Fix#58270, fix off-by-one error in error diagnostics.
This fixes#58270 by checking if two diagnostics overlap completely when we're calculating the line offset for each message.
Add codegen test for PGO instrumentation.
This PR adds a codegen test that makes sure that LLVM actually generates instrumentation code when we enable PGO instrumentation in `rustc`.
The second commit updates a test case to the new commandline option syntax introduced in #59874. Without the fix the test still works, but it confusingly creates a directory called `test.profraw`, which usually is the name of the _file_ where profiling data is collected.
tweak unresolved label suggestion
Only suggest label names in the same hygiene context, and use a
structured suggestion.
Question for reviewer: Is this the right way to check for label hygiene?
Future-proof MIR for dedicated debuginfo.
This is #56231 without the last commit (the one that actually moves to `VarDebuginfo`).
Nothing should be broken, but it should no longer depend on debuginfo for anything else.
r? @nikomatsakis
Replace HashMap implementation with SwissTable (as an external crate)
This is the same as #56241 except that it imports `hashbrown` as an external crate instead of copying the implementation into libstd.
This includes a few API changes (all unstable):
- `try_reserve` is added to `HashSet`.
- Some trait bounds have been changed in the `raw_entry` API.
- `search_bucket` has been removed from the `raw_entry` API (doesn't work with SwissTable).
Warn when ignore-tidy-linelength is present, but no lines are too long
It's easy for a `// ignore-tidy-linelength` to be added when there is a genuine need to ignore a file's line length, but then after refactoring the need is gone, but the tidy directive is not removed. This means that in the future, further editing may accidentally add unnecessarily long lines. This change forces `// ignore-tidy-linelength` to be used exactly when necessary, to make sure such changes are intentional.
Warn on unused results for operation methods on nums
From a suggestion by @llogiq
Adds a `#[must_use]` attribute to operation methods on integers that take self by value as the first operand and another value as the second. It makes it clear that these methods return the result of the operation instead of mutating `self`, which is the source of a rather embarrassing bug I had in a codebase of mine recently...
As an example:
```rust
struct Int {
value: i64,
}
impl Int {
fn add(&mut self, other: i64) {
self.value.wrapping_add(other);
}
}
```
Will produce a warning like:
```
warning: unused return value of `core::num::<impl i64>::wrapping_add` that must be used
--> src/main.rs:7:7
|
7 | self.value.wrapping_add(other);
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
= note: #[warn(unused_must_use)] on by default
= note: this returns the result of the operation, without modifying the original
```
If this is something we're on board with, we could do something similar for `f32` and `f64` too. There are probably other methods on integers that make sense.
[wg-async-await] Drop `async fn` arguments in async block
Fixes#54716.
This PR modifies the HIR lowering (and some other places to make this work) so that unused arguments to a async function are always dropped inside the async move block and not at the end of the function body.
```
async fn foo(<pattern>: <type>) {
async move {
}
} // <-- dropped as you "exit" the fn
// ...becomes...
fn foo(__arg0: <ty>) {
async move {
let <pattern>: <ty> = __arg0;
} // <-- dropped as you "exit" the async block
}
```
However, the exact ordering of drops is not the same as a regular function, [as visible in this playground example](https://play.rust-lang.org/?version=stable&mode=debug&edition=2015&gist=be39af1a58e5d430be1eb3c722cb1ec3) - I believe this to be an unrelated issue. There is a [Zulip topic](https://rust-lang.zulipchat.com/#narrow/stream/187312-t-compiler.2Fwg-async-await/topic/.2354716.20drop.20order) for this.
r? @cramertj
cc @nikomatsakis
This commit introduces a `assert_drop_order_after_poll` helper function
to the test case for this case to reduce repetitive noise and documents
what each function aims to test.
Don't stop evaluating due to errors before borrow checking
r? @oli-obk
Fix#60005. Follow up to #59903. Blocked on #53708, fixing the ICE in `src/test/ui/consts/match_ice.rs`.
Disallow double trailing newlines in tidy
This wasn't done previously in https://github.com/rust-lang/rust/pull/47064#issuecomment-354533010 as it affected too many files, but I think it's best to fix it now so that the number of files with double trailing newlines doesn't keep increasing.
r? kennytm
This commit makes the suggestion to dereference when a type implements
`Deref` only apply if the dereference would succeed (ie. the type is
`Copy`, otherwise a borrow check error would occur).
Enable NLL migrate mode on the 2015 edition
## What is in this PR?
* Remove the `-Zborrowck=ast` flag option from rustc.
* The default in the 2015 edition is now `-Zborrowck=migrate`.
* The 2018 edition default is unchanged: it's still `-Zborrowck=migrate`.
* Enable two-phase borrows (currently toggled via the `-Ztwo-phase-borrows` flag) on all editions.
* Remove most dead code that handled these options.
* Update tests for the above changes.
## What is *not* in this PR?
These are left for future PRs
* Use `-Zborrowck=mir` in NLL compare mode tests (#56993)
* Remove the `-Zborrowck=compare` option (#59193)
* Remove the `-Ztwo-phase-borrows` flag. It's kept, as a flag that does nothing so that perf.rlo has time to stop using it (cc @Mark-Simulacrum)
* Remove MIR typeck as its own MIR pass - it's now run by NLL.
* Enabling `-Zborrowck=mir` by default (#58781)
* Replace `allow_bind_by_move_patterns_with_guards` and `check_for_mutation_in_guard_via_ast_walk` with just using the feature gate. (#59192)
Soundness issues that are fixed by NLL will stay open until full NLL is emitting hard errors. However, these diagnostics and completeness issues can now be closed:
Closes#18330Closes#22323Closes#23591Closes#26736Closes#27487Closes#28092Closes#28970Closes#29733Closes#30104Closes#38915Closes#39908Closes#43407Closes#47524Closes#48540Closes#49073Closes#52614Closes#55085Closes#56093Closes#56496Closes#57804
cc #43234
r? @pnkfelix
cc @rust-lang/lang
cc @rust-lang/wg-compiler-nll
save-analysis: Use serde instead of libserialize to dump JSON data
This breaks the save-analysis infrastructure (which also includes `rls-{analysis, data, span}` crates) from depending on rustc_serialize and so we can start moving them to being supported on stable without implementing `Decodable` et al. by hand for data structures defined there.
Notable benefits:
- we drop the awkward raw byte `PathBuf` [serialization](https://gist.github.com/Xanewok/f4fe8564d0dc0c3ab1dbc244279ff895) (until now (de)serialized as `&[u8]`)
- [faster](https://github.com/serde-rs/json-benchmark) (hopefully noticeable for inner crate dependencies for the RLS workloads)
- we can easily explore the binary serialization backend (which we planned to do for save-analysis anyway)
~This should be merged together with an update to RLS (https://github.com/rust-lang/rls/pull/1435), which technically could be included right now because we can use the bundled `rls-analysis` here directly, however I'd prefer to publish this to crates.io first (https://github.com/rust-lang/rls/pull/1434, cc @nrc) and use the published version, instead.~
Includes https://github.com/rust-lang/rls/pull/1436.
@matklad @nikomatsakis This is also important for the potential RLS 1.0 - 2.0 bridge we talked about on Zulip today
This commit takes advantage of `AsyncArgument` type that was added in a
previous commit to replace the arguments of the `async fn` in the HIR
and add statements to move the bindings from the new arguments to the
pattern from the old argument.
For example, the async function `foo` below:
async fn foo((x, _y): (T, V)) {
async move {
}
}
becomes:
async fn foo(__arg0: (T, V)) {
async move {
let (x, _y) = __arg0;
}
}
