a0fa5ed3/compiler/rustc_parse/src/parser/mod.rs (L900-L901)
If parsing the item itself calls `expect_one_of`, then we will fatal because of #58903:
d1a0fa5ed3/compiler/rustc_parse/src/parser/mod.rs (L513-L516)
For `precise_capturing` feature I implemented, we do end up calling `expected_one_of`:
d1a0fa5ed3/compiler/rustc_parse/src/parser/ty.rs (L712-L714)
This leads the compiler to fatal *before* having emitted the first error, leading to absolutely no useful information for the user about what happened in the parser.
This PR makes it so that we stop doing that.
Fixes#124195
fix normalizing in different `ParamEnv`s with the same `InferCtxt`
This PR changes the key of the projection cache from just `AliasTy` to `(AliasTy, ParamEnv)` to allow normalizing in different `ParamEnv`s without resetting caches. Previously, normalizing the same alias in different param envs would always reuse the cached result from the first normalization, which is incorrect if the projection clauses in the param env have changed.
Fixing this bug allows us to get rid of `InferCtxt::clear_caches`, which was only used by the `AutoTraitFinder`, because it requires normalizing in different param envs.
r? `@fmease`
change `NormalizesTo` to fully structurally normalize
notes in https://hackmd.io/wZ016dE4QKGIhrOnHLlThQ
need to also update the dev-guide once this PR lands. in short, the setup is now as follows:
`normalizes-to` internally implements one step normalization, applying that normalization to the `goal.predicate.term` causes the projected term to get recursively normalized. With this `normalizes-to` normalizes until the projected term is rigid, meaning that we normalize as many steps necessary, but at least 1.
To handle rigid aliases, we add another candidate only if the 1 to inf step normalization failed. With this `normalizes-to` is now full structural normalization. We can now change `AliasRelate` to simply emit `normalizes-to` goals for the rhs and lhs.
This avoids the concerns from https://github.com/rust-lang/trait-system-refactor-initiative/issues/103 and generally feels cleaner
some smaller DefiningOpaqueTypes::No -> Yes switches
r? `@compiler-errors`
These are some easy cases, so let's get them out of the way first.
I added tests exercising the specialization code paths that I believe weren't tested so far.
follow-up to https://github.com/rust-lang/rust/pull/117348
Split an item bounds and an item's super predicates
This is the moral equivalent of #107614, but instead for predicates this applies to **item bounds**. This PR splits out the item bounds (i.e. *all* predicates that are assumed to hold for the alias) from the item *super predicates*, which are the subset of item bounds which share the same self type as the alias.
## Why?
Much like #107614, there are places in the compiler where we *only* care about super-predicates, and considering predicates that possibly don't have anything to do with the alias is problematic. This includes things like closure signature inference (which is at its core searching for `Self: Fn(..)` style bounds), but also lints like `#[must_use]`, error reporting for aliases, computing type outlives predicates.
Even in cases where considering all of the `item_bounds` doesn't lead to bugs, unnecessarily considering irrelevant bounds does lead to a regression (#121121) due to doing extra work in the solver.
## Example 1 - Trait Aliases
This is best explored via an example:
```
type TAIT<T> = impl TraitAlias<T>;
trait TraitAlias<T> = A + B where T: C;
```
The item bounds list for `Tait<T>` will include:
* `Tait<T>: A`
* `Tait<T>: B`
* `T: C`
While `item_super_predicates` query will include just the first two predicates.
Side-note: You may wonder why `T: C` is included in the item bounds for `TAIT`? This is because when we elaborate `TraitAlias<T>`, we will also elaborate all the predicates on the trait.
## Example 2 - Associated Type Bounds
```
type TAIT<T> = impl Iterator<Item: A>;
```
The `item_bounds` list for `TAIT<T>` will include:
* `Tait<T>: Iterator`
* `<Tait<T> as Iterator>::Item: A`
But the `item_super_predicates` will just include the first bound, since that's the only bound that is relevant to the *alias* itself.
## So what
This leads to some diagnostics duplication just like #107614, but none of it will be user-facing. We only see it in the UI test suite because we explicitly disable diagnostic deduplication.
Regarding naming, I went with `super_predicates` kind of arbitrarily; this can easily be changed, but I'd consider better names as long as we don't block this PR in perpetuity.
misc cleanups from debugging something
rename `instantiate_canonical_with_fresh_inference_vars` to `instantiate_canonical` the substs for the canonical are not solely infer vars as that would be wildly wrong and it is rather confusing to see this method called and think that the entire canonicalization setup is completely broken when it is not 👍
also update region debug printing to be more like the custom impls for Ty/Const, right now regions in debug output are horribly verbose and make it incredibly hard to read but with this atleast boundvars and placeholders when debugging the new solver do not take up excessive amounts of space.
r? `@lcnr`
Stabilize associated type bounds (RFC 2289)
This PR stabilizes associated type bounds, which were laid out in [RFC 2289]. This gives us a shorthand to express nested type bounds that would otherwise need to be expressed with nested `impl Trait` or broken into several `where` clauses.
### What are we stabilizing?
We're stabilizing the associated item bounds syntax, which allows us to put bounds in associated type position within other bounds, i.e. `T: Trait<Assoc: Bounds...>`. See [RFC 2289] for motivation.
In all position, the associated type bound syntax expands into a set of two (or more) bounds, and never anything else (see "How does this differ[...]" section for more info).
Associated type bounds are stabilized in four positions:
* **`where` clauses (and APIT)** - This is equivalent to breaking up the bound into two (or more) `where` clauses. For example, `where T: Trait<Assoc: Bound>` is equivalent to `where T: Trait, <T as Trait>::Assoc: Bound`.
* **Supertraits** - Similar to above, `trait CopyIterator: Iterator<Item: Copy> {}`. This is almost equivalent to breaking up the bound into two (or more) `where` clauses; however, the bound on the associated item is implied whenever the trait is used. See #112573/#112629.
* **Associated type item bounds** - This allows constraining the *nested* rigid projections that are associated with a trait's associated types. e.g. `trait Trait { type Assoc: Trait2<Assoc2: Copy>; }`.
* **opaque item bounds (RPIT, TAIT)** - This allows constraining associated types that are associated with the opaque without having to *name* the opaque. For example, `impl Iterator<Item: Copy>` defines an iterator whose item is `Copy` without having to actually name that item bound.
The latter three are not expressible in surface Rust (though for associated type item bounds, this will change in #120752, which I don't believe should block this PR), so this does represent a slight expansion of what can be expressed in trait bounds.
### How does this differ from the RFC?
Compared to the RFC, the current implementation *always* desugars associated type bounds to sets of `ty::Clause`s internally. Specifically, it does *not* introduce a position-dependent desugaring as laid out in [RFC 2289], and in particular:
* It does *not* desugar to anonymous associated items in associated type item bounds.
* It does *not* desugar to nested RPITs in RPIT bounds, nor nested TAITs in TAIT bounds.
This position-dependent desugaring laid out in the RFC existed simply to side-step limitations of the trait solver, which have mostly been fixed in #120584. The desugaring laid out in the RFC also added unnecessary complication to the design of the feature, and introduces its own limitations to, for example:
* Conditionally lowering to nested `impl Trait` in certain positions such as RPIT and TAIT means that we inherit the limitations of RPIT/TAIT, namely lack of support for higher-ranked opaque inference. See this code example: https://github.com/rust-lang/rust/pull/120752#issuecomment-1979412531.
* Introducing anonymous associated types makes traits no longer object safe, since anonymous associated types are not nameable, and all associated types must be named in `dyn` types.
This last point motivates why this PR is *not* stabilizing support for associated type bounds in `dyn` types, e.g, `dyn Assoc<Item: Bound>`. Why? Because `dyn` types need to have *concrete* types for all associated items, this would necessitate a distinct lowering for associated type bounds, which seems both complicated and unnecessary compared to just requiring the user to write `impl Trait` themselves. See #120719.
### Implementation history:
Limited to the significant behavioral changes and fixes and relevant PRs, ping me if I left something out--
* #57428
* #108063
* #110512
* #112629
* #120719
* #120584Closes#52662
[RFC 2289]: https://rust-lang.github.io/rfcs/2289-associated-type-bounds.html
Provide structured suggestion for `#![feature(foo)]`
```
error: `S2<'_>` is forbidden as the type of a const generic parameter
--> $DIR/lifetime-in-const-param.rs:5:23
|
LL | struct S<'a, const N: S2>(&'a ());
| ^^
|
= note: the only supported types are integers, `bool` and `char`
help: add `#![feature(adt_const_params)]` to the crate attributes to enable more complex and user defined types
|
LL + #![feature(adt_const_params)]
|
```
Fix#55941.
```
error: `S2<'_>` is forbidden as the type of a const generic parameter
--> $DIR/lifetime-in-const-param.rs:5:23
|
LL | struct S<'a, const N: S2>(&'a ());
| ^^
|
= note: the only supported types are integers, `bool` and `char`
help: add `#![feature(adt_const_params)]` to the crate attributes to enable more complex and user defined types
|
LL + #![feature(adt_const_params)]
|
```
Fix#55941.
Split refining_impl_trait lint into _reachable, _internal variants
As discussed in https://github.com/rust-lang/rust/issues/119535#issuecomment-1909352040:
> We discussed this today in triage and developed a consensus to:
>
> * Add a separate lint against impls that refine a return type defined with RPITIT even when the trait is not crate public.
> * Place that in a lint group along with the analogous crate public lint.
> * Create an issue to solicit feedback on these lints (or perhaps two separate ones).
> * Have the warnings displayed with each lint reference this issue in a similar manner to how we do that today with the required `Self: '0'` bound on GATs.
> * Make a note to review this feedback on 2-3 release cycles.
This points users to https://github.com/rust-lang/rust/issues/121718 to leave feedback.
Consolidate WF for aliases
Make RPITs/TAITs/weak (type) aliases/projections all enforce:
1. their nominal predicates
2. their args are WF
This possibly does extra work, but is also nice for consistency sake.
r? lcnr
Make `DefiningAnchor::Bind` only store the opaque types that may be constrained, instead of the current infcx root item.
This makes `Bind` almost always be empty, so we can start forwarding it to queries, allowing us to remove `Bubble` entirely (not done in this PR)
The only behaviour change is in diagnostics.
r? `@lcnr` `@compiler-errors`
Don't ICE if we collect no RPITITs unless there are no unification errors
Move an assertion in `collect_return_position_impl_trait_in_trait_tys` to after the `ObligationCtxt::eq` calls, so that we only assert and ICE if we have unification errors.
Fixes#121468
Uplift some feeding out of `associated_type_for_impl_trait_in_impl` and into queries
This PR moves the `type_of` and `generics_of` query feeding out of `associated_type_for_impl_trait_in_impl`, since eagerly feeding results in query cycles due to a subtle interaction with `resolve_bound_vars`.
Fixes#122019
r? spastorino
stricter hidden type wf-check [based on #115008]
Original work by `@aliemjay` in #115008. A huge thanks to them for originally figuring out this approach ❤️
Fixes https://github.com/rust-lang/rust/issues/114728
Fixes https://github.com/rust-lang/rust/issues/114572
Instead of adding the `WellFormed` obligations when relating opaque types, we now always emit such an obligation when defining the hidden type.
This causes nested opaque types which aren't wf to error, see the comment below for the described impact. I believe this change to be desirable as it significantly reduces complexity by removing special-cases.
It also caused an issue with RPITIT: in defaulted trait methods, we add a `Projection(synthetic_assoc, rpit_of_trait_method)` clause to the `param_env`. This clause is not added to the `ParamEnv` of the nested coroutines. This caused a normalization failure in `fn check_coroutine_obligations` with the new solver. I fixed that by using the env of the typeck root instead.
r? `@oli-obk`
When encountering trait bound errors that satisfy some heuristics that
tell us that the relevant trait for the user comes from the root
obligation and not the current obligation, we use the root predicate for
the main message.
This allows to talk about "X doesn't implement Pattern<'_>" over the
most specific case that just happened to fail, like "char doesn't
implement Fn(&mut char)" in
`tests/ui/traits/suggest-dereferences/root-obligation.rs`
The heuristics are:
- the type of the leaf predicate is (roughly) the same as the type
from the root predicate, as a proxy for "we care about the root"
- the leaf trait and the root trait are different, so as to avoid
talking about `&mut T: Trait` and instead remain talking about
`T: Trait` instead
- the root trait is not `Unsize`, as to avoid talking about it in
`tests/ui/coercion/coerce-issue-49593-box-never.rs`.
```
error[E0277]: the trait bound `&char: Pattern<'_>` is not satisfied
--> $DIR/root-obligation.rs:6:38
|
LL | .filter(|c| "aeiou".contains(c))
| -------- ^ the trait `Fn<(char,)>` is not implemented for `&char`, which is required by `&char: Pattern<'_>`
| |
| required by a bound introduced by this call
|
= note: required for `&char` to implement `FnOnce<(char,)>`
= note: required for `&char` to implement `Pattern<'_>`
note: required by a bound in `core::str::<impl str>::contains`
--> $SRC_DIR/core/src/str/mod.rs:LL:COL
help: consider dereferencing here
|
LL | .filter(|c| "aeiou".contains(*c))
| +
```
Fix#79359, fix#119983, fix#118779, cc #118415 (the suggestion needs
to change).
Now that inlining, mir validation and const eval all use reveal-all, we won't be constraining hidden types here anymore
r? `@compiler-errors`
one bubble down, two more to go
the test is unrelated, just something I noticed would be good to test in both the old solver and the new.
Count stashed errors again
Stashed diagnostics are such a pain. Their "might be emitted, might not" semantics messes with lots of things.
#120828 and #121206 made some big changes to how they work, improving some things, but still leaving some problems, as seen by the issues caused by #121206. This PR aims to fix all of them by restricting them in a way that eliminates the "might be emitted, might not" semantics while still allowing 98% of their benefit. Details in the individual commit logs.
r? `@oli-obk`
Deeply normalize obligations in `refining_impl_trait`
We somewhat awkwardly use semantic comparison when checking the `refining_impl_trait` lint. This relies on us being able to normalize bounds eagerly to avoid cases where an unnormalized alias is not considered equal to a normalized alias. Since `normalize` in the new solver is a noop, let's use `deeply_normalize` instead.
r? lcnr
cc ``@tmandry,`` this should fix your bug lol
Stashed errors used to be counted as errors, but could then be
cancelled, leading to `ErrorGuaranteed` soundness holes. #120828 changed
that, closing the soundness hole. But it introduced other difficulties
because you sometimes have to account for pending stashed errors when
making decisions about whether errors have occured/will occur and it's
easy to overlook these.
This commit aims for a middle ground.
- Stashed errors (not warnings) are counted immediately as emitted
errors, avoiding the possibility of forgetting to consider them.
- The ability to cancel (or downgrade) stashed errors is eliminated, by
disallowing the use of `steal_diagnostic` with errors, and introducing
the more restrictive methods `try_steal_{modify,replace}_and_emit_err`
that can be used instead.
Other things:
- `DiagnosticBuilder::stash` and `DiagCtxt::stash_diagnostic` now both
return `Option<ErrorGuaranteed>`, which enables the removal of two
`delayed_bug` calls and one `Ty::new_error_with_message` call. This is
possible because we store error guarantees in
`DiagCtxt::stashed_diagnostics`.
- Storing the guarantees also saves us having to maintain a counter.
- Calls to the `stashed_err_count` method are no longer necessary
alongside calls to `has_errors`, which is a nice simplification, and
eliminates two more `span_delayed_bug` calls and one FIXME comment.
- Tests are added for three of the four fixed PRs mentioned below.
- `issue-121108.rs`'s output improved slightly, omitting a non-useful
error message.
Fixes#121451.
Fixes#121477.
Fixes#121504.
Fixes#121508.
Account for RPITIT in E0310 explicit lifetime constraint suggestion
When given
```rust
trait Original {
fn f() -> impl Fn();
}
trait Erased {
fn f(&self) -> Box<dyn Fn()>;
}
impl<T: Original> Erased for T {
fn f(&self) -> Box<dyn Fn()> {
Box::new(<T as Original>::f())
}
}
```
emit do not emit an invalid suggestion restricting the `Trait::{opaque}` type in a `where` clause:
```
error[E0310]: the associated type `<T as Original>::{opaque#0}` may not live long enough
--> $DIR/missing-static-bound-from-impl.rs:11:9
|
LL | Box::new(<T as Original>::f())
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
| |
| the associated type `<T as Original>::{opaque#0}` must be valid for the static lifetime...
| ...so that the type `impl Fn()` will meet its required lifetime bounds
```
Partially address #119773. Ideally we'd suggest modifying `Erased::f` instead.
r? `@compiler-errors`
Fix more #121208 fallout
#121208 converted lots of delayed bugs to bugs. Unsurprisingly, there were a few invalid conversion found via fuzzing.
r? `@lcnr`
Provide suggestions through `rustc_confusables` annotations
Help with common API confusion, like asking for `push` when the data structure really has `append`.
```
error[E0599]: no method named `size` found for struct `Vec<{integer}>` in the current scope
--> $DIR/rustc_confusables_std_cases.rs:17:7
|
LL | x.size();
| ^^^^
|
help: you might have meant to use `len`
|
LL | x.len();
| ~~~
help: there is a method with a similar name
|
LL | x.resize();
| ~~~~~~
```
Fix#59450 (we can open subsequent tickets for specific cases).
Fix#108437:
```
error[E0599]: `Option<{integer}>` is not an iterator
--> f101.rs:3:9
|
3 | opt.flat_map(|val| Some(val));
| ^^^^^^^^ `Option<{integer}>` is not an iterator
|
::: /home/gh-estebank/rust/library/core/src/option.rs:571:1
|
571 | pub enum Option<T> {
| ------------------ doesn't satisfy `Option<{integer}>: Iterator`
|
= note: the following trait bounds were not satisfied:
`Option<{integer}>: Iterator`
which is required by `&mut Option<{integer}>: Iterator`
help: you might have meant to use `and_then`
|
3 | opt.and_then(|val| Some(val));
| ~~~~~~~~
```
On type error of method call arguments, look at confusables for suggestion. Fix#87212:
```
error[E0308]: mismatched types
--> f101.rs:8:18
|
8 | stuff.append(Thing);
| ------ ^^^^^ expected `&mut Vec<Thing>`, found `Thing`
| |
| arguments to this method are incorrect
|
= note: expected mutable reference `&mut Vec<Thing>`
found struct `Thing`
note: method defined here
--> /home/gh-estebank/rust/library/alloc/src/vec/mod.rs:2025:12
|
2025 | pub fn append(&mut self, other: &mut Self) {
| ^^^^^^
help: you might have meant to use `push`
|
8 | stuff.push(Thing);
| ~~~~
```
When given
```rust
trait Original {
fn f() -> impl Fn();
}
trait Erased {
fn f(&self) -> Box<dyn Fn()>;
}
impl<T: Original> Erased for T {
fn f(&self) -> Box<dyn Fn()> {
Box::new(<T as Original>::f())
}
}
```
avoid suggestion to restrict the `Trait::{opaque}` type in a `where` clause:
```
error[E0310]: the associated type `<T as Original>::{opaque#0}` may not live long enough
--> $DIR/missing-static-bound-from-impl.rs:11:9
|
LL | Box::new(<T as Original>::f())
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
| |
| the associated type `<T as Original>::{opaque#0}` must be valid for the static lifetime...
| ...so that the type `impl Fn()` will meet its required lifetime bounds
```
CC #119773.
Do not provide a structured suggestion when the arguments don't match.
```
error[E0599]: no method named `test_mut` found for struct `Vec<{integer}>` in the current scope
--> $DIR/auto-ref-slice-plus-ref.rs:7:7
|
LL | a.test_mut();
| ^^^^^^^^
|
= help: items from traits can only be used if the trait is implemented and in scope
note: `MyIter` defines an item `test_mut`, perhaps you need to implement it
--> $DIR/auto-ref-slice-plus-ref.rs:14:1
|
LL | trait MyIter {
| ^^^^^^^^^^^^
help: there is a method `get_mut` with a similar name, but with different arguments
--> $SRC_DIR/core/src/slice/mod.rs:LL:COL
```
Consider methods beyond inherent ones when suggesting typos.
```
error[E0599]: no method named `owned` found for reference `&dyn Foo` in the current scope
--> $DIR/object-pointer-types.rs:11:7
|
LL | fn owned(self: Box<Self>);
| --------- the method might not be found because of this arbitrary self type
...
LL | x.owned();
| ^^^^^ help: there is a method with a similar name: `to_owned`
```
Fix#101013.
Without doing so we use the same candidate cache entry
for `?0: Trait<?1>` and `?0: Trait<?0>`. These goals are different
and we must not use the same entry for them.
we don't track them when canonicalizing or when freshening,
resulting in instable caching in the old solver, and issues when
instantiating query responses in the new one.
Currently `emit_stashed_diagnostic` is called from four(!) different
places: `print_error_count`, `DiagCtxtInner::drop`, `abort_if_errors`,
and `compile_status`.
And `flush_delayed` is called from two different places:
`DiagCtxtInner::drop` and `Queries`.
This is pretty gross! Each one should really be called from a single
place, but there's a bunch of entanglements. This commit cleans up this
mess.
Specifically, it:
- Removes all the existing calls to `emit_stashed_diagnostic`, and adds
a single new call in `finish_diagnostics`.
- Removes the early `flush_delayed` call in `codegen_and_build_linker`,
replacing it with a simple early return if delayed bugs are present.
- Changes `DiagCtxtInner::drop` and `DiagCtxtInner::flush_delayed` so
they both assert that the stashed diagnostics are empty (i.e.
processed beforehand).
- Changes `interface::run_compiler` so that any errors emitted during
`finish_diagnostics` (i.e. late-emitted stashed diagnostics) are
counted and cannot be overlooked. This requires adding
`ErrorGuaranteed` return values to several functions.
- Removes the `stashed_err_count` call in `analysis`. This is possible
now that we don't have to worry about calling `flush_delayed` early
from `codegen_and_build_linker` when stashed diagnostics are pending.
- Changes the `span_bug` case in `handle_tuple_field_pattern_match` to a
`delayed_span_bug`, because it now can be reached due to the removal
of the `stashed_err_count` call in `analysis`.
- Slightly changes the expected output of three tests. If no errors are
emitted but there are delayed bugs, the error count is no longer
printed. This is because delayed bugs are now always printed after the
error count is printed (or not printed, if the error count is zero).
There is a lot going on in this commit. It's hard to break into smaller
pieces because the existing code is very tangled. It took me a long time
and a lot of effort to understand how the different pieces interact, and
I think the new code is a lot simpler and easier to understand.
return `ty::Error` when equating `ty::Error`
This helps iron out a difference in diagnostics between `Sub` and `Equate` relations, which I'm currently trying to unify.
r? oli-obk
deduplicate infer var instantiation
Having 3 separate implementations of one of the most subtle parts of our type system is not a good strategy if we want to maintain a sound type system ✨ while working on this I already found some subtle bugs in the existing code, so that's awesome 🎉 cc #121159
This was necessary as I am not confident in my nll changes in #119106, so I am first cleaning this up in a separate PR.
r? `@BoxyUwU`
we already use `instantiate_const_var`. This does lose some debugging
info for nll because we stop populating the `reg_var_to_origin` table with
`RegionCtxt::Existential(None)`, I don't think that matters however.
Supporting this adds additional complexity to one of the most involved
parts of the type system, so I really don't think it's worth it.
Continue compilation after check_mod_type_wf errors
The ICEs fixed here were probably reachable through const eval gymnastics before, but now they are easily reachable without that, too.
The new errors are often bugfixes, where useful errors were missing, because they were reported after the early abort. In other cases sometimes they are just duplication of already emitted errors, which won't be user-visible due to deduplication.
fixes https://github.com/rust-lang/rust/issues/120860
Be less confident when `dyn` suggestion is not checked for object safety
#120275 no longer checks bare traits for object safety when making a `dyn` suggestion on Rust < 2021. In this case, qualify the suggestion with a note that the trait must be object safe, to prevent user confusion as seen in #116434
r? ```@fmease```
modify alias-relate to also normalize ambiguous opaques
allows a bunch of further cleanups and generally simplifies the type system. To handle https://github.com/rust-lang/trait-system-refactor-initiative/issues/8 we'll have to add a some additional complexity to the `(Alias, Infer)` branches in alias-relate, so removing the opaque type special case here is really valuable.
It does worsen `deduce_closure_signature` and friends even more as they now receive an inference variable which is only constrained via an `AliasRelate` goal. These probably have to look into alias relate goals somehow. Leaving that for a future PR as this is something we'll have to tackle regardless.
r? `@compiler-errors`
Properly handle `async` block and `async fn` in `if` exprs without `else`
When encountering a tail expression in the then arm of an `if` expression without an `else` arm, account for `async fn` and `async` blocks to suggest `return`ing the value and pointing at the return type of the `async fn`.
We now also account for AFIT when looking for the return type to point at.
Fix#115405.
When encountering a tail expression in the then arm of an `if` expression
without an `else` arm, account for `async fn` and `async` blocks to
suggest `return`ing the value and pointing at the return type of the
`async fn`.
We now also account for AFIT when looking for the return type to point at.
Fix#115405.
Make privacy visitor use types more (instead of HIR)
r? ``@petrochenkov``
This is a prerequisite to normalizing projections, as otherwise we have too many invalid bound vars (hir_ty_to_ty is creating types that have bound vars, but no binder).
The commits are still chaotic, I'm gonna clean them up, but I just wanted to let you know about the general direction and wondering if we could land this before adding normalization, as normalization is where behavioral changes happen, and I'd like to keep that part as minimal as possible.
[context can be found on zulip](https://rust-lang.zulipchat.com/#narrow/stream/315482-t-compiler.2Fetc.2Fopaque-types/topic/weak.20type.20aliases.20and.20privacy)
Some cleanups around diagnostic levels.
Plus some refactoring in and around diagnostic levels and emission. Details in the individual commit logs.
r? ````@oli-obk````
The two kinds of delayed bug have quite different semantics so a
stronger conceptual separation is nice. (`is_error` is a good example,
because the two kinds have different behaviour.)
The commit also moves the `DelayedBug` variant after `Error` in `Level`,
to reflect the fact that it's weaker than `Error` -- it might trigger an
error but also might not. (The pre-existing `downgrade_to_delayed_bug`
function also reflects the notion that delayed bugs are lower/after
normal errors.)
Plus it condenses some of the comments on `Level` into a table, for
easier reading, and introduces `can_be_top_or_sub` to indicate which
levels can be used in top-level diagnostics vs. subdiagnostics.
Finally, it renames `DiagCtxtInner::span_delayed_bugs` as
`DiagCtxtInner::delayed_bugs`. The `span_` prefix is unnecessary because
some delayed bugs don't have a span.
```
error[E0277]: the size for values of type `[i32]` cannot be known at compilation time
--> f100.rs:2:33
|
2 | let _ = std::mem::size_of::<[i32]>();
| ^^^^^ doesn't have a size known at compile-time
|
= help: the trait `Sized` is not implemented for `[i32]`
note: required by an implicit `Sized` bound in `std::mem::size_of`
--> /home/gh-estebank/rust/library/core/src/mem/mod.rs:312:22
|
312 | pub const fn size_of<T>() -> usize {
| ^ required by the implicit `Sized` requirement on this bound in `size_of`
```
Fix#120178.
Expand the primary span of E0277 when the immediate unmet bound is not what the user wrote:
```
error[E0277]: the trait bound `i32: Bar` is not satisfied
--> f100.rs:6:6
|
6 | <i32 as Foo>::foo();
| ^^^ the trait `Bar` is not implemented for `i32`, which is required by `i32: Foo`
|
help: this trait has no implementations, consider adding one
--> f100.rs:2:1
|
2 | trait Bar {}
| ^^^^^^^^^
note: required for `i32` to implement `Foo`
--> f100.rs:3:14
|
3 | impl<T: Bar> Foo for T {}
| --- ^^^ ^
| |
| unsatisfied trait bound introduced here
```
Fix#40120.
Deduplicate more sized errors on call exprs
Change the implicit `Sized` `Obligation` `Span` for call expressions to include the whole expression. This aids the existing deduplication machinery to reduce the number of errors caused by a single unsized expression.
When encountering a type mismatch error involving `dyn Trait`, mention
the existence of boxed trait objects if the other type involved
implements `Trait`.
Partially addresses #102629.
Change the implicit `Sized` `Obligation` `Span` for call expressions to
include the whole expression. This aids the existing deduplication
machinery to reduce the number of errors caused by a single unsized
expression.
Silence some follow-up errors [3/x]
this is one piece of the requested cleanups from https://github.com/rust-lang/rust/pull/117449
Keep error types around, even in obligations.
These help silence follow-up errors, as we now figure out that some types (most notably inference variables) are equal to an error type.
But it also allows figuring out more types in the presence of errors, possibly causing more errors.
`-Ztreat-err-as-bug` treats normal errors and delayed bugs equally,
which can lead to some really surprising results.
This commit changes `-Ztreat-err-as-bug` so it ignores delayed bugs,
unless they get promoted to proper bugs and are printed.
This feels to me much simpler and more logical. And it simplifies the
implementation:
- The `-Ztreat-err-as-bug` check is removed from in
`DiagCtxt::{delayed_bug,span_delayed_bug}`.
- `treat_err_as_bug` doesn't need to count delayed bugs.
- The `-Ztreat-err-as-bug` panic message is simpler, because it doesn't
have to mention delayed bugs.
Output of delayed bugs is now more consistent. They're always printed
the same way. Previously when they triggered `-Ztreat-err-as-bug` they
would be printed slightly differently, via `span_bug` in
`span_delayed_bug` or `delayed_bug`.
A minor behaviour change: the "no errors encountered even though
`span_delayed_bug` issued" printed before delayed bugs is now a note
rather than a bug. This is done so it doesn't get counted as an error
that might trigger `-Ztreat-err-as-bug`, which would be silly.
This means that if you use `-Ztreat-err-as-bug=1` and there are no
normal errors but there are delayed bugs, the first delayed bug will be
shown (and the panic will happen after it's printed).
Also, I have added a second note saying "those delayed bugs will now be
shown as internal compiler errors". I think this makes it clearer what
is happening, because the whole concept of delayed bugs is non-obvious.
There are some test changes.
- equality-in-canonical-query.rs: Minor output changes, and the error
count reduces by one because the "no errors encountered even though
`span_delayed_bug` issued" message is no longer counted as an error.
- rpit_tait_equality_in_canonical_query.rs: Ditto.
- storage-live.rs: The query stack disappears because these delayed bugs
are now printed at the end, rather than when they are created.
- storage-return.rs, span_delayed_bug.rs: now need
`-Zeagerly-emit-delayed-bugs` because they need the delayed bugs
emitted immediately to preserve behaviour.
Remove special-casing around `AliasKind::Opaque` when structurally resolving in new solver
This fixes a few inconsistencies around where we don't eagerly resolve opaques to their (locally-defined) hidden types in the new solver. It essentially allows this code to work:
```rust
fn main() {
type Tait = impl Sized;
struct S {
i: i32,
}
let x: Tait = S { i: 0 };
println!("{}", x.i);
}
```
Since `Tait` is defined in `main`, we are able to poke through the type of `x` with deref.
r? lcnr
Errors in `DiagCtxtInner::emit_diagnostic` are never set to
`Level::Bug`, because the condition never succeeds, because
`self.treat_err_as_bug()` is called *before* the error counts are
incremented.
This commit switches to `self.treat_next_err_as_bug()`, fixing the
problem. This changes the error message output to actually say "internal
compiler error".
Add regression test for #106630
This PR adds a regression test for #106630. I was unsure where exactly to place the test or how to test it locally so please let me know if I should change something.
On borrow return type, suggest borrowing from arg or owned return type
When we encounter a function with a return type that has an anonymous lifetime with no argument to borrow from, besides suggesting the `'static` lifetime we now also suggest changing the arguments to be borrows or changing the return type to be an owned type.
```
error[E0106]: missing lifetime specifier
--> $DIR/variadic-ffi-6.rs:7:6
|
LL | ) -> &usize {
| ^ expected named lifetime parameter
|
= help: this function's return type contains a borrowed value, but there is no value for it to be borrowed from
help: consider using the `'static` lifetime, but this is uncommon unless you're returning a borrowed value from a `const` or a `static`
|
LL | ) -> &'static usize {
| +++++++
help: instead, you are more likely to want to change one of the arguments to be borrowed...
|
LL | x: &usize,
| +
help: ...or alternatively, to want to return an owned value
|
LL - ) -> &usize {
LL + ) -> usize {
|
```
Fix#85843.
When we encounter a function with a return type that has an anonymous
lifetime with no argument to borrow from, besides suggesting the
`'static` lifetime we now also suggest changing the arguments to be
borrows or changing the return type to be an owned type.
```
error[E0106]: missing lifetime specifier
--> $DIR/variadic-ffi-6.rs:7:6
|
LL | ) -> &usize {
| ^ expected named lifetime parameter
|
= help: this function's return type contains a borrowed value, but there is no value for it to be borrowed from
help: consider using the `'static` lifetime, but this is uncommon unless you're returning a borrowed value from a `const` or a `static`
|
LL | ) -> &'static usize {
| +++++++
help: instead, you are more likely to want to change one of the arguments to be borrowed...
|
LL | x: &usize,
| +
help: ...or alternatively, to want to return an owned value
|
LL - ) -> &usize {
LL + ) -> usize {
|
```
Fix#85843.
new solver normalization improvements
cool beans
At the core of this PR is a `try_normalize_ty` which stops for rigid aliases by using `commit_if_ok`.
Reworks alias-relate to fully normalize both the lhs and rhs and then equate the resulting rigid (or inference) types. This fixes https://github.com/rust-lang/trait-system-refactor-initiative/issues/68 by avoiding the exponential blowup. Also supersedes #116369 by only defining opaque types if the hidden type is rigid.
I removed the stability check in `EvalCtxt::evaluate_goal` due to https://github.com/rust-lang/trait-system-refactor-initiative/issues/75. While I personally have opinions on how to fix it, that still requires further t-types/`@nikomatsakis` buy-in, so I removed that for now. Once we've decided on our approach there, we can revert this commit.
r? `@compiler-errors`
Fix depth check in ProofTreeVisitor.
The hack to cutoff overflows and cycles in the new trait solver was incorrect. We want to inspect everything with depth [0..10].
This fix exposed a previously unseen bug, which caused the compiler to ICE when invoking `trait_ref` on a non-assoc type projection. I simply added the guard in the `AmbiguityCausesVisitor`, and updated the expected output for the `auto-trait-coherence` test which now includes the extra note:
```text
|
= note: upstream crates may add a new impl of trait `std::marker::Send` for type `OpaqueType` in future versions
```
r? `@lcnr`
When we encounter a `dyn Trait` that isn't object safe, look for its
implementors. If there's one, mention using it directly If there are
less than 9, mention the possibility of creating a new enum and using
that instead.
Account for object unsafe `impl Trait on dyn Trait {}`. Make a
distinction between public and sealed traits.
Fix#80194.
Consider alias bounds when computing liveness in NLL (but this time sound hopefully)
This is a revival of #116040, except removing the changes to opaque lifetime captures check to make sure that we're not triggering any unsoundness due to the lack of general existential regions and the currently-existing `ReErased` hack we use instead.
r? `@aliemjay` -- I appreciate you pointing out the unsoundenss in the previous iteration of this PR, and I'd like to hear that you're happy with this iteration of this PR before this goes back into FCP :>
Fixes#116794 as well
---
(mostly copied from #116040 and reworked slightly)
# Background
Right now, liveness analysis in NLL is a bit simplistic. It simply walks through all of the regions of a type and marks them as being live at points. This is problematic in the case of aliases, since it requires that we mark **all** of the regions in their args[^1] as live, leading to bugs like #42940.
In reality, we may be able to deduce that fewer regions are allowed to be present in the projected type (or "hidden type" for opaques) via item bounds or where clauses, and therefore ideally, we should be able to soundly require fewer regions to be live in the alias.
For example:
```rust
trait Captures<'a> {}
impl<T> Captures<'_> for T {}
fn capture<'o>(_: &'o mut ()) -> impl Sized + Captures<'o> + 'static {}
fn test_two_mut(mut x: ()) {
let _f1 = capture(&mut x);
let _f2 = capture(&mut x);
//~^ ERROR cannot borrow `x` as mutable more than once at a time
}
```
In the example above, we should be able to deduce from the `'static` bound on `capture`'s opaque that even though `'o` is a captured region, it *can never* show up in the opaque's hidden type, and can soundly be ignored for liveness purposes.
# The Fix
We apply a simple version of RFC 1214's `OutlivesProjectionEnv` and `OutlivesProjectionTraitDef` rules to NLL's `make_all_regions_live` computation.
Specifically, when we encounter an alias type, we:
1. Look for a unique outlives bound in the param-env or item bounds for that alias. If there is more than one unique region, bail, unless any of the outlives bound's regions is `'static`, and in that case, prefer `'static`. If we find such a unique region, we can mark that outlives region as live and skip walking through the args of the opaque.
2. Otherwise, walk through the alias's args recursively, as we do today.
## Limitation: Multiple choices
This approach has some limitations. Firstly, since liveness doesn't use the same type-test logic as outlives bounds do, we can't really try several options when we're faced with a choice.
If we encounter two unique outlives regions in the param-env or bounds, we simply fall back to walking the opaque via its args. I expect this to be mostly mitigated by the special treatment of `'static`, and can be fixed in a forwards-compatible by a more sophisticated analysis in the future.
## Limitation: Opaque hidden types
Secondly, we do not employ any of these rules when considering whether the regions captured by a hidden type are valid. That causes this code (cc #42940) to fail:
```rust
trait Captures<'a> {}
impl<T> Captures<'_> for T {}
fn a() -> impl Sized + 'static {
b(&vec![])
}
fn b<'o>(_: &'o Vec<i32>) -> impl Sized + Captures<'o> + 'static {}
```
We need to have existential regions to avoid [unsoundness](https://github.com/rust-lang/rust/pull/116040#issuecomment-1751628189) when an opaque captures a region which is not represented in its own substs but which outlives a region that does.
## Read more
Context: https://github.com/rust-lang/rust/pull/115822#issuecomment-1731153952 (for the liveness case)
More context: https://github.com/rust-lang/rust/issues/42940#issuecomment-455198309 (for the opaque capture case, which this does not fix)
[^1]: except for bivariant region args in opaques, which will become less relevant when we move onto edition 2024 capture semantics for opaques.
Stash and cancel cycle errors for auto trait leakage in opaques
We don't need to emit a traditional cycle error when we have a selection error that explains what's going on but in more detail.
We may want to augment this error to actually point out the cycle, now that the cycle error is not being emitted. We could do that by storing the set of opaques that was in the `CyclePlaceholder` that gets returned from `type_of_opaque`.
r? `@oli-obk` cc `@estebank` #117235
Rework negative coherence to properly consider impls that only partly overlap
This PR implements a modified negative coherence that handles impls that only have partial overlap.
It does this by:
1. taking both impl trait refs, instantiating them with infer vars
2. equating both trait refs
3. taking the equated trait ref (which represents the two impls' intersection), and resolving any vars
4. plugging all remaining infer vars with placeholder types
these placeholder-plugged trait refs can then be used normally with the new trait solver, since we no longer have to worry about the issue with infer vars in param-envs.
We use the **new trait solver** to reason correctly about unnormalized trait refs (due to deferred projection equality), since this avoid having to normalize anything under param-envs with infer vars in them.
This PR then additionally:
* removes the `FnPtr` knowable hack by implementing proper negative `FnPtr` trait bounds for rigid types.
---
An example:
Consider these two partially overlapping impls:
```
impl<T, U> PartialEq<&U> for &T where T: PartialEq<U> {}
impl<F> PartialEq<F> for F where F: FnPtr {}
```
Under the old algorithm, we would take one of these impls and replace it with infer vars, then try unifying it with the other impl under identity substitutions. This is not possible in either direction, since it either sets `T = U`, or tries to equate `F = &?0`.
Under the new algorithm, we try to unify `?0: PartialEq<?0>` with `&?1: PartialEq<&?2>`. This gives us `?0 = &?1 = &?2` and thus `?1 = ?2`. The intersection of these two trait refs therefore looks like: `&?1: PartialEq<&?1>`. After plugging this with placeholders, we get a trait ref that looks like `&!0: PartialEq<&!0>`, with the first impl having substs `?T = ?U = !0` and the second having substs `?F = &!0`[^1].
Then we can take the param-env from the first impl, and try to prove the negated where clause of the second.
We know that `&!0: !FnPtr` never holds, since it's a rigid type that is also not a fn ptr, we successfully detect that these impls may never overlap.
[^1]: For the purposes of this example, I just ignored lifetimes, since it doesn't really matter.
