Projection types in user annotations may contain inference variables.
This makes the normalization depend on the unification with the actual
type and thus requires a separate TypeOp to track the obligations.
Otherwise simply calling `TypeChecker::normalize` would ICE with
"unexpected ambiguity"
We delay projection normalization to further stages in order to
register user type annotations before normalization in HIR typeck.
There are two consumers of astconv: ItemCtxt and FnCtxt.
The former already expects unnormalized types from astconv, see its
AstConv trait impl.
The latter needs `RawTy` for a cleaner interface.
Unfortunately astconv still needs the normalization machinery in
order to resolve enum variants that have projections in the self type,
e.g. `<<T as Trait>::Assoc>::StructVariant {}`.
This is why `AstConv::normalize_ty_2` is necessary.
Fix rustdoc source code rendering for `#[path = "../path/to/mod.rs"]` links
Fixes#103517
While generating the location for modules source HTML to be saved at, a `..` path component appeared to be translated to `/up/`.
Additionally, while generating the navigation sidebar, `..` path components were ignored. This means that (as in the issue above), a *real* directory structure of:
```
sys/
unix/
mod.rs <-- contains #![path = "../unix/mod.rs]
cmath.rs
```
was rendered as:
```
sys/
unix/
mod.rs
unix/
cmath.rs <-- links to sys/unix/unix/cmath.rs.html, 404
```
While the *files* were stored as
```
sys/
unix/
mod.rs.html
up/
unix/
cmath.rs.html
```
Suggest adding named lifetime when the return contains value borrowed from more than one lifetimes of function inputs
fix for #105227.
The problem: The suggestion of adding an explicit `'_` lifetime bound is **incorrect** when the function's return type contains a value which could be borrowed from more than one lifetimes of the function's inputs. Instead, a named lifetime parameter can be introduced in such a case.
The solution: Checking the number of elided lifetimes in the function signature. If more than one lifetimes found in the function inputs when the suggestion of adding explicit `'_` lifetime, change it to using named lifetime parameter `'a` instead.
Rollup of 5 pull requests
Successful merges:
- #106400 (Point at expressions where inference refines an unexpected type)
- #106491 (Fix error-index redirect to work with the back button.)
- #106494 (Add regression test for #58355)
- #106499 (fix [type error] for error E0029 and E0277)
- #106502 (rustdoc: remove legacy user-select CSS)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
fix [type error] for error E0029 and E0277
check explicitly for the type references error
if ty.references_error() is true change the error to be err.delay_as_bug() and prevent the error E0029 and E0277 from emitting out this fix#105946
Point at expressions where inference refines an unexpected type
Fix#106355. Fix#14007. (!)
```
error[E0308]: mismatched types
--> src/test/ui/type/type-check/point-at-inference.rs:12:9
|
9 | foo.push(i);
| - this is of type `&{integer}`, which makes `foo` to be inferred as `Vec<&{integer}>`
...
12 | bar(foo);
| --- ^^^ expected `i32`, found `&{integer}`
| |
| arguments to this function are incorrect
|
= note: expected struct `Vec<i32>`
found struct `Vec<&{integer}>`
note: function defined here
--> src/test/ui/type/type-check/point-at-inference.rs:2:4
|
2 | fn bar(_: Vec<i32>) {}
| ^^^ -----------
help: consider dereferencing the borrow
|
9 | foo.push(*i);
| +
```
Add vendor to Fuchsia's target triple
Historically, Rust's Fuchsia targets have been labeled x86_64-fuchsia and aarch64-fuchsia. However, they should technically contain vendor information. This CL changes Fuchsia's target triples to include the "unknown" vendor since Clang now does normalization and handles all triple spellings.
This was previously attempted in #90510, which was closed due to inactivity.
check explicitly for the type references error
if ty.references_error() is true change the error to be err.delay_as_bug()
and prevent the error E0029 and E0277 from emitting out
this fix#105946
```rust
fn main() {
let v = Vec::new();
v.push(0);
v.push(0);
v.push("");
}
```
now produces
```
error[E0308]: mismatched types
--> $DIR/point-at-inference-3.rs:6:12
|
LL | v.push(0);
| - this is of type `{integer}`, which makes `v` to be inferred as `Vec<{integer}>`
...
LL | v.push("");
| ---- ^^ expected integer, found `&str`
| |
| arguments to this function are incorrect
|
note: associated function defined here
--> $SRC_DIR/alloc/src/vec/mod.rs:LL:COL
```
- Only point at a the single expression where the found type was first
inferred.
- Find method call argument that might have caused the found type to be
inferred.
- Provide structured suggestion.
- Apply some review comments.
- Tweak wording.
Historically, Rust's Fuchsia targets have been labeled x86_64-fuchsia
and aarch64-fuchsia. However, they should technically contain vendor
information. This CL changes Fuchsia's target triples to include the
"unknown" vendor since Clang now does normalization and handles all
triple spellings.
This was previously attempted in #90510, which was closed due to
inactivity.
Rollup of 6 pull requests
Successful merges:
- #105846 (Account for return-position `impl Trait` in trait in `opt_suggest_box_span`)
- #106385 (Split `-Zchalk` flag into `-Ztrait-solver=(classic|chalk|next)` flag)
- #106403 (Rename `hir::Map::{get_,find_}parent_node` to `hir::Map::{,opt_}parent_id`, and add `hir::Map::{get,find}_parent`)
- #106462 (rustdoc: remove unnecessary wrapper around sidebar and mobile logos)
- #106464 (Update Fuchsia walkthrough with new configs)
- #106478 (Tweak wording of fn call with wrong number of args)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Split `-Zchalk` flag into `-Ztrait-solver=(classic|chalk|next)` flag
We'll eventually need a way to select more than chalk + not-chalk.
Does this need an MCP since it's touching a `-Z` flag? Or perhaps I should preserve `-Zchalk` for the time being... maybe I could make it a warning to use that flag? cc ``@rust-lang/types``
r? types
Account for return-position `impl Trait` in trait in `opt_suggest_box_span`
RPITITs are the only types where their opaque bounds might normalize to some other self type than the opaque type itself. To avoid needing to do normalization, let's just match on either alias kind.
Ideally, we'd just get rid of `opt_suggest_box_span`. It's kind of a wart on type-checking `if`/`match`. I've recently refactored this expression for being confusing/wrong, but moving it into the error path is pretty hard.
Fixes#105838
Don't deduce a signature that makes a closure cyclic
Sometimes when elaborating supertrait bounds for closure signature inference, we end up deducing a closure signature that is cyclical because either a parameter or the return type references a projection mentioning `Self` that also has escaping bound vars, which means that it's not eagerly replaced with an inference variable.
Interestingly, this is not *just* related to my PR that elaborates supertrait bounds for closure signature deduction. The committed test `supertrait-hint-cycle-3.rs` shows **stable** code that is fixed by this PR:
```rust
trait Foo<'a> {
type Input;
}
impl<F: Fn(u32)> Foo<'_> for F {
type Input = u32;
}
fn needs_super<F: for<'a> Fn(<F as Foo<'a>>::Input) + for<'a> Foo<'a>>(_: F) {}
fn main() {
needs_super(|_: u32| {});
}
```
Fixes#105401Fixes#105396
r? types
Fix link generation for local primitive types in rustdoc JSON output
Fixes https://github.com/rust-lang/rust/issues/104064.
As mentioned in the issue, I'm not super happy about this fix which is more a hack rather than a sound-proof solution. However I couldn't find a better way to fix it.
r? `@aDotInTheVoid`
rustdoc: allow popover child links to work
No need to prevent default click behavior on a `<div>`, it will also disable all child click behavior.
Closes#106390
