fix [`cast_possible_truncation`] offering wrong suggestion for casting float to integer
fixes: #10366
---
changelog: [`cast_possible_truncation`] Fix incorrect suggestions when casting from float types or to `_`
Wrap `transmutes_expressible_as_ptr_casts` suggestions in parentheses
changelog: [`transmutes_expressible_as_ptr_casts`]: Fix suggestion missing wrapping parentheses
Fixes#10449
r? `@Jarcho`
Is this the best way to go about this? `unused_parens` will catch the unnecessary ones but emitting them in the first place isn't ideal
suggest `try_into` when casting to wildcard type;
fix [`cast_possible_truncation`] suggesting useless parenthesis;
remove suggesting for float to float conversion in [`cast_possible_truncation`]
style nit
New lint: detect unnecessary struct building
Fixes#10476.
Running this lint on the top 500 crates produced one hit (in `rust-lang/rust-bindgen`) and [a PR has been submitted there](https://github.com/rust-lang/rust-bindgen/pull/2440).
changelog: [`unnecessary_struct_initialization`]: new lint
Do not propose to simplify a not expression coming from a macro
Fixes#10523
changelog: FP [`nonminimal_bool`]: do not propose to change code coming from a macro
Do not propose to remove `async move` if variables are captured by ref
Fixes#10482
changelog: FP [`redundant_async_block`] Do not propose to remove `async move` if variables are captured by ref
rustc has proper heuristics for actually checking whether a type allows
being left uninitialized (by asking CTFE). We can now use this for our
helper instead of rolling our own bad version with false positives.
Issue function modifiers in the right order in manual_async_fn lint
Fixes#10450
changelog: [`manual_async_fn`] output function modifiers in correct order
fix `almost_swapped`: Ignore external macros
Fixes#10421 ; Related to #10499 (Fixing points *1* and *3* from #10421)
changelog: [`almost_swapped`]: Add a check to ignore external macros
Fix `almost_swapped` false positive (`let mut a = b; a = a`)
Fixes `2` in #10421
changelog: [`almost_swapped`]: Fix false positive when a variable is changed to itself. (`a = a`)
Remove `identity_future` indirection
This was previously needed because the indirection used to hide some unexplained lifetime errors, which it turned out were related to the `min_choice` algorithm.
Removing the indirection also solves a couple of cycle errors, large moves and makes async blocks support the `#[track_caller]`annotation.
Fixes https://github.com/rust-lang/rust/issues/104826.
enhance [`ifs_same_cond`] to warn same immutable method calls as well
fixes: #10272
---
changelog: Enhancement: [`ifs_same_cond`]: Now also detects immutable method calls.
[#10350](https://github.com/rust-lang/rust-clippy/pull/10350)
<!-- changelog_checked -->
Remove `box_syntax`
r? `@Nilstrieb`
This removes the feature `box_syntax`, which allows the use of `box <expr>` to create a Box, and finalises removing use of the feature from the compiler. `box_patterns` (allowing the use of `box <pat>` in a pattern) is unaffected.
It also removes `ast::ExprKind::Box` - the only way to create a 'box' expression now is with the rustc-internal `#[rustc_box]` attribute.
As a temporary measure to help users move away, `box <expr>` now parses the inner expression, and emits a `MachineApplicable` lint to replace it with `Box::new`
Closes#49733
Strengthen state tracking in const-prop
Some/many of the changes are replicated between both the const-prop lint and the const-prop optimization.
Behaviour changes:
- const-prop opt does not give a span to propagated values. This was useless as that span's primary purpose is to diagnose evaluation failure in codegen.
- we remove the `OnlyPropagateInto` mode. It was only used for function arguments, which are better modeled by a write before entry.
- the tracking of assignments and discriminants make clearer that we do nothing in `NoPropagation` mode or on indirect places.
Add utility macros to help with writing tests.
Adds two utility macros to help with testing:
* `external` expands to it's argument tokens, but makes them appear to come from an external macro. Helps make tests for `in_external_macro` much more readable.
* `inline_macros` is an attribute macro which allows the use of a pseudo `inline!` macro which expands to it's argument tokens, but makes them appear to be from a crate-local macro expansion. This removes the need to write `macro_rules` boilerplate when testing how lints interact with macros.
---
`external`'s usage is simple. `external!(struct Foo { x: u32});` will make the struct appear as though it came from an external macro. Individual tokens can be escaped if needed. `external!($x + 0 / 10)` will make everything except `x` appear as though it came from an external macro. Can also use `$literal` and `$(tokens...)` as well.
---
`inline_macros` is more complicated due to compiler constraints. Given:
```rust
#[inline_macros]
fn foo() {
inline!(5 + 5 / 10);
}
```
`inline!(5 + 5 / 10)` will be replace with a call to a generated macro which expands to the contained tokens.
Tokens can be escaped by prefixing them with `$`:
```rust
#[inline_macros]
fn foo() {
let x = 5;
inline!($x + 5 / $10);
}
```
This will pass `x` as an `ident` argument and `10` as a `literal` argument.
Token sequences can also be passed with `$(...)`:
```rust
#[inline_macros]
fn foo() {
let mut x = 5;
inline!(if $(x >= 5) {
$x = 5;
});
}
```
This will pass `x >= 5` as `tt` arguments, and `x` as an `ident` argument.
---
Not 100% sure `inline_macros` is actually worth having. It does make the tests a little easier to read once you're used to it and it becomes more useful once there are multiple macro tests. The verbosity of declaring single use macros starts to hurt at that point.
changelog: None
Improve diagnostic of `no_mangle_with_rust_abi`
fixes#10409
Pending rust-lang/rustfmt#5701
This rewords the message to focus on the error being an implicit ABI, rather than the `Rust` ABI. Also downgrades the suggestion to `MaybeIncorrect` and changes the suggestion span to better highlight the change.
---
changelog: None
<!-- changelog_checked -->
Migrate `write.rs` to `rustc_ast::FormatArgs`
changelog: none
Part 1 of #10233
The additions to `clippy_utils` are the main novelty of this PR, there's no removals yet since other parts still rely on `FormatArgsExpn`
The changes to `write.rs` itself are relatively straightforward this time around, as there's no lints in it that rely on type checking format params
r? `@flip1995`
Include async functions in the len_without_is_empty
fixes#7232
Changes done to the functionality:
Allowing different error types for the functions was disallowed. So the following was linted before but is not after this change
```
impl Foo {
pub len(&self) -> Result<usize, Error1> { todo!(); }
pub is_empty(&self) -> Result<bool, Error2> { todo!(); }
}
```
---
changelog: Enhancement: [`len_without_is_empty`]: Now also detects `async` functions
[#10359](https://github.com/rust-lang/rust-clippy/pull/10359)
<!-- changelog_checked -->
[arithmetic_side_effects] Fix#10252Fix#10252
At least for integers, shifts are already handled by the compiler.
----
changelog: [`arithmetic_side_effects`]: No longer lints on right or left shifts with constant integers, as the compiler warns about them.
[#10309](https://github.com/rust-lang/rust-clippy/pull/10309)
<!-- changelog_checked-->
This was previously needed because the indirection used to hide some unexplained lifetime errors, which it turned out were related to the `min_choice` algorithm.
Removing the indirection also solves a couple of cycle errors, large moves and makes async blocks support the `#[track_caller]` annotation.
Add `collection_is_never_read`
Fixes#9267
`@flip1995` and `@llogiq,` I talked with you about this one at Rust Nation in London last week. :-)
This is my first contribution to Clippy, so lots of feedback would be greatly appreciated.
- \[ ] Followed [lint naming conventions][lint_naming]
- \[x] Added passing UI tests (including committed `.stderr` file)
- \[x] `cargo test` passes locally
- \[x] Executed `cargo dev update_lints`
- \[x] Added lint documentation
- \[x] Run `cargo dev fmt`
`dogfood` found one true positive (see #9509) and no false positives.
`lintcheck` found no (true or false) positives, even when running on an extended set of crates.
---
changelog: new lint [`collection_is_never_read`]
[#10415](https://github.com/rust-lang/rust-clippy/pull/10415)
<!-- changelog_checked -->
Scope `missing_docs_in_private_items` to only private items
`missing_docs_in_private_items` currently detects missing docs for public items as well as private. Since `missing_docs`already covers public items, this PR updates `missing_docs_in_private_items` to only cover private items.
Fixes#1895
changelog: [`missing_docs_in_private_items`]: Apply lint only to private items (used to be public and private)
Fix ICE in `multiple_unsafe_ops_per_block`
fixes#10367
changelog: [`multiple_unsafe_ops_per_block`]: Fix ICE when calling a function-like object in an unsafe block
Do not suggest to derive `Default` on generics with implicit arguments
Fixes#10396
changelog: FP: [`derivable_impls`]: do not suggest to derive `Default` on generics with implicit arguments
chore: remove unneeded rustfmt skip
---
The associated rustfmt bug that originally necessitated these skips was resolved a while back, so these are no longer necessary
changelog: none
Normalize projections types when checking `explicit_auto_deref`
fixes#10384
changelog: [`explicit_auto_deref`]: Better consider projection types when checking if auto deref is applicable
Ignore lifetimes from differing contexts in `needless_lifetimes`
Fixes#10379
changelog: [`needless_lifetimes`]: Don't lint signatures in macros if the lifetime is a metavariable
Add `impl_trait_in_params` lint
As this is a lint about style, and using `impl Trait` is purely cosmetical (even with downsides), a lot of unrelated files needed to allow this lint.
---
Resolves#10030
changelog: New lint: [`impl_trait_in_params`]
[10197](https://github.com/rust-lang/rust-clippy/pull/10197)
<!-- changelog_checked -->
Within a larger expression, when the type of `Box::new(T::default())` is
`Box<dyn Trait>`, the concrete type `T` cannot be omitted in the
proposed replacement `Box::<T>::default()`.
[significant_drop_tightening] Ignore inexpensive statements
Not all statements that follow the last use of a lock guard are expensive and can therefore be ignored by the lint.
```rust
pub fn foo() -> i32 {
let mutex = Mutex::new(1);
let lock = mutex.lock().unwrap();
let rslt = *lock;
let another = rslt;
another
}
```
---
changelog: [`significant_drop_tightening`]: No longer lints for inexpensive statements after the lock guard
[#10363](https://github.com/rust-lang/rust-clippy/pull/10363)
<!-- changelog_checked -->
manual_let_else: do not suggest semantically different replacements
The problem is that this lint does not consider the possibility that the divergent branch can come first and that the patterns may overlap. This led to incorrect suggestions, previously registered as correct in the tests themselves:
```rust
let v = match build_enum() {
_ => continue,
Variant::Bar(v) | Variant::Baz(v) => v,
};
```
had a `let Variant::Bar(v) | Variant::Baz(v) = v else { continue; }` suggestion, which is obviously wrong as the original code `continue`s in any case. Issue #10241 gives another example.
The code now checks that the divergent branch comes second. It could be extended later (I've added a TODO) to check for non-overlapping patterns.
Fixes#10241.
changelog: [`manual_let_else`] do not suggest non equivalent replacements in `match`
Stop bytes_nth from suggesting code that does not compile
Fixes#10151
As discussed in the issue, this PR changes the lint in 2 ways
1. Replace `bytes().nth(n).unwrap()` with `as_bytes()[n]`
2. Replace other `bytes().nth(n)` with `as_bytes().get(n).copied()`
---
changelog: Stop bytes_nth from suggesting code that does not compile in some cases
Stop doc_markdown requiring backticks on links to external websites
Fixes#10302
This lint currently checks that any link should be enclosed with `backticks` if the title looks like a lang item. This PR changes the lint to only run on internal links. External links, indicated by `http` or `https`, are skipped.
This PR also reorganises `pulldown_cmark` imports to bypass the clippy lint enforcing 100 line functions.
---
changelog: Stop doc_markdown requiring backticks on links to external websites
Add question-mark-used lint
This lint complains when the question mark operator (try operator) is used. This is a restriction lint that can be useful on local scopes where a custom error handling macro is supposed to be used to augment the error based on local scope data before returning.
Fixes#10340
---
changelog: New lint [`question_mark_used`]
[#10342](https://github.com/rust-lang/rust-clippy/pull/10342)
<!-- changelog_checked -->
Add `let_underscore_untyped`
Fixes#6842
This adds a new pedantic `let_underscore_untyped` lint which checks for `let _ = <expr>`, and suggests to either provide a type annotation, or to remove the `let` keyword. That way the author is forced to specify the type they intended to ignore, and thus get forced to re-visit the decision should the type of `<expr>` change. Alternatively, they can drop the `let` keyword to truly just ignore the value no matter what.
r? `@llogiq`
changelog: New lint: [let_underscore_untyped]
[significant_drop_tightening] Add MVP
cc #9399
Creates the lint with minimum functionalities, which is a good start IMO.
---
changelog: new lint: [`significant_drop_tightening`]
[#10163](https://github.com/rust-lang/rust-clippy/pull/10163)
<!-- changelog_checked -->
As this is a lint about "style", and a purely cosmetical choice (using `<A: Trait>` over `impl Trait`), a lot of other files needed to be allowed this lint.
Fix false positives for `extra_unused_type_parameters`
Don't lint external macros. Also, if the function body is empty, any type parameters with bounds on them are not linted. Note that only the body needs be empty - this rule still applies if the function takes any arguments.
fixes#10318fixes#10319
changelog: none
<!-- changelog_checked -->
uninlined_format_args: do not inline argument with generic parameters
Fix#10339
---
changelog: FP: [`uninlined_format_args`]: No longer lints for arguments with generic parameters
[#10343](https://github.com/rust-lang/rust-clippy/pull/10343)
<!-- changelog_checked -->
fix [`needless_return`] incorrect suggestion when returning if sequence
fixes: #10049
---
changelog: [`needless_return`]: fix incorrect suggestion on if sequence
Liberate late-bound regions rather than erasing them in `needless_pass_by_value`
changelog: [`needless_pass_by_value`]: fixes an ICE when there are late-bound regions in function arguments that are needlessly passed by value
Fixesrust-lang/rust#107147
r? `@matthiaskrgr`
It is not sufficient to ignore break from a block inside the loop.
Instructions after the break must be ignored, as they are unreachable.
This is also true for all instructions in outer blocks and loops
until the right block is reached.
This lint complains when the question mark operator (try operator)
is used. This is a restriction lint that can be useful on local
scopes where a custom error handling macro is supposed to be used
to augment the error based on local scope data before returning.
Negate suggestions when needed in `bool_assert_comparison`
changelog: none assuming this gets into the same release as #10218Fixes#10291
r? `@dswij`
Thanks to `@black-puppydog` for spotting it early
wildcard_enum_match_arm lint takes the enum origin into account
fixes#7419
---
changelog: Enhancement: [`wildcard_enum_match_arm`]: Now lints missing private variants, for local enums
[#10250](https://github.com/rust-lang/rust-clippy/pull/10250)
<!-- changelog_checked -->
[`unused_io_amount`]: Lint with `is_ok` and `is_err`
Fixes#10132
changelog: Apply [`unused_io_amount`] lint to `is_ok` and `is_err` without checking read/write amount
prevents `len_without_is_empty` from yielding positive when `len` takes arguments besides `&self`
Fixes#9520
---
changelog: FP [`len_without_is_empty`]: No longer lints, if `len` as a non-default signature
[#10255](https://github.com/rust-lang/rust-clippy/pull/10255)
<!-- changelog_checked -->
more than just `&self` in non-standard implementations.
changelog: Fix [`len_without_is_empty`] false positive when len has a
non-standard method signature
Fixes#9520
`invalid_regex`: Show full error when string value doesn't match source
changelog: [`invalid_regex`]: Show full error when parsing non-literals or regular strings containing escape sequences
Fixes#4170, the escape sequence there causes the span to be incorrect which will have caused most of the confusion
Add `multiple_unsafe_ops_per_block` lint
Adds a lint, which restricts an `unsafe` block to only one unsafe operation.
Closes#10064
---
changelog: New lint: [`multiple_unsafe_ops_per_block`]
[#10206](https://github.com/rust-lang/rust-clippy/pull/10206)
<!-- changelog_checked -->
Fix suggestion in `transmutes_expressible_as_ptr_casts` when the source type is a borrow.
fixes#9894
changelog: `transmutes_expressible_as_ptr_casts`: Fix suggestion when the source type is a borrow.
[needless_return]: Remove all semicolons on suggestion
Closes#10182
Multiple semicolons currently breaks autofix for `needless_return` suggestions. Any semicolons left after removing return means that the return type will always be `()`, and thus fail to compile.
This PR allows `needless_return` to remove multiple semicolons.
The change won't cover the case where there is multiple line yet.
i.e.
```rust
fn needless_return() -> bool {
return true;
;;
}
```
---
changelog: Sugg: [`needless_return`]: Now removes all semicolons on the same line
[#10187](https://github.com/rust-lang/rust-clippy/pull/10187)
<!-- changelog_checked -->
`cast_possible_truncation` Suggest TryFrom when truncation possible
This fixes the last issues from https://github.com/rust-lang/rust-clippy/pull/9664 as the author seems to be inactive. The PR author was sadly kept during the rebase, due to the conflict resolution.
IDK if it's worth it do to a full review, I only added the last commit, everything else remained the same, besides a rebase.
---
changelog: Sugg: [`cast_possible_truncation`]: Now suggests using `try_from` or allowing the lint
[#10038](https://github.com/rust-lang/rust-clippy/pull/10038)
<!-- changelog_checked -->
closes: https://github.com/rust-lang/rust-clippy/issues/9231
Render missing generics suggestion verbosely
It's a bit easier to read like this, especially ones that are appending new generics onto an existing list, like ": `, T`" which render somewhat poorly inline.
Also don't suggest `dyn` as a type parameter to add, even if technically that's valid in edition 2015.
Allow implementing `Hash` with derived `PartialEq` (`derive_hash_xor_eq`
This is a common pattern and is totally allowed by the `Hash` trait.
Fixes#2627
changelog: Move: Renamed `derive_hash_xor_eq` to [`derived_hash_with_manual_eq`]
[#10184](https://github.com/rust-lang/rust-clippy/pull/10184)
changelog: Enhancement: [`derived_hash_with_manual_eq`]: Now allows `#[derive(PartialEq)]` with custom `Hash` implementations
[#10184](https://github.com/rust-lang/rust-clippy/pull/10184)
<!-- changelog_checked -->
trim paths in `suspicious_to_owned`
This continues my path trimming spree. I'm not going to add yet another changelog entry, we should have one "trim paths in some applicable lints" entry instead.
---
changelog: none
unused_self: Don't trigger if the method body contains todo!()
If the author is using todo!(), presumably they intend to use self at some point later, so we don't have a good basis to recommend factoring out to an associated function.
Fixes#10117.
---
changelog: Enhancement: [`unused_self`]: No longer lints, if the method body contains a `todo!()` call
[#10166](https://github.com/rust-lang/rust-clippy/pull/10166)
<!-- changelog_checked -->
[arithmetic_side_effects] Add more tests related to custom types
Add tests to ensure that custom types are triggered with any type of arithmetic operation as well as combinations with or without references.
---
changelog: none
<!-- changelog_checked -->
trim paths in `box_default`
This might help with #10089, though I have not tested that yet. In any event, it keeps the suggestion short and to the point.
---
changelog: Trim paths in [`box_default`] suggestion
trim paths in `default_trait_access`/`clone_on_copy` suggestions
This should help making the suggestions more palatable. Similar to #10153.
---
changelog: trim paths in [`default_trait_access`]/[`clone_on_copy`] suggestions
If the author is using todo!(), presumably they intend to use self at
some point later, so we don't have a good basis to recommend factoring
out to an associated function.
Fixes#10117.
changelog: Don't trigger [`unused_self`] if the method body contains a `todo!()` call
[`drop_ref`]: don't lint idiomatic in match arm
fixes#10122
As established in issue #9482, it is idiomatic to use a single `drop()` expression in a match arm to achieve a side-effect of a function while discarding its output. This should also apply to cases where the function returns a reference.
The change to the lint's code was less than 1 line, because all the heavy lifting was done in PR #9491.
---
changelog: FP: [`drop_ref`]: No longer lints idiomatic expression in `match` arms
[#10142](https://github.com/rust-lang/rust-clippy/pull/10142)
<!-- changelog_checked -->
Make the iter_kv_map lint handle ref/mut annotations.
For the degenerate (`map(|(k, _)| k)`/`map(|(_, v)| v)`) cases a mut annotation is superfluous and a ref annotation won't compile, so no additional handling is required. For cases where the `map` call must be preserved ref/mut annotations should also be presereved so that the map body continues to work as expected.
*Please write a short comment explaining your change (or "none" for internal only changes)*
changelog: [`iter_kv_map`]: handle ref/mut annotations
For the degenerate (`map(|(k, _)| k)`/`map(|(_, v)| v)`) cases a mut annotation is superfluous and a ref annotation won't compile, so no additional handling is required. For cases where the `map` call must be preserved ref/mut annotations should also be presereved so that the map body continues to work as expected.
don't lint field_reassign when field in closure
fixes#10136
This change makes the ContainsName struct visit all interior expressions, which means that ContainsName will return true even if `name` is used in a closure within `expr`.
---
changelog: FP: [`field_reassign_with_default`]: No longer lints cases, where values are initializes from closures capturing struct values
[#10143](https://github.com/rust-lang/rust-clippy/pull/10143)
<!-- changelog_checked -->
This commit makes the ContainsName struct visit all interior
expressions, which means that ContainsName will return true
even if `name` is used in a closure within `expr`.
fix incorrect suggestion in `suboptimal_flops`
fixes#10003
There was an error when trying to negate an expression like `x - 1.0`. We used to format it as `-x - 1.0` whereas a proper negation would be `-(x - 1.0)`.
Therefore, we add parentheses around the expression when it is `ExprKind::Binary`.
We also add parentheses around multiply and divide expressions, even though this is not strictly necessary.
changelog: [`suboptimal_flops`]: fix incorrect suggestion caused by an incorrect negation of floating point expressions.
There was an error when trying to negate an expression
like `x - 1.0`. We used to format it as `-x - 1.0` whereas
a proper negation would be `-(x - 1.0)`.
Therefore, we add parentheses around the expression when it is a
Binary ExprKind.
We also add parentheses around multiply and divide expressions,
even though this is not strictly necessary.
Add size_of_ref lint
This addresses #9995, which is likely raising a valid point about `std::mem::size_of_val()`: It's [very easy to use double-references as the argument](https://github.com/apache/arrow-datafusion/pull/4371#discussion_r1032385224), which the function will happily accept and give back the size of _the reference_, not the size of the value _behind_ the reference. In the worst case, if the value matches the programmer's expectation, this seems to work, while in fact, everything will go horribly wrong e.g. on a different platform.
The size of a `&T` is independent of what `T` is, and people might want to use `std::mem::size_of_val()` to actually get the size of _any_ reference (e.g. via `&&()`). I would rather suggest that this is always bad behavior, though ([instead](https://doc.rust-lang.org/reference/type-layout.html#pointers-and-references-layout), [and](https://doc.rust-lang.org/stable/std/primitive.usize.html#associatedconstant.BITS)). I, therefore, put this lint into `correctness`.
Since the problem is usually easily fixed by removing extra `&`, I went light on suggesting code.
---
changelog: New lint: [`size_of_ref`]
[#10098](https://github.com/rust-lang/rust-clippy/pull/10098)
<!-- changelog_checked -->
Improve `possible_borrower`
This PR makes several improvements to `clippy_uitls::mir::possible_borrower`. These changes benefit both `needless_borrow` and `redundant clone`.
1. **Use the compiler's `MaybeStorageLive` analysis**
I could spot not functional differences between the one in the compiler and the one in Clippy's repository. So, I removed the latter in favor of the the former.
2. **Make `PossibleBorrower` a dataflow analysis instead of a visitor**
The main benefit of this change is that allows `possible_borrower` to take advantage of statements' relative locations, which is easier to do in an analysis than in a visitor.
This is easier to illustrate with an example, so consider this one:
```rust
fn foo(cx: &LateContext<'_>, lint: &'static Lint) {
cx.struct_span_lint(lint, rustc_span::Span::default(), "", |diag| diag.note(&String::new()));
// ^
}
```
We would like to flag the `&` pointed to by the `^` for removal. `foo`'s MIR begins like this:
```rust
fn span_lint::foo::{closure#0}(_1: [closure@$DIR/needless_borrow.rs:396:68: 396:74], _2: &mut rustc_errors::diagnostic_builder::DiagnosticBuilder<'_, ()>) -> &mut rustc_errors::diagnostic_builder::DiagnosticBuilder<'_, ()> {
debug diag => _2; // in scope 0 at $DIR/needless_borrow.rs:396:69: 396:73
let mut _0: &mut rustc_errors::diagnostic_builder::DiagnosticBuilder<'_, ()>; // return place in scope 0 at $DIR/needless_borrow.rs:396:75: 396:75
let mut _3: &mut rustc_errors::diagnostic_builder::DiagnosticBuilder<'_, ()>; // in scope 0 at $DIR/needless_borrow.rs:396:75: 396:100
let mut _4: &mut rustc_errors::diagnostic_builder::DiagnosticBuilder<'_, ()>; // in scope 0 at $DIR/needless_borrow.rs:396:75: 396:100
let mut _5: &std::string::String; // in scope 0 at $DIR/needless_borrow.rs:396:85: 396:99
let _6: std::string::String; // in scope 0 at $DIR/needless_borrow.rs:396:86: 396:99
bb0: {
StorageLive(_3); // scope 0 at $DIR/needless_borrow.rs:396:75: 396:100
StorageLive(_4); // scope 0 at $DIR/needless_borrow.rs:396:75: 396:100
_4 = &mut (*_2); // scope 0 at $DIR/needless_borrow.rs:396:75: 396:100
StorageLive(_5); // scope 0 at $DIR/needless_borrow.rs:396:85: 396:99
StorageLive(_6); // scope 0 at $DIR/needless_borrow.rs:396:86: 396:99
_6 = std::string::String::new() -> bb1; // scope 0 at $DIR/needless_borrow.rs:396:86: 396:99
// mir::Constant
// + span: $DIR/needless_borrow.rs:396:86: 396:97
// + literal: Const { ty: fn() -> std::string::String {std::string::String::new}, val: Value(<ZST>) }
}
bb1: {
_5 = &_6; // scope 0 at $DIR/needless_borrow.rs:396:85: 396:99
_3 = rustc_errors::diagnostic_builder::DiagnosticBuilder::<'_, ()>::note::<&std::string::String>(move _4, move _5) -> [return: bb2, unwind: bb4]; // scope 0 at $DIR/needless_borrow.rs:396:75: 396:100
// mir::Constant
// + span: $DIR/needless_borrow.rs:396:80: 396:84
// + literal: Const { ty: for<'a> fn(&'a mut rustc_errors::diagnostic_builder::DiagnosticBuilder<'_, ()>, &std::string::String) -> &'a mut rustc_errors::diagnostic_builder::DiagnosticBuilder<'_, ()> {rustc_errors::diagnostic_builder::DiagnosticBuilder::<'_, ()>::note::<&std::string::String>}, val: Value(<ZST>) }
}
```
The call to `diag.note` appears in `bb1` on the line beginning with `_3 =`. The `String` is owned by `_6`. So, in the call to `diag.note`, we would like to know whether there are any references to `_6` besides `_5`.
The old, visitor approach did not consider the relative locations of statements. So all borrows were treated the same, *even if they occurred after the location of interest*.
For example, before the `_3 = ...` call, the possible borrowers of `_6` would be just `_5`. But after the call, the possible borrowers would include `_2`, `_3`, and `_4`.
So, in a sense, the call from which we are try to remove the needless borrow is trying to prevent us from removing the needless borrow(!).
With an analysis, things do not get so muddled. We can determine the set of possible borrowers at any specific location, e.g., using a `ResultsCursor`.
3. **Change `only_borrowers` to `at_most_borrowers`**
`possible_borrowers` exposed a function `only_borrowers` that determined whether the borrowers of some local were *exactly* some set `S`. But, from what I can tell, this was overkill. For the lints that currently use `possible_borrower` (`needless_borrow` and `redundant_clone`), all we really want to know is whether there are borrowers *other than* those in `S`. (Put another way, we only care about the subset relation in one direction.) The new function `at_most_borrowers` takes this more tailored approach.
4. **Compute relations "on the fly" rather than using `transitive_relation`**
The visitor would compute and store the transitive closure of the possible borrower relation for an entire MIR body.
But with an analysis, there is effectively a different possible borrower relation at each location in the body. Computing and storing a transitive closure at each location would not be practical.
So the new approach is to compute the transitive closure on the fly, as needed. But the new approach might actually be more efficient, as I now explain.
In all current uses of `at_most_borrowers` (previously `only_borrowers`), the size of the set of borrowers `S` is at most 2. So you need only check at most three borrowers to determine whether the subset relation holds. That is, once you have found a third borrower, you can stop, since you know the relation cannot hold.
Note that `transitive_relation` is still used by `clippy_uitls::mir::possible_origin` (a kind of "subroutine" of `possible_borrower`).
cc: `@Jarcho`
---
changelog: [`needless_borrow`], [`redundant_clone`]: Now track references better and detect more cases
[#9701](https://github.com/rust-lang/rust-clippy/pull/9701)
<!-- changelog_checked -->
