When encountering a fn call that has a path to another fn being passed
in, where an `Fn` impl is expected, and the arguments differ, suggest
wrapping the argument with a closure with the appropriate arguments.
When trying to create an inaccessible ADT due to private fields, handle
the case when no fields were passed.
```
error: cannot construct `Foo` with struct literal syntax due to private fields
--> $DIR/issue-76077.rs:8:5
|
LL | foo::Foo {};
| ^^^^^^^^
|
= note: private field `you_cant_use_this_field` that was not provided
```
This restriction made sense back when spaces separated function
parameters, but now that they separate path components, there's
no real ambiguity any more.
Additionally, the Rust language allows it.
ConstProp: Correctly remove const if unknown value assigned to it.
Closes#118328
The problematic sequence of MIR is:
```rust
_1 = const 0_usize;
_1 = const _; // This is an associated constant we can't know before monomorphization.
_0 = _1;
```
1. When `ConstProp::visit_assign` happens on `_1 = const 0_usize;`, it records that `0x0usize` is the value for `_1`.
2. Next `visit_assign` happens on `_1 = const _;`. Because the rvalue `.has_param()`, it can't be const evaled.
3. Finaly, `visit_assign` happens on `_0 = _1;`. Here it would think the value of `_1` was `0x0usize` from step 1.
The solution is to remove consts when checking the RValue fails, as they may have contained values that should now be invalidated, as that local was overwritten.
This should probably be back-ported to beta. Stable is more iffy, as it's gone unidentified since 1.70, so I only think it's worthwhile if there's another reason for a 1.74.1 release anyway.
Suggest `let` or `==` on typo'd let-chain
When encountering a bare assignment in a let-chain, suggest turning the
assignment into a `let` expression or an equality check.
```
error: expected expression, found `let` statement
--> $DIR/bad-if-let-suggestion.rs:5:8
|
LL | if let x = 1 && i = 2 {}
| ^^^^^^^^^
|
= note: only supported directly in conditions of `if` and `while` expressions
help: you might have meant to continue the let-chain
|
LL | if let x = 1 && let i = 2 {}
| +++
help: you might have meant to compare for equality
|
LL | if let x = 1 && i == 2 {}
| +
```
Add `never_patterns` feature gate
This PR adds the feature gate and most basic parsing for the experimental `never_patterns` feature. See the tracking issue (https://github.com/rust-lang/rust/issues/118155) for details on the experiment.
`@scottmcm` has agreed to be my lang-team liaison for this experiment.
Fix coroutine validation for mixed panic strategy
Validation introduced in #113124 allows `UnwindAction::Continue` and `TerminatorKind::Resume` to occur only in functions with ABI that can unwind. The function ABI depends on the panic strategy, which can vary across crates.
Usually MIR is built and validated in the same crate. The coroutine drop glue thus far was an exception. As a result validation could fail when mixing different panic strategies.
Avoid the problem by executing `AbortUnwindingCalls` along with the validation.
Fixes#116953.
Eagerly return `ExprKind::Err` on `yield`/`await` in wrong coroutine context
This PR does 2 things:
1. Refuses to lower `.await` or `yield` when we are outside of the right coroutine context for the operator. Instead, we lower to `hir::ExprKind::Err`, to silence subsequent redundant errors.
2. Reworks a bit of the span tracking in `LoweringContext` to fix a bad span when we have something like `let x = [0; async_fn().await]` where the `await` is inside of an anon const. The span for the "item" still kinda sucks, since it overlaps with the `await` span, but at least it's accurate.
Rollup of 6 pull requests
Successful merges:
- #118193 (Add missing period in `std::process::Command` docs)
- #118222 (unify read_to_end and io::copy impls for reading into a Vec)
- #118323 (give dev-friendly error message for incorrect config profiles)
- #118378 (Perform LTO optimisations with wasm-ld + -Clinker-plugin-lto)
- #118399 (Clean dead codes in miri)
- #118410 (update test for new LLVM 18 codegen)
r? `@ghost`
`@rustbot` modify labels: rollup
Validation introduced in #113124 allows UnwindAction::Continue and
TerminatorKind::Resume to occur only in functions with ABI that can
unwind. The function ABI depends on the panic strategy, which can vary
across crates.
Usually MIR is built and validated in the same crate. The coroutine drop
glue thus far was an exception. As a result validation could fail when
mixing different panic strategies.
Avoid the problem by executing AbortUnwindingCalls along with the
validation.
When encountering a bare assignment in a let-chain, suggest turning the
assignment into a `let` expression or an equality check.
```
error: expected expression, found `let` statement
--> $DIR/bad-if-let-suggestion.rs:5:8
|
LL | if let x = 1 && i = 2 {}
| ^^^^^^^^^
|
= note: only supported directly in conditions of `if` and `while` expressions
help: you might have meant to continue the let-chain
|
LL | if let x = 1 && let i = 2 {}
| +++
help: you might have meant to compare for equality
|
LL | if let x = 1 && i == 2 {}
| +
```
effects: Run `enforce_context_effects` for all method calls
So that we also perform checks when overloaded `PartialEq`s are called.
r? `@compiler-errors`
rustc_span: Use correct edit distance start length for suggestions
Otherwise the suggestions can be off-base for non-ASCII identifiers. For example suggesting that `Ok` is a name similar to `读文`.
Closes https://github.com/rust-lang/rust/issues/72553.
Added linker_arg(s) Linker trait methods for link-arg to be prefixed "-Wl," for cc-like linker args and not verbatim
https://github.com/rust-lang/rust/issues/99427#issuecomment-1234443468
> here's one possible improvement to -l link-arg making it more portable between linkers and useful - befriending it with the verbatim modifier (https://github.com/rust-lang/rust/issues/99425).
>
> -l link-arg:-verbatim=-foo would add -Wl,-foo (or equivalent) when C compiler is used as a linker, and just -foo when bare linker is used.
> -l link-arg:+verbatim=-bar on the other hand would always pass just -bar.
Account for `!` arm in tail `match` expr
On functions with a default return type that influences the coerced type of `match` arms, check if the failing arm is actually of type `!`. If so, suggest changing the return type so the coercion against the prior arms is successful.
```
error[E0308]: `match` arms have incompatible types
--> $DIR/match-tail-expr-never-type-error.rs:9:13
|
LL | fn bar(a: bool) {
| - help: try adding a return type: `-> i32`
LL | / match a {
LL | | true => 1,
| | - this is found to be of type `{integer}`
LL | | false => {
LL | | never()
| | ^^^^^^^
| | |
| | expected integer, found `()`
| | this expression is of type `!`, but it get's coerced to `()` due to its surrounding expression
LL | | }
LL | | }
| |_____- `match` arms have incompatible types
```
Fix#24157.
Do not erase late bound regions when selecting inherent associated types
In the fix for #97156 we would want the following code:
```rust
#![feature(inherent_associated_types)]
#![allow(incomplete_features)]
struct Foo<T>(T);
impl Foo<fn(&'static ())> {
type Assoc = u32;
}
trait Other {}
impl Other for u32 {}
// FIXME(inherent_associated_types): Avoid emitting two diagnostics (they only differ in span).
// FIXME(inherent_associated_types): Enhancement: Spruce up the diagnostic by saying something like
// "implementation is not general enough" as is done for traits via
// `try_report_trait_placeholder_mismatch`.
fn bar(_: Foo<for<'a> fn(&'a ())>::Assoc) {}
//~^ ERROR mismatched types
//~| ERROR mismatched types
fn main() {}
```
to fail with ...
```
error[E0220]: associated type `Assoc` not found for `Foo<for<'a> fn(&'a ())>` in the current scope
--> tests/ui/associated-inherent-types/issue-109789.rs:18:36
|
4 | struct Foo<T>(T);
| ------------- associated item `Assoc` not found for this struct
...
18 | fn bar(_: Foo<for<'a> fn(&'a ())>::Assoc) {}
| ^^^^^ associated item not found in `Foo<for<'a> fn(&'a ())>`
|
= note: the associated type was found for
- `Foo<fn(&'static ())>`
error: aborting due to previous error
For more information about this error, try `rustc --explain E0220`.
```
This PR fixes the ICE we are currently getting "was a subtype of Foo<Binder(fn(&ReStatic ()), [])> during selection but now it is not"
Also fixes#112631
r? `@lcnr`
On functions with a default return type that influences the coerced type
of `match` arms, check if the failing arm is actually of type `!`. If
so, suggest changing the return type so the coercion against the prior
arms is successful.
```
error[E0308]: `match` arms have incompatible types
--> $DIR/match-tail-expr-never-type-error.rs:9:13
|
LL | fn bar(a: bool) {
| - help: try adding a return type: `-> i32`
LL | / match a {
LL | | true => 1,
| | - this is found to be of type `{integer}`
LL | | false => {
LL | | never()
| | ^^^^^^^
| | |
| | expected integer, found `()`
| | this expression is of type `!`, but it get's coerced to `()` due to its surrounding expression
LL | | }
LL | | }
| |_____- `match` arms have incompatible types
```
Fix#24157.
Suggest swapping the order of `ref` and `box`
It is not valid grammar to write `ref box <ident>` in patterns, but `box ref <ident>` is.
This patch adds a diagnostic to suggest swapping them, analogous to what we do for `mut let`.
Add `debug_assert_nounwind` and convert `assert_unsafe_precondition`
`assert_unsafe_precondition` checks non-CTFE-evaluable conditions in runtime and performs no-op in compile time, while many of its current usage can be checked during const eval.
Rewrite exhaustiveness in one pass
This is at least my 4th attempt at this in as many years x) Previous attempts were all too complicated or too slow. But we're finally here!
The previous version of the exhaustiveness algorithm computed reachability for each arm then exhaustiveness of the whole match. Since each of these steps does roughly the same things, this rewrites the algorithm to do them all in one go. I also think this makes things much simpler.
I also rewrote the documentation of the algorithm in depth. Hopefully it's up-to-date and easier to follow now. Plz comment if anything's unclear.
r? `@oli-obk` I think you're one of the rare other people to understand the exhaustiveness algorithm?
cc `@varkor` I know you're not active anymore, but if you feel like having a look you might enjoy this :D
Fixes https://github.com/rust-lang/rust/issues/79307
Don't ICE when encountering placeholders in implied bounds computation
I *could* fix this the right way, though I don't really want to think about the implications of the change. This should have minimal side-effects.
r? `@aliemjay`
Fixes#118286
coverage: Simplify building coverage expressions based on sums
This is a combination of some interlinked changes to the code that creates coverage counters/expressions for nodes and edges in the coverage graph:
- Some preparatory cleanups in `MakeBcbCounters::make_branch_counters`
- Use `BcbCounter` (instead of `CovTerm`) when building coverage expressions
- This makes it easier to introduce a fold for building sums
- Simplify the creation of coverage expressions based on sums, by having `Iterator::fold` do much of the work
- Get rid of the awkward `BcbBranch` enum, and replace it with graph edges represented as `(from_bcb, to_bcb)`
- This further simplifies the body of the fold
Currently we always do this:
```
use rustc_fluent_macro::fluent_messages;
...
fluent_messages! { "./example.ftl" }
```
But there is no need, we can just do this everywhere:
```
rustc_fluent_macro::fluent_messages! { "./example.ftl" }
```
which is shorter.
The `fluent_messages!` macro produces uses of
`crate::{D,Subd}iagnosticMessage`, which means that every crate using
the macro must have this import:
```
use rustc_errors::{DiagnosticMessage, SubdiagnosticMessage};
```
This commit changes the macro to instead use
`rustc_errors::{D,Subd}iagnosticMessage`, which avoids the need for the
imports.
In some cases we need to prepare a coverage expression that is the sum of an
arbitrary number of other terms. This patch simplifies the code paths that
build those sums.
This causes some churn in the mappings, because the previous code was building
its sums in a somewhat idiosyncratic order.
Fixes error count display is different when there's only one error left
Supersedes #114759
### What did I do?
I did the small change in `rustc_errors` by hand. Then I did the other changes in `/compiler` by hand, those were just find replace on `*.rs` in the workspace. The changes in run-make are find replace for `run-make` in the workspace.
All other changes are blessed using `x test TEST --bless`. I blessed the tests that were blessed in #114759.
### how to review this nightmare
ping bors with an `r+`. You should check that my logic is sound and maybe quickly scroll through the diff, but fully verifying it seems fairly hard to impossible. I did my best to do this correctly.
Thank you `@adrianEffe` for bringing this up and your initial implementation.
cc `@flip1995,` you said you want to do a subtree sync asap
cc `@RalfJung` maybe you want to do a quick subtree sync afterwards as well for Miri
r? `@WaffleLapkin`
Add `Span` to `TraitBoundModifier`
This improves diagnostics for the message "`~const` is not allowed here", and also fixes the span that we use when desugaring `~const Tr` into `Tr<host>` in effects desugaring.
Add common trait for crate definitions
In stable mir, we specialize DefId, however some functionality is the same for every definition, such as def paths, and getting their crate. Use a trait to implement those.
feat: make `let_binding_suggestion` more reasonable
This is my first PR for rustc, which trying to fix https://github.com/rust-lang/rust/issues/117894, I am not familiar with some internal api so maybe some modification here isn't the way to go, appreciated for any review suggestion.
Don't print "private fields" on empty tuple structs
Closes#118180.
While working on this I also noticed that empty struct variants are also rendered rather awkwardly. I'll make another issue for that, since I don't know what the correct rendering would be.
Rework supertrait lint once again
I accidentally pushed the wrong commits because I totally didn't check I was on the right computer when updating #118026.
Sorry, this should address all the nits in #118026.
r? lcnr
improve tool-only help for multiple `#[default]` variants
When defining an enum with multiple `#[default]` variants, we emit a tool-only suggestion for every `#[default]`ed variant to remove all other `#[default]`s. This PR improves the suggestion to correctly handle the cases where one variant has multiple `#[default]`s and where different `#[default]`s have the same span due to macro expansions.
fixes https://github.com/rust-lang/rust/issues/118119
We currently provide only a `help` message, this PR introduces the last
two structured suggestions instead:
```
error[E0308]: mismatched types
--> $DIR/issue-98982.rs:2:5
|
LL | fn foo() -> i32 {
| --- expected `i32` because of return type
LL | / for i in 0..0 {
LL | | return i;
LL | | }
| |_____^ expected `i32`, found `()`
|
note: the function expects a value to always be returned, but loops might run zero times
--> $DIR/issue-98982.rs:2:5
|
LL | for i in 0..0 {
| ^^^^^^^^^^^^^ this might have zero elements to iterate on
LL | return i;
| -------- if the loop doesn't execute, this value would never get returned
help: return a value for the case when the loop has zero elements to iterate on
|
LL ~ }
LL ~ /* `i32` value */
|
help: otherwise consider changing the return type to account for that possibility
|
LL ~ fn foo() -> Option<i32> {
LL | for i in 0..0 {
LL ~ return Some(i);
LL ~ }
LL ~ None
|
```
Fix#98982.
Rollup of 6 pull requests
Successful merges:
- #118012 (Add support for global allocation in smir)
- #118013 (Enable Rust to use the EHCont security feature of Windows)
- #118100 (Enable profiler in dist-powerpc64-linux)
- #118142 (Tighten up link attributes for llvm-wrapper bindings)
- #118147 (Fix some unnecessary casts)
- #118161 (Allow defining opaques in `check_coroutine_obligations`)
r? `@ghost`
`@rustbot` modify labels: rollup
Allow defining opaques in `check_coroutine_obligations`
In the new trait solver, when an obligation stalls on an unresolved coroutine witness, we will stash away the *root* obligation, even if the stalled obligation is only a distant descendent of the root obligation, since the new solver is purely recursive.
This means that we may need to reprocess alias-relate obligations (and others) which may define opaque types in the new solver. Currently, we use the coroutine's def id as the defining anchor in `check_coroutine_obligations`, which will allow defining no opaque types, resulting in errors like:
```
error[E0271]: type mismatch resolving `{coroutine@<source>:6:5: 6:17} <: impl Clone`
--> <source>:6:5
|
6 | / move |_: ()| {
7 | | let () = yield ();
8 | | }
| |_____^ types differ
```
So this PR fixes the defining anchor and does the same trick as `check_opaque_well_formed`, where we manually compare opaques that were defined against their hidden types to make sure they weren't defined differently when processing these stalled coroutine obligations.
r? `@lcnr` cc `@cjgillot`
Enable Rust to use the EHCont security feature of Windows
In the future Windows will enable Control-flow Enforcement Technology (CET aka Shadow Stacks). To protect the path where the context is updated during exception handling, the binary is required to enumerate valid unwind entrypoints in a dedicated section which is validated when the context is being set during exception handling.
The required support for EHCONT Guard has already been merged into LLVM, long ago. This change simply adds the Rust codegen option to enable it.
Relevant LLVM change: https://reviews.llvm.org/D40223
This also adds a new `ehcont-guard` option to the bootstrap config which enables EHCont Guard when building std.
We at Microsoft have been using this feature for a significant period of time; we are confident that the LLVM feature, when enabled, generates well-formed code.
We currently enable EHCONT using a codegen feature, but I'm certainly open to refactoring this to be a target feature instead, or to use any appropriate mechanism to enable it.
Add support for global allocation in smir
Add APIs to StableMir to support global allocation. Before this change, StableMir users had no API available to retrieve Allocation provenance information. They had to resource to internal APIs instead.
One example is retrieving the Allocation of an `&str`. See test for an example on how the API can be used.
Add allow-by-default lint for unit bindings
### Example
```rust
#![warn(unit_bindings)]
macro_rules! owo {
() => {
let whats_this = ();
}
}
fn main() {
// No warning if user explicitly wrote `()` on either side.
let expr = ();
let () = expr;
let _ = ();
let _ = expr; //~ WARN binding has unit type
let pat = expr; //~ WARN binding has unit type
let _pat = expr; //~ WARN binding has unit type
// No warning for let bindings with unit type in macro expansions.
owo!();
// No warning if user explicitly annotates the unit type on the binding.
let pat: () = expr;
}
```
outputs
```
warning: binding has unit type `()`
--> $DIR/unit-bindings.rs:17:5
|
LL | let _ = expr;
| ^^^^-^^^^^^^^
| |
| this pattern is inferred to be the unit type `()`
|
note: the lint level is defined here
--> $DIR/unit-bindings.rs:3:9
|
LL | #![warn(unit_bindings)]
| ^^^^^^^^^^^^^
warning: binding has unit type `()`
--> $DIR/unit-bindings.rs:18:5
|
LL | let pat = expr;
| ^^^^---^^^^^^^^
| |
| this pattern is inferred to be the unit type `()`
warning: binding has unit type `()`
--> $DIR/unit-bindings.rs:19:5
|
LL | let _pat = expr;
| ^^^^----^^^^^^^^
| |
| this pattern is inferred to be the unit type `()`
warning: 3 warnings emitted
```
This lint is not triggered if any of the following conditions are met:
- The user explicitly annotates the binding with the `()` type.
- The binding is from a macro expansion.
- The user explicitly wrote `let () = init;`
- The user explicitly wrote `let pat = ();`. This is allowed for local lifetimes.
### Known Issue
It is known that this lint can trigger on some proc-macro generated code whose span returns false for `Span::from_expansion` because e.g. the proc-macro simply forwards user code spans, and otherwise don't have distinguishing syntax context compared to non-macro-generated code. For those kind of proc-macros, I believe the correct way to fix them is to instead emit identifers with span like `Span::mixed_site().located_at(user_span)`.
Closes#71432.
Remove `feature` from the list of well known check-cfg name
This PR removes `feature` from the list of well known check-cfg.
This is done for multiple reasons:
- Cargo is the source of truth, rustc shouldn't have any knowledge of it
- It creates a conflict between Cargo and rustc when there are no features defined.
In this case Cargo won't pass any `--check-cfg` for `feature` since no feature will ever be passed, but rustc by having in it's list adds a implicit `cfg(feature, values(any()))` which is completely wrong. Having any cfg `feature` is unexpected not allow any `feature` value.
While doing this, I took the opportunity to specialise the diagnostic a bit for the case above.
r? `@petrochenkov`
Don't ICE when ambiguity is found when selecting `Index` implementation in typeck
Fixes#118111
The problem here is when we're manually "selecting" an impl for `base_ty: Index<?0>`, we don't consider placeholder region errors (leak check) or ambiguous predicates. Those can lead to us not actually emitting any fulfillment errors on line 3131.
In the future Windows will enable Control-flow Enforcement Technology
(CET aka Shadow Stacks). To protect the path where the context is
updated during exception handling, the binary is required to enumerate
valid unwind entrypoints in a dedicated section which is validated when
the context is being set during exception handling.
The required support for EHCONT has already been merged into LLVM,
long ago. This change adds the Rust codegen option to enable it.
Reference:
* https://reviews.llvm.org/D40223
This also adds a new `ehcont-guard` option to the bootstrap config which
enables EHCont Guard when building std.
The search sorting code already sorts by item type discriminant,
putting things with smaller discriminants first. There was
also a special case for sorting keywords and primitives earlier,
and this commit removes it by giving them lower discriminants.
The sorting code has another criteria where items with descriptions
appear earlier than items without, and that criteria has higher
priority than the item type. This shouldn't matter, though,
because primitives and keywords normally only appear in the
standard library, and it always gives them descriptions.
Fix early param lifetimes in generic_const_exprs
In cases like below, we never actually be able to capture region name for two reasons, first `'static` becomes anonymous lifetime and second we never capture region if it doesn't have a name so this results in ICE.
```
struct DataWrapper<'static> {
data: &'a [u8; Self::SIZE],
}
impl DataWrapper<'a> {
```
Fixes https://github.com/rust-lang/rust/issues/118021
Remove `--check-cfg` checking of command line `--cfg` args
Back in https://github.com/rust-lang/rust/pull/100574 we added to the `unexpected_cfgs` lint the checking of `--cfg` CLI arguments and emitted unexpected names and values for them.
The implementation works as expected, but it's usability in particular when using it in combination with Cargo+`RUSTFLAGS` as people who set `RUSTFLAGS=--cfg=tokio_unstable` (or whatever) have `unexpected_cfgs` warnings on all of their crates is debatable. ~~To fix this issue this PR proposes that we split the CLI argument checking into it's own separate allow-by-default lint: `unexpected_cli_cfgs`.~~
~~This has the advantage of letting people who want CLI warnings have them (although not by default anymore), while still linting on every unexpected cfg name and values in the code.~~
After some discussion with the Cargo team ([Zulip thread](https://rust-lang.zulipchat.com/#narrow/stream/246057-t-cargo/topic/check-cfg.20and.20RUSTFLAGS.20interaction)) and member of the compiler team (see below), I propose that we follow the suggestion from `@epage:` never check `--cfg` arguments, but still reserve us the possibility to do it later.
We would still lint on unexpected cfgs found in the source code no matter the `--cfg` args passed. This mean reverting https://github.com/rust-lang/rust/pull/100574 but NOT https://github.com/rust-lang/rust/pull/99519.
r? `@petrochenkov`
rustdoc-search: add support for traits and associated types
# Summary
Trait associated type queries work in rustdoc's type driven search. The data is included in the search-index.js file, and the queries are designed to "do what I mean" when users type them in, so, for example, `Iterator<Item=T> -> Option<T>` includes `Iterator::next` in the SERP[^SERP], and `Iterator<T> -> Option<T>` also includes `Iterator::next` in the SERP.
[^SERP]: search engine results page
## Sample searches
* [`iterator<Item=T>, fnmut -> T`][iterreduce]
* [`iterator<T>, fnmut -> T`][iterreduceterse]
[iterreduce]: http://notriddle.com/rustdoc-html-demo-5/associated-types/std/index.html?search=iterator%3CItem%3DT%3E%2C%20fnmut%20-%3E%20T&filter-crate=std
[iterreduceterse]: http://notriddle.com/rustdoc-html-demo-5/associated-types/std/index.html?search=iterator%3CT%3E%2C%20fnmut%20-%3E%20T&filter-crate=std
# Motivation
My primary motivation for working on search.js at all is to make it easier to use highly generic APIs, like the Iterator API. The type signature describes these functions pretty well, while the names are almost arbitrary.
Before this PR, type bindings were not consistently included in search-index.js at all (you couldn't find Iterator::next by typing in its function signature) and you couldn't explicitly search for them. This PR fixes both of these problems.
# Guide-level explanation
*Excerpt from [the Rustdoc book](http://notriddle.com/rustdoc-html-demo-5/associated-types/rustdoc/read-documentation/search.html), included in this PR.*
> Function signature searches can query generics, wrapped in angle brackets, and traits will be normalized like types in the search engine if no type parameters match them. For example, a function with the signature `fn my_function<I: Iterator<Item=u32>>(input: I) -> usize` can be matched with the following queries:
>
> * `Iterator<Item=u32> -> usize`
> * `Iterator<u32> -> usize` (you can leave out the `Item=` part)
> * `Iterator -> usize` (you can leave out iterator's generic entirely)
> * `T -> usize` (you can match with a generic parameter)
>
> Each of the above queries is progressively looser, except the last one would not match `dyn Iterator`, since that's not a type parameter.
# Reference-level explanation
Inside the angle brackets, you can choose whether to write a name before the parameter and the equal sign. This syntax is called [`GenericArgsBinding`](https://doc.rust-lang.org/reference/paths.html#paths-in-expressions) in the Rust Reference, and it allows you to constrain a trait's associated type.
As a convenience, you don't actually have to put the name in (Rust requires it, but Rustdoc Search doesn't). This works about the same way unboxing already works in Search: the terse `Iterator<u32>` is a match for `Iterator<Item=u32>`, but the opposite is not true, just like `u32` is a match for `Iterator<u32>`.
When converting a trait method for the search index, the trait is substituted for `Self`, and all associated types are bound to generics. This way, if you have the following trait definition:
```rust
pub trait MyTrait {
type Output;
fn method(self) -> Self::Output;
}
```
The following queries will match its method:
* `MyTrait<Output=T> -> T`
* `MyTrait<T> -> T`
* `MyTrait -> T`
But these queries will not match it:
* <i>`MyTrait<Output=u32> -> u32`</i>
* <i>`MyTrait<Output> -> Output`</i>
* <i>`MyTrait -> MyTrait::Output`</i>
# Drawbacks
It's a little bit bigger:
```console
$ du before/search-index1.74.0.js after/search-index1.74.0.js
4020 before/search-index1.74.0.js
4068 after/search-index1.74.0.js
```
# Rationale and alternatives
I don't want to just not do this. On it's own, it's not terribly useful, but in addition to searching by normal traits, this is also intended as a desugaring target for closures. That's why it needs to actually distinguish the two: it allows the future desugaring to distinguish function output and input.
The other alternative would be to not allow users to leave out the name, so `iterator<u32>` doesn't work. That would be unfortunate, because mixing up which ones have out params and which ones are plain generics is an easy enough mistake that the Rust compiler itself helps people out with it.
# Prior art
* <http://neilmitchell.blogspot.com/2020/06/hoogle-searching-overview.html>
The current Rustdoc algorithm, both before this PR and after it, has a fairly expensive matching algorithm over a fairly simple file format. Luckily, we aren't trying to scale to all of crates.io, so it's usable, but it's not great when I throw it at docs.servo.org
# Unresolved questions
Okay, but *how do we want to handle closures?* I know the system will desugar `FnOnce(T) -> U` into `trait:FnOnce<Output=U, primitive:tuple<T>>`, but what if I don't know what trait I'm looking for? This PR can merge with nothing, but it'd be nice to have a plan.
Specifically, how should the special form used to handle all varieties of basic callable: primitive:fn (function pointers), and trait:Fn, trait:FnOnce, and trait:FnMut should all be searchable using a single syntax, because I'm always forgetting which one is used in the function I'm looking for.
The essential question is how closely we want to copy Rust's own syntax. The tersest way to expression Option::map might be:
Option<T>, (T -> U) -> Option<U>
That's the approach I would prefer, but nobody's going to attempt it without being told, so maybe this would be better?
Option<T>, (fn(T) -> U) -> Option<U>
It does require double parens, but at least it's mostly unambiguous. Unfortunately, it looks like the syntax you'd use for function pointers, implying that if you specifically wanted to limit your search to function pointers, you'd need to use `primitive:fn(T) -> U`. Then again, searching is normally case-insensitive, so you'd want that anyway to disambiguate from `trait:Fn(T) -> U`.
# Future possibilities
## This thing really needs a ranking algorithm
That is, this PR increases the number of matches for some type-based queries. They're usually pretty good matches, but there's still more of them, and it's evident that if you have two functions, `foo(MyTrait<u8>)` and `bar(MyTrait<Item=u8>)`, if the user typed `MyTrait<u8>` then `foo` should show up first.
A design choice that these PRs have followed is that adding more stuff to the search query always reduces the number of functions that get matched. The advantage of doing it that way is that you can rank them by just counting how many atoms are in the function's signature (lowest score goes on top). Since it's impossible for a matching function to have fewer atoms than the search query, if there's a function with exactly the same set of atoms in it, then that'll be on top.
More complicated ranking algos tend to penalize long documents anyway, if the [distance metrics](https://www.benfrederickson.com/distance-metrics/?utm_source=flipboard&utm_content=other) I found through [Flipboard](https://flipboard.com/`@arnie0426/building-recommender-systems-nvue3iqtgrn10t45)` (and postgresql's `ts_rank_cd`) are anything to go by. Real-world data sets tend to have weird outliers, like they have God Functions with zillions of arguments of all sorts of types, and Rustdoc ought to put such a function at the bottom.
The other natural choice would be interleaving with `unifyFunctionTypes` to count the number of unboxings and reorderings. This would compute a distance function, and would do a fine job of ranking the results, as [described here](https://ndmitchell.com/downloads/slides-hoogle_finding_functions_from_types-16_may_2011.pdf) by the Hoogle dev, but is more complicated than it sounds. The current algorithm returns when it finds a result that *exists at all*, but a distance function should find an *optimal solution* to find the smallest sequence of edits.
## This could also use a benchmark suite and some optimization
This approach also lends itself to layering a bloom filter in front of the backtracking unification engine.
* At load time, hash the typeNameIdMap ID for each atom and set the matching entry in a fixed-size byte array for each function to 1. Call it `fnType.bloomFilter`
* At search time, do the same for the atoms in the query (excluding special forms like `[]` that can match more than one thing). Call it `parsedQuery.bloomFilter`
* For each function, `if (fnType.bloomFilter | (~parsedQuery.bloomFilter) !== ~0) { return false; }`
There's also room to optimize the unification engine itself, by using stacks and persistent data structures instead of copying arrays around, or by using hashing instead of linear scans (the current algorithm was rewritten from one that tried to do that, but was too much to fit in my head and had a bunch of bugs). The advantage of Just Backtracking Better over the bloom filter is that it doesn't require the engine to retain any special algebraic properties.
But, first, we need a set of benchmarks to be able to judge if such a thing will actually help.
## Referring to associated types by path
*I don't want to implement this one, but if I did, this is how I'd do it.*
In Rust, this is represented by a structure called a qualified path, or QPath. They look like this:
<Self as Iterator>::Item
<F as FnOnce>::Output
They can also, if it's unambiguous, use a plain path and just let the system figure it out:
Self::Item
F::Output
In Rustdoc Type-Driven Search, we don't want to force people to be unambiguous. Instead, we should try *all reasonable interpretations*, return results whenever any of them match, and let users make their query more specific if too many results are matches.
To enable associated type path searches in Rustdoc, we need to:
1. When lowering a trait method to a search-index.js function signature, Self should be explicitly represented as a generic argument. It should always be assigned `-1`, so that if the user uses `Self` in their search query, we can ensure it always matches the real Self and not something else. Any functions that don't *have* a Self should drop a `0` into the first position of the where clause, to express that there isn't one and reserve the `-1` position.
* Reminder: generics are negative, concrete types are positive, and zero is a reserved sentinel.
* Right now, `Iterator::next` is lowered as if it were `fn next<T>(self: Iterator<Item=T>) -> Option<T>`.
It should become `fn next<Self, T>(self: Self) -> Option<T> where Self: Iterator<Item=T>` instead.
3. Add another backtracking edge to the unification engine, so that when the user writes something like `some::thing`, the interpretation where `some` is a module and `thing` is a standalone item becomes one possible match candidate, while the interpretation where `some` is a trait and `thing` is an associated type is a separate match candidate. The backtracking engine is basically powerful enough to do this already, since unboxing generic type parameters into their traits already requires the ability to do this kind of thing.
* When interpreting `some::thing` where `some` is a trait and `thing` is an associated type, it should be treated equivalently to `<self as some>::thing`. If you want to bind it to some generic parameter other than `Self`, you need to explicitly say so.
* If no trait called `some` actually exists, treat it as a generic type parameter instead. Track every trait mentioned in the current working function signature, and add a match candidate for each one.
* A user that explicitly wants the trait-associated-type interpretation could write a qpath (like `<self as trait>::type`), and a user that explicitly wants the module-item interpretation should use an item type filter (like `struct:module::type`).
4. To actually do the matching, maintain a `Map<(QueryGenericParamId, TraitId), FnGenericParamId>` alongside the existing `Map<QueryGenericParamId, FnGenericParamId>` that is already used to handle plain generic parameters. This works, because, when a trait function signature is lowered to search-index.js, the `rustdoc` backend always generates an FnGenericParamId for every trait associated type it sees mentioned in the function's signature.
5. Parse QPaths. Specifically,
* QueryElem adds three new fields. `isQPath` is a boolean flag, and `traitNameId` contains an entry for `typeNameIdMap` corresponding to the trait part of a qpath, and `parentId` may contain either a concrete type ID or a negative number referring to a generic type parameter. The actual `id` of the query elem will always be a negative number, because this is essentially a funny way to add a generic type constraint.
* If it's a QPath, then both of those IDs get filled in with the respective parts of the map. The unification engine will check the where clause to ensure the trait actually applies to the generic parameter in question, will check the type parameter constraint, and will add a mapping to `mgens` recording this as a solution.
* If it's just a regular path, then `isQPath` is false, and the parser will fill in both `traitNameId` and `parentId` based on the same path. The unification engine, seeing isQPath is false and that these IDs were filled in, will try all three solutions: the path might be part of a concrete type name, or it might be referring to a trait, or it might be referring to a generic type parameter.
### Why not implement QPath searches?
I'm not sure if anybody really wants to write such complicated queries. You can do a pretty good job of describing the generic functions in the standard library without resorting to FQPs.
These two queries, for example, would both match the Iterator::map function if we added support for higher order function queries and a rule that allows a type to match its *notable traits*.
// I like this version, because it's identical to how `Option::map` would be written.
// There's a reason why Iterator::map and Option::map have the same name.
Iterator<T>, (T -> U) -> Iterator<U>
// This version explicitly uses the type parameter constraints.
Iterator<Item=T>, (T -> U) -> Iterator<Item=U>
If I try to write this one using FQP, however, the results seem worse:
// This one is less expressive than the versions that don't use associated type paths.
// It matches `Iterator::filter`, while the above two example queries don't.
Iterator, (Iterator::Item -> Iterator::Item) -> Iterator
// This doesn't work, because the return type of `Iterator::map` is not a generic
// parameter with an `Iterator` trait bound. It's a concrete type that
// implements `Iterator`. Return-Position-Impl-Trait is the same way.
//
// There's a difference between something like `map`, whose return value
// implements Iterator, and something like `collect`, where the caller
// gets to decide what the concrete type is going to be.
//Self, (Self::Item -> I::Item) -> I where Self: Iterator, I: Iterator
// This works, but it seems subjectively ugly, complex, and counterintuitive to me.
Self, (<Self as Iterator>::Item -> T) -> Iterator<Item=T>
Add `$message_type` field to distinguish json diagnostic outputs
Currently the json-formatted outputs have no way to unambiguously determine which kind of message is being output. A consumer can look for specific fields in the json object (eg "message"), but there's no guarantee that in future some other kind of output will have a field of the same name.
This PR adds a `"type"` field to add json outputs which can be used to unambiguously determine which kind of output it is. The mapping is:
`diagnostic`: regular compiler diagnostics
`artifact`: artifact notifications
`future_incompat`: Future incompatibility report
`unused_extern`: Unused crate warnings/errors
This matches the "internally tagged" representation for serde enums.
Add place.ty() and Ty build from a kind to smir
Add a method to retrieve the type of a place and a few utility functions needed to build the projection type. I decided to return a result to avoid panicking if the user passes invalid inputs, such as wrong list of locals.
r? `@spastorino`
Don't consider regions in `deref_into_dyn_supertrait` lint
I actually wonder if we should just warn on *any* deref impl with a target type that matches a supertrait by *def-id*.
cc #89460
r? types
Typeck break expr even if break is illegal
Fixes#117821
We were returning immediately when encountering an illegal break. However, this caused problems later when the expr that the break was returning was evaluated during writeback. So now we don't return and instead simply set tainted by error. This lets typeck of break expr to occur even though we've encountered an illegal break.
Don't require intercrate mode for negative coherence
Negative coherence needs to be *sound*, but does not need to be *complete*, since it's looking for the *existence* of a negative goal, not the non-existence of a positive goal.
This removes some trivial and annoying ambiguities when a negative impl has region constraints.
r? lcnr idk if this needs an fcp but if it does, pls kick it off
Note about object lifetime defaults in does not live long enough error
This is a aspect of Rust that frequently trips up people who are not aware of it yet. This diagnostic attempts to explain what's happening and why the lifetime constraint, that was never mentioned in the source, arose.
The implementation feels a bit questionable, I'm not sure whether there are better ways to do this. There probably are.
fixes#117835
r? types
Add a test to ensure issue #89699 does not show up again. This test
emits an `async move` closure in a proc macro, which is used in a
test program compiled with edition 2015. We make sure the error message
is nice and shows up properly.
Rollup of 8 pull requests
Successful merges:
- #117828 (Avoid iterating over hashmaps in astconv)
- #117832 (interpret: simplify handling of shifts by no longer trying to handle signed and unsigned shift amounts in the same branch)
- #117891 (Recover `dyn` and `impl` after `for<...>`)
- #117957 (if available use a Child's pidfd for kill/wait)
- #117988 (Handle attempts to have multiple `cfg`d tail expressions)
- #117994 (Ignore but do not assume region obligations from unifying headers in negative coherence)
- #118000 (Make regionck care about placeholders in outlives components)
- #118068 (subtree update cg_gcc 2023/11/17)
r? `@ghost`
`@rustbot` modify labels: rollup
We were earlier returning immediately when encountering an illegal break. However, this caused problems later
when the expr that the break was returning was evaluated during writeback. So now we don't return and instead
simply set tainted by error. This lets typeck of break expr to occur even though we've encountered an illegal break.
Fix insertion of statements to be executed along return edge in inlining
Inlining creates additional statements to be executed along the return
edge: an assignment to the destination, storage end for temporaries.
Previously those statements where inserted directly into a call target,
but this is incorrect when the target has other predecessors.
Avoid the issue by creating a new dedicated block for those statements.
When the block happens to be redundant it will be removed by CFG
simplification that follows inlining.
Fixes#117355
Inlining creates additional statements to be executed along the return
edge: an assignment to the destination, storage end for temporaries.
Previously those statements where inserted directly into a call target,
but this is incorrect when the target has other predecessors.
Avoid the issue by creating a new dedicated block for those statements.
When the block happens to be redundant it will be removed by CFG
simplification that follows inlining.
Fixes#117355
This lint is not triggered if any of the following conditions are met:
- The user explicitly annotates the binding with the `()` type.
- The binding is from a macro expansion.
- The user explicitly wrote `let () = init;`
- The user explicitly wrote `let pat = ();`. This is allowed for local
lifetimes.
Make regionck care about placeholders in outlives components
Currently, we don't consider a placeholder type `!T` to be a type component when it comes to processing type-outlives obligations. This means that they are essentially treated like unit values with no sub-components, and always outlive any region. This is problematic for `non_lifetime_binders`, and even more problematic for `with_negative_coherence`, since negative coherence uses placeholders as universals.
This PR adds `Component::Placeholder` which acts much like `Component::Param`. This currently causes a regression in some non-lifetime-binders tests because `for<T> T: 'static` doesn't imply itself when processing outlives obligations, so code like this will fail:
```
fn foo() where for<T> T: 'static {
foo() //~ fails
}
```
Since the where clause doesn't imply itself. This requires making the `MatchAgainstHigherRankedOutlives` relation smarter when it comes to binders.
r? types
Ignore but do not assume region obligations from unifying headers in negative coherence
Partly addresses a FIXME that was added in #112875. Just as we can throw away the nested trait/projection obligations from unifying two impl headers, we can also just throw away the region obligations too.
I removed part of the FIXME that was incorrect, namely:
> Given that the only region constraints we get are involving inference regions in the root, it shouldn't matter, but still sus.
This is not true when unifying `fn(A)` and `for<'b> fn(&'b B)` which ends up with placeholder region outlives from non-root universes. I'm pretty sure this is okay, though it would be nice if we were to use them as assumptions. See the `explicit` revision of the test I committed, which still fails.
Fixes#117986
r? lcnr, feel free to reassign tho.
Recover `dyn` and `impl` after `for<...>`
Recover `dyn` and `impl` after `for<...>` in types. Reuses the logic for parsing bare trait objects, so it doesn't fix cases like `for<'a> dyn Trait + dyn Trait` or anything, but that seems somewhat of a different issue.
Parsing recovery logic is a bit involved, but I couldn't find a way to simplify it.
Fixes#117882
interpret: simplify handling of shifts by no longer trying to handle signed and unsigned shift amounts in the same branch
While we're at it, also update comments in codegen and MIR building related to shifts, and fix the overflow error printed by Miri on negative shift amounts.
When encountering struct fn call literal with private fields, suggest all builders
When encountering code like `Box(42)`, suggest `Box::new(42)` and *all* other associated functions that return `-> Box<T>`.
Add a way to give pre-sorted suggestions.
