solver cycles are coinductive once they have one coinductive step
Implements the new cycle semantics in the new solver, dealing with the fallout from https://github.com/rust-lang/trait-system-refactor-initiative/issues/10.
The first commit has been extensively fuzzed via https://github.com/lcnr/search_graph_fuzz.
A trait solver cycle is now coinductive if it has at least one *coinductive step*. A step is only considered coinductive if it's a where-clause of an impl of a coinductive trait. The only coinductive traits are `Sized` and auto traits.
This differs from the current stable because where a cycle had to consist of exclusively coinductive goals. This is overly limiting and wasn't properly enforced as it (mostly) ignored all non-trait goals.
A more in-depth explanation of my reasoning can be found in this separate doc: https://gist.github.com/lcnr/c49d887bbd34f5d05c36d1cf7a1bf5a5. A summary:
- imagine using dictionary passing style: map where-bounds to additional "dictonary" fn arguments instead of monomorphization
- impls are the only source of truth and introduce a *constructor* of the dictionary type
- a trait goal holds if mapping its proof tree to dictionary passing style results in a valid corecursive function
- a corecursive function is valid if it is guarded: matching on it should result in a constructor in a finite amount of time. This property should recursively hold for all fields of the constructor
- a function is guarded if the recursive call is *behind* a constructor
- **and** this constructor is not *moved out of*, e.g. by accessing a field of the dictionary
- the "not moved out of" condition is difficult to guarantee in general, e.g. for item bounds of associated types. However, there is no way to *move out* of an auto trait as there is no information you can get from *the inside of* an auto trait bound in the trait system
- if we encounter a cycle/recursive call which involves an auto trait, we can always convert the proof tree into a non-recursive function which calls a corecursive function whose first step is the construction of the auto trait dict and which only recursively depends on itself (by inlining the original function until they reach the uses of the auto trait)
**we can therefore make any cycle during which we step into an auto trait (or `Sized`) impl coinductive**
----
To fix https://github.com/rust-lang/trait-system-refactor-initiative/issues/10 we could go with a more restrictive version which tries to restrict cycles to only allow code already supported on stable, potentially forcing cycles to be ambiguous if they step through an impl-where clause of a non-coinductive trait.
`PathKind` should be a strictly ordered set to allow merging paths without worry. We could therefore add another variant `PathKind::ForceUnknown` which is greater than `PathKind::Coinductive`. We already have to add such a third `PathKind` in #137314 anyways.
I am not doing this here due to multiple reasons:
- I cannot think of a principled reason why cycles using an impl to normalize differ in any way from simply using that impl to prove a trait bound. It feels unnecessary and like it makes it more difficult to reason about our cycle semantics :<
- This PR does not affect stable as coherence doesn't care about whether a goal holds or is ambiguous. So we don't yet have to make a final decision
r? `@compiler-errors` `@nikomatsakis`
fix: overflowing bin hex
**Overview:**
- This PR fixes#135404.
**Testing**
- Tested the updated functionality.
- previously emitted diagnostics:
```bash
error: literal out of range for `i32`
--> src/main.rs:2:9
|
2 | _ = 0x8FFF_FFFF_FFFF_FFFE;
| ^^^^^^^^^^^^^^^^^^^^^
|
= note: the literal `0x8FFF_FFFF_FFFF_FFFE` (decimal `10376293541461622782`) does not fit into the type `i32` and will become `-2i32`
= help: consider using the type `i128` instead
= note: `#[deny(overflowing_literals)]` on by default
help: to use as a negative number (decimal `-2`), consider using the type `u32` for the literal and cast it to `i32`
|
2 | _ = 0x8FFF_FFFF_FFFF_FFFEu32 as i32;
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
```
- current diagnostics:
```bash
error: literal out of range for `i32`
--> ../temp.rs:2:13
|
2 | let x = 0x8FFF_FFFF_FFFF_FFFE;
| ^^^^^^^^^^^^^^^^^^^^^
|
= note: the literal `0x8FFF_FFFF_FFFF_FFFE` (decimal `10376293541461622782`) does not fit into the type `i32` and will become `-2i32`
= help: consider using the type `u64` instead
= note: `#[deny(overflowing_literals)]` on by default
help: to use as a negative number (decimal `-2`), consider using the type `u64` for the literal and cast it to `i32`
|
2 | let x = 0x8FFF_FFFF_FFFF_FFFEu64 as i32;
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
```
Rollup of 8 pull requests
Successful merges:
- #136542 ([`compiletest`-related cleanups 4/7] Make the distinction between root build directory vs test suite specific build directory in compiletest less confusing)
- #136579 (Fix UB in ThinVec::flat_map_in_place)
- #136688 (require trait impls to have matching const stabilities as the traits)
- #136846 (Make `AssocOp` more like `ExprKind`)
- #137304 (add `IntoBounds::intersect` and `RangeBounds::is_empty`)
- #137455 (Reuse machinery from `tail_expr_drop_order` for `if_let_rescope`)
- #137480 (Return unexpected termination error instead of panicing in `Thread::join`)
- #137694 (Spruce up `AttributeKind` docs)
r? `@ghost`
`@rustbot` modify labels: rollup
Don't infer attributes of virtual calls based on the function body
Fixes (after backport) #137646.
Since we don't know the exact implementation of the virtual call, it might write to parameters, we can't infer the readonly attribute.
Reuse machinery from `tail_expr_drop_order` for `if_let_rescope`
Namely, it defines its own `extract_component_with_significant_dtor` which is a bit more accurate than `Ty::has_significant_drop`, since it has a hard-coded list of types from the ecosystem which are opted out of the lint.[^a]
Also, since we extract the dtors themselves, adopt the same *label* we use in `tail_expr_drop_order` to point out the destructor impl. This makes it much clear what's actually being dropped, so it should be clearer to know when it's a false positive.
This conflicts with #137444, but I will rebase whichever lands first.
[^a]: Side-note, it's kinda a shame that now there are two functions that presumably do the same thing. But this isn't my circus, nor are these my monkeys.
Make -Z unpretty=mir suggest -Z dump-mir as well for discoverability
While debugging something else, I got quite annoyed with `-Z unpretty=mir` showing me post-processed MIR instead of the one just after it is built. I ended up asking on Zulip and got pointed to `-Z dump-mir`. While this feature is documented in the rustc dev guide, I think it'd be good if the possibility of making use of it was staring you in the face while you need it.
Don't suggest constraining unstable associated types
Fixes#137624
This could be made a bit more specific, considering the local crate's stability or nightly status or something, but I think in general we should not be suggesting associated type bounds on unstable associated items.
Teach structured errors to display short `Ty<'_>`
Make it so that in every structured error annotated with `#[derive(Diagnostic)]` that has a field of type `Ty<'_>`, the printing of that value into a `String` will look at the thread-local storage `TyCtxt` in order to shorten to a length appropriate with the terminal width. When this happen, the resulting error will have a note with the file where the full type name was written to.
```
error[E0618]: expected function, found `((..., ..., ..., ...), ..., ..., ...)``
--> long.rs:7:5
|
6 | fn foo(x: D) { //~ `x` has type `(...
| - `x` has type `((..., ..., ..., ...), ..., ..., ...)`
7 | x(); //~ ERROR expected function, found `(...
| ^--
| |
| call expression requires function
|
= note: the full name for the type has been written to 'long.long-type-14182675702747116984.txt'
= note: consider using `--verbose` to print the full type name to the console
```
Follow up to and response to the comments on #136898.
r? ``@oli-obk``
hir_analysis: skip self type of host effect preds in variances_of
Discovered as part of an implementation of rust-lang/rfcs#3729 - w/out this then when introducing const trait bounds: many more interesting tests change with different output, missing errors, new errors, etc related to this but they all depend on feature flags and are much more complex than this test.
r? ``@oli-obk``
trait_sel: resolve vars in host effects
In the standard library, the `Extend` impl for `Iterator` (specialised with `TrustedLen`) has a parameter which is constrained by a projection predicate. This projection predicate provides a value for an inference variable but - if the default bound is `const Sized` instead of `Sized` - host effect evaluation wasn't resolving variables first. Added a test that doesn't depend on a rust-lang/rfcs#3729 implementation.
Adding the extra resolve can the number of errors in some tests when they gain host effect predicates, but this is not unexpected as calls to `resolve_vars_if_possible` can cause more error tainting to happen.
run some tests on emscripten again
these were ignored because of #45351, but that issue has long been fixed. Let's see if these pass, or if there is some issue lurking still
I believe this is the try-job for emscripten? probably a good idea to run that first.
~~try-job: test-various~~
try-job: dist-various-1
Rollup of 8 pull requests
Successful merges:
- #137370 (adjust_abi: make fallback logic for ABIs a bit easier to read)
- #137444 (Improve behavior of `IF_LET_RESCOPE` around temporaries and place expressions)
- #137464 (Fix invalid suggestion from type error for derive macro)
- #137539 ( Add rustdoc-gui regression test for #137082 )
- #137576 (Don't doc-comment BTreeMap<K, SetValZST, A>)
- #137595 (remove `simd_fpow` and `simd_fpowi`)
- #137600 (type_ir: remove redundant part of comment)
- #137602 (feature: fix typo in attribute description)
r? `@ghost`
`@rustbot` modify labels: rollup
```
error[E0614]: type `(..., ..., ..., ...)` cannot be dereferenced
--> $DIR/long-E0614.rs:10:5
|
LL | *x;
| ^^ can't be dereferenced
|
= note: the full name for the type has been written to '$TEST_BUILD_DIR/$FILE.long-type-hash.txt'
= note: consider using `--verbose` to print the full type name to the console
```
Make it so that every structured error annotated with `#[derive(Diagnostic)]` that has a field of type `Ty<'_>`, the printing of that value into a `String` will look at the thread-local storage `TyCtxt` in order to shorten to a length appropriate with the terminal width. When this happen, the resulting error will have a note with the file where the full type name was written to.
```
error[E0618]: expected function, found `((..., ..., ..., ...), ..., ..., ...)``
--> long.rs:7:5
|
6 | fn foo(x: D) { //~ `x` has type `(...
| - `x` has type `((..., ..., ..., ...), ..., ..., ...)`
7 | x(); //~ ERROR expected function, found `(...
| ^--
| |
| call expression requires function
|
= note: the full name for the type has been written to 'long.long-type-14182675702747116984.txt'
= note: consider using `--verbose` to print the full type name to the console
```
remove `simd_fpow` and `simd_fpowi`
Discussed in https://github.com/rust-lang/rust/issues/137555
These functions are not exposed from `std::intrinsics::simd`, and not used anywhere outside of the compiler. They also don't lower to particularly good code at least on the major ISAs (I checked x86_64, aarch64, s390x, powerpc), where the vector is just spilled to the stack and scalar functions are used for the actual logic.
r? `@RalfJung`
Add rustdoc-gui regression test for #137082
Fixes https://github.com/rust-lang/rust/issues/137082.
Added new commands in `browser-ui-test` allowing us to add a regression test for #137082 and also another to copy code examples content.
r? `@notriddle`
Improve behavior of `IF_LET_RESCOPE` around temporaries and place expressions
Heavily reworks the `IF_LET_RESCOPE` to be more sensitive around 1. temporaries that get consumed/terminated and therefore should not trigger the lint, and 2. borrows of place expressions, which are not temporary values.
Fixes#137411
Make `#[used]` work when linking with `ld64`
To make `#[used]` work in static libraries, we use the `symbols.o` trick introduced in https://github.com/rust-lang/rust/pull/95604.
However, the linker shipped with Xcode, ld64, works a bit differently from other linkers; in particular, [it completely ignores undefined symbols by themselves](https://github.com/apple-oss-distributions/ld64/blob/ld64-954.16/src/ld/parsers/macho_relocatable_file.cpp#L2455-L2468), and only consider them if they have relocations (something something atoms something fixups, I don't know the details).
So to make the `symbols.o` file work on ld64, we need to actually insert a relocation. That's kinda cumbersome to do though, since the relocation must be valid, and hence must point to a valid piece of machine code, and is hence very architecture-specific.
Fixes https://github.com/rust-lang/rust/issues/133491, see that for investigation.
---
Another option would be to pass `-u _foo` to the final linker invocation. This has the problem that `-u` causes the linker to not be able to dead-strip the symbol, which is undesirable. (If we did this, we would possibly also want to do it by putting the arguments in a file by itself, and passing that file via ``@`,` e.g. ``@undefined_symbols.txt`,` similar to https://github.com/rust-lang/rust/issues/52699, though that [is only supported since Xcode 12](https://developer.apple.com/documentation/xcode-release-notes/xcode-12-release-notes#Linking), and I'm not sure we wanna bump that).
Various other options that are probably all undesirable as they affect link time performance:
- Pass `-all_load` to the linker.
- Pass `-ObjC` to the linker (the Objective-C support in the linker has different code paths that load more of the binary), and instrument the binaries that contain `#[used]` symbols.
- Pass `-force_load` to libraries that contain `#[used]` symbols.
Failed attempt: Embed `-u _foo` in the object file with `LC_LINKER_OPTION`, akin to https://github.com/rust-lang/rust/issues/121293. Doesn't work, both because `ld64` doesn't read that from archive members unless it already has a reason to load the member (which is what this PR is trying to make it do), and because `ld64` only support the `-l`, `-needed-l`, `-framework` and `-needed_framework` flags in there.
---
TODO:
- [x] Support all Apple architectures.
- [x] Ensure that this works regardless of the actual type of the symbol.
- [x] Write up more docs.
- [x] Wire up a few proper tests.
`@rustbot` label O-apple
Don't immediately panic if dropck fails without returning errors
This span_bug was a little too optimistic. I've decided that matching on the ErrorGuaranteed is a little more sensible than a delay bug that will always be ignored.
closes#137329
r? `@compiler-errors`
