| Age | Commit message (Collapse) | Author | Lines |
|---|
|
find the generic container rather than simply looking up for the assoc with const arg
Fixes #132534
This issue is caused by mismatched generic parameters. Previously, it tried to find `T` in `trait X`, but after this change, it will find `T` in `fn a`.
r? `@compiler-errors` as this assertion was introduced by you.
|
|
r=compiler-errors
Use backticks instead of single quotes for library feature names in diagnostics
This PR changes the text of library feature errors for using unstable or body-unstable items. Displaying library feature names in backticks is consistent with other diagnostics (e.g. those from `rustc_passes`) and with the `reason`s on unstable attributes in the library. Additionally, this simplifies diagnostics when supporting multiple unstable attributes on items (see #131824) since `DiagSymbolList` also displays symbols using backticks.
|
|
Register `~const` preds for `Deref` adjustments in HIR typeck
This doesn't *do* anything yet, since `Deref` and `DerefMut` aren't constified, and we explicitly don't error on calling non-const trait methods in HIR yet -- presumably that will wait until std is re-constified. But I'm confident this logic is correct, and this (afaict?) is the only major hole left in enforcing `~const` in HIR typeck.
r? fee1-dead
|
|
|
|
const arg
|
|
Reduce dependence on the target name
The target name can be anything with custom target specs. Matching on fields inside the target spec is much more robust than matching on the target name.
Also remove the unused is_builtin target spec field.
|
|
Add bad-reg inline assembly ui test for RISC-V and s390x
https://github.com/rust-lang/rust/pull/131341#discussion_r1826555431
> Btw, such unsupported registers are present in most architectures, but only aarch64/arm64ec, x86_64, and not yet merged [sparc/sparc64](https://github.com/rust-lang/rust/pull/132472/files#diff-02aebda3376c2b020265137f9ce2c387669ca5cfecd7d60494275c2387db5114) (and powerpc/powerpc64 by this PR) currently have ui tests for them. I plan to add tests for other arches later.
Starting with RISC-V and s390x, which I'm familiar with and relatively easy to check for correctness.
(Relevant rustc code are supported_types/def_regs/overlapping_regs in [compiler/rustc_target/src/asm/riscv.rs](https://github.com/rust-lang/rust/blob/588a4203508ed7c76750c96b482641261630ed36/compiler/rustc_target/src/asm/riscv.rs) and [compiler/rustc_target/src/asm/s390x.rs](https://github.com/rust-lang/rust/blob/588a4203508ed7c76750c96b482641261630ed36/compiler/rustc_target/src/asm/s390x.rs).)
r? workingjubilee
`@rustbot` label +A-inline-assembly
|
|
This commit can be reverted once the error message changes are in the stage 0 compiler.
|
|
library feature"
This is consistent with all other diagnostics I could find containing
features and enables the use of `DiagSymbolList` for generalizing
diagnostics for unstable library features to multiple features.
|
|
It is unused.
|
|
|
|
|
|
|
|
Rollup of 15 pull requests
Successful merges:
- #129329 (Implement `From<&mut {slice}>` for `Box/Rc/Arc<{slice}>`)
- #131377 (Add LowerExp and UpperExp implementations to NonZero)
- #132393 (Docs: added brief colon explanation)
- #132437 (coverage: Regression test for inlining into an uninstrumented crate)
- #132499 (unicode_data.rs: show command for generating file)
- #132503 (better test for const HashMap; remove const_hash leftovers)
- #132511 (stabilize const_arguments_as_str)
- #132520 (NFC add known bug nr to test)
- #132522 (make codegen help output more consistent)
- #132523 (Added regression test for generics index out of bounds)
- #132528 (Use `*_opt` typeck results fns to not ICE in fallback suggestion)
- #132537 (PassWrapper: adapt for llvm/llvm-project@5445edb5d)
- #132540 (Do not format generic consts)
- #132543 (add and update some crashtests)
- #132550 (compiler: Continue introducing rustc_abi to the compiler)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
Port most of `--print=target-cpus` to Rust
The logic and formatting needed by `--print=target-cpus` has historically been carried out in C++ code. Originally it used `printf` to write directly to the console, but later it switched over to writing to a `std::ostringstream` and then passing its buffer to a callback function pointer.
This PR replaces that C++ code with a very simple function that writes a list of CPU names to a `&RustString`, with the rest of the logic and formatting being handled by ordinary safe Rust code.
|
|
add and update some crashtests
Adds tests for #126268 and #132126, updates it for #131050.
|
|
Use `*_opt` typeck results fns to not ICE in fallback suggestion
Self-explanatory. Fixes #132517.
|
|
Added regression test for generics index out of bounds
Added a regression test for #117446
This ICE was fixed in Rust 1.75 but a regression test was never added.
This PR adds a UI test with a reduced version of the original bug report that does not rely on external crates.
|
|
NFC add known bug nr to test
r? ```````@jieyouxu```````
|
|
coverage: Regression test for inlining into an uninstrumented crate
Regression test for #132395, after I was able to figure out a simple way to reproduce it. See also #132436.
In addition to confirming that there is no ICE, this test also demonstrates that the affected code is undercounted, because executing the inlined copy doesn't increment coverage counters.
|
|
This also adds comments explaining the difference to bad-reg.rs.
|
|
Co-authored-by: Jubilee <workingjubilee@gmail.com>
|
|
Add `--print host-tuple` to print host target tuple
People often parse `-vV` output to get to the host tuple, which is annoying to do. It's easier to just get it directly.
I called it "host-tuple" instead of "host" because it's clearer that it's just the target name. I'm open to different names, but I think this one is fine.
a quick GitHub search for `'^host` reveals many instances of people doing the parsing, for example:
https://github.com/japaric/xargo/blob/68e0ca57cd90837fe02f262f074182f9cfeb6227/README.md?plain=1#L369
https://github.com/taiki-e/setup-cross-toolchain-action/blob/0e38473b0c562d6db19a98d3ec20a80f7ac189ae/main.sh#L96
https://github.com/taiki-e/cargo-llvm-cov/blob/8a3553b86551eabf9c30c060b1f72a5bbccb98c6/README.md?plain=1#L625
https://github.com/SiliconLabs/cpc-nvm3/blob/43f3ec39709b30700ef7f39d91fa647974323bf1/do.sh#L35
needs a compiler FCP. I could also do an MCP but I think just an FCP here makes the most sense.
|
|
|
|
People often parse `-vV` output to get to the host triple, which is
annoying to do. It's easier to just get it directly.
|
|
|
|
Tweak E0277 output when a candidate is available
*Follow up to #132086.*
Go from
```
error[E0277]: the trait bound `Then<Ignored<chumsky::combinator::Filter<chumsky::primitive::Any<&str, chumsky::extra::Full<EmptyErr, (), ()>>, {closure@src/main.rs:9:17: 9:27}>, char>, chumsky::combinator::Map<impl CSTParser<'a, O>, O, {closure@src/main.rs:11:24: 11:27}>, (), (), chumsky::extra::Full<EmptyErr, (), ()>>: CSTParser<'a>` is not satisfied
--> src/main.rs:7:50
|
7 | fn leaf<'a, O>(parser: impl CSTParser<'a, O>) -> impl CSTParser<'a, ()> {
| ^^^^^^^^^^^^^^^^^^^^^^ the trait `chumsky::private::ParserSealed<'_, &str, (), chumsky::extra::Full<EmptyErr, (), ()>>` is not implemented for `Then<Ignored<Filter<Any<&str, ...>, ...>, ...>, ..., ..., ..., ...>`, which is required by `Then<Ignored<chumsky::combinator::Filter<chumsky::primitive::Any<&str, chumsky::extra::Full<EmptyErr, (), ()>>, {closure@src/main.rs:9:17: 9:27}>, char>, chumsky::combinator::Map<impl CSTParser<'a, O>, O, {closure@src/main.rs:11:24: 11:27}>, (), (), chumsky::extra::Full<EmptyErr, (), ()>>: CSTParser<'a>`
|
= help: the trait `chumsky::private::ParserSealed<'_, &'a str, ((), ()), chumsky::extra::Full<EmptyErr, (), ()>>` is implemented for `Then<Ignored<chumsky::combinator::Filter<chumsky::primitive::Any<&str, chumsky::extra::Full<EmptyErr, (), ()>>, {closure@src/main.rs:9:17: 9:27}>, char>, chumsky::combinator::Map<impl CSTParser<'a, O>, O, {closure@src/main.rs:11:24: 11:27}>, (), (), chumsky::extra::Full<EmptyErr, (), ()>>`
= help: for that trait implementation, expected `((), ())`, found `()`
= note: required for `Then<Ignored<Filter<Any<&str, ...>, ...>, ...>, ..., ..., ..., ...>` to implement `Parser<'_, &str, ()>`
note: required for `Then<Ignored<Filter<Any<&str, ...>, ...>, ...>, ..., ..., ..., ...>` to implement `CSTParser<'a>`
--> src/main.rs:5:16
|
5 | impl<'a, O, T> CSTParser<'a, O> for T where T: Parser<'a, &'a str, O> {}
| ^^^^^^^^^^^^^^^^ ^ ---------------------- unsatisfied trait bound introduced here
= note: the full name for the type has been written to '/home/gh-estebank/longlong/target/debug/deps/longlong-0008f9a4f2023b08.long-type-13239977239800463552.txt'
= note: consider using `--verbose` to print the full type name to the console
= note: the full name for the type has been written to '/home/gh-estebank/longlong/target/debug/deps/longlong-0008f9a4f2023b08.long-type-13239977239800463552.txt'
= note: consider using `--verbose` to print the full type name to the console
```
to
```
error[E0277]: the trait bound `Then<Ignored<chumsky::combinator::Filter<chumsky::primitive::Any<&str, chumsky::extra::Full<EmptyErr, (), ()>>, {closure@src/main.rs:9:17: 9:27}>, char>, chumsky::combinator::Map<impl CSTParser<'a, O>, O, {closure@src/main.rs:11:24: 11:27}>, (), (), chumsky::extra::Full<EmptyErr, (), ()>>: CSTParser<'a>` is not satisfied
--> src/main.rs:7:50
|
7 | fn leaf<'a, O>(parser: impl CSTParser<'a, O>) -> impl CSTParser<'a, ()> {
| ^^^^^^^^^^^^^^^^^^^^^^ unsatisfied trait bound
...
11 | ws.then(parser.map(|_| ()))
| --------------------------- return type was inferred to be `Then<Ignored<..., ...>, ..., ..., ..., ...>` here
|
= help: the trait `ParserSealed<'_, &_, (), Full<_, _, _>>` is not implemented for `Then<Ignored<..., ...>, ..., ..., ..., ...>`
but trait `ParserSealed<'_, &'a _, ((), ()), Full<_, _, _>>` is implemented for it
= help: for that trait implementation, expected `((), ())`, found `()`
= note: required for `Then<Ignored<..., ...>, ..., ..., ..., ...>` to implement `Parser<'_, &str, ()>`
note: required for `Then<Ignored<..., ...>, ..., ..., ..., ...>` to implement `CSTParser<'a>`
--> src/main.rs:5:16
|
5 | impl<'a, O, T> CSTParser<'a, O> for T where T: Parser<'a, &'a str, O> {}
| ^^^^^^^^^^^^^^^^ ^ ---------------------- unsatisfied trait bound introduced here
= note: the full name for the type has been written to '/home/gh-estebank/longlong/target/debug/deps/longlong-df9d52be87eada65.long-type-1337037744507305372.txt'
= note: consider using `--verbose` to print the full type name to the console
```
* Remove redundant wording
* Introduce trait diff highlighting logic and use it
* Fix incorrect "long type written to path" logic (can be split off)
* Point at tail expression in more cases in E0277
* Avoid long primary span labels in E0277 by moving them to a `help`
Fix #132013.
There are individual commits that can be their own PR. If the review load is too big, happy to split them off.
|
|
Paths in CI can be longer than in devs' machines.
|
|
|
|
|
|
`f32`, `f64` and `asm!` macro.
|
|
|
|
|
|
|
|
|
|
|
|
rustdoc: skip stability inheritance for some item kinds
For some item kinds it's incorrect to inherit their parent's stability, because they might be accessible without referring to the parent directly -- This PR removes the stability inheritance for these items and reverts their displayed stability to that before https://github.com/rust-lang/rust/pull/130798.
Impl items, both inherent and trait impls, have a stability, but it is ignored when checking for enabled features. However, impl items are automatically unstable if they're nested inside an unstable module -- this caused the children of impl to inherit the instability and lead to https://github.com/rust-lang/rust/issues/132440.
Furthermore, for associated items only the stability of the associated item itself is checked and not that of its parent impl. This is true even for trait impls and we have [relied on this behavior in the standard library in the past](https://doc.rust-lang.org/1.37.0/std/slice/trait.SliceConcatExt.html#tymethod.concat), so these also shouldn't inherit the impl's stability.
I've also removed the stability inheritance for primitives and keywords so that viewing e.g. [the `i32` docs on `core`](https://doc.rust-lang.org/nightly/core/primitive.i32.html) will no longer show "since 1.6.0". Note that we currently don't annotate stability for the keyword docs, but if we start doing so in the future then this is probably more correct.
fixes (after backport) https://github.com/rust-lang/rust/issues/132440
|
|
Rollup of 11 pull requests
Successful merges:
- #131037 (Move versioned Apple LLVM targets from `rustc_target` to `rustc_codegen_ssa`)
- #132170 (Add a Few Codegen Tests)
- #132333 (rust_analyzer_helix.toml: add library/ manifest)
- #132398 (Add a couple of intra-doc links to str)
- #132411 (CI: switch dist-x86_64-musl to free runner)
- #132453 (Also treat `impl` definition parent as transparent regarding modules)
- #132457 (Remove needless #![feature(asm_experimental_arch)] from loongarch64 inline assembly test)
- #132465 (refactor(config): remove FIXME statement in comment of `omit-git-hash`)
- #132466 (Account for late-bound depth when capturing all opaque lifetimes.)
- #132471 (Add a bunch of mailmap entries)
- #132488 (Remove or fix some more `FIXME(async_closure)`)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
Remove or fix some more `FIXME(async_closure)`
Self-explanatory
|
|
Account for late-bound depth when capturing all opaque lifetimes.
Fixes https://github.com/rust-lang/rust/issues/132429
r? ````@compiler-errors````
|
|
Remove needless #![feature(asm_experimental_arch)] from loongarch64 inline assembly test
inline assembly is already stable on this architecture.
|
|
Also treat `impl` definition parent as transparent regarding modules
This PR changes the `non_local_definitions` lint logic to also consider `impl` definition parent as transparent regarding modules.
See tests and explanation in the changes.
``````@rustbot`````` label +L-non_local_definitions
Fixes *(after beta-backport)* #132427
cc ``````@leighmcculloch``````
r? ``````@jieyouxu``````
|
|
Add a Few Codegen Tests
Closes #86109
Closes #64219
Those issues somehow got fixed over time.
So, this PR adds a couple of codegen tests to ensure we don't regress in the future.
|
|
|
|
Update LLVM to 19.1.3
Closes #131031.
r? nikic
|
|
|
|
Do not suggest `#[derive(Copy)]` when we wanted a `!Copy` type.
Do not say "`Copy` is not implemented for `T` but `Copy` is".
Do not talk about `Trait` having no implementations when `!Trait` was desired.
|
|
Long span labels don't read well.
|
|
```
error[E0277]: the trait bound `{gen block@$DIR/gen_block_is_coro.rs:7:5: 7:8}: Coroutine` is not satisfied
--> $DIR/gen_block_is_coro.rs:6:13
|
LL | fn foo() -> impl Coroutine<Yield = u32, Return = ()> {
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ the trait `Coroutine` is not implemented for `{gen block@$DIR/gen_block_is_coro.rs:7:5: 7:8}`
LL | gen { yield 42 }
| ---------------- return type was inferred to be `{gen block@$DIR/gen_block_is_coro.rs:7:5: 7:8}` here
```
The secondary span label is new.
|
|
Remove default note for "trait is not implemented" in favor of the
more colorful diff output from the previous commit. Removes
duplicated output.
|
