Theme

What it does

Checks for usage of items through absolute paths, like std::env::current_dir.

Why restrict this?

Many codebases have their own style when it comes to importing, but one that is seldom used is using absolute paths everywhere. This is generally considered unidiomatic, and you should add a use statement.

The default maximum segments (2) is pretty strict, you may want to increase this in clippy.toml.

Note: One exception to this is code from macro expansion - this does not lint such cases, as using absolute paths is the proper way of referencing items in one.

Known issues

There are currently a few cases which are not caught by this lint:

  • Macro calls. e.g. path::to::macro!()
  • Derive macros. e.g. #[derive(path::to::macro)]
  • Attribute macros. e.g. #[path::to::macro]

Example

let x = std::f64::consts::PI;

Use any of the below instead, or anything else:

use std::f64;
use std::f64::consts;
use std::f64::consts::PI;
let x = f64::consts::PI;
let x = consts::PI;
let x = PI;
use std::f64::consts as f64_consts;
let x = f64_consts::PI;

Configuration

  • absolute-paths-allowed-crates: Which crates to allow absolute paths from

    (default: [])

  • absolute-paths-max-segments: The maximum number of segments a path can have before being linted, anything above this will be linted.

    (default: 2)

Applicability: Unspecified(?)
Added in: 1.73.0

What it does

Checks for comparisons where one side of the relation is either the minimum or maximum value for its type and warns if it involves a case that is always true or always false. Only integer and boolean types are checked.

Why is this bad?

An expression like min <= x may misleadingly imply that it is possible for x to be less than the minimum. Expressions like max < x are probably mistakes.

Known problems

For usize the size of the current compile target will be assumed (e.g., 64 bits on 64 bit systems). This means code that uses such a comparison to detect target pointer width will trigger this lint. One can use mem::sizeof and compare its value or conditional compilation attributes like #[cfg(target_pointer_width = "64")] .. instead.

Example

let vec: Vec<isize> = Vec::new();
if vec.len() <= 0 {}
if 100 > i32::MAX {}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Finds items imported through alloc when available through core.

Why restrict this?

Crates which have no_std compatibility and may optionally require alloc may wish to ensure types are imported from core to ensure disabling alloc does not cause the crate to fail to compile. This lint is also useful for crates migrating to become no_std compatible.

Known problems

The lint is only partially aware of the required MSRV for items that were originally in std but moved to core.

Example

use alloc::slice::from_ref;

Use instead:

use core::slice::from_ref;
Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

Checks for usage of the #[allow] attribute and suggests replacing it with the #[expect] (See RFC 2383)

This lint only warns outer attributes (#[allow]), as inner attributes (#![allow]) are usually used to enable or disable lints on a global scale.

Why is this bad?

#[expect] attributes suppress the lint emission, but emit a warning, if the expectation is unfulfilled. This can be useful to be notified when the lint is no longer triggered.

Example

#[allow(unused_mut)]
fn foo() -> usize {
    let mut a = Vec::new();
    a.len()
}

Use instead:

#[expect(unused_mut)]
fn foo() -> usize {
    let mut a = Vec::new();
    a.len()
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.70.0

What it does

Checks for attributes that allow lints without a reason.

Why restrict this?

Justifying each allow helps readers understand the reasoning, and may allow removing allow attributes if their purpose is obsolete.

Example

#![allow(clippy::some_lint)]

Use instead:

#![allow(clippy::some_lint, reason = "False positive rust-lang/rust-clippy#1002020")]

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: Unspecified(?)
Added in: 1.61.0

What it does

Checks for ranges which almost include the entire range of letters from ‘a’ to ‘z’ or digits from ‘0’ to ‘9’, but don’t because they’re a half open range.

Why is this bad?

This ('a'..'z') is almost certainly a typo meant to include all letters.

Example

let _ = 'a'..'z';

Use instead:

let _ = 'a'..='z';

Past names

  • almost_complete_letter_range

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MaybeIncorrect(?)
Added in: 1.68.0

What it does

Checks for foo = bar; bar = foo sequences.

Why is this bad?

This looks like a failed attempt to swap.

Example

a = b;
b = a;

If swapping is intended, use swap() instead:

std::mem::swap(&mut a, &mut b);
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for floating point literals that approximate constants which are defined in std::f32::consts or std::f64::consts, respectively, suggesting to use the predefined constant.

Why is this bad?

Usually, the definition in the standard library is more precise than what people come up with. If you find that your definition is actually more precise, please file a Rust issue.

Example

let x = 3.14;
let y = 1_f64 / x;

Use instead:

let x = std::f32::consts::PI;
let y = std::f64::consts::FRAC_1_PI;

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Confirms that items are sorted in source files as per configuration.

Why restrict this?

Keeping a consistent ordering throughout the codebase helps with working as a team, and possibly improves maintainability of the codebase. The idea is that by defining a consistent and enforceable rule for how source files are structured, less time will be wasted during reviews on a topic that is (under most circumstances) not relevant to the logic implemented in the code. Sometimes this will be referred to as “bikeshedding”.

Default Ordering and Configuration

As there is no generally applicable rule, and each project may have different requirements, the lint can be configured with high granularity. The configuration is split into two stages:

  1. Which item kinds that should have an internal order enforced.
  2. Individual ordering rules per item kind.

The item kinds that can be linted are:

  • Module (with customized groupings, alphabetical within)
  • Trait (with customized order of associated items, alphabetical within)
  • Enum, Impl, Struct (purely alphabetical)

Module Item Order

Due to the large variation of items within modules, the ordering can be configured on a very granular level. Item kinds can be grouped together arbitrarily, items within groups will be ordered alphabetically. The following table shows the default groupings:

GroupItem Kinds
modules“mod”, “foreign_mod”
use“use”
macros“macro”
global_asm“global_asm”
UPPER_SNAKE_CASE“static”, “const”
PascalCase“ty_alias”, “opaque_ty”, “enum”, “struct”, “union”, “trait”, “trait_alias”, “impl”
lower_snake_case“fn”

All item kinds must be accounted for to create an enforceable linting rule set.

Known Problems

Performance Impact

Keep in mind, that ordering source code alphabetically can lead to reduced performance in cases where the most commonly used enum variant isn’t the first entry anymore, and similar optimizations that can reduce branch misses, cache locality and such. Either don’t use this lint if that’s relevant, or disable the lint in modules or items specifically where it matters. Other solutions can be to use profile guided optimization (PGO), post-link optimization (e.g. using BOLT for LLVM), or other advanced optimization methods. A good starting point to dig into optimization is cargo-pgo.

Lints on a Contains basis

The lint can be disabled only on a “contains” basis, but not per element within a “container”, e.g. the lint works per-module, per-struct, per-enum, etc. but not for “don’t order this particular enum variant”.

Module documentation

Module level rustdoc comments are not part of the resulting syntax tree and as such cannot be linted from within check_mod. Instead, the rustdoc::missing_documentation lint may be used.

Module Tests

This lint does not implement detection of module tests (or other feature dependent elements for that matter). To lint the location of mod tests, the lint items_after_test_module can be used instead.

Example

trait TraitUnordered {
    const A: bool;
    const C: bool;
    const B: bool;

    type SomeType;

    fn a();
    fn c();
    fn b();
}

Use instead:

trait TraitOrdered {
    const A: bool;
    const B: bool;
    const C: bool;

    type SomeType;

    fn a();
    fn b();
    fn c();
}

Configuration

  • module-item-order-groupings: The named groupings of different source item kinds within modules.

    (default: [["modules", ["extern_crate", "mod", "foreign_mod"]], ["use", ["use"]], ["macros", ["macro"]], ["global_asm", ["global_asm"]], ["UPPER_SNAKE_CASE", ["static", "const"]], ["PascalCase", ["ty_alias", "enum", "struct", "union", "trait", "trait_alias", "impl"]], ["lower_snake_case", ["fn"]]])

  • source-item-ordering: Which kind of elements should be ordered internally, possible values being enum, impl, module, struct, trait.

    (default: ["enum", "impl", "module", "struct", "trait"])

  • trait-assoc-item-kinds-order: The order of associated items in traits.

    (default: ["const", "type", "fn"])

Applicability: Unspecified(?)
Added in: 1.82.0

What it does.

This lint warns when you use Arc with a type that does not implement Send or Sync.

Why is this bad?

Arc<T> is a thread-safe Rc<T> and guarantees that updates to the reference counter use atomic operations. To send an Arc<T> across thread boundaries and share ownership between multiple threads, T must be both Send and Sync, so either T should be made Send + Sync or an Rc should be used instead of an Arc.

Example


fn main() {
    // This is fine, as `i32` implements `Send` and `Sync`.
    let a = Arc::new(42);

    // `RefCell` is `!Sync`, so either the `Arc` should be replaced with an `Rc`
    // or the `RefCell` replaced with something like a `RwLock`
    let b = Arc::new(RefCell::new(42));
}
Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Checks any kind of arithmetic operation of any type.

Operators like +, -, * or << are usually capable of overflowing according to the Rust Reference, or can panic (/, %).

Known safe built-in types like Wrapping or Saturating, floats, operations in constant environments, allowed types and non-constant operations that won’t overflow are ignored.

Why restrict this?

For integers, overflow will trigger a panic in debug builds or wrap the result in release mode; division by zero will cause a panic in either mode. As a result, it is desirable to explicitly call checked, wrapping or saturating arithmetic methods.

Example

// `n` can be any number, including `i32::MAX`.
fn foo(n: i32) -> i32 {
    n + 1
}

Third-party types can also overflow or present unwanted side-effects.

Example

use rust_decimal::Decimal;
let _n = Decimal::MAX + Decimal::MAX;

Past names

  • integer_arithmetic

Configuration

  • arithmetic-side-effects-allowed: Suppress checking of the passed type names in all types of operations.

If a specific operation is desired, consider using arithmetic_side_effects_allowed_binary or arithmetic_side_effects_allowed_unary instead.

Example

arithmetic-side-effects-allowed = ["SomeType", "AnotherType"]

Noteworthy

A type, say SomeType, listed in this configuration has the same behavior of ["SomeType" , "*"], ["*", "SomeType"] in arithmetic_side_effects_allowed_binary.

(default: [])

  • arithmetic-side-effects-allowed-binary: Suppress checking of the passed type pair names in binary operations like addition or multiplication.

Supports the “*” wildcard to indicate that a certain type won’t trigger the lint regardless of the involved counterpart. For example, ["SomeType", "*"] or ["*", "AnotherType"].

Pairs are asymmetric, which means that ["SomeType", "AnotherType"] is not the same as ["AnotherType", "SomeType"].

Example

arithmetic-side-effects-allowed-binary = [["SomeType" , "f32"], ["AnotherType", "*"]]

(default: [])

  • arithmetic-side-effects-allowed-unary: Suppress checking of the passed type names in unary operations like “negation” (-).

Example

arithmetic-side-effects-allowed-unary = ["SomeType", "AnotherType"]

(default: [])

Applicability: Unspecified(?)
Added in: 1.64.0

What it does

Checks for usage of as conversions.

Note that this lint is specialized in linting every single use of as regardless of whether good alternatives exist or not. If you want more precise lints for as, please consider using these separate lints: unnecessary_cast, cast_lossless/cast_possible_truncation/cast_possible_wrap/cast_precision_loss/cast_sign_loss, fn_to_numeric_cast(_with_truncation), char_lit_as_u8, ref_to_mut and ptr_as_ptr. There is a good explanation the reason why this lint should work in this way and how it is useful in this issue.

Why restrict this?

as conversions will perform many kinds of conversions, including silently lossy conversions and dangerous coercions. There are cases when it makes sense to use as, so the lint is Allow by default.

Example

let a: u32;
...
f(a as u16);

Use instead:

f(a.try_into()?);

// or

f(a.try_into().expect("Unexpected u16 overflow in f"));
Applicability: Unspecified(?)
Added in: 1.41.0

What it does

Checks for the result of a &self-taking as_ptr being cast to a mutable pointer.

Why is this bad?

Since as_ptr takes a &self, the pointer won’t have write permissions unless interior mutability is used, making it unlikely that having it as a mutable pointer is correct.

Example

let mut vec = Vec::<u8>::with_capacity(1);
let ptr = vec.as_ptr() as *mut u8;
unsafe { ptr.write(4) }; // UNDEFINED BEHAVIOUR

Use instead:

let mut vec = Vec::<u8>::with_capacity(1);
let ptr = vec.as_mut_ptr();
unsafe { ptr.write(4) };
Applicability: MaybeIncorrect(?)
Added in: 1.66.0

What it does

Checks for the usage of as _ conversion using inferred type.

Why restrict this?

The conversion might include lossy conversion or a dangerous cast that might go undetected due to the type being inferred.

The lint is allowed by default as using _ is less wordy than always specifying the type.

Example

fn foo(n: usize) {}
let n: u16 = 256;
foo(n as _);

Use instead:

fn foo(n: usize) {}
let n: u16 = 256;
foo(n as usize);
Applicability: MachineApplicable(?)
Added in: 1.63.0

What it does

Checks for assert!(true) and assert!(false) calls.

Why is this bad?

Will be optimized out by the compiler or should probably be replaced by a panic!() or unreachable!()

Example

assert!(false)
assert!(true)
const B: bool = false;
assert!(B)
Applicability: Unspecified(?)
Added in: 1.34.0

What it does

Checks for assert!(r.is_ok()) or assert!(r.is_err()) calls.

Why restrict this?

This form of assertion does not show any of the information present in the Result other than which variant it isn’t.

Known problems

The suggested replacement decreases the readability of code and log output.

Example

assert!(r.is_ok());
assert!(r.is_err());

Use instead:

r.unwrap();
r.unwrap_err();
Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

Checks for a = a op b or a = b commutative_op a patterns.

Why is this bad?

These can be written as the shorter a op= b.

Known problems

While forbidden by the spec, OpAssign traits may have implementations that differ from the regular Op impl.

Example

let mut a = 5;
let b = 0;
// ...

a = a + b;

Use instead:

let mut a = 5;
let b = 0;
// ...

a += b;
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

Compound operators are harmless and linting on them is not in scope for clippy.

Applicability: Unspecified(?)
Deprecated in: 1.30.0

What it does

Checks for code like foo = bar.clone();

Why is this bad?

Custom Clone::clone_from() or ToOwned::clone_into implementations allow the objects to share resources and therefore avoid allocations.

Example

struct Thing;

impl Clone for Thing {
    fn clone(&self) -> Self { todo!() }
    fn clone_from(&mut self, other: &Self) { todo!() }
}

pub fn assign_to_ref(a: &mut Thing, b: Thing) {
    *a = b.clone();
}

Use instead:

struct Thing;

impl Clone for Thing {
    fn clone(&self) -> Self { todo!() }
    fn clone_from(&mut self, other: &Self) { todo!() }
}

pub fn assign_to_ref(a: &mut Thing, b: Thing) {
    a.clone_from(&b);
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: Unspecified(?)
Added in: 1.78.0

What it does

Checks for async blocks that yield values of types that can themselves be awaited.

Why is this bad?

An await is likely missing.

Example

async fn foo() {}

fn bar() {
  let x = async {
    foo()
  };
}

Use instead:

async fn foo() {}

fn bar() {
  let x = async {
    foo().await
  };
}
Applicability: MaybeIncorrect(?)
Added in: 1.48.0

What it does

Allows users to configure types which should not be held across await suspension points.

Why is this bad?

There are some types which are perfectly safe to use concurrently from a memory access perspective, but that will cause bugs at runtime if they are held in such a way.

Example

await-holding-invalid-types = [
  # You can specify a type name
  "CustomLockType",
  # You can (optionally) specify a reason
  { path = "OtherCustomLockType", reason = "Relies on a thread local" }
]
struct CustomLockType;
struct OtherCustomLockType;
async fn foo() {
  let _x = CustomLockType;
  let _y = OtherCustomLockType;
  baz().await; // Lint violation
}

Configuration

  • await-holding-invalid-types: The list of types which may not be held across an await point.

    (default: [])

Applicability: Unspecified(?)
Added in: 1.62.0

What it does

Checks for calls to await while holding a non-async-aware MutexGuard.

Why is this bad?

The Mutex types found in std::sync and parking_lot are not designed to operate in an async context across await points.

There are two potential solutions. One is to use an async-aware Mutex type. Many asynchronous foundation crates provide such a Mutex type. The other solution is to ensure the mutex is unlocked before calling await, either by introducing a scope or an explicit call to Drop::drop.

Known problems

Will report false positive for explicitly dropped guards (#6446). A workaround for this is to wrap the .lock() call in a block instead of explicitly dropping the guard.

Example

async fn foo(x: &Mutex<u32>) {
  let mut guard = x.lock().unwrap();
  *guard += 1;
  baz().await;
}

async fn bar(x: &Mutex<u32>) {
  let mut guard = x.lock().unwrap();
  *guard += 1;
  drop(guard); // explicit drop
  baz().await;
}

Use instead:

async fn foo(x: &Mutex<u32>) {
  {
    let mut guard = x.lock().unwrap();
    *guard += 1;
  }
  baz().await;
}

async fn bar(x: &Mutex<u32>) {
  {
    let mut guard = x.lock().unwrap();
    *guard += 1;
  } // guard dropped here at end of scope
  baz().await;
}
Applicability: Unspecified(?)
Added in: 1.45.0

What it does

Checks for calls to await while holding a RefCell, Ref, or RefMut.

Why is this bad?

RefCell refs only check for exclusive mutable access at runtime. Holding a RefCell ref across an await suspension point risks panics from a mutable ref shared while other refs are outstanding.

Known problems

Will report false positive for explicitly dropped refs (#6353). A workaround for this is to wrap the .borrow[_mut]() call in a block instead of explicitly dropping the ref.

Example

async fn foo(x: &RefCell<u32>) {
  let mut y = x.borrow_mut();
  *y += 1;
  baz().await;
}

async fn bar(x: &RefCell<u32>) {
  let mut y = x.borrow_mut();
  *y += 1;
  drop(y); // explicit drop
  baz().await;
}

Use instead:

async fn foo(x: &RefCell<u32>) {
  {
     let mut y = x.borrow_mut();
     *y += 1;
  }
  baz().await;
}

async fn bar(x: &RefCell<u32>) {
  {
    let mut y = x.borrow_mut();
    *y += 1;
  } // y dropped here at end of scope
  baz().await;
}
Applicability: Unspecified(?)
Added in: 1.49.0

What it does

Checks for incompatible bit masks in comparisons.

The formula for detecting if an expression of the type _ <bit_op> m <cmp_op> c (where <bit_op> is one of {&, |} and <cmp_op> is one of {!=, >=, >, !=, >=, >}) can be determined from the following table:

ComparisonBit OpExampleis alwaysFormula
== or !=&x & 2 == 3falsec & m != c
< or >=&x & 2 < 3truem < c
> or <=&x & 1 > 1falsem <= c
== or !=|x | 1 == 0falsec | m != c
< or >=|x | 1 < 1falsem >= c
<= or >|x | 1 > 0truem > c

Why is this bad?

If the bits that the comparison cares about are always set to zero or one by the bit mask, the comparison is constant true or false (depending on mask, compared value, and operators).

So the code is actively misleading, and the only reason someone would write this intentionally is to win an underhanded Rust contest or create a test-case for this lint.

Example

if (x & 1 == 2) { }
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for the usage of the to_be_bytes method and/or the function from_be_bytes.

Why restrict this?

To ensure use of little-endian or the target’s endianness rather than big-endian.

Example

let _x = 2i32.to_be_bytes();
let _y = 2i64.to_be_bytes();
Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Checks for usage of _.and_then(|x| Some(y)), _.and_then(|x| Ok(y)) or _.or_else(|x| Err(y)).

Why is this bad?

This can be written more concisely as _.map(|x| y) or _.map_err(|x| y).

Example

let _ = opt().and_then(|s| Some(s.len()));
let _ = res().and_then(|s| if s.len() == 42 { Ok(10) } else { Ok(20) });
let _ = res().or_else(|s| if s.len() == 42 { Err(10) } else { Err(20) });

The correct use would be:

let _ = opt().map(|s| s.len());
let _ = res().map(|s| if s.len() == 42 { 10 } else { 20 });
let _ = res().map_err(|s| if s.len() == 42 { 10 } else { 20 });

Past names

  • option_and_then_some
Applicability: MachineApplicable(?)
Added in: 1.45.0

What it does

Checks for warn/deny/forbid attributes targeting the whole clippy::restriction category.

Why is this bad?

Restriction lints sometimes are in contrast with other lints or even go against idiomatic rust. These lints should only be enabled on a lint-by-lint basis and with careful consideration.

Example

#![deny(clippy::restriction)]

Use instead:

#![deny(clippy::as_conversions)]
Applicability: Unspecified(?)
Added in: 1.47.0

What it does

Checks for if and match conditions that use blocks containing an expression, statements or conditions that use closures with blocks.

Why is this bad?

Style, using blocks in the condition makes it hard to read.

Examples

if { true } { /* ... */ }

if { let x = somefunc(); x } { /* ... */ }

match { let e = somefunc(); e } {
    // ...
}

Use instead:

if true { /* ... */ }

let res = { let x = somefunc(); x };
if res { /* ... */ }

let res = { let e = somefunc(); e };
match res {
    // ...
}

Past names

  • block_in_if_condition_expr
  • block_in_if_condition_stmt
  • blocks_in_if_conditions
Applicability: MachineApplicable(?)
Added in: 1.45.0

What it does

This lint warns about boolean comparisons in assert-like macros.

Why is this bad?

It is shorter to use the equivalent.

Example

assert_eq!("a".is_empty(), false);
assert_ne!("a".is_empty(), true);

Use instead:

assert!(!"a".is_empty());
Applicability: MachineApplicable(?)
Added in: 1.53.0

What it does

Checks for expressions of the form x == true, x != true and order comparisons such as x < true (or vice versa) and suggest using the variable directly.

Why is this bad?

Unnecessary code.

Example

if x == true {}
if y == false {}

use x directly:

if x {}
if !y {}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Instead of using an if statement to convert a bool to an int, this lint suggests using a from() function or an as coercion.

Why is this bad?

Coercion or from() is another way to convert bool to a number. Both methods are guaranteed to return 1 for true, and 0 for false.

See https://doc.rust-lang.org/std/primitive.bool.html#impl-From%3Cbool%3E

Example

if condition {
    1_i64
} else {
    0
};

Use instead:

i64::from(condition);

or

condition as i64;
Applicability: MaybeIncorrect(?)
Added in: 1.65.0

What it does

Checks for the usage of &expr as *const T or &mut expr as *mut T, and suggest using ptr::addr_of or ptr::addr_of_mut instead.

Why is this bad?

This would improve readability and avoid creating a reference that points to an uninitialized value or unaligned place. Read the ptr::addr_of docs for more information.

Example

let val = 1;
let p = &val as *const i32;

let mut val_mut = 1;
let p_mut = &mut val_mut as *mut i32;

Use instead:

let val = 1;
let p = std::ptr::addr_of!(val);

let mut val_mut = 1;
let p_mut = std::ptr::addr_of_mut!(val_mut);

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.60.0

What it does

Checks for &*(&T).

Why is this bad?

Dereferencing and then borrowing a reference value has no effect in most cases.

Known problems

False negative on such code:

let x = &12;
let addr_x = &x as *const _ as usize;
let addr_y = &&*x as *const _ as usize; // assert ok now, and lint triggered.
                                        // But if we fix it, assert will fail.
assert_ne!(addr_x, addr_y);

Example

let s = &String::new();

let a: &String = &* s;

Use instead:

let a: &String = s;
Applicability: MachineApplicable(?)
Added in: 1.63.0

What it does

Checks if const items which is interior mutable (e.g., contains a Cell, Mutex, AtomicXxxx, etc.) has been borrowed directly.

Why is this bad?

Consts are copied everywhere they are referenced, i.e., every time you refer to the const a fresh instance of the Cell or Mutex or AtomicXxxx will be created, which defeats the whole purpose of using these types in the first place.

The const value should be stored inside a static item.

Known problems

When an enum has variants with interior mutability, use of its non interior mutable variants can generate false positives. See issue #3962

Types that have underlying or potential interior mutability trigger the lint whether the interior mutable field is used or not. See issues #5812 and #3825

Example

use std::sync::atomic::{AtomicUsize, Ordering::SeqCst};
const CONST_ATOM: AtomicUsize = AtomicUsize::new(12);

CONST_ATOM.store(6, SeqCst); // the content of the atomic is unchanged
assert_eq!(CONST_ATOM.load(SeqCst), 12); // because the CONST_ATOM in these lines are distinct

Use instead:

use std::sync::atomic::{AtomicUsize, Ordering::SeqCst};
const CONST_ATOM: AtomicUsize = AtomicUsize::new(12);

static STATIC_ATOM: AtomicUsize = CONST_ATOM;
STATIC_ATOM.store(9, SeqCst);
assert_eq!(STATIC_ATOM.load(SeqCst), 9); // use a `static` item to refer to the same instance

Configuration

  • ignore-interior-mutability: A list of paths to types that should be treated as if they do not contain interior mutability

    (default: ["bytes::Bytes"])

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of &Box<T> anywhere in the code. Check the Box documentation for more information.

Why is this bad?

A &Box<T> parameter requires the function caller to box T first before passing it to a function. Using &T defines a concrete type for the parameter and generalizes the function, this would also auto-deref to &T at the function call site if passed a &Box<T>.

Example

fn foo(bar: &Box<T>) { ... }

Better:

fn foo(bar: &T) { ... }
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of Box<T> where T is a collection such as Vec anywhere in the code. Check the Box documentation for more information.

Why is this bad?

Collections already keeps their contents in a separate area on the heap. So if you Box them, you just add another level of indirection without any benefit whatsoever.

Example

struct X {
    values: Box<Vec<Foo>>,
}

Better:

struct X {
    values: Vec<Foo>,
}

Past names

  • box_vec

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: 1.57.0

What it does

checks for Box::new(Default::default()), which can be written as Box::default().

Why is this bad?

Box::default() is equivalent and more concise.

Example

let x: Box<String> = Box::new(Default::default());

Use instead:

let x: Box<String> = Box::default();
Applicability: MachineApplicable(?)
Added in: 1.66.0

What it does

Checks for usage of Box<T> where an unboxed T would work fine.

Why is this bad?

This is an unnecessary allocation, and bad for performance. It is only necessary to allocate if you wish to move the box into something.

Example

fn foo(x: Box<u32>) {}

Use instead:

fn foo(x: u32) {}

Configuration

  • too-large-for-stack: The maximum size of objects (in bytes) that will be linted. Larger objects are ok on the heap

    (default: 200)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks if the if and else block contain shared code that can be moved out of the blocks.

Why is this bad?

Duplicate code is less maintainable.

Known problems

  • The lint doesn’t check if the moved expressions modify values that are being used in the if condition. The suggestion can in that case modify the behavior of the program. See rust-clippy#7452

Example

let foo = if … {
    println!("Hello World");
    13
} else {
    println!("Hello World");
    42
};

Use instead:

println!("Hello World");
let foo = if … {
    13
} else {
    42
};
Applicability: Unspecified(?)
Added in: 1.53.0

What it does

Warns if a generic shadows a built-in type.

Why is this bad?

This gives surprising type errors.

Example

impl<u32> Foo<u32> {
    fn impl_func(&self) -> u32 {
        42
    }
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for hard to read slices of byte characters, that could be more easily expressed as a byte string.

Why is this bad?

Potentially makes the string harder to read.

Example

&[b'H', b'e', b'l', b'l', b'o'];

Use instead:

b"Hello"
Applicability: MachineApplicable(?)
Added in: 1.81.0

What it does

It checks for str::bytes().count() and suggests replacing it with str::len().

Why is this bad?

str::bytes().count() is longer and may not be as performant as using str::len().

Example

"hello".bytes().count();
String::from("hello").bytes().count();

Use instead:

"hello".len();
String::from("hello").len();
Applicability: MachineApplicable(?)
Added in: 1.62.0

What it does

Checks for the use of .bytes().nth().

Why is this bad?

.as_bytes().get() is more efficient and more readable.

Example

"Hello".bytes().nth(3);

Use instead:

"Hello".as_bytes().get(3);
Applicability: MachineApplicable(?)
Added in: 1.52.0

What it does

Checks to see if all common metadata is defined in Cargo.toml. See: https://rust-lang-nursery.github.io/api-guidelines/documentation.html#cargotoml-includes-all-common-metadata-c-metadata

Why is this bad?

It will be more difficult for users to discover the purpose of the crate, and key information related to it.

Example

[package]
name = "clippy"
version = "0.0.212"
repository = "https://github.com/rust-lang/rust-clippy"
readme = "README.md"
license = "MIT OR Apache-2.0"
keywords = ["clippy", "lint", "plugin"]
categories = ["development-tools", "development-tools::cargo-plugins"]

Should include a description field like:

[package]
name = "clippy"
version = "0.0.212"
description = "A bunch of helpful lints to avoid common pitfalls in Rust"
repository = "https://github.com/rust-lang/rust-clippy"
readme = "README.md"
license = "MIT OR Apache-2.0"
keywords = ["clippy", "lint", "plugin"]
categories = ["development-tools", "development-tools::cargo-plugins"]

Configuration

  • cargo-ignore-publish: For internal testing only, ignores the current publish settings in the Cargo manifest.

    (default: false)

Applicability: Unspecified(?)
Added in: 1.32.0

What it does

Checks for calls to ends_with with possible file extensions and suggests to use a case-insensitive approach instead.

Why is this bad?

ends_with is case-sensitive and may not detect files with a valid extension.

Example

fn is_rust_file(filename: &str) -> bool {
    filename.ends_with(".rs")
}

Use instead:

fn is_rust_file(filename: &str) -> bool {
    let filename = std::path::Path::new(filename);
    filename.extension()
        .map_or(false, |ext| ext.eq_ignore_ascii_case("rs"))
}
Applicability: MaybeIncorrect(?)
Added in: 1.51.0

What it does

Checks for usage of the abs() method that cast the result to unsigned.

Why is this bad?

The unsigned_abs() method avoids panic when called on the MIN value.

Example

let x: i32 = -42;
let y: u32 = x.abs() as u32;

Use instead:

let x: i32 = -42;
let y: u32 = x.unsigned_abs();

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.62.0

What it does

Checks for casts from an enum tuple constructor to an integer.

Why is this bad?

The cast is easily confused with casting a c-like enum value to an integer.

Example

enum E { X(i32) };
let _ = E::X as usize;
Applicability: Unspecified(?)
Added in: 1.61.0

What it does

Checks for casts from an enum type to an integral type that will definitely truncate the value.

Why is this bad?

The resulting integral value will not match the value of the variant it came from.

Example

enum E { X = 256 };
let _ = E::X as u8;
Applicability: Unspecified(?)
Added in: 1.61.0

What it does

Checks for casts between numeric types that can be replaced by safe conversion functions.

Why is this bad?

Rust’s as keyword will perform many kinds of conversions, including silently lossy conversions. Conversion functions such as i32::from will only perform lossless conversions. Using the conversion functions prevents conversions from becoming silently lossy if the input types ever change, and makes it clear for people reading the code that the conversion is lossless.

Example

fn as_u64(x: u8) -> u64 {
    x as u64
}

Using ::from would look like this:

fn as_u64(x: u8) -> u64 {
    u64::from(x)
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for a known NaN float being cast to an integer

Why is this bad?

NaNs are cast into zero, so one could simply use this and make the code more readable. The lint could also hint at a programmer error.

Example

let _ = (0.0_f32 / 0.0) as u64;

Use instead:

let _ = 0_u64;
Applicability: Unspecified(?)
Added in: 1.66.0

What it does

Checks for casts between numeric types that may truncate large values. This is expected behavior, so the cast is Allow by default. It suggests user either explicitly ignore the lint, or use try_from() and handle the truncation, default, or panic explicitly.

Why is this bad?

In some problem domains, it is good practice to avoid truncation. This lint can be activated to help assess where additional checks could be beneficial.

Example

fn as_u8(x: u64) -> u8 {
    x as u8
}

Use instead:

fn as_u8(x: u64) -> u8 {
    if let Ok(x) = u8::try_from(x) {
        x
    } else {
        todo!();
    }
}
// Or
#[allow(clippy::cast_possible_truncation)]
fn as_u16(x: u64) -> u16 {
    x as u16
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for casts from an unsigned type to a signed type of the same size, or possibly smaller due to target-dependent integers. Performing such a cast is a no-op for the compiler (that is, nothing is changed at the bit level), and the binary representation of the value is reinterpreted. This can cause wrapping if the value is too big for the target signed type. However, the cast works as defined, so this lint is Allow by default.

Why is this bad?

While such a cast is not bad in itself, the results can be surprising when this is not the intended behavior:

Example

u32::MAX as i32; // will yield a value of `-1`
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for casts from any numeric type to a float type where the receiving type cannot store all values from the original type without rounding errors. This possible rounding is to be expected, so this lint is Allow by default.

Basically, this warns on casting any integer with 32 or more bits to f32 or any 64-bit integer to f64.

Why is this bad?

It’s not bad at all. But in some applications it can be helpful to know where precision loss can take place. This lint can help find those places in the code.

Example

let x = u64::MAX;
x as f64;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for casts, using as or pointer::cast, from a less strictly aligned pointer to a more strictly aligned pointer.

Why is this bad?

Dereferencing the resulting pointer may be undefined behavior.

Known problems

Using std::ptr::read_unaligned and std::ptr::write_unaligned or similar on the resulting pointer is fine. Is over-zealous: casts with manual alignment checks or casts like u64 -> u8 -> u16 can be fine. Miri is able to do a more in-depth analysis.

Example

let _ = (&1u8 as *const u8) as *const u16;
let _ = (&mut 1u8 as *mut u8) as *mut u16;

(&1u8 as *const u8).cast::<u16>();
(&mut 1u8 as *mut u8).cast::<u16>();
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for casts from a signed to an unsigned numeric type. In this case, negative values wrap around to large positive values, which can be quite surprising in practice. However, since the cast works as defined, this lint is Allow by default.

Why is this bad?

Possibly surprising results. You can activate this lint as a one-time check to see where numeric wrapping can arise.

Example

let y: i8 = -1;
y as u128; // will return 18446744073709551615
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for as casts between raw pointers to slices with differently sized elements.

Why is this bad?

The produced raw pointer to a slice does not update its length metadata. The produced pointer will point to a different number of bytes than the original pointer because the length metadata of a raw slice pointer is in elements rather than bytes. Producing a slice reference from the raw pointer will either create a slice with less data (which can be surprising) or create a slice with more data and cause Undefined Behavior.

Example

// Missing data

let a = [1_i32, 2, 3, 4];
let p = &a as *const [i32] as *const [u8];
unsafe {
    println!("{:?}", &*p);
}

// Undefined Behavior (note: also potential alignment issues)

let a = [1_u8, 2, 3, 4];
let p = &a as *const [u8] as *const [u32];
unsafe {
    println!("{:?}", &*p);
}

Instead use ptr::slice_from_raw_parts to construct a slice from a data pointer and the correct length

let a = [1_i32, 2, 3, 4];
let old_ptr = &a as *const [i32];
// The data pointer is cast to a pointer to the target `u8` not `[u8]`
// The length comes from the known length of 4 i32s times the 4 bytes per i32
let new_ptr = core::ptr::slice_from_raw_parts(old_ptr as *const u8, 16);
unsafe {
    println!("{:?}", &*new_ptr);
}
Applicability: HasPlaceholders(?)
Added in: 1.61.0

What it does

Checks for a raw slice being cast to a slice pointer

Why is this bad?

This can result in multiple &mut references to the same location when only a pointer is required. ptr::slice_from_raw_parts is a safe alternative that doesn’t require the same safety requirements to be upheld.

Example

let _: *const [u8] = std::slice::from_raw_parts(ptr, len) as *const _;
let _: *mut [u8] = std::slice::from_raw_parts_mut(ptr, len) as *mut _;

Use instead:

let _: *const [u8] = std::ptr::slice_from_raw_parts(ptr, len);
let _: *mut [u8] = std::ptr::slice_from_raw_parts_mut(ptr, len);
Applicability: MachineApplicable(?)
Added in: 1.65.0

What it does

Checks for usage of cfg that excludes code from test builds. (i.e., #[cfg(not(test))])

Why is this bad?

This may give the false impression that a codebase has 100% coverage, yet actually has untested code. Enabling this also guards against excessive mockery as well, which is an anti-pattern.

Example

#[cfg(not(test))]
important_check(); // I'm not actually tested, but not including me will falsely increase coverage!

Use instead:

important_check();
Applicability: Unspecified(?)
Added in: 1.81.0

What it does

Checks for expressions where a character literal is cast to u8 and suggests using a byte literal instead.

Why is this bad?

In general, casting values to smaller types is error-prone and should be avoided where possible. In the particular case of converting a character literal to u8, it is easy to avoid by just using a byte literal instead. As an added bonus, b'a' is also slightly shorter than 'a' as u8.

Example

'x' as u8

A better version, using the byte literal:

b'x'
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of _.chars().last() or _.chars().next_back() on a str to check if it ends with a given char.

Why is this bad?

Readability, this can be written more concisely as _.ends_with(_).

Example

name.chars().last() == Some('_') || name.chars().next_back() == Some('-');

Use instead:

name.ends_with('_') || name.ends_with('-');
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of .chars().next() on a str to check if it starts with a given char.

Why is this bad?

Readability, this can be written more concisely as _.starts_with(_).

Example

let name = "foo";
if name.chars().next() == Some('_') {};

Use instead:

let name = "foo";
if name.starts_with('_') {};
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for explicit bounds checking when casting.

Why is this bad?

Reduces the readability of statements & is error prone.

Example

foo <= i32::MAX as u32;

Use instead:

i32::try_from(foo).is_ok();

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.37.0

What it does

Checks for usage of .drain(..) for the sole purpose of clearing a container.

Why is this bad?

This creates an unnecessary iterator that is dropped immediately.

Calling .clear() also makes the intent clearer.

Example

let mut v = vec![1, 2, 3];
v.drain(..);

Use instead:

let mut v = vec![1, 2, 3];
v.clear();
Applicability: MachineApplicable(?)
Added in: 1.70.0

What it does

Checks for usage of .clone() on a Copy type.

Why is this bad?

The only reason Copy types implement Clone is for generics, not for using the clone method on a concrete type.

Example

42u64.clone();
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of .clone() on a ref-counted pointer, (Rc, Arc, rc::Weak, or sync::Weak), and suggests calling Clone via unified function syntax instead (e.g., Rc::clone(foo)).

Why restrict this?

Calling .clone() on an Rc, Arc, or Weak can obscure the fact that only the pointer is being cloned, not the underlying data.

Example

let x = Rc::new(1);

x.clone();

Use instead:

Rc::clone(&x);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of cloned() on an Iterator or Option where copied() could be used instead.

Why is this bad?

copied() is better because it guarantees that the type being cloned implements Copy.

Example

[1, 2, 3].iter().cloned();

Use instead:

[1, 2, 3].iter().copied();

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.53.0

What it does

This lint checks for equality comparisons with ptr::null

Why is this bad?

It’s easier and more readable to use the inherent .is_null() method instead

Example

use std::ptr;

if x == ptr::null {
    // ..
}

Use instead:

if x.is_null() {
    // ..
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for conversions to owned values just for the sake of a comparison.

Why is this bad?

The comparison can operate on a reference, so creating an owned value effectively throws it away directly afterwards, which is needlessly consuming code and heap space.

Example

if x.to_owned() == y {}

Use instead:

if x == y {}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for methods with high cognitive complexity.

Why is this bad?

Methods of high cognitive complexity tend to be hard to both read and maintain. Also LLVM will tend to optimize small methods better.

Known problems

Sometimes it’s hard to find a way to reduce the complexity.

Example

You’ll see it when you get the warning.

Past names

  • cyclomatic_complexity

Configuration

  • cognitive-complexity-threshold: The maximum cognitive complexity a function can have

    (default: 25)

Applicability: Unspecified(?)
Added in: 1.35.0

What it does

Checks for collapsible else { if ... } expressions that can be collapsed to else if ....

Why is this bad?

Each if-statement adds one level of nesting, which makes code look more complex than it really is.

Example


if x {
    …
} else {
    if y {
        …
    }
}

Should be written:

if x {
    …
} else if y {
    …
}
Applicability: MachineApplicable(?)
Added in: 1.51.0

What it does

Checks for nested if statements which can be collapsed by &&-combining their conditions.

Why is this bad?

Each if-statement adds one level of nesting, which makes code look more complex than it really is.

Example

if x {
    if y {
        // …
    }
}

Use instead:

if x && y {
    // …
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Finds nested match or if let expressions where the patterns may be “collapsed” together without adding any branches.

Note that this lint is not intended to find all cases where nested match patterns can be merged, but only cases where merging would most likely make the code more readable.

Why is this bad?

It is unnecessarily verbose and complex.

Example

fn func(opt: Option<Result<u64, String>>) {
    let n = match opt {
        Some(n) => match n {
            Ok(n) => n,
            _ => return,
        }
        None => return,
    };
}

Use instead:

fn func(opt: Option<Result<u64, String>>) {
    let n = match opt {
        Some(Ok(n)) => n,
        _ => return,
    };
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: Unspecified(?)
Added in: 1.50.0

What it does

Checks for consecutive calls to str::replace (2 or more) that can be collapsed into a single call.

Why is this bad?

Consecutive str::replace calls scan the string multiple times with repetitive code.

Example

let hello = "hesuo worpd"
    .replace('s', "l")
    .replace("u", "l")
    .replace('p', "l");

Use instead:

let hello = "hesuo worpd".replace(['s', 'u', 'p'], "l");

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.65.0

What it does

Checks for collections that are never queried.

Why is this bad?

Putting effort into constructing a collection but then never querying it might indicate that the author forgot to do whatever they intended to do with the collection. Example: Clone a vector, sort it for iteration, but then mistakenly iterate the original vector instead.

Example

let mut sorted_samples = samples.clone();
sorted_samples.sort();
for sample in &samples { // Oops, meant to use `sorted_samples`.
    println!("{sample}");
}

Use instead:

let mut sorted_samples = samples.clone();
sorted_samples.sort();
for sample in &sorted_samples {
    println!("{sample}");
}
Applicability: Unspecified(?)
Added in: 1.70.0

What it does

Checks comparison chains written with if that can be rewritten with match and cmp.

Why is this bad?

if is not guaranteed to be exhaustive and conditionals can get repetitive

Known problems

The match statement may be slower due to the compiler not inlining the call to cmp. See issue #5354

Example

fn f(x: u8, y: u8) {
    if x > y {
        a()
    } else if x < y {
        b()
    } else {
        c()
    }
}

Use instead:

use std::cmp::Ordering;
fn f(x: u8, y: u8) {
     match x.cmp(&y) {
         Ordering::Greater => a(),
         Ordering::Less => b(),
         Ordering::Equal => c()
     }
}
Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Checks for comparing to an empty slice such as "" or [], and suggests using .is_empty() where applicable.

Why is this bad?

Some structures can answer .is_empty() much faster than checking for equality. So it is good to get into the habit of using .is_empty(), and having it is cheap. Besides, it makes the intent clearer than a manual comparison in some contexts.

Example

if s == "" {
    ..
}

if arr == [] {
    ..
}

Use instead:

if s.is_empty() {
    ..
}

if arr.is_empty() {
    ..
}
Applicability: MachineApplicable(?)
Added in: 1.49.0

What it does

It identifies calls to .is_empty() on constant values.

Why is this bad?

String literals and constant values are known at compile time. Checking if they are empty will always return the same value. This might not be the intention of the expression.

Example

let value = "";
if value.is_empty() {
    println!("the string is empty");
}

Use instead:

println!("the string is empty");
Applicability: Unspecified(?)
Added in: 1.79.0

What it does

Checks for types that implement Copy as well as Iterator.

Why is this bad?

Implicit copies can be confusing when working with iterator combinators.

Example

#[derive(Copy, Clone)]
struct Countdown(u8);

impl Iterator for Countdown {
    // ...
}

let a: Vec<_> = my_iterator.take(1).collect();
let b: Vec<_> = my_iterator.collect();
Applicability: Unspecified(?)
Added in: 1.30.0

What it does

Checks for usage of crate as opposed to $crate in a macro definition.

Why is this bad?

crate refers to the macro call’s crate, whereas $crate refers to the macro definition’s crate. Rarely is the former intended. See: https://doc.rust-lang.org/reference/macros-by-example.html#hygiene

Example

#[macro_export]
macro_rules! print_message {
    () => {
        println!("{}", crate::MESSAGE);
    };
}
pub const MESSAGE: &str = "Hello!";

Use instead:

#[macro_export]
macro_rules! print_message {
    () => {
        println!("{}", $crate::MESSAGE);
    };
}
pub const MESSAGE: &str = "Hello!";

Note that if the use of crate is intentional, an allow attribute can be applied to the macro definition, e.g.:

#[allow(clippy::crate_in_macro_def)]
macro_rules! ok { ... crate::foo ... }
Applicability: MachineApplicable(?)
Added in: 1.62.0

What it does

Checks usage of std::fs::create_dir and suggest using std::fs::create_dir_all instead.

Why restrict this?

Sometimes std::fs::create_dir is mistakenly chosen over std::fs::create_dir_all, resulting in failure when more than one directory needs to be created or when the directory already exists. Crates which never need to specifically create a single directory may wish to prevent this mistake.

Example

std::fs::create_dir("foo");

Use instead:

std::fs::create_dir_all("foo");
Applicability: MaybeIncorrect(?)
Added in: 1.48.0

What it does

Checks for transmutes between a type T and *T.

Why is this bad?

It’s easy to mistakenly transmute between a type and a pointer to that type.

Example

core::intrinsics::transmute(t) // where the result type is the same as
                               // `*t` or `&t`'s
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of the dbg! macro.

Why restrict this?

The dbg! macro is intended as a debugging tool. It should not be present in released software or committed to a version control system.

Example

dbg!(true)

Use instead:

true

Configuration

  • allow-dbg-in-tests: Whether dbg! should be allowed in test functions or #[cfg(test)]

    (default: false)

Applicability: MachineApplicable(?)
Added in: 1.34.0

What it does

Checks for function/method calls with a mutable parameter in debug_assert!, debug_assert_eq! and debug_assert_ne! macros.

Why is this bad?

In release builds debug_assert! macros are optimized out by the compiler. Therefore mutating something in a debug_assert! macro results in different behavior between a release and debug build.

Example

debug_assert_eq!(vec![3].pop(), Some(3));

// or

debug_assert!(takes_a_mut_parameter(&mut x));
Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Warns if there is a better representation for a numeric literal.

Why restrict this?

Especially for big powers of 2, a hexadecimal representation is usually more readable than a decimal representation.

Example

`255` => `0xFF`
`65_535` => `0xFFFF`
`4_042_322_160` => `0xF0F0_F0F0`

Configuration

  • literal-representation-threshold: The lower bound for linting decimal literals

    (default: 16384)

Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for declaration of const items which is interior mutable (e.g., contains a Cell, Mutex, AtomicXxxx, etc.).

Why is this bad?

Consts are copied everywhere they are referenced, i.e., every time you refer to the const a fresh instance of the Cell or Mutex or AtomicXxxx will be created, which defeats the whole purpose of using these types in the first place.

The const should better be replaced by a static item if a global variable is wanted, or replaced by a const fn if a constructor is wanted.

Known problems

A “non-constant” const item is a legacy way to supply an initialized value to downstream static items (e.g., the std::sync::ONCE_INIT constant). In this case the use of const is legit, and this lint should be suppressed.

Even though the lint avoids triggering on a constant whose type has enums that have variants with interior mutability, and its value uses non interior mutable variants (see #3962 and #3825 for examples); it complains about associated constants without default values only based on its types; which might not be preferable. There’re other enums plus associated constants cases that the lint cannot handle.

Types that have underlying or potential interior mutability trigger the lint whether the interior mutable field is used or not. See issue #5812

Example

use std::sync::atomic::{AtomicUsize, Ordering::SeqCst};

const CONST_ATOM: AtomicUsize = AtomicUsize::new(12);
CONST_ATOM.store(6, SeqCst); // the content of the atomic is unchanged
assert_eq!(CONST_ATOM.load(SeqCst), 12); // because the CONST_ATOM in these lines are distinct

Use instead:

static STATIC_ATOM: AtomicUsize = AtomicUsize::new(15);
STATIC_ATOM.store(9, SeqCst);
assert_eq!(STATIC_ATOM.load(SeqCst), 9); // use a `static` item to refer to the same instance

Configuration

  • ignore-interior-mutability: A list of paths to types that should be treated as if they do not contain interior mutability

    (default: ["bytes::Bytes"])

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for construction on unit struct using default.

Why is this bad?

This adds code complexity and an unnecessary function call.

Example

#[derive(Default)]
struct S<T> {
    _marker: PhantomData<T>
}

let _: S<i32> = S {
    _marker: PhantomData::default()
};

Use instead:

struct S<T> {
    _marker: PhantomData<T>
}

let _: S<i32> = S {
    _marker: PhantomData
};
Applicability: MachineApplicable(?)
Added in: 1.71.0

What it does

It checks for std::iter::Empty::default() and suggests replacing it with std::iter::empty().

Why is this bad?

std::iter::empty() is the more idiomatic way.

Example

let _ = std::iter::Empty::<usize>::default();
let iter: std::iter::Empty<usize> = std::iter::Empty::default();

Use instead:

let _ = std::iter::empty::<usize>();
let iter: std::iter::Empty<usize> = std::iter::empty();
Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

Checks for usage of unconstrained numeric literals which may cause default numeric fallback in type inference.

Default numeric fallback means that if numeric types have not yet been bound to concrete types at the end of type inference, then integer type is bound to i32, and similarly floating type is bound to f64.

See RFC0212 for more information about the fallback.

Why restrict this?

To ensure that every numeric type is chosen explicitly rather than implicitly.

Known problems

This lint can only be allowed at the function level or above.

Example

let i = 10;
let f = 1.23;

Use instead:

let i = 10i32;
let f = 1.23f64;
Applicability: MaybeIncorrect(?)
Added in: 1.52.0

What it does

Checks for literal calls to Default::default().

Why is this bad?

It’s easier for the reader if the name of the type is used, rather than the generic Default.

Example

let s: String = Default::default();

Use instead:

let s = String::default();
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Displays a warning when a union is declared with the default representation (without a #[repr(C)] attribute).

Why restrict this?

Unions in Rust have unspecified layout by default, despite many people thinking that they lay out each field at the start of the union (like C does). That is, there are no guarantees about the offset of the fields for unions with multiple non-ZST fields without an explicitly specified layout. These cases may lead to undefined behavior in unsafe blocks.

Example

union Foo {
    a: i32,
    b: u32,
}

fn main() {
    let _x: u32 = unsafe {
        Foo { a: 0_i32 }.b // Undefined behavior: `b` is allowed to be padding
    };
}

Use instead:

#[repr(C)]
union Foo {
    a: i32,
    b: u32,
}

fn main() {
    let _x: u32 = unsafe {
        Foo { a: 0_i32 }.b // Now defined behavior, this is just an i32 -> u32 transmute
    };
}
Applicability: Unspecified(?)
Added in: 1.60.0

What it does

Checks for #[cfg_attr(rustfmt, rustfmt_skip)] and suggests to replace it with #[rustfmt::skip].

Why is this bad?

Since tool_attributes (rust-lang/rust#44690) are stable now, they should be used instead of the old cfg_attr(rustfmt) attributes.

Known problems

This lint doesn’t detect crate level inner attributes, because they get processed before the PreExpansionPass lints get executed. See #3123

Example

#[cfg_attr(rustfmt, rustfmt_skip)]
fn main() { }

Use instead:

#[rustfmt::skip]
fn main() { }

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.32.0

What it does

Checks for #[cfg_attr(feature = "cargo-clippy", ...)] and for #[cfg(feature = "cargo-clippy")] and suggests to replace it with #[cfg_attr(clippy, ...)] or #[cfg(clippy)].

Why is this bad?

This feature has been deprecated for years and shouldn’t be used anymore.

Example

#[cfg(feature = "cargo-clippy")]
struct Bar;

Use instead:

#[cfg(clippy)]
struct Bar;
Applicability: MachineApplicable(?)
Added in: 1.78.0

What it does

Checks for #[deprecated] annotations with a since field that is not a valid semantic version. Also allows “TBD” to signal future deprecation.

Why is this bad?

For checking the version of the deprecation, it must be a valid semver. Failing that, the contained information is useless.

Example

#[deprecated(since = "forever")]
fn something_else() { /* ... */ }
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of *& and *&mut in expressions.

Why is this bad?

Immediately dereferencing a reference is no-op and makes the code less clear.

Known problems

Multiple dereference/addrof pairs are not handled so the suggested fix for x = **&&y is x = *&y, which is still incorrect.

Example

let a = f(*&mut b);
let c = *&d;

Use instead:

let a = f(b);
let c = d;
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for slicing expressions which are equivalent to dereferencing the value.

Why restrict this?

Some people may prefer to dereference rather than slice.

Example

let vec = vec![1, 2, 3];
let slice = &vec[..];

Use instead:

let vec = vec![1, 2, 3];
let slice = &*vec;
Applicability: MachineApplicable(?)
Added in: 1.61.0

What it does

Detects manual std::default::Default implementations that are identical to a derived implementation.

Why is this bad?

It is less concise.

Example

struct Foo {
    bar: bool
}

impl Default for Foo {
    fn default() -> Self {
        Self {
            bar: false
        }
    }
}

Use instead:

#[derive(Default)]
struct Foo {
    bar: bool
}

Known problems

Derive macros sometimes use incorrect bounds in generic types and the user defined impl may be more generalized or specialized than what derive will produce. This lint can’t detect the manual impl has exactly equal bounds, and therefore this lint is disabled for types with generic parameters.

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.57.0

What it does

Lints against manual PartialOrd and Ord implementations for types with a derived Ord or PartialOrd implementation.

Why is this bad?

The implementation of these traits must agree (for example for use with sort) so it’s probably a bad idea to use a default-generated Ord implementation with an explicitly defined PartialOrd. In particular, the following must hold for any type implementing Ord:

k1.cmp(&k2) == k1.partial_cmp(&k2).unwrap()

Example

#[derive(Ord, PartialEq, Eq)]
struct Foo;

impl PartialOrd for Foo {
    ...
}

Use instead:

#[derive(PartialEq, Eq)]
struct Foo;

impl PartialOrd for Foo {
    fn partial_cmp(&self, other: &Foo) -> Option<Ordering> {
       Some(self.cmp(other))
    }
}

impl Ord for Foo {
    ...
}

or, if you don’t need a custom ordering:

#[derive(Ord, PartialOrd, PartialEq, Eq)]
struct Foo;
Applicability: Unspecified(?)
Added in: 1.47.0

What it does

Checks for types that derive PartialEq and could implement Eq.

Why is this bad?

If a type T derives PartialEq and all of its members implement Eq, then T can always implement Eq. Implementing Eq allows T to be used in APIs that require Eq types. It also allows structs containing T to derive Eq themselves.

Example

#[derive(PartialEq)]
struct Foo {
    i_am_eq: i32,
    i_am_eq_too: Vec<String>,
}

Use instead:

#[derive(PartialEq, Eq)]
struct Foo {
    i_am_eq: i32,
    i_am_eq_too: Vec<String>,
}
Applicability: MachineApplicable(?)
Added in: 1.63.0

What it does

Lints against manual PartialEq implementations for types with a derived Hash implementation.

Why is this bad?

The implementation of these traits must agree (for example for use with HashMap) so it’s probably a bad idea to use a default-generated Hash implementation with an explicitly defined PartialEq. In particular, the following must hold for any type:

k1 == k2 ⇒ hash(k1) == hash(k2)

Example

#[derive(Hash)]
struct Foo;

impl PartialEq for Foo {
    ...
}

Past names

  • derive_hash_xor_eq
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Denies the configured macros in clippy.toml

Note: Even though this lint is warn-by-default, it will only trigger if macros are defined in the clippy.toml file.

Why is this bad?

Some macros are undesirable in certain contexts, and it’s beneficial to lint for them as needed.

Example

An example clippy.toml configuration:

disallowed-macros = [
    # Can use a string as the path of the disallowed macro.
    "std::print",
    # Can also use an inline table with a `path` key.
    { path = "std::println" },
    # When using an inline table, can add a `reason` for why the macro
    # is disallowed.
    { path = "serde::Serialize", reason = "no serializing" },
]
use serde::Serialize;

// Example code where clippy issues a warning
println!("warns");

// The diagnostic will contain the message "no serializing"
#[derive(Serialize)]
struct Data {
    name: String,
    value: usize,
}

Configuration

  • disallowed-macros: The list of disallowed macros, written as fully qualified paths.

    (default: [])

Applicability: Unspecified(?)
Added in: 1.66.0

What it does

Denies the configured methods and functions in clippy.toml

Note: Even though this lint is warn-by-default, it will only trigger if methods are defined in the clippy.toml file.

Why is this bad?

Some methods are undesirable in certain contexts, and it’s beneficial to lint for them as needed.

Example

An example clippy.toml configuration:

disallowed-methods = [
    # Can use a string as the path of the disallowed method.
    "std::boxed::Box::new",
    # Can also use an inline table with a `path` key.
    { path = "std::time::Instant::now" },
    # When using an inline table, can add a `reason` for why the method
    # is disallowed.
    { path = "std::vec::Vec::leak", reason = "no leaking memory" },
]
// Example code where clippy issues a warning
let xs = vec![1, 2, 3, 4];
xs.leak(); // Vec::leak is disallowed in the config.
// The diagnostic contains the message "no leaking memory".

let _now = Instant::now(); // Instant::now is disallowed in the config.

let _box = Box::new(3); // Box::new is disallowed in the config.

Use instead:

// Example code which does not raise clippy warning
let mut xs = Vec::new(); // Vec::new is _not_ disallowed in the config.
xs.push(123); // Vec::push is _not_ disallowed in the config.

Past names

  • disallowed_method

Configuration

  • disallowed-methods: The list of disallowed methods, written as fully qualified paths.

    (default: [])

Applicability: Unspecified(?)
Added in: 1.49.0

What it does

Checks for usage of disallowed names for variables, such as foo.

Why is this bad?

These names are usually placeholder names and should be avoided.

Example

let foo = 3.14;

Past names

  • blacklisted_name

Configuration

  • disallowed-names: The list of disallowed names to lint about. NB: bar is not here since it has legitimate uses. The value ".." can be used as part of the list to indicate that the configured values should be appended to the default configuration of Clippy. By default, any configuration will replace the default value.

    (default: ["foo", "baz", "quux"])

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of unicode scripts other than those explicitly allowed by the lint config.

This lint doesn’t take into account non-text scripts such as Unknown and Linear_A. It also ignores the Common script type. While configuring, be sure to use official script name aliases from the list of supported scripts.

See also: non_ascii_idents.

Why restrict this?

It may be not desired to have many different scripts for identifiers in the codebase.

Note that if you only want to allow typical English, you might want to use built-in non_ascii_idents lint instead.

Example

// Assuming that `clippy.toml` contains the following line:
// allowed-scripts = ["Latin", "Cyrillic"]
let counter = 10; // OK, latin is allowed.
let счётчик = 10; // OK, cyrillic is allowed.
let zähler = 10; // OK, it's still latin.
let カウンタ = 10; // Will spawn the lint.

Configuration

  • allowed-scripts: The list of unicode scripts allowed to be used in the scope.

    (default: ["Latin"])

Applicability: Unspecified(?)
Added in: 1.55.0

What it does

Denies the configured types in clippy.toml.

Note: Even though this lint is warn-by-default, it will only trigger if types are defined in the clippy.toml file.

Why is this bad?

Some types are undesirable in certain contexts.

Example:

An example clippy.toml configuration:

disallowed-types = [
    # Can use a string as the path of the disallowed type.
    "std::collections::BTreeMap",
    # Can also use an inline table with a `path` key.
    { path = "std::net::TcpListener" },
    # When using an inline table, can add a `reason` for why the type
    # is disallowed.
    { path = "std::net::Ipv4Addr", reason = "no IPv4 allowed" },
]
use std::collections::BTreeMap;
// or its use
let x = std::collections::BTreeMap::new();

Use instead:

// A similar type that is allowed by the config
use std::collections::HashMap;

Past names

  • disallowed_type

Configuration

  • disallowed-types: The list of disallowed types, written as fully qualified paths.

    (default: [])

Applicability: Unspecified(?)
Added in: 1.55.0

What it does

Checks for diverging calls that are not match arms or statements.

Why is this bad?

It is often confusing to read. In addition, the sub-expression evaluation order for Rust is not well documented.

Known problems

Someone might want to use some_bool || panic!() as a shorthand.

Example

let a = b() || panic!() || c();
// `c()` is dead, `panic!()` is only called if `b()` returns `false`
let x = (a, b, c, panic!());
// can simply be replaced by `panic!()`
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

In CommonMark Markdown, the language used to write doc comments, a paragraph nested within a list or block quote does not need any line after the first one to be indented or marked. The specification calls this a “lazy paragraph continuation.”

Why is this bad?

This is easy to write but hard to read. Lazy continuations makes unintended markers hard to see, and make it harder to deduce the document’s intended structure.

Example

This table is probably intended to have two rows, but it does not. It has zero rows, and is followed by a block quote.

/// Range | Description
/// ----- | -----------
/// >= 1  | fully opaque
/// < 1   | partially see-through
fn set_opacity(opacity: f32) {}

Fix it by escaping the marker:

/// Range | Description
/// ----- | -----------
/// \>= 1 | fully opaque
/// < 1   | partially see-through
fn set_opacity(opacity: f32) {}

This example is actually intended to be a list:

/// * Do nothing.
/// * Then do something. Whatever it is needs done,
/// it should be done right now.

Fix it by indenting the list contents:

/// * Do nothing.
/// * Then do something. Whatever it is needs done,
///   it should be done right now.
Applicability: MachineApplicable(?)
Added in: 1.80.0

What it does

Checks for the presence of _, :: or camel-case words outside ticks in documentation.

Why is this bad?

Rustdoc supports markdown formatting, _, :: and camel-case probably indicates some code which should be included between ticks. _ can also be used for emphasis in markdown, this lint tries to consider that.

Known problems

Lots of bad docs won’t be fixed, what the lint checks for is limited, and there are still false positives. HTML elements and their content are not linted.

In addition, when writing documentation comments, including [] brackets inside a link text would trip the parser. Therefore, documenting link with [SmallVec<[T; INLINE_CAPACITY]>] and then [SmallVec<[T; INLINE_CAPACITY]>]: SmallVec would fail.

Examples

/// Do something with the foo_bar parameter. See also
/// that::other::module::foo.
// ^ `foo_bar` and `that::other::module::foo` should be ticked.
fn doit(foo_bar: usize) {}
// Link text with `[]` brackets should be written as following:
/// Consume the array and return the inner
/// [`SmallVec<[T; INLINE_CAPACITY]>`][SmallVec].
/// [SmallVec]: SmallVec
fn main() {}

Configuration

  • doc-valid-idents: The list of words this lint should not consider as identifiers needing ticks. The value ".." can be used as part of the list to indicate, that the configured values should be appended to the default configuration of Clippy. By default, any configuration will replace the default value. For example:
  • doc-valid-idents = ["ClipPy"] would replace the default list with ["ClipPy"].

  • doc-valid-idents = ["ClipPy", ".."] would append ClipPy to the default list.

    (default: ["KiB", "MiB", "GiB", "TiB", "PiB", "EiB", "MHz", "GHz", "THz", "AccessKit", "CoAP", "CoreFoundation", "CoreGraphics", "CoreText", "DevOps", "Direct2D", "Direct3D", "DirectWrite", "DirectX", "ECMAScript", "GPLv2", "GPLv3", "GitHub", "GitLab", "IPv4", "IPv6", "ClojureScript", "CoffeeScript", "JavaScript", "PostScript", "PureScript", "TypeScript", "WebAssembly", "NaN", "NaNs", "OAuth", "GraphQL", "OCaml", "OpenAL", "OpenDNS", "OpenGL", "OpenMP", "OpenSSH", "OpenSSL", "OpenStreetMap", "OpenTelemetry", "OpenType", "WebGL", "WebGL2", "WebGPU", "WebRTC", "WebSocket", "WebTransport", "WebP", "OpenExr", "YCbCr", "sRGB", "TensorFlow", "TrueType", "iOS", "macOS", "FreeBSD", "NetBSD", "OpenBSD", "TeX", "LaTeX", "BibTeX", "BibLaTeX", "MinGW", "CamelCase"])

Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for double comparisons that could be simplified to a single expression.

Why is this bad?

Readability.

Example

if x == y || x < y {}

Use instead:

if x <= y {}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for a #[must_use] attribute without further information on functions and methods that return a type already marked as #[must_use].

Why is this bad?

The attribute isn’t needed. Not using the result will already be reported. Alternatively, one can add some text to the attribute to improve the lint message.

Examples

#[must_use]
fn double_must_use() -> Result<(), ()> {
    unimplemented!();
}
Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Detects expressions of the form --x.

Why is this bad?

It can mislead C/C++ programmers to think x was decremented.

Example

let mut x = 3;
--x;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for unnecessary double parentheses.

Why is this bad?

This makes code harder to read and might indicate a mistake.

Example

fn simple_double_parens() -> i32 {
    ((0))
}

foo((0));

Use instead:

fn simple_no_parens() -> i32 {
    0
}

foo(0);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for calls to .drain() that clear the collection, immediately followed by a call to .collect().

“Collection” in this context refers to any type with a drain method: Vec, VecDeque, BinaryHeap, HashSet,HashMap, String

Why is this bad?

Using mem::take is faster as it avoids the allocation. When using mem::take, the old collection is replaced with an empty one and ownership of the old collection is returned.

Known issues

mem::take(&mut vec) is almost equivalent to vec.drain(..).collect(), except that it also moves the capacity. The user might have explicitly written it this way to keep the capacity on the original Vec.

Example

fn remove_all(v: &mut Vec<i32>) -> Vec<i32> {
    v.drain(..).collect()
}

Use instead:

use std::mem;
fn remove_all(v: &mut Vec<i32>) -> Vec<i32> {
    mem::take(v)
}
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Checks for calls to std::mem::drop with a value that does not implement Drop.

Why is this bad?

Calling std::mem::drop is no different than dropping such a type. A different value may have been intended.

Example

struct Foo;
let x = Foo;
std::mem::drop(x);
Applicability: Unspecified(?)
Added in: 1.62.0

What it does

Checks for files that are included as modules multiple times.

Why is this bad?

Loading a file as a module more than once causes it to be compiled multiple times, taking longer and putting duplicate content into the module tree.

Example

// lib.rs
mod a;
mod b;
// a.rs
#[path = "./b.rs"]
mod b;

Use instead:

// lib.rs
mod a;
mod b;
// a.rs
use crate::b;
Applicability: Unspecified(?)
Added in: 1.63.0

What it does

Checks for function arguments having the similar names differing by an underscore.

Why is this bad?

It affects code readability.

Example

fn foo(a: i32, _a: i32) {}

Use instead:

fn bar(a: i32, _b: i32) {}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for attributes that appear two or more times.

Why is this bad?

Repeating an attribute on the same item (or globally on the same crate) is unnecessary and doesn’t have an effect.

Example

#[allow(dead_code)]
#[allow(dead_code)]
fn foo() {}

Use instead:

#[allow(dead_code)]
fn foo() {}
Applicability: Unspecified(?)
Added in: 1.79.0

What it does

Checks for calculation of subsecond microseconds or milliseconds from other Duration methods.

Why is this bad?

It’s more concise to call Duration::subsec_micros() or Duration::subsec_millis() than to calculate them.

Example

let micros = duration.subsec_nanos() / 1_000;
let millis = duration.subsec_nanos() / 1_000_000;

Use instead:

let micros = duration.subsec_micros();
let millis = duration.subsec_millis();
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for integer validity checks, followed by a transmute that is (incorrectly) evaluated eagerly (e.g. using bool::then_some).

Why is this bad?

Eager evaluation means that the transmute call is executed regardless of whether the condition is true or false. This can introduce unsoundness and other subtle bugs.

Example

Consider the following function which is meant to convert an unsigned integer to its enum equivalent via transmute.

#[repr(u8)]
enum Opcode {
    Add = 0,
    Sub = 1,
    Mul = 2,
    Div = 3
}

fn int_to_opcode(op: u8) -> Option<Opcode> {
    (op < 4).then_some(unsafe { std::mem::transmute(op) })
}

This may appear fine at first given that it checks that the u8 is within the validity range of the enum, however the transmute is evaluated eagerly, meaning that it executes even if op >= 4!

This makes the function unsound, because it is possible for the caller to cause undefined behavior (creating an enum with an invalid bitpattern) entirely in safe code only by passing an incorrect value, which is normally only a bug that is possible in unsafe code.

One possible way in which this can go wrong practically is that the compiler sees it as:

let temp: Foo = unsafe { std::mem::transmute(op) };
(0 < 4).then_some(temp)

and optimizes away the (0 < 4) check based on the assumption that since a Foo was created from op with the validity range 0..3, it is impossible for this condition to be false.

In short, it is possible for this function to be optimized in a way that makes it never return None, even if passed the value 4.

This can be avoided by instead using lazy evaluation. For the example above, this should be written:

fn int_to_opcode(op: u8) -> Option<Opcode> {
    (op < 4).then(|| unsafe { std::mem::transmute(op) })
             ^^^^ ^^ `bool::then` only executes the closure if the condition is true!
}
Applicability: MaybeIncorrect(?)
Added in: 1.77.0

What it does

Checks for usage of if expressions with an else if branch, but without a final else branch.

Why restrict this?

Some coding guidelines require this (e.g., MISRA-C:2004 Rule 14.10).

Example

if x.is_positive() {
    a();
} else if x.is_negative() {
    b();
}

Use instead:

if x.is_positive() {
    a();
} else if x.is_negative() {
    b();
} else {
    // We don't care about zero.
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Detects documentation that is empty.

Why is this bad?

Empty docs clutter code without adding value, reducing readability and maintainability.

Example

///
fn returns_true() -> bool {
    true
}

Use instead:

fn returns_true() -> bool {
    true
}
Applicability: Unspecified(?)
Added in: 1.78.0

What it does

Checks for empty Drop implementations.

Why restrict this?

Empty Drop implementations have no effect when dropping an instance of the type. They are most likely useless. However, an empty Drop implementation prevents a type from being destructured, which might be the intention behind adding the implementation as a marker.

Example

struct S;

impl Drop for S {
    fn drop(&mut self) {}
}

Use instead:

struct S;
Applicability: MaybeIncorrect(?)
Added in: 1.62.0

What it does

Checks for enums with no variants, which therefore are uninhabited types (cannot be instantiated).

As of this writing, the never_type is still a nightly-only experimental API. Therefore, this lint is only triggered if #![feature(never_type)] is enabled.

Why is this bad?

  • If you only want a type which can’t be instantiated, you should use ! (the primitive type “never”), because ! has more extensive compiler support (type inference, etc.) and implementations of common traits.

  • If you need to introduce a distinct type, consider using a newtype struct containing ! instead (struct MyType(pub !)), because it is more idiomatic to use a struct rather than an enum when an enum is unnecessary.

    If you do this, note that the visibility of the ! field determines whether the uninhabitedness is visible in documentation, and whether it can be pattern matched to mark code unreachable. If the field is not visible, then the struct acts like any other struct with private fields.

  • If the enum has no variants only because all variants happen to be disabled by conditional compilation, then it would be appropriate to allow the lint, with #[allow(empty_enum)].

For further information, visit the never type’s documentation.

Example

enum CannotExist {}

Use instead:

#![feature(never_type)]

/// Use the `!` type directly...
type CannotExist = !;

/// ...or define a newtype which is distinct.
struct CannotExist2(pub !);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Finds enum variants without fields that are declared with empty brackets.

Why restrict this?

Empty brackets after a enum variant declaration are redundant and can be omitted, and it may be desirable to do so consistently for style.

However, removing the brackets also introduces a public constant named after the variant, so this is not just a syntactic simplification but an API change, and adding them back is a breaking API change.

Example

enum MyEnum {
    HasData(u8),
    HasNoData(),       // redundant parentheses
    NoneHereEither {}, // redundant braces
}

Use instead:

enum MyEnum {
    HasData(u8),
    HasNoData,
    NoneHereEither,
}
Applicability: MaybeIncorrect(?)
Added in: 1.77.0

What it does

Checks for empty lines after doc comments.

Why is this bad?

The doc comment may have meant to be an inner doc comment, regular comment or applied to some old code that is now commented out. If it was intended to be a doc comment, then the empty line should be removed.

Example

/// Some doc comment with a blank line after it.

fn f() {}

/// Docs for `old_code`
// fn old_code() {}

fn new_code() {}

Use instead:

//! Convert it to an inner doc comment

// Or a regular comment

/// Or remove the empty line
fn f() {}

// /// Docs for `old_code`
// fn old_code() {}

fn new_code() {}
Applicability: MaybeIncorrect(?)
Added in: 1.70.0

What it does

Checks for empty lines after outer attributes

Why is this bad?

The attribute may have meant to be an inner attribute (#![attr]). If it was meant to be an outer attribute (#[attr]) then the empty line should be removed

Example

#[allow(dead_code)]

fn not_quite_good_code() {}

Use instead:

// Good (as inner attribute)
#![allow(dead_code)]

fn this_is_fine() {}

// or

// Good (as outer attribute)
#[allow(dead_code)]
fn this_is_fine_too() {}
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for empty loop expressions.

Why is this bad?

These busy loops burn CPU cycles without doing anything. It is almost always a better idea to panic! than to have a busy loop.

If panicking isn’t possible, think of the environment and either:

  • block on something
  • sleep the thread for some microseconds
  • yield or pause the thread

For std targets, this can be done with std::thread::sleep or std::thread::yield_now.

For no_std targets, doing this is more complicated, especially because #[panic_handler]s can’t panic. To stop/pause the thread, you will probably need to invoke some target-specific intrinsic. Examples include:

Example

loop {}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Finds structs without fields (a so-called “empty struct”) that are declared with brackets.

Why restrict this?

Empty brackets after a struct declaration can be omitted, and it may be desirable to do so consistently for style.

However, removing the brackets also introduces a public constant named after the struct, so this is not just a syntactic simplification but an API change, and adding them back is a breaking API change.

Example

struct Cookie {}
struct Biscuit();

Use instead:

struct Cookie;
struct Biscuit;
Applicability: Unspecified(?)
Added in: 1.62.0

What it does

Checks for C-like enumerations that are repr(isize/usize) and have values that don’t fit into an i32.

Why is this bad?

This will truncate the variant value on 32 bit architectures, but works fine on 64 bit.

Example

#[repr(usize)]
enum NonPortable {
    X = 0x1_0000_0000,
    Y = 0,
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for use Enum::*.

Why is this bad?

It is usually better style to use the prefixed name of an enumeration variant, rather than importing variants.

Known problems

Old-style enumerations that prefix the variants are still around.

Example

use std::cmp::Ordering::*;

foo(Less);

Use instead:

use std::cmp::Ordering;

foo(Ordering::Less)
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Detects enumeration variants that are prefixed or suffixed by the same characters.

Why is this bad?

Enumeration variant names should specify their variant, not repeat the enumeration name.

Limitations

Characters with no casing will be considered when comparing prefixes/suffixes This applies to numbers and non-ascii characters without casing e.g. Foo1 and Foo2 is considered to have different prefixes (the prefixes are Foo1 and Foo2 respectively), as also Bar螃, Bar蟹

Example

enum Cake {
    BlackForestCake,
    HummingbirdCake,
    BattenbergCake,
}

Use instead:

enum Cake {
    BlackForest,
    Hummingbird,
    Battenberg,
}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

  • enum-variant-name-threshold: The minimum number of enum variants for the lints about variant names to trigger

    (default: 3)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for equal operands to comparison, logical and bitwise, difference and division binary operators (==, >, etc., &&, ||, &, |, ^, - and /).

Why is this bad?

This is usually just a typo or a copy and paste error.

Known problems

False negatives: We had some false positives regarding calls (notably racer had one instance of x.pop() && x.pop()), so we removed matching any function or method calls. We may introduce a list of known pure functions in the future.

Example

if x + 1 == x + 1 {}

// or

assert_eq!(a, a);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for pattern matchings that can be expressed using equality.

Why is this bad?

  • It reads better and has less cognitive load because equality won’t cause binding.
  • It is a Yoda condition. Yoda conditions are widely criticized for increasing the cognitive load of reading the code.
  • Equality is a simple bool expression and can be merged with && and || and reuse if blocks

Example

if let Some(2) = x {
    do_thing();
}

Use instead:

if x == Some(2) {
    do_thing();
}
Applicability: MachineApplicable(?)
Added in: 1.57.0

What it does

Checks for erasing operations, e.g., x * 0.

Why is this bad?

The whole expression can be replaced by zero. This is most likely not the intended outcome and should probably be corrected

Example

let x = 1;
0 / x;
0 * x;
x & 0;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for .err().expect() calls on the Result type.

Why is this bad?

.expect_err() can be called directly to avoid the extra type conversion from err().

Example

let x: Result<u32, &str> = Ok(10);
x.err().expect("Testing err().expect()");

Use instead:

let x: Result<u32, &str> = Ok(10);
x.expect_err("Testing expect_err");

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.62.0

What it does

Checks for types named Error that implement Error.

Why restrict this?

It can become confusing when a codebase has 20 types all named Error, requiring either aliasing them in the use statement or qualifying them like my_module::Error. This hinders comprehension, as it requires you to memorize every variation of importing Error used across a codebase.

Example

#[derive(Debug)]
pub enum Error { ... }

impl std::fmt::Display for Error { ... }

impl std::error::Error for Error { ... }
Applicability: Unspecified(?)
Added in: 1.73.0

What it does

Checks for blocks which are nested beyond a certain threshold.

Note: Even though this lint is warn-by-default, it will only trigger if a maximum nesting level is defined in the clippy.toml file.

Why is this bad?

It can severely hinder readability.

Example

An example clippy.toml configuration:

excessive-nesting-threshold = 3
// lib.rs
pub mod a {
    pub struct X;
    impl X {
        pub fn run(&self) {
            if true {
                // etc...
            }
        }
    }
}

Use instead:

// a.rs
fn private_run(x: &X) {
    if true {
        // etc...
    }
}

pub struct X;
impl X {
    pub fn run(&self) {
        private_run(self);
    }
}
// lib.rs
pub mod a;

Configuration

  • excessive-nesting-threshold: The maximum amount of nesting a block can reside in

    (default: 0)

Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Checks for float literals with a precision greater than that supported by the underlying type.

Why is this bad?

Rust will truncate the literal silently.

Example

let v: f32 = 0.123_456_789_9;
println!("{}", v); //  0.123_456_789

Use instead:

let v: f64 = 0.123_456_789_9;
println!("{}", v); //  0.123_456_789_9
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Warns on any exported enums that are not tagged #[non_exhaustive]

Why restrict this?

Making an enum exhaustive is a stability commitment: adding a variant is a breaking change. A project may wish to ensure that there are no exhaustive enums or that every exhaustive enum is explicitly #[allow]ed.

Example

enum Foo {
    Bar,
    Baz
}

Use instead:

#[non_exhaustive]
enum Foo {
    Bar,
    Baz
}
Applicability: MaybeIncorrect(?)
Added in: 1.51.0

What it does

Warns on any exported structs that are not tagged #[non_exhaustive]

Why restrict this?

Making a struct exhaustive is a stability commitment: adding a field is a breaking change. A project may wish to ensure that there are no exhaustive structs or that every exhaustive struct is explicitly #[allow]ed.

Example

struct Foo {
    bar: u8,
    baz: String,
}

Use instead:

#[non_exhaustive]
struct Foo {
    bar: u8,
    baz: String,
}
Applicability: MaybeIncorrect(?)
Added in: 1.51.0

What it does

Detects calls to the exit() function which terminates the program.

Why restrict this?

exit() immediately terminates the program with no information other than an exit code. This provides no means to troubleshoot a problem, and may be an unexpected side effect.

Codebases may use this lint to require that all exits are performed either by panicking (which produces a message, a code location, and optionally a backtrace) or by returning from main() (which is a single place to look).

Example

std::process::exit(0)

Use instead:

// To provide a stacktrace and additional information
panic!("message");

// or a main method with a return
fn main() -> Result<(), i32> {
    Ok(())
}
Applicability: Unspecified(?)
Added in: 1.41.0

What it does

Checks for calls to .expect(&format!(...)), .expect(foo(..)), etc., and suggests to use unwrap_or_else instead

Why is this bad?

The function will always be called.

Known problems

If the function has side-effects, not calling it will change the semantics of the program, but you shouldn’t rely on that anyway.

Example

foo.expect(&format!("Err {}: {}", err_code, err_msg));

// or

foo.expect(format!("Err {}: {}", err_code, err_msg).as_str());

Use instead:

foo.unwrap_or_else(|| panic!("Err {}: {}", err_code, err_msg));
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for .expect() or .expect_err() calls on Results and .expect() call on Options.

Why restrict this?

Usually it is better to handle the None or Err case. Still, for a lot of quick-and-dirty code, expect is a good choice, which is why this lint is Allow by default.

result.expect() will let the thread panic on Err values. Normally, you want to implement more sophisticated error handling, and propagate errors upwards with ? operator.

Examples

option.expect("one");
result.expect("one");

Use instead:

option?;

// or

result?;

Past names

  • option_expect_used
  • result_expect_used

Configuration

  • allow-expect-in-tests: Whether expect should be allowed in test functions or #[cfg(test)]

    (default: false)

Applicability: Unspecified(?)
Added in: 1.45.0

What it does

Checks for explicit Clone implementations for Copy types.

Why is this bad?

To avoid surprising behavior, these traits should agree and the behavior of Copy cannot be overridden. In almost all situations a Copy type should have a Clone implementation that does nothing more than copy the object, which is what #[derive(Copy, Clone)] gets you.

Example

#[derive(Copy)]
struct Foo;

impl Clone for Foo {
    // ..
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for dereferencing expressions which would be covered by auto-deref.

Why is this bad?

This unnecessarily complicates the code.

Example

let x = String::new();
let y: &str = &*x;

Use instead:

let x = String::new();
let y: &str = &x;
Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

Checks for loops over slices with an explicit counter and suggests the use of .enumerate().

Why is this bad?

Using .enumerate() makes the intent more clear, declutters the code and may be faster in some instances.

Example

let mut i = 0;
for item in &v {
    bar(i, *item);
    i += 1;
}

Use instead:

for (i, item) in v.iter().enumerate() { bar(i, *item); }
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for explicit deref() or deref_mut() method calls.

Why is this bad?

Dereferencing by &*x or &mut *x is clearer and more concise, when not part of a method chain.

Example

use std::ops::Deref;
let a: &mut String = &mut String::from("foo");
let b: &str = a.deref();

Use instead:

let a: &mut String = &mut String::from("foo");
let b = &*a;

This lint excludes all of:

let _ = d.unwrap().deref();
let _ = Foo::deref(&foo);
let _ = <Foo as Deref>::deref(&foo);
Applicability: MachineApplicable(?)
Added in: 1.44.0

What it does

Checks for loops on y.into_iter() where y will do, and suggests the latter.

Why is this bad?

Readability.

Example

// with `y` a `Vec` or slice:
for x in y.into_iter() {
    // ..
}

can be rewritten to

for x in y {
    // ..
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for loops on x.iter() where &x will do, and suggests the latter.

Why is this bad?

Readability.

Known problems

False negatives. We currently only warn on some known types.

Example

// with `y` a `Vec` or slice:
for x in y.iter() {
    // ..
}

Use instead:

for x in &y {
    // ..
}

Configuration

  • enforce-iter-loop-reborrow: Whether to recommend using implicit into iter for reborrowed values.

Example

let mut vec = vec![1, 2, 3];
let rmvec = &mut vec;
for _ in rmvec.iter() {}
for _ in rmvec.iter_mut() {}

Use instead:

let mut vec = vec![1, 2, 3];
let rmvec = &mut vec;
for _ in &*rmvec {}
for _ in &mut *rmvec {}

(default: false)

Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of write!() / writeln()! which can be replaced with (e)print!() / (e)println!()

Why is this bad?

Using (e)println! is clearer and more concise

Example

writeln!(&mut std::io::stderr(), "foo: {:?}", bar).unwrap();
writeln!(&mut std::io::stdout(), "foo: {:?}", bar).unwrap();

Use instead:

eprintln!("foo: {:?}", bar);
println!("foo: {:?}", bar);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

Vec::extend_from_slice is no longer faster than Vec::extend due to specialization.

Applicability: Unspecified(?)
Deprecated in: pre 1.29.0

What it does

Checks for occurrences where one vector gets extended instead of append

Why is this bad?

Using append instead of extend is more concise and faster

Example

let mut a = vec![1, 2, 3];
let mut b = vec![4, 5, 6];

a.extend(b.drain(..));

Use instead:

let mut a = vec![1, 2, 3];
let mut b = vec![4, 5, 6];

a.append(&mut b);
Applicability: MachineApplicable(?)
Added in: 1.55.0

What it does

Checks for lifetimes in generics that are never used anywhere else.

Why is this bad?

The additional lifetimes make the code look more complicated, while there is nothing out of the ordinary going on. Removing them leads to more readable code.

Example

// unnecessary lifetimes
fn unused_lifetime<'a>(x: u8) {
    // ..
}

Use instead:

fn no_lifetime(x: u8) {
    // ...
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for type parameters in generics that are never used anywhere else.

Why is this bad?

Functions cannot infer the value of unused type parameters; therefore, calling them requires using a turbofish, which serves no purpose but to satisfy the compiler.

Example

fn unused_ty<T>(x: u8) {
    // ..
}

Use instead:

fn no_unused_ty(x: u8) {
    // ..
}
Applicability: MachineApplicable(?)
Added in: 1.69.0

What it does

Checks for impls of From<..> that contain panic!() or unwrap()

Why is this bad?

TryFrom should be used if there’s a possibility of failure.

Example

struct Foo(i32);

impl From<String> for Foo {
    fn from(s: String) -> Self {
        Foo(s.parse().unwrap())
    }
}

Use instead:

struct Foo(i32);

impl TryFrom<String> for Foo {
    type Error = ();
    fn try_from(s: String) -> Result<Self, Self::Error> {
        if let Ok(parsed) = s.parse() {
            Ok(Foo(parsed))
        } else {
            Err(())
        }
    }
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for immediate reassignment of fields initialized with Default::default().

Why is this bad?

It’s more idiomatic to use the functional update syntax.

Known problems

Assignments to patterns that are of tuple type are not linted.

Example

let mut a: A = Default::default();
a.i = 42;

Use instead:

let a = A {
    i: 42,
    .. Default::default()
};
Applicability: Unspecified(?)
Added in: 1.49.0

What it does

Checks for usage of scoped visibility modifiers, like pub(crate), on fields. These make a field visible within a scope between public and private.

Why restrict this?

Scoped visibility modifiers cause a field to be accessible within some scope between public and private, potentially within an entire crate. This allows for fields to be non-private while upholding internal invariants, but can be a code smell. Scoped visibility requires checking a greater area, potentially an entire crate, to verify that an invariant is upheld, and global analysis requires a lot of effort.

Example

pub mod public_module {
    struct MyStruct {
        pub(crate) first_field: bool,
        pub(super) second_field: bool
    }
}

Use instead:

pub mod public_module {
    struct MyStruct {
        first_field: bool,
        second_field: bool
    }
    impl MyStruct {
        pub(crate) fn get_first_field(&self) -> bool {
            self.first_field
        }
        pub(super) fn get_second_field(&self) -> bool {
            self.second_field
        }
    }
}
Applicability: Unspecified(?)
Added in: 1.81.0

What it does

Checks for FileType::is_file().

Why restrict this?

When people testing a file type with FileType::is_file they are testing whether a path is something they can get bytes from. But is_file doesn’t cover special file types in unix-like systems, and doesn’t cover symlink in windows. Using !FileType::is_dir() is a better way to that intention.

Example

let metadata = std::fs::metadata("foo.txt")?;
let filetype = metadata.file_type();

if filetype.is_file() {
    // read file
}

should be written as:

let metadata = std::fs::metadata("foo.txt")?;
let filetype = metadata.file_type();

if !filetype.is_dir() {
    // read file
}
Applicability: Unspecified(?)
Added in: 1.42.0

What it does

Checks for usage of bool::then in Iterator::filter_map.

Why is this bad?

This can be written with filter then map instead, which would reduce nesting and separates the filtering from the transformation phase. This comes with no cost to performance and is just cleaner.

Limitations

Does not lint bool::then_some, as it eagerly evaluates its arguments rather than lazily. This can create differing behavior, so better safe than sorry.

Example

_ = v.into_iter().filter_map(|i| (i % 2 == 0).then(|| really_expensive_fn(i)));

Use instead:

_ = v.into_iter().filter(|i| i % 2 == 0).map(|i| really_expensive_fn(i));
Applicability: MachineApplicable(?)
Added in: 1.73.0

What it does

Checks for usage of filter_map(|x| x).

Why is this bad?

Readability, this can be written more concisely by using flatten.

Example

iter.filter_map(|x| x);

Use instead:

iter.flatten();
Applicability: MachineApplicable(?)
Added in: 1.52.0

What it does

Checks for usage of _.filter_map(_).next().

Why is this bad?

Readability, this can be written more concisely as _.find_map(_).

Example

 (0..3).filter_map(|x| if x == 2 { Some(x) } else { None }).next();

Can be written as

 (0..3).find_map(|x| if x == 2 { Some(x) } else { None });

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.36.0

What it does

Checks for usage of _.filter(_).next().

Why is this bad?

Readability, this can be written more concisely as _.find(_).

Example

vec.iter().filter(|x| **x == 0).next();

Use instead:

vec.iter().find(|x| **x == 0);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of flat_map(|x| x).

Why is this bad?

Readability, this can be written more concisely by using flatten.

Example

iter.flat_map(|x| x);

Can be written as

iter.flatten();
Applicability: MachineApplicable(?)
Added in: 1.39.0

What it does

Checks for usage of Iterator::flat_map() where filter_map() could be used instead.

Why is this bad?

filter_map() is known to always produce 0 or 1 output items per input item, rather than however many the inner iterator type produces. Therefore, it maintains the upper bound in Iterator::size_hint(), and communicates to the reader that the input items are not being expanded into multiple output items without their having to notice that the mapping function returns an Option.

Example

let nums: Vec<i32> = ["1", "2", "whee!"].iter().flat_map(|x| x.parse().ok()).collect();

Use instead:

let nums: Vec<i32> = ["1", "2", "whee!"].iter().filter_map(|x| x.parse().ok()).collect();
Applicability: MachineApplicable(?)
Added in: 1.53.0

What it does

Checks for float arithmetic.

Why restrict this?

For some embedded systems or kernel development, it can be useful to rule out floating-point numbers.

Example

a + 1.0;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for (in-)equality comparisons on floating-point values (apart from zero), except in functions called *eq* (which probably implement equality for a type involving floats).

Why is this bad?

Floating point calculations are usually imprecise, so asking if two values are exactly equal is asking for trouble because arriving at the same logical result via different routes (e.g. calculation versus constant) may yield different values.

Example

let a: f64 = 1000.1;
let b: f64 = 0.2;
let x = a + b;
let y = 1000.3; // Expected value.

// Actual value: 1000.3000000000001
println!("{x}");

let are_equal = x == y;
println!("{are_equal}"); // false

The correct way to compare floating point numbers is to define an allowed error margin. This may be challenging if there is no “natural” error margin to permit. Broadly speaking, there are two cases:

  1. If your values are in a known range and you can define a threshold for “close enough to be equal”, it may be appropriate to define an absolute error margin. For example, if your data is “length of vehicle in centimeters”, you may consider 0.1 cm to be “close enough”.
  2. If your code is more general and you do not know the range of values, you should use a relative error margin, accepting e.g. 0.1% of error regardless of specific values.

For the scenario where you can define a meaningful absolute error margin, consider using:

let a: f64 = 1000.1;
let b: f64 = 0.2;
let x = a + b;
let y = 1000.3; // Expected value.

const ALLOWED_ERROR_VEHICLE_LENGTH_CM: f64 = 0.1;
let within_tolerance = (x - y).abs() < ALLOWED_ERROR_VEHICLE_LENGTH_CM;
println!("{within_tolerance}"); // true

NB! Do not use f64::EPSILON - while the error margin is often called “epsilon”, this is a different use of the term that is not suitable for floating point equality comparison. Indeed, for the example above using f64::EPSILON as the allowed error would return false.

For the scenario where no meaningful absolute error can be defined, refer to the floating point guide for a reference implementation of relative error based comparison of floating point values. MIN_NORMAL in the reference implementation is equivalent to MIN_POSITIVE in Rust.

Applicability: HasPlaceholders(?)
Added in: pre 1.29.0

What it does

Checks for (in-)equality comparisons on constant floating-point values (apart from zero), except in functions called *eq* (which probably implement equality for a type involving floats).

Why restrict this?

Floating point calculations are usually imprecise, so asking if two values are exactly equal is asking for trouble because arriving at the same logical result via different routes (e.g. calculation versus constant) may yield different values.

Example

let a: f64 = 1000.1;
let b: f64 = 0.2;
let x = a + b;
const Y: f64 = 1000.3; // Expected value.

// Actual value: 1000.3000000000001
println!("{x}");

let are_equal = x == Y;
println!("{are_equal}"); // false

The correct way to compare floating point numbers is to define an allowed error margin. This may be challenging if there is no “natural” error margin to permit. Broadly speaking, there are two cases:

  1. If your values are in a known range and you can define a threshold for “close enough to be equal”, it may be appropriate to define an absolute error margin. For example, if your data is “length of vehicle in centimeters”, you may consider 0.1 cm to be “close enough”.
  2. If your code is more general and you do not know the range of values, you should use a relative error margin, accepting e.g. 0.1% of error regardless of specific values.

For the scenario where you can define a meaningful absolute error margin, consider using:

let a: f64 = 1000.1;
let b: f64 = 0.2;
let x = a + b;
const Y: f64 = 1000.3; // Expected value.

const ALLOWED_ERROR_VEHICLE_LENGTH_CM: f64 = 0.1;
let within_tolerance = (x - Y).abs() < ALLOWED_ERROR_VEHICLE_LENGTH_CM;
println!("{within_tolerance}"); // true

NB! Do not use f64::EPSILON - while the error margin is often called “epsilon”, this is a different use of the term that is not suitable for floating point equality comparison. Indeed, for the example above using f64::EPSILON as the allowed error would return false.

For the scenario where no meaningful absolute error can be defined, refer to the floating point guide for a reference implementation of relative error based comparison of floating point values. MIN_NORMAL in the reference implementation is equivalent to MIN_POSITIVE in Rust.

Applicability: HasPlaceholders(?)
Added in: pre 1.29.0

What it does

Checks for statements of the form (a - b) < f32::EPSILON or (a - b) < f64::EPSILON. Notes the missing .abs().

Why is this bad?

The code without .abs() is more likely to have a bug.

Known problems

If the user can ensure that b is larger than a, the .abs() is technically unnecessary. However, it will make the code more robust and doesn’t have any large performance implications. If the abs call was deliberately left out for performance reasons, it is probably better to state this explicitly in the code, which then can be done with an allow.

Example

pub fn is_roughly_equal(a: f32, b: f32) -> bool {
    (a - b) < f32::EPSILON
}

Use instead:

pub fn is_roughly_equal(a: f32, b: f32) -> bool {
    (a - b).abs() < f32::EPSILON
}
Applicability: MaybeIncorrect(?)
Added in: 1.48.0

What it does

Checks for comparisons with an address of a function item.

Why is this bad?

Function item address is not guaranteed to be unique and could vary between different code generation units. Furthermore different function items could have the same address after being merged together.

Example

type F = fn();
fn a() {}
let f: F = a;
if f == a {
    // ...
}
Applicability: Unspecified(?)
Added in: 1.44.0

What it does

Checks for excessive use of bools in function definitions.

Why is this bad?

Calls to such functions are confusing and error prone, because it’s hard to remember argument order and you have no type system support to back you up. Using two-variant enums instead of bools often makes API easier to use.

Example

fn f(is_round: bool, is_hot: bool) { ... }

Use instead:

enum Shape {
    Round,
    Spiky,
}

enum Temperature {
    Hot,
    IceCold,
}

fn f(shape: Shape, temperature: Temperature) { ... }

Configuration

  • max-fn-params-bools: The maximum number of bool parameters a function can have

    (default: 3)

Applicability: Unspecified(?)
Added in: 1.43.0

What it does

Checks for casts of function pointers to something other than usize.

Why is this bad?

Casting a function pointer to anything other than usize/isize is not portable across architectures. If the target type is too small the address would be truncated, and target types larger than usize are unnecessary.

Casting to isize also doesn’t make sense, since addresses are never signed.

Example

fn fun() -> i32 { 1 }
let _ = fun as i64;

Use instead:

let _ = fun as usize;
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for casts of a function pointer to any integer type.

Why restrict this?

Casting a function pointer to an integer can have surprising results and can occur accidentally if parentheses are omitted from a function call. If you aren’t doing anything low-level with function pointers then you can opt out of casting functions to integers in order to avoid mistakes. Alternatively, you can use this lint to audit all uses of function pointer casts in your code.

Example

// fn1 is cast as `usize`
fn fn1() -> u16 {
    1
};
let _ = fn1 as usize;

Use instead:

// maybe you intended to call the function?
fn fn2() -> u16 {
    1
};
let _ = fn2() as usize;

// or

// maybe you intended to cast it to a function type?
fn fn3() -> u16 {
    1
}
let _ = fn3 as fn() -> u16;
Applicability: MaybeIncorrect(?)
Added in: 1.58.0

What it does

Checks for casts of a function pointer to a numeric type not wide enough to store an address.

Why is this bad?

Such a cast discards some bits of the function’s address. If this is intended, it would be more clearly expressed by casting to usize first, then casting the usize to the intended type (with a comment) to perform the truncation.

Example

fn fn1() -> i16 {
    1
};
let _ = fn1 as i32;

Use instead:

// Cast to usize first, then comment with the reason for the truncation
fn fn1() -> i16 {
    1
};
let fn_ptr = fn1 as usize;
let fn_ptr_truncated = fn_ptr as i32;
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for iterating a map (HashMap or BTreeMap) and ignoring either the keys or values.

Why is this bad?

Readability. There are keys and values methods that can be used to express that don’t need the values or keys.

Example

for (k, _) in &map {
    ..
}

could be replaced by

for k in map.keys() {
    ..
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for calls to std::mem::forget with a value that does not implement Drop.

Why is this bad?

Calling std::mem::forget is no different than dropping such a type. A different value may have been intended.

Example

struct Foo;
let x = Foo;
std::mem::forget(x);
Applicability: Unspecified(?)
Added in: 1.62.0

What it does

Checks for usage of .map(|_| format!(..)).collect::<String>().

Why is this bad?

This allocates a new string for every element in the iterator. This can be done more efficiently by creating the String once and appending to it in Iterator::fold, using either the write! macro which supports exactly the same syntax as the format! macro, or concatenating with + in case the iterator yields &str/String.

Note also that write!-ing into a String can never fail, despite the return type of write! being std::fmt::Result, so it can be safely ignored or unwrapped.

Example

fn hex_encode(bytes: &[u8]) -> String {
    bytes.iter().map(|b| format!("{b:02X}")).collect()
}

Use instead:

use std::fmt::Write;
fn hex_encode(bytes: &[u8]) -> String {
    bytes.iter().fold(String::new(), |mut output, b| {
        let _ = write!(output, "{b:02X}");
        output
    })
}
Applicability: Unspecified(?)
Added in: 1.73.0

What it does

Detects format! within the arguments of another macro that does formatting such as format! itself, write! or println!. Suggests inlining the format! call.

Why is this bad?

The recommended code is both shorter and avoids a temporary allocation.

Example

println!("error: {}", format!("something failed at {}", Location::caller()));

Use instead:

println!("error: something failed at {}", Location::caller());
Applicability: Unspecified(?)
Added in: 1.58.0

What it does

Detects cases where the result of a format! call is appended to an existing String.

Why restrict this?

Introduces an extra, avoidable heap allocation.

Known problems

format! returns a String but write! returns a Result. Thus you are forced to ignore the Err variant to achieve the same API.

While using write! in the suggested way should never fail, this isn’t necessarily clear to the programmer.

Example

let mut s = String::new();
s += &format!("0x{:X}", 1024);
s.push_str(&format!("0x{:X}", 1024));

Use instead:

use std::fmt::Write as _; // import without risk of name clashing

let mut s = String::new();
let _ = write!(s, "0x{:X}", 1024);
Applicability: Unspecified(?)
Added in: 1.62.0

What it does

Checks for outer doc comments written with 4 forward slashes (////).

Why is this bad?

This is (probably) a typo, and results in it not being a doc comment; just a regular comment.

Example

//// My amazing data structure
pub struct Foo {
    // ...
}

Use instead:

/// My amazing data structure
pub struct Foo {
    // ...
}
Applicability: MachineApplicable(?)
Added in: 1.73.0

What it does

Checks for from_iter() function calls on types that implement the FromIterator trait.

Why is this bad?

If it’s needed to create a collection from the contents of an iterator, the Iterator::collect(_) method is preferred. However, when it’s needed to specify the container type, Vec::from_iter(_) can be more readable than using a turbofish (e.g. _.collect::<Vec<_>>()). See FromIterator documentation

Example

let five_fives = std::iter::repeat(5).take(5);

let v = Vec::from_iter(five_fives);

assert_eq!(v, vec![5, 5, 5, 5, 5]);

Use instead:

let five_fives = std::iter::repeat(5).take(5);

let v: Vec<i32> = five_fives.collect();

assert_eq!(v, vec![5, 5, 5, 5, 5]);

but prefer to use

let numbers: Vec<i32> = FromIterator::from_iter(1..=5);

instead of

let numbers = (1..=5).collect::<Vec<_>>();
Applicability: MaybeIncorrect(?)
Added in: 1.49.0

What it does

Searches for implementations of the Into<..> trait and suggests to implement From<..> instead.

Why is this bad?

According the std docs implementing From<..> is preferred since it gives you Into<..> for free where the reverse isn’t true.

Example

struct StringWrapper(String);

impl Into<StringWrapper> for String {
    fn into(self) -> StringWrapper {
        StringWrapper(self)
    }
}

Use instead:

struct StringWrapper(String);

impl From<String> for StringWrapper {
    fn from(s: String) -> StringWrapper {
        StringWrapper(s)
    }
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.51.0

What it does

Checks if we’re passing a c_void raw pointer to {Box,Rc,Arc,Weak}::from_raw(_)

Why is this bad?

When dealing with c_void raw pointers in FFI, it is easy to run into the pitfall of calling from_raw with the c_void pointer. The type signature of Box::from_raw is fn from_raw(raw: *mut T) -> Box<T>, so if you pass a *mut c_void you will get a Box<c_void> (and similarly for Rc, Arc and Weak). For this to be safe, c_void would need to have the same memory layout as the original type, which is often not the case.

Example

let ptr = Box::into_raw(Box::new(42usize)) as *mut c_void;
let _ = unsafe { Box::from_raw(ptr) };

Use instead:

let _ = unsafe { Box::from_raw(ptr as *mut usize) };
Applicability: Unspecified(?)
Added in: 1.67.0

What it does

Checks for function invocations of the form primitive::from_str_radix(s, 10)

Why is this bad?

This specific common use case can be rewritten as s.parse::<primitive>() (and in most cases, the turbofish can be removed), which reduces code length and complexity.

Known problems

This lint may suggest using (&<expression>).parse() instead of <expression>.parse() directly in some cases, which is correct but adds unnecessary complexity to the code.

Example

let input: &str = get_input();
let num = u16::from_str_radix(input, 10)?;

Use instead:

let input: &str = get_input();
let num: u16 = input.parse()?;
Applicability: MaybeIncorrect(?)
Added in: 1.52.0

What it does

This lint requires Future implementations returned from functions and methods to implement the Send marker trait. It is mostly used by library authors (public and internal) that target an audience where multithreaded executors are likely to be used for running these Futures.

Why is this bad?

A Future implementation captures some state that it needs to eventually produce its final value. When targeting a multithreaded executor (which is the norm on non-embedded devices) this means that this state may need to be transported to other threads, in other words the whole Future needs to implement the Send marker trait. If it does not, then the resulting Future cannot be submitted to a thread pool in the end user’s code.

Especially for generic functions it can be confusing to leave the discovery of this problem to the end user: the reported error location will be far from its cause and can in many cases not even be fixed without modifying the library where the offending Future implementation is produced.

Example

async fn not_send(bytes: std::rc::Rc<[u8]>) {}

Use instead:

async fn is_send(bytes: std::sync::Arc<[u8]>) {}
Applicability: Unspecified(?)
Added in: 1.44.0

What it does

Checks for usage of x.get(0) instead of x.first() or x.front().

Why is this bad?

Using x.first() for Vecs and slices or x.front() for VecDeques is easier to read and has the same result.

Example

let x = vec![2, 3, 5];
let first_element = x.get(0);

Use instead:

let x = vec![2, 3, 5];
let first_element = x.first();
Applicability: MachineApplicable(?)
Added in: 1.63.0

What it does

Checks for usage of x.get(x.len() - 1) instead of x.last().

Why is this bad?

Using x.last() is easier to read and has the same result.

Note that using x[x.len() - 1] is semantically different from x.last(). Indexing into the array will panic on out-of-bounds accesses, while x.get() and x.last() will return None.

There is another lint (get_unwrap) that covers the case of using x.get(index).unwrap() instead of x[index].

Example

let x = vec![2, 3, 5];
let last_element = x.get(x.len() - 1);

Use instead:

let x = vec![2, 3, 5];
let last_element = x.last();
Applicability: MachineApplicable(?)
Added in: 1.37.0

What it does

Checks for usage of .get().unwrap() (or .get_mut().unwrap) on a standard library type which implements Index

Why restrict this?

Using the Index trait ([]) is more clear and more concise.

Known problems

Not a replacement for error handling: Using either .unwrap() or the Index trait ([]) carries the risk of causing a panic if the value being accessed is None. If the use of .get().unwrap() is a temporary placeholder for dealing with the Option type, then this does not mitigate the need for error handling. If there is a chance that .get() will be None in your program, then it is advisable that the None case is handled in a future refactor instead of using .unwrap() or the Index trait.

Example

let mut some_vec = vec![0, 1, 2, 3];
let last = some_vec.get(3).unwrap();
*some_vec.get_mut(0).unwrap() = 1;

The correct use would be:

let mut some_vec = vec![0, 1, 2, 3];
let last = some_vec[3];
some_vec[0] = 1;
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for the usage of the to_ne_bytes method and/or the function from_ne_bytes.

Why restrict this?

To ensure use of explicitly chosen endianness rather than the target’s endianness, such as when implementing network protocols or file formats rather than FFI.

Example

let _x = 2i32.to_ne_bytes();
let _y = 2i64.to_ne_bytes();
Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Checks for identity operations, e.g., x + 0.

Why is this bad?

This code can be removed without changing the meaning. So it just obscures what’s going on. Delete it mercilessly.

Example

x / 1 + 0 * 1 - 0 | 0;
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for Mutex::lock calls in if let expression with lock calls in any of the else blocks.

Why is this bad?

The Mutex lock remains held for the whole if let ... else block and deadlocks.

Example

if let Ok(thing) = mutex.lock() {
    do_thing();
} else {
    mutex.lock();
}

Should be written

let locked = mutex.lock();
if let Ok(thing) = locked {
    do_thing(thing);
} else {
    use_locked(locked);
}
Applicability: Unspecified(?)
Added in: 1.45.0

What it does

Checks for usage of ! or != in an if condition with an else branch.

Why is this bad?

Negations reduce the readability of statements.

Example

if !v.is_empty() {
    a()
} else {
    b()
}

Could be written:

if v.is_empty() {
    b()
} else {
    a()
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for if/else with the same body as the then part and the else part.

Why is this bad?

This is probably a copy & paste error.

Example

let foo = if … {
    42
} else {
    42
};
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for if-else that could be written using either bool::then or bool::then_some.

Why restrict this?

Looks a little redundant. Using bool::then is more concise and incurs no loss of clarity. For simple calculations and known values, use bool::then_some, which is eagerly evaluated in comparison to bool::then.

Example

let a = if v.is_empty() {
    println!("true!");
    Some(42)
} else {
    None
};

Could be written:

let a = v.is_empty().then(|| {
    println!("true!");
    42
});

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: Unspecified(?)
Added in: 1.53.0

What it does

Checks for consecutive ifs with the same condition.

Why is this bad?

This is probably a copy & paste error.

Example

if a == b {
    …
} else if a == b {
    …
}

Note that this lint ignores all conditions with a function call as it could have side effects:

if foo() {
    …
} else if foo() { // not linted
    …
}

Configuration

  • ignore-interior-mutability: A list of paths to types that should be treated as if they do not contain interior mutability

    (default: ["bytes::Bytes"])

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of _ in patterns of type ().

Why is this bad?

Matching with () explicitly instead of _ outlines the fact that the pattern contains no data. Also it would detect a type change that _ would ignore.

Example

match std::fs::create_dir("tmp-work-dir") {
   Ok(_) => println!("Working directory created"),
   Err(s) => eprintln!("Could not create directory: {s}"),
}

Use instead:

match std::fs::create_dir("tmp-work-dir") {
   Ok(()) => println!("Working directory created"),
   Err(s) => eprintln!("Could not create directory: {s}"),
}
Applicability: MachineApplicable(?)
Added in: 1.73.0

What it does

This lint is concerned with the semantics of Borrow and Hash for a type that implements all three of Hash, Borrow<str> and Borrow<[u8]> as it is impossible to satisfy the semantics of Borrow and Hash for both Borrow<str> and Borrow<[u8]>.

Why is this bad?

When providing implementations for Borrow<T>, one should consider whether the different implementations should act as facets or representations of the underlying type. Generic code typically uses Borrow<T> when it relies on the identical behavior of these additional trait implementations. These traits will likely appear as additional trait bounds.

In particular Eq, Ord and Hash must be equivalent for borrowed and owned values: x.borrow() == y.borrow() should give the same result as x == y. It follows then that the following equivalence must hold: hash(x) == hash((x as Borrow<[u8]>).borrow()) == hash((x as Borrow<str>).borrow())

Unfortunately it doesn’t hold as hash("abc") != hash("abc".as_bytes()). This happens because the Hash impl for str passes an additional 0xFF byte to the hasher to avoid collisions. For example, given the tuples ("a", "bc"), and ("ab", "c"), the two tuples would have the same hash value if the 0xFF byte was not added.

Example

use std::borrow::Borrow;
use std::hash::{Hash, Hasher};

struct ExampleType {
    data: String
}

impl Hash for ExampleType {
    fn hash<H: Hasher>(&self, state: &mut H) {
        self.data.hash(state);
    }
}

impl Borrow<str> for ExampleType {
    fn borrow(&self) -> &str {
        &self.data
    }
}

impl Borrow<[u8]> for ExampleType {
    fn borrow(&self) -> &[u8] {
        self.data.as_bytes()
    }
}

As a consequence, hashing a &ExampleType and hashing the result of the two borrows will result in different values.

Applicability: Unspecified(?)
Added in: 1.76.0

What it does

Lints when impl Trait is being used in a function’s parameters.

Why restrict this?

Turbofish syntax (::<>) cannot be used to specify the type of an impl Trait parameter, making impl Trait less powerful. Readability may also be a factor.

Example

trait MyTrait {}
fn foo(a: impl MyTrait) {
	// [...]
}

Use instead:

trait MyTrait {}
fn foo<T: MyTrait>(a: T) {
	// [...]
}
Applicability: HasPlaceholders(?)
Added in: 1.69.0

What it does

Checks for the usage of _.to_owned(), vec.to_vec(), or similar when calling _.clone() would be clearer.

Why is this bad?

These methods do the same thing as _.clone() but may be confusing as to why we are calling to_vec on something that is already a Vec or calling to_owned on something that is already owned.

Example

let a = vec![1, 2, 3];
let b = a.to_vec();
let c = a.to_owned();

Use instead:

let a = vec![1, 2, 3];
let b = a.clone();
let c = a.clone();
Applicability: MachineApplicable(?)
Added in: 1.52.0

What it does

Checks for public impl or fn missing generalization over different hashers and implicitly defaulting to the default hashing algorithm (SipHash).

Why is this bad?

HashMap or HashSet with custom hashers cannot be used with them.

Known problems

Suggestions for replacing constructors can contain false-positives. Also applying suggestions can require modification of other pieces of code, possibly including external crates.

Example

impl<K: Hash + Eq, V> Serialize for HashMap<K, V> { }

pub fn foo(map: &mut HashMap<i32, i32>) { }

could be rewritten as

impl<K: Hash + Eq, V, S: BuildHasher> Serialize for HashMap<K, V, S> { }

pub fn foo<S: BuildHasher>(map: &mut HashMap<i32, i32, S>) { }
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for missing return statements at the end of a block.

Why restrict this?

Omitting the return keyword whenever possible is idiomatic Rust code, but:

  • Programmers coming from other languages might prefer the expressiveness of return.
  • It’s possible to miss the last returning statement because the only difference is a missing ;.
  • Especially in bigger code with multiple return paths, having a return keyword makes it easier to find the corresponding statements.

Example

fn foo(x: usize) -> usize {
    x
}

add return

fn foo(x: usize) -> usize {
    return x;
}
Applicability: MachineApplicable(?)
Added in: 1.33.0

What it does

Checks for implicit saturating addition.

Why is this bad?

The built-in function is more readable and may be faster.

Example

let mut u:u32 = 7000;

if u != u32::MAX {
    u += 1;
}

Use instead:

let mut u:u32 = 7000;

u = u.saturating_add(1);
Applicability: MachineApplicable(?)
Added in: 1.66.0

What it does

Checks for implicit saturating subtraction.

Why is this bad?

Simplicity and readability. Instead we can easily use an builtin function.

Example

let mut i: u32 = end - start;

if i != 0 {
    i -= 1;
}

Use instead:

let mut i: u32 = end - start;

i = i.saturating_sub(1);
Applicability: MachineApplicable(?)
Added in: 1.44.0

What it does

Looks for bounds in impl Trait in return position that are implied by other bounds. This can happen when a trait is specified that another trait already has as a supertrait (e.g. fn() -> impl Deref + DerefMut<Target = i32> has an unnecessary Deref bound, because Deref is a supertrait of DerefMut)

Why is this bad?

Specifying more bounds than necessary adds needless complexity for the reader.

Limitations

This lint does not check for implied bounds transitively. Meaning that it doesn’t check for implied bounds from supertraits of supertraits (e.g. trait A {} trait B: A {} trait C: B {}, then having an fn() -> impl A + C)

Example

fn f() -> impl Deref<Target = i32> + DerefMut<Target = i32> {
//             ^^^^^^^^^^^^^^^^^^^ unnecessary bound, already implied by the `DerefMut` trait bound
    Box::new(123)
}

Use instead:

fn f() -> impl DerefMut<Target = i32> {
    Box::new(123)
}
Applicability: MachineApplicable(?)
Added in: 1.74.0

What it does

Checks for double comparisons that can never succeed

Why is this bad?

The whole expression can be replaced by false, which is probably not the programmer’s intention

Example

if status_code <= 400 && status_code > 500 {}
Applicability: Unspecified(?)
Added in: 1.73.0

What it does

Looks for floating-point expressions that can be expressed using built-in methods to improve accuracy at the cost of performance.

Why is this bad?

Negatively impacts accuracy.

Example

let a = 3f32;
let _ = a.powf(1.0 / 3.0);
let _ = (1.0 + a).ln();
let _ = a.exp() - 1.0;

Use instead:

let a = 3f32;
let _ = a.cbrt();
let _ = a.ln_1p();
let _ = a.exp_m1();
Applicability: MachineApplicable(?)
Added in: 1.43.0

What it does

This lint checks that no function newer than the defined MSRV (minimum supported rust version) is used in the crate.

Why is this bad?

It would prevent the crate to be actually used with the specified MSRV.

Example

// MSRV of 1.3.0
use std::thread::sleep;
use std::time::Duration;

// Sleep was defined in `1.4.0`.
sleep(Duration::new(1, 0));

To fix this problem, either increase your MSRV or use another item available in your current MSRV.

Applicability: Unspecified(?)
Added in: 1.78.0

What it does

Warns if an integral or floating-point constant is grouped inconsistently with underscores.

Why is this bad?

Readers may incorrectly interpret inconsistently grouped digits.

Example

618_64_9189_73_511

Use instead:

61_864_918_973_511
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for struct constructors where all fields are shorthand and the order of the field init shorthand in the constructor is inconsistent with the order in the struct definition.

Why is this bad?

Since the order of fields in a constructor doesn’t affect the resulted instance as the below example indicates,

#[derive(Debug, PartialEq, Eq)]
struct Foo {
    x: i32,
    y: i32,
}
let x = 1;
let y = 2;

// This assertion never fails:
assert_eq!(Foo { x, y }, Foo { y, x });

inconsistent order can be confusing and decreases readability and consistency.

Example

struct Foo {
    x: i32,
    y: i32,
}
let x = 1;
let y = 2;

Foo { y, x };

Use instead:

Foo { x, y };
Applicability: MachineApplicable(?)
Added in: 1.52.0

What it does

The lint checks for slice bindings in patterns that are only used to access individual slice values.

Why is this bad?

Accessing slice values using indices can lead to panics. Using refutable patterns can avoid these. Binding to individual values also improves the readability as they can be named.

Limitations

This lint currently only checks for immutable access inside if let patterns.

Example

let slice: Option<&[u32]> = Some(&[1, 2, 3]);

if let Some(slice) = slice {
    println!("{}", slice[0]);
}

Use instead:

let slice: Option<&[u32]> = Some(&[1, 2, 3]);

if let Some(&[first, ..]) = slice {
    println!("{}", first);
}

Configuration

  • max-suggested-slice-pattern-length: When Clippy suggests using a slice pattern, this is the maximum number of elements allowed in the slice pattern that is suggested. If more elements are necessary, the lint is suppressed. For example, [_, _, _, e, ..] is a slice pattern with 4 elements.

    (default: 3)

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MaybeIncorrect(?)
Added in: 1.59.0

What it does

Checks for usage of indexing or slicing. Arrays are special cases, this lint does report on arrays if we can tell that slicing operations are in bounds and does not lint on constant usize indexing on arrays because that is handled by rustc’s const_err lint.

Why restrict this?

To avoid implicit panics from indexing and slicing. There are “checked” alternatives which do not panic, and can be used with unwrap() to make an explicit panic when it is desired.

Example

// Vector
let x = vec![0; 5];

x[2];
&x[2..100];

// Array
let y = [0, 1, 2, 3];

&y[10..100];
&y[10..];

Use instead:

x.get(2);
x.get(2..100);

y.get(10);
y.get(10..100);

Configuration

  • suppress-restriction-lint-in-const: Whether to suppress a restriction lint in constant code. In same cases the restructured operation might not be unavoidable, as the suggested counterparts are unavailable in constant code. This configuration will cause restriction lints to trigger even if no suggestion can be made.

    (default: false)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for bit masks in comparisons which can be removed without changing the outcome. The basic structure can be seen in the following table:

ComparisonBit OpExampleequals
> / <=| / ^x | 2 > 3x > 3
< / >=| / ^x ^ 1 < 4x < 4

Why is this bad?

Not equally evil as bad_bit_mask, but still a bit misleading, because the bit mask is ineffective.

Known problems

False negatives: This lint will only match instances where we have figured out the math (which is for a power-of-two compared value). This means things like x | 1 >= 7 (which would be better written as x >= 6) will not be reported (but bit masks like this are fairly uncommon).

Example

if (x | 1 > 3) {  }

Use instead:

if (x >= 2) {  }
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks if both .write(true) and .append(true) methods are called on a same OpenOptions.

Why is this bad?

.append(true) already enables write(true), making this one superfluous.

Example

let _ = OpenOptions::new()
           .write(true)
           .append(true)
           .create(true)
           .open("file.json");

Use instead:

let _ = OpenOptions::new()
           .append(true)
           .create(true)
           .open("file.json");
Applicability: MachineApplicable(?)
Added in: 1.76.0

What it does

Checks for usage of .to_string() on an &&T where T implements ToString directly (like &&str or &&String).

Why is this bad?

This bypasses the specialized implementation of ToString and instead goes through the more expensive string formatting facilities.

Example

// Generic implementation for `T: Display` is used (slow)
["foo", "bar"].iter().map(|s| s.to_string());

// OK, the specialized impl is used
["foo", "bar"].iter().map(|&s| s.to_string());
Applicability: MachineApplicable(?)
Added in: 1.40.0

What it does

Checks for matches being used to destructure a single-variant enum or tuple struct where a let will suffice.

Why is this bad?

Just readability – let doesn’t nest, whereas a match does.

Example

enum Wrapper {
    Data(i32),
}

let wrapper = Wrapper::Data(42);

let data = match wrapper {
    Wrapper::Data(i) => i,
};

The correct use would be:

enum Wrapper {
    Data(i32),
}

let wrapper = Wrapper::Data(42);
let Wrapper::Data(data) = wrapper;
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for iteration that is guaranteed to be infinite.

Why is this bad?

While there may be places where this is acceptable (e.g., in event streams), in most cases this is simply an error.

Example

use std::iter;

iter::repeat(1_u8).collect::<Vec<_>>();
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for infinite loops in a function where the return type is not ! and lint accordingly.

Why restrict this?

Making the return type ! serves as documentation that the function does not return. If the function is not intended to loop infinitely, then this lint may detect a bug.

Example

fn run_forever() {
    loop {
        // do something
    }
}

If infinite loops are as intended:

fn run_forever() -> ! {
    loop {
        // do something
    }
}

Otherwise add a break or return condition:

fn run_forever() {
    loop {
        // do something
        if condition {
            break;
        }
    }
}
Applicability: MaybeIncorrect(?)
Added in: 1.76.0

What it does

Checks for the definition of inherent methods with a signature of to_string(&self) -> String.

Why is this bad?

This method is also implicitly defined if a type implements the Display trait. As the functionality of Display is much more versatile, it should be preferred.

Example

pub struct A;

impl A {
    pub fn to_string(&self) -> String {
        "I am A".to_string()
    }
}

Use instead:

use std::fmt;

pub struct A;

impl fmt::Display for A {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        write!(f, "I am A")
    }
}
Applicability: Unspecified(?)
Added in: 1.38.0

What it does

Checks for the definition of inherent methods with a signature of to_string(&self) -> String and if the type implementing this method also implements the Display trait.

Why is this bad?

This method is also implicitly defined if a type implements the Display trait. The less versatile inherent method will then shadow the implementation introduced by Display.

Example

use std::fmt;

pub struct A;

impl A {
    pub fn to_string(&self) -> String {
        "I am A".to_string()
    }
}

impl fmt::Display for A {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        write!(f, "I am A, too")
    }
}

Use instead:

use std::fmt;

pub struct A;

impl fmt::Display for A {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        write!(f, "I am A")
    }
}
Applicability: Unspecified(?)
Added in: 1.38.0

What it does

Checks for tuple structs initialized with field syntax. It will however not lint if a base initializer is present. The lint will also ignore code in macros.

Why is this bad?

This may be confusing to the uninitiated and adds no benefit as opposed to tuple initializers

Example

struct TupleStruct(u8, u16);

let _ = TupleStruct {
    0: 1,
    1: 23,
};

// should be written as
let base = TupleStruct(1, 23);

// This is OK however
let _ = TupleStruct { 0: 42, ..base };
Applicability: MachineApplicable(?)
Added in: 1.59.0

What it does

Checks for items annotated with #[inline(always)], unless the annotated function is empty or simply panics.

Why is this bad?

While there are valid uses of this annotation (and once you know when to use it, by all means allow this lint), it’s a common newbie-mistake to pepper one’s code with it.

As a rule of thumb, before slapping #[inline(always)] on a function, measure if that additional function call really affects your runtime profile sufficiently to make up for the increase in compile time.

Known problems

False positives, big time. This lint is meant to be deactivated by everyone doing serious performance work. This means having done the measurement.

Example

#[inline(always)]
fn not_quite_hot_code(..) { ... }
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of AT&T x86 assembly syntax.

Why restrict this?

To enforce consistent use of Intel x86 assembly syntax.

Example

asm!("lea ({}), {}", in(reg) ptr, lateout(reg) _, options(att_syntax));

Use instead:

asm!("lea {}, [{}]", lateout(reg) _, in(reg) ptr);
Applicability: Unspecified(?)
Added in: 1.49.0

What it does

Checks for usage of Intel x86 assembly syntax.

Why restrict this?

To enforce consistent use of AT&T x86 assembly syntax.

Example

asm!("lea {}, [{}]", lateout(reg) _, in(reg) ptr);

Use instead:

asm!("lea ({}), {}", in(reg) ptr, lateout(reg) _, options(att_syntax));
Applicability: Unspecified(?)
Added in: 1.49.0

What it does

Checks for #[inline] on trait methods without bodies

Why is this bad?

Only implementations of trait methods may be inlined. The inline attribute is ignored for trait methods without bodies.

Example

trait Animal {
    #[inline]
    fn name(&self) -> &'static str;
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of inspect().for_each().

Why is this bad?

It is the same as performing the computation inside inspect at the beginning of the closure in for_each.

Example

[1,2,3,4,5].iter()
.inspect(|&x| println!("inspect the number: {}", x))
.for_each(|&x| {
    assert!(x >= 0);
});

Can be written as

[1,2,3,4,5].iter()
.for_each(|&x| {
    println!("inspect the number: {}", x);
    assert!(x >= 0);
});
Applicability: Unspecified(?)
Added in: 1.51.0

What it does

Checks for usage of x >= y + 1 or x - 1 >= y (and <=) in a block

Why is this bad?

Readability – better to use > y instead of >= y + 1.

Example

if x >= y + 1 {}

Use instead:

if x > y {}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for division of integers

Why restrict this?

When outside of some very specific algorithms, integer division is very often a mistake because it discards the remainder.

Example

let x = 3 / 2;
println!("{}", x);

Use instead:

let x = 3f32 / 2f32;
println!("{}", x);
Applicability: Unspecified(?)
Added in: 1.37.0

What it does

Checks for the usage of division (/) and remainder (%) operations when performed on any integer types using the default Div and Rem trait implementations.

Why restrict this?

In cryptographic contexts, division can result in timing sidechannel vulnerabilities, and needs to be replaced with constant-time code instead (e.g. Barrett reduction).

Example

let my_div = 10 / 2;

Use instead:

let my_div = 10 >> 1;
Applicability: Unspecified(?)
Added in: 1.79.0

What it does

Checks for into_iter calls on references which should be replaced by iter or iter_mut.

Why is this bad?

Readability. Calling into_iter on a reference will not move out its content into the resulting iterator, which is confusing. It is better just call iter or iter_mut directly.

Example

(&vec).into_iter();

Use instead:

(&vec).iter();
Applicability: MachineApplicable(?)
Added in: 1.32.0

What it does

This is the opposite of the iter_without_into_iter lint. It looks for IntoIterator for (&|&mut) Type implementations without an inherent iter or iter_mut method on the type or on any of the types in its Deref chain.

Why is this bad?

It’s not bad, but having them is idiomatic and allows the type to be used in iterator chains by just calling .iter(), instead of the more awkward <&Type>::into_iter or (&val).into_iter() syntax in case of ambiguity with another IntoIterator impl.

Limitations

This lint focuses on providing an idiomatic API. Therefore, it will only lint on types which are accessible outside of the crate. For internal types, these methods can be added on demand if they are actually needed. Otherwise, it would trigger the dead_code lint for the unused method.

Example

struct MySlice<'a>(&'a [u8]);
impl<'a> IntoIterator for &MySlice<'a> {
    type Item = &'a u8;
    type IntoIter = std::slice::Iter<'a, u8>;
    fn into_iter(self) -> Self::IntoIter {
        self.0.iter()
    }
}

Use instead:

struct MySlice<'a>(&'a [u8]);
impl<'a> MySlice<'a> {
    pub fn iter(&self) -> std::slice::Iter<'a, u8> {
        self.into_iter()
    }
}
impl<'a> IntoIterator for &MySlice<'a> {
    type Item = &'a u8;
    type IntoIter = std::slice::Iter<'a, u8>;
    fn into_iter(self) -> Self::IntoIter {
        self.0.iter()
    }
}
Applicability: Unspecified(?)
Added in: 1.75.0

What it does

This lint checks for invalid usages of ptr::null.

Why is this bad?

This causes undefined behavior.

Example

// Undefined behavior
unsafe { std::slice::from_raw_parts(ptr::null(), 0); }

Use instead:

unsafe { std::slice::from_raw_parts(NonNull::dangling().as_ptr(), 0); }
Applicability: MachineApplicable(?)
Added in: 1.53.0

What it does

Checks regex creation (with Regex::new, RegexBuilder::new, or RegexSet::new) for correct regex syntax.

Why is this bad?

This will lead to a runtime panic.

Example

Regex::new("(")
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for comparisons where the relation is always either true or false, but where one side has been upcast so that the comparison is necessary. Only integer types are checked.

Why is this bad?

An expression like let x : u8 = ...; (x as u32) > 300 will mistakenly imply that it is possible for x to be outside the range of u8.

Known problems

https://github.com/rust-lang/rust-clippy/issues/886

Example

let x: u8 = 1;
(x as u32) > 300;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for comparisons between integers, followed by subtracting the greater value from the lower one.

Why is this bad?

This could result in an underflow and is most likely not what the user wants. If this was intended to be a saturated subtraction, consider using the saturating_sub method directly.

Example

let a = 12u32;
let b = 13u32;

let result = if a > b { b - a } else { 0 };

Use instead:

let a = 12u32;
let b = 13u32;

let result = a.saturating_sub(b);
Applicability: MaybeIncorrect(?)
Added in: 1.44.0

What it does

Checks for invisible Unicode characters in the code.

Why is this bad?

Having an invisible character in the code makes for all sorts of April fools, but otherwise is very much frowned upon.

Example

You don’t see it, but there may be a zero-width space or soft hyphen some­where in this text.

Past names

  • zero_width_space
Applicability: MachineApplicable(?)
Added in: 1.49.0

What it does

Finds usages of char::is_digit that can be replaced with is_ascii_digit or is_ascii_hexdigit.

Why is this bad?

is_digit(..) is slower and requires specifying the radix.

Example

let c: char = '6';
c.is_digit(10);
c.is_digit(16);

Use instead:

let c: char = '6';
c.is_ascii_digit();
c.is_ascii_hexdigit();
Applicability: MachineApplicable(?)
Added in: 1.62.0

What it does

Checks for items declared after some statement in a block.

Why is this bad?

Items live for the entire scope they are declared in. But statements are processed in order. This might cause confusion as it’s hard to figure out which item is meant in a statement.

Example

fn foo() {
    println!("cake");
}

fn main() {
    foo(); // prints "foo"
    fn foo() {
        println!("foo");
    }
    foo(); // prints "foo"
}

Use instead:

fn foo() {
    println!("cake");
}

fn main() {
    fn foo() {
        println!("foo");
    }
    foo(); // prints "foo"
    foo(); // prints "foo"
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Triggers if an item is declared after the testing module marked with #[cfg(test)].

Why is this bad?

Having items declared after the testing module is confusing and may lead to bad test coverage.

Example

#[cfg(test)]
mod tests {
    // [...]
}

fn my_function() {
    // [...]
}

Use instead:

fn my_function() {
    // [...]
}

#[cfg(test)]
mod tests {
    // [...]
}
Applicability: MachineApplicable(?)
Added in: 1.71.0

What it does

Checks for the use of .cloned().collect() on slice to create a Vec.

Why is this bad?

.to_vec() is clearer

Example

let s = [1, 2, 3, 4, 5];
let s2: Vec<isize> = s[..].iter().cloned().collect();

The better use would be:

let s = [1, 2, 3, 4, 5];
let s2: Vec<isize> = s.to_vec();
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for the use of .iter().count().

Why is this bad?

.len() is more efficient and more readable.

Example

let some_vec = vec![0, 1, 2, 3];

some_vec.iter().count();
&some_vec[..].iter().count();

Use instead:

let some_vec = vec![0, 1, 2, 3];

some_vec.len();
&some_vec[..].len();
Applicability: MachineApplicable(?)
Added in: 1.52.0

What it does

Checks for usage of .filter(Result::is_ok) that may be replaced with a .flatten() call. This lint will require additional changes to the follow-up calls as it affects the type.

Why is this bad?

This pattern is often followed by manual unwrapping of Result. The simplification results in more readable and succinct code without the need for manual unwrapping.

Example

// example code where clippy issues a warning
vec![Ok::<i32, String>(1)].into_iter().filter(Result::is_ok);

Use instead:

// example code which does not raise clippy warning
vec![Ok::<i32, String>(1)].into_iter().flatten();
Applicability: HasPlaceholders(?)
Added in: 1.77.0

What it does

Checks for usage of .filter(Option::is_some) that may be replaced with a .flatten() call. This lint will require additional changes to the follow-up calls as it affects the type.

Why is this bad?

This pattern is often followed by manual unwrapping of the Option. The simplification results in more readable and succinct code without the need for manual unwrapping.

Example

// example code where clippy issues a warning
vec![Some(1)].into_iter().filter(Option::is_some);

Use instead:

// example code which does not raise clippy warning
vec![Some(1)].into_iter().flatten();
Applicability: HasPlaceholders(?)
Added in: 1.77.0

What it does

Checks for iterating a map (HashMap or BTreeMap) and ignoring either the keys or values.

Why is this bad?

Readability. There are keys and values methods that can be used to express that we only need the keys or the values.

Example

let map: HashMap<u32, u32> = HashMap::new();
let values = map.iter().map(|(_, value)| value).collect::<Vec<_>>();

Use instead:

let map: HashMap<u32, u32> = HashMap::new();
let values = map.values().collect::<Vec<_>>();

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.66.0

What it does

Checks for loops on x.next().

Why is this bad?

next() returns either Some(value) if there was a value, or None otherwise. The insidious thing is that Option<_> implements IntoIterator, so that possibly one value will be iterated, leading to some hard to find bugs. No one will want to write such code except to win an Underhanded Rust Contest.

Example

for x in y.next() {
    ..
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of iter().next() on a Slice or an Array

Why is this bad?

These can be shortened into .get()

Example

a[2..].iter().next();
b.iter().next();

should be written as:

a.get(2);
b.get(0);
Applicability: MachineApplicable(?)
Added in: 1.46.0

What it does

Detects methods named iter or iter_mut that do not have a return type that implements Iterator.

Why is this bad?

Methods named iter or iter_mut conventionally return an Iterator.

Example

// `String` does not implement `Iterator`
struct Data {}
impl Data {
    fn iter(&self) -> String {
        todo!()
    }
}

Use instead:

use std::str::Chars;
struct Data {}
impl Data {
   fn iter(&self) -> Chars<'static> {
       todo!()
   }
}
Applicability: Unspecified(?)
Added in: 1.57.0

What it does

Checks for usage of .iter().nth()/.iter_mut().nth() on standard library types that have equivalent .get()/.get_mut() methods.

Why is this bad?

.get() and .get_mut() are equivalent but more concise.

Example

let some_vec = vec![0, 1, 2, 3];
let bad_vec = some_vec.iter().nth(3);
let bad_slice = &some_vec[..].iter().nth(3);

The correct use would be:

let some_vec = vec![0, 1, 2, 3];
let bad_vec = some_vec.get(3);
let bad_slice = &some_vec[..].get(3);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for the use of iter.nth(0).

Why is this bad?

iter.next() is equivalent to iter.nth(0), as they both consume the next element, but is more readable.

Example

let x = s.iter().nth(0);

Use instead:

let x = s.iter().next();
Applicability: MachineApplicable(?)
Added in: 1.42.0

What it does

Checks for calls to iter, iter_mut or into_iter on empty collections

Why is this bad?

It is simpler to use the empty function from the standard library:

Example

use std::{slice, option};
let a: slice::Iter<i32> = [].iter();
let f: option::IntoIter<i32> = None.into_iter();

Use instead:

use std::iter;
let a: iter::Empty<i32> = iter::empty();
let b: iter::Empty<i32> = iter::empty();

Known problems

The type of the resulting iterator might become incompatible with its usage

Applicability: MaybeIncorrect(?)
Added in: 1.65.0

What it does

Checks for calls to iter, iter_mut or into_iter on collections containing a single item

Why is this bad?

It is simpler to use the once function from the standard library:

Example

let a = [123].iter();
let b = Some(123).into_iter();

Use instead:

use std::iter;
let a = iter::once(&123);
let b = iter::once(123);

Known problems

The type of the resulting iterator might become incompatible with its usage

Applicability: MaybeIncorrect(?)
Added in: 1.65.0

What it does

Looks for iterator combinator calls such as .take(x) or .skip(x) where x is greater than the amount of items that an iterator will produce.

Why is this bad?

Taking or skipping more items than there are in an iterator either creates an iterator with all items from the original iterator or an iterator with no items at all. This is most likely not what the user intended to do.

Example

for _ in [1, 2, 3].iter().take(4) {}

Use instead:

for _ in [1, 2, 3].iter() {}
Applicability: Unspecified(?)
Added in: 1.74.0

What it does

This is a restriction lint which prevents the use of hash types (i.e., HashSet and HashMap) in for loops.

Why restrict this?

Because hash types are unordered, when iterated through such as in a for loop, the values are returned in an undefined order. As a result, on redundant systems this may cause inconsistencies and anomalies. In addition, the unknown order of the elements may reduce readability or introduce other undesired side effects.

Example

    let my_map = std::collections::HashMap::<i32, String>::new();
    for (key, value) in my_map { /* ... */ }

Use instead:

    let my_map = std::collections::HashMap::<i32, String>::new();
    let mut keys = my_map.keys().clone().collect::<Vec<_>>();
    keys.sort();
    for key in keys {
        let value = &my_map[key];
    }
Applicability: Unspecified(?)
Added in: 1.76.0

What it does

Checks for usage of _.cloned().<func>() where call to .cloned() can be postponed.

Why is this bad?

It’s often inefficient to clone all elements of an iterator, when eventually, only some of them will be consumed.

Known Problems

This lint removes the side of effect of cloning items in the iterator. A code that relies on that side-effect could fail.

Examples

vec.iter().cloned().take(10);
vec.iter().cloned().last();

Use instead:

vec.iter().take(10).cloned();
vec.iter().last().cloned();
Applicability: MachineApplicable(?)
Added in: 1.60.0

What it does

Checks for usage of .skip(x).next() on iterators.

Why is this bad?

.nth(x) is cleaner

Example

let some_vec = vec![0, 1, 2, 3];
let bad_vec = some_vec.iter().skip(3).next();
let bad_slice = &some_vec[..].iter().skip(3).next();

The correct use would be:

let some_vec = vec![0, 1, 2, 3];
let bad_vec = some_vec.iter().nth(3);
let bad_slice = &some_vec[..].iter().nth(3);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of .skip(0) on iterators.

Why is this bad?

This was likely intended to be .skip(1) to skip the first element, as .skip(0) does nothing. If not, the call should be removed.

Example

let v = vec![1, 2, 3];
let x = v.iter().skip(0).collect::<Vec<_>>();
let y = v.iter().collect::<Vec<_>>();
assert_eq!(x, y);
Applicability: MaybeIncorrect(?)
Added in: 1.73.0

What it does

Checks for usage of .drain(..) on Vec and VecDeque for iteration.

Why is this bad?

.into_iter() is simpler with better performance.

Example

let mut foo = vec![0, 1, 2, 3];
let bar: HashSet<usize> = foo.drain(..).collect();

Use instead:

let foo = vec![0, 1, 2, 3];
let bar: HashSet<usize> = foo.into_iter().collect();
Applicability: MaybeIncorrect(?)
Added in: 1.61.0

What it does

Looks for iter and iter_mut methods without an associated IntoIterator for (&|&mut) Type implementation.

Why is this bad?

It’s not bad, but having them is idiomatic and allows the type to be used in for loops directly (for val in &iter {}), without having to first call iter() or iter_mut().

Limitations

This lint focuses on providing an idiomatic API. Therefore, it will only lint on types which are accessible outside of the crate. For internal types, the IntoIterator trait can be implemented on demand if it is actually needed.

Example

struct MySlice<'a>(&'a [u8]);
impl<'a> MySlice<'a> {
    pub fn iter(&self) -> std::slice::Iter<'a, u8> {
        self.0.iter()
    }
}

Use instead:

struct MySlice<'a>(&'a [u8]);
impl<'a> MySlice<'a> {
    pub fn iter(&self) -> std::slice::Iter<'a, u8> {
        self.0.iter()
    }
}
impl<'a> IntoIterator for &MySlice<'a> {
    type Item = &'a u8;
    type IntoIter = std::slice::Iter<'a, u8>;
    fn into_iter(self) -> Self::IntoIter {
        self.iter()
    }
}
Applicability: Unspecified(?)
Added in: 1.75.0

What it does

Checks for calling .step_by(0) on iterators which panics.

Why is this bad?

This very much looks like an oversight. Use panic!() instead if you actually intend to panic.

Example

for x in (0..100).step_by(0) {
    //..
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for calls to Path::join that start with a path separator (\\ or /).

Why is this bad?

If the argument to Path::join starts with a separator, it will overwrite the original path. If this is intentional, prefer using Path::new instead.

Note the behavior is platform dependent. A leading \\ will be accepted on unix systems as part of the file name

See Path::join

Example

let path = Path::new("/bin");
let joined_path = path.join("/sh");
assert_eq!(joined_path, PathBuf::from("/sh"));

Use instead;

let path = Path::new("/bin");

// If this was unintentional, remove the leading separator
let joined_path = path.join("sh");
assert_eq!(joined_path, PathBuf::from("/bin/sh"));

// If this was intentional, create a new path instead
let new = Path::new("/sh");
assert_eq!(new, PathBuf::from("/sh"));
Applicability: Unspecified(?)
Added in: 1.76.0

What it does

Checks if you have variables whose name consists of just underscores and digits.

Why is this bad?

It’s hard to memorize what a variable means without a descriptive name.

Example

let _1 = 1;
let ___1 = 1;
let __1___2 = 11;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for large const arrays that should be defined as static instead.

Why is this bad?

Performance: const variables are inlined upon use. Static items result in only one instance and has a fixed location in memory.

Example

pub const a = [0u32; 1_000_000];

Use instead:

pub static a = [0u32; 1_000_000];

Configuration

  • array-size-threshold: The maximum allowed size for arrays on the stack

    (default: 16384)

Applicability: MachineApplicable(?)
Added in: 1.44.0

What it does

Warns if the digits of an integral or floating-point constant are grouped into groups that are too large.

Why is this bad?

Negatively impacts readability.

Example

let x: u64 = 6186491_8973511;
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for large size differences between variants on enums.

Why is this bad?

Enum size is bounded by the largest variant. Having one large variant can penalize the memory layout of that enum.

Known problems

This lint obviously cannot take the distribution of variants in your running program into account. It is possible that the smaller variants make up less than 1% of all instances, in which case the overhead is negligible and the boxing is counter-productive. Always measure the change this lint suggests.

For types that implement Copy, the suggestion to Box a variant’s data would require removing the trait impl. The types can of course still be Clone, but that is worse ergonomically. Depending on the use case it may be possible to store the large data in an auxiliary structure (e.g. Arena or ECS).

The lint will ignore the impact of generic types to the type layout by assuming every type parameter is zero-sized. Depending on your use case, this may lead to a false positive.

Example

enum Test {
    A(i32),
    B([i32; 8000]),
}

Use instead:

// Possibly better
enum Test2 {
    A(i32),
    B(Box<[i32; 8000]>),
}

Configuration

  • enum-variant-size-threshold: The maximum size of an enum’s variant to avoid box suggestion

    (default: 200)

Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

It checks for the size of a Future created by async fn or async {}.

Why is this bad?

Due to the current unideal implementation of Coroutine, large size of a Future may cause stack overflows.

Example

async fn large_future(_x: [u8; 16 * 1024]) {}

pub async fn trigger() {
    large_future([0u8; 16 * 1024]).await;
}

Box::pin the big future instead.

async fn large_future(_x: [u8; 16 * 1024]) {}

pub async fn trigger() {
    Box::pin(large_future([0u8; 16 * 1024])).await;
}

Configuration

  • future-size-threshold: The maximum byte size a Future can have, before it triggers the clippy::large_futures lint

    (default: 16384)

Applicability: Unspecified(?)
Added in: 1.70.0

What it does

Checks for the inclusion of large files via include_bytes!() or include_str!().

Why restrict this?

Including large files can undesirably increase the size of the binary produced by the compiler. This lint may be used to catch mistakes where an unexpectedly large file is included, or temporarily to obtain a list of all large files.

Example

let included_str = include_str!("very_large_file.txt");
let included_bytes = include_bytes!("very_large_file.txt");

Use instead:

use std::fs;

// You can load the file at runtime
let string = fs::read_to_string("very_large_file.txt")?;
let bytes = fs::read("very_large_file.txt")?;

Configuration

  • max-include-file-size: The maximum size of a file included via include_bytes!() or include_str!(), in bytes

    (default: 1000000)

Applicability: Unspecified(?)
Added in: 1.62.0

What it does

Checks for local arrays that may be too large.

Why is this bad?

Large local arrays may cause stack overflow.

Example

let a = [0u32; 1_000_000];

Configuration

  • array-size-threshold: The maximum allowed size for arrays on the stack

    (default: 16384)

Applicability: Unspecified(?)
Added in: 1.41.0

What it does

Checks for functions that use a lot of stack space.

This often happens when constructing a large type, such as an array with a lot of elements, or constructing many smaller-but-still-large structs, or copying around a lot of large types.

This lint is a more general version of large_stack_arrays that is intended to look at functions as a whole instead of only individual array expressions inside of a function.

Why is this bad?

The stack region of memory is very limited in size (usually much smaller than the heap) and attempting to use too much will result in a stack overflow and crash the program. To avoid this, you should consider allocating large types on the heap instead (e.g. by boxing them).

Keep in mind that the code path to construction of large types does not even need to be reachable; it purely needs to exist inside of the function to contribute to the stack size. For example, this causes a stack overflow even though the branch is unreachable:

fn main() {
    if false {
        let x = [0u8; 10000000]; // 10 MB stack array
        black_box(&x);
    }
}

Known issues

False positives. The stack size that clippy sees is an estimated value and can be vastly different from the actual stack usage after optimizations passes have run (especially true in release mode). Modern compilers are very smart and are able to optimize away a lot of unnecessary stack allocations. In debug mode however, it is usually more accurate.

This lint works by summing up the size of all variables that the user typed, variables that were implicitly introduced by the compiler for temporaries, function arguments and the return value, and comparing them against a (configurable, but high-by-default).

Example

This function creates four 500 KB arrays on the stack. Quite big but just small enough to not trigger large_stack_arrays. However, looking at the function as a whole, it’s clear that this uses a lot of stack space.

struct QuiteLargeType([u8; 500_000]);
fn foo() {
    // ... some function that uses a lot of stack space ...
    let _x1 = QuiteLargeType([0; 500_000]);
    let _x2 = QuiteLargeType([0; 500_000]);
    let _x3 = QuiteLargeType([0; 500_000]);
    let _x4 = QuiteLargeType([0; 500_000]);
}

Instead of doing this, allocate the arrays on the heap. This currently requires going through a Vec first and then converting it to a Box:

struct NotSoLargeType(Box<[u8]>);

fn foo() {
    let _x1 = NotSoLargeType(vec![0; 500_000].into_boxed_slice());
//                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^  Now heap allocated.
//                                                                The size of `NotSoLargeType` is 16 bytes.
//  ...
}

Configuration

  • stack-size-threshold: The maximum allowed stack size for functions in bytes

    (default: 512000)

Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Checks for functions taking arguments by value, where the argument type is Copy and large enough to be worth considering passing by reference. Does not trigger if the function is being exported, because that might induce API breakage, if the parameter is declared as mutable, or if the argument is a self.

Why is this bad?

Arguments passed by value might result in an unnecessary shallow copy, taking up more space in the stack and requiring a call to memcpy, which can be expensive.

Example

#[derive(Clone, Copy)]
struct TooLarge([u8; 2048]);

fn foo(v: TooLarge) {}

Use instead:

fn foo(v: &TooLarge) {}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

  • pass-by-value-size-limit: The minimum size (in bytes) to consider a type for passing by reference instead of by value.

    (default: 256)

Applicability: MaybeIncorrect(?)
Added in: 1.49.0

What it does

Checks for usage of <integer>::max_value(), std::<integer>::MAX, std::<float>::EPSILON, etc.

Why is this bad?

All of these have been superseded by the associated constants on their respective types, such as i128::MAX. These legacy items may be deprecated in a future version of rust.

Example

let eps = std::f32::EPSILON;

Use instead:

let eps = f32::EPSILON;

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MaybeIncorrect(?)
Added in: 1.79.0

What it does

Checks for items that implement .len() but not .is_empty().

Why is this bad?

It is good custom to have both methods, because for some data structures, asking about the length will be a costly operation, whereas .is_empty() can usually answer in constant time. Also it used to lead to false positives on the len_zero lint – currently that lint will ignore such entities.

Example

impl X {
    pub fn len(&self) -> usize {
        ..
    }
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for getting the length of something via .len() just to compare to zero, and suggests using .is_empty() where applicable.

Why is this bad?

Some structures can answer .is_empty() much faster than calculating their length. So it is good to get into the habit of using .is_empty(), and having it is cheap. Besides, it makes the intent clearer than a manual comparison in some contexts.

Example

if x.len() == 0 {
    ..
}
if y.len() != 0 {
    ..
}

instead use

if x.is_empty() {
    ..
}
if !y.is_empty() {
    ..
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for let-bindings, which are subsequently returned.

Why is this bad?

It is just extraneous code. Remove it to make your code more rusty.

Known problems

In the case of some temporaries, e.g. locks, eliding the variable binding could lead to deadlocks. See this issue. This could become relevant if the code is later changed to use the code that would have been bound without first assigning it to a let-binding.

Example

fn foo() -> String {
    let x = String::new();
    x
}

instead, use

fn foo() -> String {
    String::new()
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for let _ = <expr> where the resulting type of expr implements Future

Why is this bad?

Futures must be polled for work to be done. The original intention was most likely to await the future and ignore the resulting value.

Example

async fn foo() -> Result<(), ()> {
    Ok(())
}
let _ = foo();

Use instead:

async fn foo() -> Result<(), ()> {
    Ok(())
}
let _ = foo().await;
Applicability: Unspecified(?)
Added in: 1.67.0

What it does

Checks for let _ = sync_lock. This supports mutex and rwlock in parking_lot. For std locks see the rustc lint let_underscore_lock

Why is this bad?

This statement immediately drops the lock instead of extending its lifetime to the end of the scope, which is often not intended. To extend lock lifetime to the end of the scope, use an underscore-prefixed name instead (i.e. _lock). If you want to explicitly drop the lock, std::mem::drop conveys your intention better and is less error-prone.

Example

let _ = mutex.lock();

Use instead:

let _lock = mutex.lock();
Applicability: Unspecified(?)
Added in: 1.43.0

What it does

Checks for let _ = <expr> where expr is #[must_use]

Why restrict this?

To ensure that all #[must_use] types are used rather than ignored.

Example

fn f() -> Result<u32, u32> {
    Ok(0)
}

let _ = f();
// is_ok() is marked #[must_use]
let _ = f().is_ok();
Applicability: Unspecified(?)
Added in: 1.42.0

What it does

Checks for let _ = <expr> without a type annotation, and suggests to either provide one, or remove the let keyword altogether.

Why restrict this?

The let _ = <expr> expression ignores the value of <expr>, but will continue to do so even if the type were to change, thus potentially introducing subtle bugs. By supplying a type annotation, one will be forced to re-visit the decision to ignore the value in such cases.

Known problems

The _ = <expr> is not properly supported by some tools (e.g. IntelliJ) and may seem odd to many developers. This lint also partially overlaps with the other let_underscore_* lints.

Example

fn foo() -> Result<u32, ()> {
    Ok(123)
}
let _ = foo();

Use instead:

fn foo() -> Result<u32, ()> {
    Ok(123)
}
// Either provide a type annotation:
let _: Result<u32, ()> = foo();
// …or drop the let keyword:
_ = foo();
Applicability: Unspecified(?)
Added in: 1.69.0

What it does

Checks for binding a unit value.

Why is this bad?

A unit value cannot usefully be used anywhere. So binding one is kind of pointless.

Example

let x = {
    1;
};
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Detects when a variable is declared with an explicit type of _.

Why is this bad?

It adds noise, : _ provides zero clarity or utility.

Example

let my_number: _ = 1;

Use instead:

let my_number = 1;
Applicability: Unspecified(?)
Added in: 1.70.0

What it does

Checks for usage of lines.filter_map(Result::ok) or lines.flat_map(Result::ok) when lines has type std::io::Lines.

Why is this bad?

Lines instances might produce a never-ending stream of Err, in which case filter_map(Result::ok) will enter an infinite loop while waiting for an Ok variant. Calling next() once is sufficient to enter the infinite loop, even in the absence of explicit loops in the user code.

This situation can arise when working with user-provided paths. On some platforms, std::fs::File::open(path) might return Ok(fs) even when path is a directory, but any later attempt to read from fs will return an error.

Known problems

This lint suggests replacing filter_map() or flat_map() applied to a Lines instance in all cases. There are two cases where the suggestion might not be appropriate or necessary:

  • If the Lines instance can never produce any error, or if an error is produced only once just before terminating the iterator, using map_while() is not necessary but will not do any harm.
  • If the Lines instance can produce intermittent errors then recover and produce successful results, using map_while() would stop at the first error.

Example

let mut lines = BufReader::new(File::open("some-path")?).lines().filter_map(Result::ok);
// If "some-path" points to a directory, the next statement never terminates:
let first_line: Option<String> = lines.next();

Use instead:

let mut lines = BufReader::new(File::open("some-path")?).lines().map_while(Result::ok);
let first_line: Option<String> = lines.next();
Applicability: MaybeIncorrect(?)
Added in: 1.70.0

What it does

Checks for usage of any LinkedList, suggesting to use a Vec or a VecDeque (formerly called RingBuf).

Why is this bad?

Gankra says:

The TL;DR of LinkedList is that it’s built on a massive amount of pointers and indirection. It wastes memory, it has terrible cache locality, and is all-around slow. RingBuf, while “only” amortized for push/pop, should be faster in the general case for almost every possible workload, and isn’t even amortized at all if you can predict the capacity you need.

LinkedLists are only really good if you’re doing a lot of merging or splitting of lists. This is because they can just mangle some pointers instead of actually copying the data. Even if you’re doing a lot of insertion in the middle of the list, RingBuf can still be better because of how expensive it is to seek to the middle of a LinkedList.

Known problems

False positives – the instances where using a LinkedList makes sense are few and far between, but they can still happen.

Example

let x: LinkedList<usize> = LinkedList::new();

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for lint groups with the same priority as lints in the Cargo.toml [lints] table.

This lint will be removed once cargo#12918 is resolved.

Why is this bad?

The order of lints in the [lints] is ignored, to have a lint override a group the priority field needs to be used, otherwise the sort order is undefined.

Known problems

Does not check lints inherited using lints.workspace = true

Example

[lints.clippy]
pedantic = "warn"
similar_names = "allow"

Use instead:

[lints.clippy]
pedantic = { level = "warn", priority = -1 }
similar_names = "allow"
Applicability: Unspecified(?)
Added in: 1.78.0

What it does

Checks for the usage of the to_le_bytes method and/or the function from_le_bytes.

Why restrict this?

To ensure use of big-endian or the target’s endianness rather than little-endian.

Example

let _x = 2i32.to_le_bytes();
let _y = 2i64.to_le_bytes();
Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Checks for whole number float literals that cannot be represented as the underlying type without loss.

Why restrict this?

If the value was intended to be exact, it will not be. This may be especially surprising when the lost precision is to the left of the decimal point.

Example

let _: f32 = 16_777_217.0; // 16_777_216.0

Use instead:

let _: f32 = 16_777_216.0;
let _: f64 = 16_777_217.0;
Applicability: MachineApplicable(?)
Added in: 1.43.0

What it does

Looks for macros that expand metavariables in an unsafe block.

Why is this bad?

This hides an unsafe block and allows the user of the macro to write unsafe code without an explicit unsafe block at callsite, making it possible to perform unsafe operations in seemingly safe code.

The macro should be restructured so that these metavariables are referenced outside of unsafe blocks and that the usual unsafety checks apply to the macro argument.

This is usually done by binding it to a variable outside of the unsafe block and then using that variable inside of the block as shown in the example, or by referencing it a second time in a safe context, e.g. if false { $expr }.

Known limitations

Due to how macros are represented in the compiler at the time Clippy runs its lints, it’s not possible to look for metavariables in macro definitions directly.

Instead, this lint looks at expansions of macros. This leads to false negatives for macros that are never actually invoked.

By default, this lint is rather conservative and will only emit warnings on publicly-exported macros from the same crate, because oftentimes private internal macros are one-off macros where this lint would just be noise (e.g. macros that generate impl blocks). The default behavior should help with preventing a high number of such false positives, however it can be configured to also emit warnings in private macros if desired.

Example

/// Gets the first element of a slice
macro_rules! first {
    ($slice:expr) => {
        unsafe {
            let slice = $slice; // ⚠️ expansion inside of `unsafe {}`

            assert!(!slice.is_empty());
            // SAFETY: slice is checked to have at least one element
            slice.first().unwrap_unchecked()
        }
    }
}

assert_eq!(*first!(&[1i32]), 1);

// This will compile as a consequence (note the lack of `unsafe {}`)
assert_eq!(*first!(std::hint::unreachable_unchecked() as &[i32]), 1);

Use instead:

macro_rules! first {
    ($slice:expr) => {{
        let slice = $slice; // ✅ outside of `unsafe {}`
        unsafe {
            assert!(!slice.is_empty());
            // SAFETY: slice is checked to have at least one element
            slice.first().unwrap_unchecked()
        }
    }}
}

assert_eq!(*first!(&[1]), 1);

// This won't compile:
assert_eq!(*first!(std::hint::unreachable_unchecked() as &[i32]), 1);

Configuration

  • warn-unsafe-macro-metavars-in-private-macros: Whether to also emit warnings for unsafe blocks with metavariable expansions in private macros.

    (default: false)

Applicability: Unspecified(?)
Added in: 1.80.0

What it does

Checks for #[macro_use] use....

Why is this bad?

Since the Rust 2018 edition you can import macro’s directly, this is considered idiomatic.

Example

#[macro_use]
use some_macro;
Applicability: MaybeIncorrect(?)
Added in: 1.44.0

What it does

Checks for recursion using the entrypoint.

Why is this bad?

Apart from special setups (which we could detect following attributes like #![no_std]), recursing into main() seems like an unintuitive anti-pattern we should be able to detect.

Example

fn main() {
    main();
}
Applicability: Unspecified(?)
Added in: 1.38.0

What it does

Detects if-then-panic! that can be replaced with assert!.

Why is this bad?

assert! is simpler than if-then-panic!.

Example

let sad_people: Vec<&str> = vec![];
if !sad_people.is_empty() {
    panic!("there are sad people: {:?}", sad_people);
}

Use instead:

let sad_people: Vec<&str> = vec![];
assert!(sad_people.is_empty(), "there are sad people: {:?}", sad_people);
Applicability: MachineApplicable(?)
Added in: 1.57.0

What it does

It checks for manual implementations of async functions.

Why is this bad?

It’s more idiomatic to use the dedicated syntax.

Example

use std::future::Future;

fn foo() -> impl Future<Output = i32> { async { 42 } }

Use instead:

async fn foo() -> i32 { 42 }
Applicability: MachineApplicable(?)
Added in: 1.45.0

What it does

Checks for usage of std::mem::size_of::<T>() * 8 when T::BITS is available.

Why is this bad?

Can be written as the shorter T::BITS.

Example

std::mem::size_of::<usize>() * 8;

Use instead:

usize::BITS as usize;

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.60.0

What it does

Checks for the manual creation of C strings (a string with a NUL byte at the end), either through one of the CStr constructor functions, or more plainly by calling .as_ptr() on a (byte) string literal with a hardcoded \0 byte at the end.

Why is this bad?

This can be written more concisely using c"str" literals and is also less error-prone, because the compiler checks for interior NUL bytes and the terminating NUL byte is inserted automatically.

Example

fn needs_cstr(_: &CStr) {}

needs_cstr(CStr::from_bytes_with_nul(b"Hello\0").unwrap());
unsafe { libc::puts("World\0".as_ptr().cast()) }

Use instead:

fn needs_cstr(_: &CStr) {}

needs_cstr(c"Hello");
unsafe { libc::puts(c"World".as_ptr()) }

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.78.0

What it does

Identifies good opportunities for a clamp function from std or core, and suggests using it.

Why is this bad?

clamp is much shorter, easier to read, and doesn’t use any control flow.

Limitations

This lint will only trigger if max and min are known at compile time, and max is greater than min.

Known issue(s)

If the clamped variable is NaN this suggestion will cause the code to propagate NaN rather than returning either max or min.

clamp functions will panic if max < min, max.is_nan(), or min.is_nan(). Some may consider panicking in these situations to be desirable, but it also may introduce panicking where there wasn’t any before.

See also the discussion in the PR.

Examples

if input > max {
    max
} else if input < min {
    min
} else {
    input
}
input.max(min).min(max)
match input {
    x if x > max => max,
    x if x < min => min,
    x => x,
}
let mut x = input;
if x < min { x = min; }
if x > max { x = max; }

Use instead:

input.clamp(min, max)

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MaybeIncorrect(?)
Added in: 1.66.0

What it does

Checks for an expression like (x + (y - 1)) / y which is a common manual reimplementation of x.div_ceil(y).

Why is this bad?

It’s simpler, clearer and more readable.

Example

let x: i32 = 7;
let y: i32 = 4;
let div = (x + (y - 1)) / y;

Use instead:

#![feature(int_roundings)]
let x: i32 = 7;
let y: i32 = 4;
let div = x.div_ceil(y);
Applicability: MachineApplicable(?)
Added in: 1.81.0

What it does

Checks for usage of match which could be implemented using filter

Why is this bad?

Using the filter method is clearer and more concise.

Example

match Some(0) {
    Some(x) => if x % 2 == 0 {
                    Some(x)
               } else {
                    None
                },
    None => None,
};

Use instead:

Some(0).filter(|&x| x % 2 == 0);
Applicability: MachineApplicable(?)
Added in: 1.66.0

What it does

Checks for usage of _.filter(_).map(_) that can be written more simply as filter_map(_).

Why is this bad?

Redundant code in the filter and map operations is poor style and less performant.

Example

(0_i32..10)
    .filter(|n| n.checked_add(1).is_some())
    .map(|n| n.checked_add(1).unwrap());

Use instead:

(0_i32..10).filter_map(|n| n.checked_add(1));

Past names

  • filter_map
Applicability: MachineApplicable(?)
Added in: 1.51.0

What it does

Checks for manual implementations of Iterator::find

Why is this bad?

It doesn’t affect performance, but using find is shorter and easier to read.

Example

fn example(arr: Vec<i32>) -> Option<i32> {
    for el in arr {
        if el == 1 {
            return Some(el);
        }
    }
    None
}

Use instead:

fn example(arr: Vec<i32>) -> Option<i32> {
    arr.into_iter().find(|&el| el == 1)
}
Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

Checks for usage of _.find(_).map(_) that can be written more simply as find_map(_).

Why is this bad?

Redundant code in the find and map operations is poor style and less performant.

Example

(0_i32..10)
    .find(|n| n.checked_add(1).is_some())
    .map(|n| n.checked_add(1).unwrap());

Use instead:

(0_i32..10).find_map(|n| n.checked_add(1));

Past names

  • find_map
Applicability: MachineApplicable(?)
Added in: 1.51.0

What it does

Checks for unnecessary if let usage in a for loop where only the Some or Ok variant of the iterator element is used.

Why is this bad?

It is verbose and can be simplified by first calling the flatten method on the Iterator.

Example

let x = vec![Some(1), Some(2), Some(3)];
for n in x {
    if let Some(n) = n {
        println!("{}", n);
    }
}

Use instead:

let x = vec![Some(1), Some(2), Some(3)];
for n in x.into_iter().flatten() {
    println!("{}", n);
}
Applicability: MaybeIncorrect(?)
Added in: 1.52.0

What it does

Checks for cases where BuildHasher::hash_one can be used.

Why is this bad?

It is more concise to use the hash_one method.

Example

use std::hash::{BuildHasher, Hash, Hasher};
use std::collections::hash_map::RandomState;

let s = RandomState::new();
let value = vec![1, 2, 3];

let mut hasher = s.build_hasher();
value.hash(&mut hasher);
let hash = hasher.finish();

Use instead:

use std::hash::BuildHasher;
use std::collections::hash_map::RandomState;

let s = RandomState::new();
let value = vec![1, 2, 3];

let hash = s.hash_one(&value);

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.75.0

What it does

Checks for manual case-insensitive ASCII comparison.

Why is this bad?

The eq_ignore_ascii_case method is faster because it does not allocate memory for the new strings, and it is more readable.

Example

fn compare(a: &str, b: &str) -> bool {
    a.to_ascii_lowercase() == b.to_ascii_lowercase() || a.to_ascii_lowercase() == "abc"
}

Use instead:

fn compare(a: &str, b: &str) -> bool {
   a.eq_ignore_ascii_case(b) || a.eq_ignore_ascii_case("abc")
}
Applicability: MachineApplicable(?)
Added in: 1.82.0

What it does

Checks for uses of map which return the original item.

Why is this bad?

inspect is both clearer in intent and shorter.

Example

let x = Some(0).map(|x| { println!("{x}"); x });

Use instead:

let x = Some(0).inspect(|x| println!("{x}"));
Applicability: MachineApplicable(?)
Added in: 1.81.0

What it does

Lints subtraction between Instant::now() and another Instant.

Why is this bad?

It is easy to accidentally write prev_instant - Instant::now(), which will always be 0ns as Instant subtraction saturates.

prev_instant.elapsed() also more clearly signals intention.

Example

use std::time::Instant;
let prev_instant = Instant::now();
let duration = Instant::now() - prev_instant;

Use instead:

use std::time::Instant;
let prev_instant = Instant::now();
let duration = prev_instant.elapsed();
Applicability: MachineApplicable(?)
Added in: 1.65.0

What it does

Suggests to use dedicated built-in methods, is_ascii_(lowercase|uppercase|digit|hexdigit) for checking on corresponding ascii range

Why is this bad?

Using the built-in functions is more readable and makes it clear that it’s not a specific subset of characters, but all ASCII (lowercase|uppercase|digit|hexdigit) characters.

Example

fn main() {
    assert!(matches!('x', 'a'..='z'));
    assert!(matches!(b'X', b'A'..=b'Z'));
    assert!(matches!('2', '0'..='9'));
    assert!(matches!('x', 'A'..='Z' | 'a'..='z'));
    assert!(matches!('C', '0'..='9' | 'a'..='f' | 'A'..='F'));

    ('0'..='9').contains(&'0');
    ('a'..='z').contains(&'a');
    ('A'..='Z').contains(&'A');
}

Use instead:

fn main() {
    assert!('x'.is_ascii_lowercase());
    assert!(b'X'.is_ascii_uppercase());
    assert!('2'.is_ascii_digit());
    assert!('x'.is_ascii_alphabetic());
    assert!('C'.is_ascii_hexdigit());

    '0'.is_ascii_digit();
    'a'.is_ascii_lowercase();
    'A'.is_ascii_uppercase();
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.67.0

What it does

Checks for manual is_finite reimplementations (i.e., x != <float>::INFINITY && x != <float>::NEG_INFINITY).

Why is this bad?

The method is_finite is shorter and more readable.

Example

if x != f32::INFINITY && x != f32::NEG_INFINITY {}
if x.abs() < f32::INFINITY {}

Use instead:

if x.is_finite() {}
if x.is_finite() {}
Applicability: MaybeIncorrect(?)
Added in: 1.73.0

What it does

Checks for manual is_infinite reimplementations (i.e., x == <float>::INFINITY || x == <float>::NEG_INFINITY).

Why is this bad?

The method is_infinite is shorter and more readable.

Example

if x == f32::INFINITY || x == f32::NEG_INFINITY {}

Use instead:

if x.is_infinite() {}
Applicability: MachineApplicable(?)
Added in: 1.73.0

What it does

Checks for expressions like x.count_ones() == 1 or x & (x - 1) == 0, with x and unsigned integer, which may be manual reimplementations of x.is_power_of_two().

Why is this bad?

Manual reimplementations of is_power_of_two increase code complexity for little benefit.

Example

let a: u32 = 4;
let result = a.count_ones() == 1;

Use instead:

let a: u32 = 4;
let result = a.is_power_of_two();
Applicability: MachineApplicable(?)
Added in: 1.82.0

What it does

Checks for usage of option.map(f).unwrap_or_default() and result.map(f).unwrap_or_default() where f is a function or closure that returns the bool type.

Why is this bad?

Readability. These can be written more concisely as option.is_some_and(f) and result.is_ok_and(f).

Example

option.map(|a| a > 10).unwrap_or_default();
result.map(|a| a > 10).unwrap_or_default();

Use instead:

option.is_some_and(|a| a > 10);
result.is_ok_and(|a| a > 10);
Applicability: MachineApplicable(?)
Added in: 1.77.0

What it does

Warn of cases where let...else could be used

Why is this bad?

let...else provides a standard construct for this pattern that people can easily recognize. It’s also more compact.

Example

let v = if let Some(v) = w { v } else { return };

Could be written:

let Some(v) = w else { return };

Configuration

  • matches-for-let-else: Whether the matches should be considered by the lint, and whether there should be filtering for common types.

    (default: "WellKnownTypes")

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: HasPlaceholders(?)
Added in: 1.67.0

What it does

Checks for references on std::path::MAIN_SEPARATOR.to_string() used to build a &str.

Why is this bad?

There exists a std::path::MAIN_SEPARATOR_STR which does not require an extra memory allocation.

Example

let s: &str = &std::path::MAIN_SEPARATOR.to_string();

Use instead:

let s: &str = std::path::MAIN_SEPARATOR_STR;
Applicability: MachineApplicable(?)
Added in: 1.70.0

What it does

Checks for usage of match which could be implemented using map

Why is this bad?

Using the map method is clearer and more concise.

Example

match Some(0) {
    Some(x) => Some(x + 1),
    None => None,
};

Use instead:

Some(0).map(|x| x + 1);
Applicability: MachineApplicable(?)
Added in: 1.52.0

What it does

Checks for for-loops that manually copy items between slices that could be optimized by having a memcpy.

Why is this bad?

It is not as fast as a memcpy.

Example

for i in 0..src.len() {
    dst[i + 64] = src[i];
}

Use instead:

dst[64..(src.len() + 64)].clone_from_slice(&src[..]);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for .rev().next() on a DoubleEndedIterator

Why is this bad?

.next_back() is cleaner.

Example

foo.iter().rev().next();

Use instead:

foo.iter().next_back();
Applicability: MachineApplicable(?)
Added in: 1.71.0

What it does

Checks for manual implementations of the non-exhaustive pattern.

Why is this bad?

Using the #[non_exhaustive] attribute expresses better the intent and allows possible optimizations when applied to enums.

Example

struct S {
    pub a: i32,
    pub b: i32,
    _c: (),
}

enum E {
    A,
    B,
    #[doc(hidden)]
    _C,
}

struct T(pub i32, pub i32, ());

Use instead:

#[non_exhaustive]
struct S {
    pub a: i32,
    pub b: i32,
}

#[non_exhaustive]
enum E {
    A,
    B,
}

#[non_exhaustive]
struct T(pub i32, pub i32);

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MaybeIncorrect(?)
Added in: 1.45.0

What it does

Finds patterns that reimplement Option::ok_or.

Why is this bad?

Concise code helps focusing on behavior instead of boilerplate.

Examples

let foo: Option<i32> = None;
foo.map_or(Err("error"), |v| Ok(v));

Use instead:

let foo: Option<i32> = None;
foo.ok_or("error");
Applicability: MachineApplicable(?)
Added in: 1.49.0

What it does

Checks for manual char comparison in string patterns

Why is this bad?

This can be written more concisely using a char or an array of char. This is more readable and more optimized when comparing to only one char.

Example

"Hello World!".trim_end_matches(|c| c == '.' || c == ',' || c == '!' || c == '?');

Use instead:

"Hello World!".trim_end_matches(['.', ',', '!', '?']);

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.81.0

What it does

Checks for expressions like x >= 3 && x < 8 that could be more readably expressed as (3..8).contains(x).

Why is this bad?

contains expresses the intent better and has less failure modes (such as fencepost errors or using || instead of &&).

Example

// given
let x = 6;

assert!(x >= 3 && x < 8);

Use instead:

assert!((3..8).contains(&x));

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.49.0

What it does

Looks for combined OR patterns that are all contained in a specific range, e.g. 6 | 4 | 5 | 9 | 7 | 8 can be rewritten as 4..=9.

Why is this bad?

Using an explicit range is more concise and easier to read.

Known issues

This lint intentionally does not handle numbers greater than i128::MAX for u128 literals in order to support negative numbers.

Example

let x = 6;
let foo = matches!(x, 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10);

Use instead:

let x = 6;
let foo = matches!(x, 1..=10);
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Checks for an expression like ((x % 4) + 4) % 4 which is a common manual reimplementation of x.rem_euclid(4).

Why is this bad?

It’s simpler and more readable.

Example

let x: i32 = 24;
let rem = ((x % 4) + 4) % 4;

Use instead:

let x: i32 = 24;
let rem = x.rem_euclid(4);

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

Checks for code to be replaced by .retain().

Why is this bad?

.retain() is simpler and avoids needless allocation.

Example

let mut vec = vec![0, 1, 2];
vec = vec.iter().filter(|&x| x % 2 == 0).copied().collect();
vec = vec.into_iter().filter(|x| x % 2 == 0).collect();

Use instead:

let mut vec = vec![0, 1, 2];
vec.retain(|x| x % 2 == 0);
vec.retain(|x| x % 2 == 0);

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

It detects manual bit rotations that could be rewritten using standard functions rotate_left or rotate_right.

Why is this bad?

Calling the function better conveys the intent.

Known issues

Currently, the lint only catches shifts by constant amount.

Example

let x = 12345678_u32;
let _ = (x >> 8) | (x << 24);

Use instead:

let x = 12345678_u32;
let _ = x.rotate_right(8);
Applicability: MachineApplicable(?)
Added in: 1.81.0

What it does

Checks for .checked_add/sub(x).unwrap_or(MAX/MIN).

Why is this bad?

These can be written simply with saturating_add/sub methods.

Example

let add = x.checked_add(y).unwrap_or(u32::MAX);
let sub = x.checked_sub(y).unwrap_or(u32::MIN);

can be written using dedicated methods for saturating addition/subtraction as:

let add = x.saturating_add(y);
let sub = x.saturating_sub(y);
Applicability: MachineApplicable(?)
Added in: 1.39.0

What it does

When a is &[T], detect a.len() * size_of::<T>() and suggest size_of_val(a) instead.

Why is this better?

  • Shorter to write
  • Removes the need for the human and the compiler to worry about overflow in the multiplication
  • Potentially faster at runtime as rust emits special no-wrapping flags when it calculates the byte length
  • Less turbofishing

Example

let newlen = data.len() * std::mem::size_of::<i32>();

Use instead:

let newlen = std::mem::size_of_val(data);
Applicability: MachineApplicable(?)
Added in: 1.70.0

What it does

Checks for usage of str::splitn(2, _)

Why is this bad?

split_once is both clearer in intent and slightly more efficient.

Example

let s = "key=value=add";
let (key, value) = s.splitn(2, '=').next_tuple()?;
let value = s.splitn(2, '=').nth(1)?;

let mut parts = s.splitn(2, '=');
let key = parts.next()?;
let value = parts.next()?;

Use instead:

let s = "key=value=add";
let (key, value) = s.split_once('=')?;
let value = s.split_once('=')?.1;

let (key, value) = s.split_once('=')?;

Limitations

The multiple statement variant currently only detects iter.next()?/iter.next().unwrap() in two separate let statements that immediately follow the splitn()

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.57.0

What it does

Checks for manual implementations of str::repeat

Why is this bad?

These are both harder to read, as well as less performant.

Example

let x: String = std::iter::repeat('x').take(10).collect();

Use instead:

let x: String = "x".repeat(10);

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.54.0

What it does

Checks for usage of "" to create a String, such as "".to_string(), "".to_owned(), String::from("") and others.

Why is this bad?

Different ways of creating an empty string makes your code less standardized, which can be confusing.

Example

let a = "".to_string();
let b: String = "".into();

Use instead:

let a = String::new();
let b = String::new();
Applicability: MachineApplicable(?)
Added in: 1.65.0

What it does

Suggests using strip_{prefix,suffix} over str::{starts,ends}_with and slicing using the pattern’s length.

Why is this bad?

Using str:strip_{prefix,suffix} is safer and may have better performance as there is no slicing which may panic and the compiler does not need to insert this panic code. It is also sometimes more readable as it removes the need for duplicating or storing the pattern used by str::{starts,ends}_with and in the slicing.

Example

let s = "hello, world!";
if s.starts_with("hello, ") {
    assert_eq!(s["hello, ".len()..].to_uppercase(), "WORLD!");
}

Use instead:

let s = "hello, world!";
if let Some(end) = s.strip_prefix("hello, ") {
    assert_eq!(end.to_uppercase(), "WORLD!");
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: HasPlaceholders(?)
Added in: 1.48.0

What it does

Checks for manual swapping.

Note that the lint will not be emitted in const blocks, as the suggestion would not be applicable.

Why is this bad?

The std::mem::swap function exposes the intent better without deinitializing or copying either variable.

Example

let mut a = 42;
let mut b = 1337;

let t = b;
b = a;
a = t;

Use std::mem::swap():

let mut a = 1;
let mut b = 2;
std::mem::swap(&mut a, &mut b);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of Iterator::fold with a type that implements Try.

Why is this bad?

The code should use try_fold instead, which short-circuits on failure, thus opening the door for additional optimizations not possible with fold as rustc can guarantee the function is never called on None, Err, etc., alleviating otherwise necessary checks. It’s also slightly more idiomatic.

Known issues

This lint doesn’t take into account whether a function does something on the failure case, i.e., whether short-circuiting will affect behavior. Refactoring to try_fold is not desirable in those cases.

Example

vec![1, 2, 3].iter().fold(Some(0i32), |sum, i| sum?.checked_add(*i));

Use instead:

vec![1, 2, 3].iter().try_fold(0i32, |sum, i| sum.checked_add(*i));

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: HasPlaceholders(?)
Added in: 1.72.0

What it does

Finds patterns that reimplement Option::unwrap_or or Result::unwrap_or.

Why is this bad?

Concise code helps focusing on behavior instead of boilerplate.

Example

let foo: Option<i32> = None;
match foo {
    Some(v) => v,
    None => 1,
};

Use instead:

let foo: Option<i32> = None;
foo.unwrap_or(1);
Applicability: MachineApplicable(?)
Added in: 1.49.0

What it does

Checks if a match or if let expression can be simplified using .unwrap_or_default().

Why is this bad?

It can be done in one call with .unwrap_or_default().

Example

let x: Option<String> = Some(String::new());
let y: String = match x {
    Some(v) => v,
    None => String::new(),
};

let x: Option<Vec<String>> = Some(Vec::new());
let y: Vec<String> = if let Some(v) = x {
    v
} else {
    Vec::new()
};

Use instead:

let x: Option<String> = Some(String::new());
let y: String = x.unwrap_or_default();

let x: Option<Vec<String>> = Some(Vec::new());
let y: Vec<String> = x.unwrap_or_default();
Applicability: MachineApplicable(?)
Added in: 1.79.0

What it does

Looks for loops that check for emptiness of a Vec in the condition and pop an element in the body as a separate operation.

Why is this bad?

Such loops can be written in a more idiomatic way by using a while-let loop and directly pattern matching on the return value of Vec::pop().

Example

let mut numbers = vec![1, 2, 3, 4, 5];
while !numbers.is_empty() {
    let number = numbers.pop().unwrap();
    // use `number`
}

Use instead:

let mut numbers = vec![1, 2, 3, 4, 5];
while let Some(number) = numbers.pop() {
    // use `number`
}
Applicability: MachineApplicable(?)
Added in: 1.71.0

What it does

Checks for too many variables whose name consists of a single character.

Why is this bad?

It’s hard to memorize what a variable means without a descriptive name.

Example

let (a, b, c, d, e, f, g) = (...);

Configuration

  • single-char-binding-names-threshold: The maximum number of single char bindings a scope may have

    (default: 4)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of .map(…), followed by .all(identity) or .any(identity).

Why is this bad?

The .all(…) or .any(…) methods can be called directly in place of .map(…).

Example

let e1 = v.iter().map(|s| s.is_empty()).all(|a| a);
let e2 = v.iter().map(|s| s.is_empty()).any(std::convert::identity);

Use instead:

let e1 = v.iter().all(|s| s.is_empty());
let e2 = v.iter().any(|s| s.is_empty());
Applicability: MachineApplicable(?)
Added in: 1.84.0

What it does

Checks for usage of map(|x| x.clone()) or dereferencing closures for Copy types, on Iterator or Option, and suggests cloned() or copied() instead

Why is this bad?

Readability, this can be written more concisely

Example

let x = vec![42, 43];
let y = x.iter();
let z = y.map(|i| *i);

The correct use would be:

let x = vec![42, 43];
let y = x.iter();
let z = y.cloned();

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of _.map(_).collect::<Result<(), _>().

Why is this bad?

Using try_for_each instead is more readable and idiomatic.

Example

(0..3).map(|t| Err(t)).collect::<Result<(), _>>();

Use instead:

(0..3).try_for_each(|t| Err(t));
Applicability: MachineApplicable(?)
Added in: 1.49.0

What it does

Checks for usage of contains_key + insert on HashMap or BTreeMap.

Why is this bad?

Using entry is more efficient.

Known problems

The suggestion may have type inference errors in some cases. e.g.

let mut map = std::collections::HashMap::new();
let _ = if !map.contains_key(&0) {
    map.insert(0, 0)
} else {
    None
};

Example

if !map.contains_key(&k) {
    map.insert(k, v);
}

Use instead:

map.entry(k).or_insert(v);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for instances of map_err(|_| Some::Enum)

Why restrict this?

This map_err throws away the original error rather than allowing the enum to contain and report the cause of the error.

Example

Before:

use std::fmt;

#[derive(Debug)]
enum Error {
    Indivisible,
    Remainder(u8),
}

impl fmt::Display for Error {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        match self {
            Error::Indivisible => write!(f, "could not divide input by three"),
            Error::Remainder(remainder) => write!(
                f,
                "input is not divisible by three, remainder = {}",
                remainder
            ),
        }
    }
}

impl std::error::Error for Error {}

fn divisible_by_3(input: &str) -> Result<(), Error> {
    input
        .parse::<i32>()
        .map_err(|_| Error::Indivisible)
        .map(|v| v % 3)
        .and_then(|remainder| {
            if remainder == 0 {
                Ok(())
            } else {
                Err(Error::Remainder(remainder as u8))
            }
        })
}

After:

use std::{fmt, num::ParseIntError};

#[derive(Debug)]
enum Error {
    Indivisible(ParseIntError),
    Remainder(u8),
}

impl fmt::Display for Error {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        match self {
            Error::Indivisible(_) => write!(f, "could not divide input by three"),
            Error::Remainder(remainder) => write!(
                f,
                "input is not divisible by three, remainder = {}",
                remainder
            ),
        }
    }
}

impl std::error::Error for Error {
    fn source(&self) -> Option<&(dyn std::error::Error + 'static)> {
        match self {
            Error::Indivisible(source) => Some(source),
            _ => None,
        }
    }
}

fn divisible_by_3(input: &str) -> Result<(), Error> {
    input
        .parse::<i32>()
        .map_err(Error::Indivisible)
        .map(|v| v % 3)
        .and_then(|remainder| {
            if remainder == 0 {
                Ok(())
            } else {
                Err(Error::Remainder(remainder as u8))
            }
        })
}
Applicability: Unspecified(?)
Added in: 1.48.0

What it does

Checks for usage of _.map(_).flatten(_) on Iterator and Option

Why is this bad?

Readability, this can be written more concisely as _.flat_map(_) for Iterator or _.and_then(_) for Option

Example

let vec = vec![vec![1]];
let opt = Some(5);

vec.iter().map(|x| x.iter()).flatten();
opt.map(|x| Some(x * 2)).flatten();

Use instead:

vec.iter().flat_map(|x| x.iter());
opt.and_then(|x| Some(x * 2));
Applicability: MachineApplicable(?)
Added in: 1.31.0

What it does

Checks for instances of map(f) where f is the identity function.

Why is this bad?

It can be written more concisely without the call to map.

Example

let x = [1, 2, 3];
let y: Vec<_> = x.iter().map(|x| x).map(|x| 2*x).collect();

Use instead:

let x = [1, 2, 3];
let y: Vec<_> = x.iter().map(|x| 2*x).collect();
Applicability: MachineApplicable(?)
Added in: 1.47.0

What it does

Checks for usage of option.map(_).unwrap_or(_) or option.map(_).unwrap_or_else(_) or result.map(_).unwrap_or_else(_).

Why is this bad?

Readability, these can be written more concisely (resp.) as option.map_or(_, _), option.map_or_else(_, _) and result.map_or_else(_, _).

Known problems

The order of the arguments is not in execution order

Examples

option.map(|a| a + 1).unwrap_or(0);
option.map(|a| a > 10).unwrap_or(false);
result.map(|a| a + 1).unwrap_or_else(some_function);

Use instead:

option.map_or(0, |a| a + 1);
option.is_some_and(|a| a > 10);
result.map_or_else(some_function, |a| a + 1);

Past names

  • option_map_unwrap_or
  • option_map_unwrap_or_else
  • result_map_unwrap_or_else

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.45.0

What it does

Checks for Iterator::map over ranges without using the parameter which could be more clearly expressed using std::iter::repeat(...).take(...) or std::iter::repeat_n.

Why is this bad?

It expresses the intent more clearly to take the correct number of times from a generating function than to apply a closure to each number in a range only to discard them.

Example

let random_numbers : Vec<_> = (0..10).map(|_| { 3 + 1 }).collect();

Use instead:

let f : Vec<_> = std::iter::repeat( 3 + 1 ).take(10).collect();

Known Issues

This lint may suggest replacing a Map<Range> with a Take<RepeatWith>. The former implements some traits that the latter does not, such as DoubleEndedIterator.

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MaybeIncorrect(?)
Added in: 1.84.0

What it does

Checks for match which is used to add a reference to an Option value.

Why is this bad?

Using as_ref() or as_mut() instead is shorter.

Example

let x: Option<()> = None;

let r: Option<&()> = match x {
    None => None,
    Some(ref v) => Some(v),
};

Use instead:

let x: Option<()> = None;

let r: Option<&()> = x.as_ref();
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for matches where match expression is a bool. It suggests to replace the expression with an if...else block.

Why is this bad?

It makes the code less readable.

Example

let condition: bool = true;
match condition {
    true => foo(),
    false => bar(),
}

Use if/else instead:

let condition: bool = true;
if condition {
    foo();
} else {
    bar();
}
Applicability: HasPlaceholders(?)
Added in: pre 1.29.0

What it does

Checks for match or if let expressions producing a bool that could be written using matches!

Why is this bad?

Readability and needless complexity.

Known problems

This lint falsely triggers, if there are arms with cfg attributes that remove an arm evaluating to false.

Example

let x = Some(5);

let a = match x {
    Some(0) => true,
    _ => false,
};

let a = if let Some(0) = x {
    true
} else {
    false
};

Use instead:

let x = Some(5);
let a = matches!(x, Some(0));

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MaybeIncorrect(?)
Added in: 1.47.0

What it does

Checks for match vec[idx] or match vec[n..m].

Why is this bad?

This can panic at runtime.

Example

let arr = vec![0, 1, 2, 3];
let idx = 1;

match arr[idx] {
    0 => println!("{}", 0),
    1 => println!("{}", 3),
    _ => {},
}

Use instead:

let arr = vec![0, 1, 2, 3];
let idx = 1;

match arr.get(idx) {
    Some(0) => println!("{}", 0),
    Some(1) => println!("{}", 3),
    _ => {},
}
Applicability: MaybeIncorrect(?)
Added in: 1.45.0

What it does

Checks for overlapping match arms.

Why is this bad?

It is likely to be an error and if not, makes the code less obvious.

Example

let x = 5;
match x {
    1..=10 => println!("1 ... 10"),
    5..=15 => println!("5 ... 15"),
    _ => (),
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for matches where all arms match a reference, suggesting to remove the reference and deref the matched expression instead. It also checks for if let &foo = bar blocks.

Why is this bad?

It just makes the code less readable. That reference destructuring adds nothing to the code.

Example

match x {
    &A(ref y) => foo(y),
    &B => bar(),
    _ => frob(&x),
}

Use instead:

match *x {
    A(ref y) => foo(y),
    B => bar(),
    _ => frob(x),
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for unnecessary ok() in while let.

Why is this bad?

Calling ok() in while let is unnecessary, instead match on Ok(pat)

Example

while let Some(value) = iter.next().ok() {
    vec.push(value)
}

if let Some(value) = iter.next().ok() {
    vec.push(value)
}

Use instead:

while let Ok(value) = iter.next() {
    vec.push(value)
}

if let Ok(value) = iter.next() {
       vec.push(value)
}

Past names

  • if_let_some_result
Applicability: MachineApplicable(?)
Added in: 1.57.0

What it does

Checks for match with identical arm bodies.

Note: Does not lint on wildcards if the non_exhaustive_omitted_patterns_lint feature is enabled and disallowed.

Why is this bad?

This is probably a copy & paste error. If arm bodies are the same on purpose, you can factor them using |.

Known problems

False positive possible with order dependent match (see issue #860).

Example

match foo {
    Bar => bar(),
    Quz => quz(),
    Baz => bar(), // <= oops
}

This should probably be

match foo {
    Bar => bar(),
    Quz => quz(),
    Baz => baz(), // <= fixed
}

or if the original code was not a typo:

match foo {
    Bar | Baz => bar(), // <= shows the intent better
    Quz => quz(),
}
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for useless match that binds to only one value.

Why is this bad?

Readability and needless complexity.

Known problems

Suggested replacements may be incorrect when match is actually binding temporary value, bringing a ‘dropped while borrowed’ error.

Example

match (a, b) {
    (c, d) => {
        // useless match
    }
}

Use instead:

let (c, d) = (a, b);
Applicability: MachineApplicable(?)
Added in: 1.43.0

What it does

Checks for match expressions modifying the case of a string with non-compliant arms

Why is this bad?

The arm is unreachable, which is likely a mistake

Example

match &*text.to_ascii_lowercase() {
    "foo" => {},
    "Bar" => {},
    _ => {},
}

Use instead:

match &*text.to_ascii_lowercase() {
    "foo" => {},
    "bar" => {},
    _ => {},
}
Applicability: MachineApplicable(?)
Added in: 1.58.0

What it does

Checks for arm which matches all errors with Err(_) and take drastic actions like panic!.

Why is this bad?

It is generally a bad practice, similar to catching all exceptions in java with catch(Exception)

Example

let x: Result<i32, &str> = Ok(3);
match x {
    Ok(_) => println!("ok"),
    Err(_) => panic!("err"),
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for wildcard enum matches for a single variant.

Why is this bad?

New enum variants added by library updates can be missed.

Known problems

Suggested replacements may not use correct path to enum if it’s not present in the current scope.

Example

match x {
    Foo::A => {},
    Foo::B => {},
    _ => {},
}

Use instead:

match x {
    Foo::A => {},
    Foo::B => {},
    Foo::C => {},
}
Applicability: MaybeIncorrect(?)
Added in: 1.45.0

What it does

Checks for iteration that may be infinite.

Why is this bad?

While there may be places where this is acceptable (e.g., in event streams), in most cases this is simply an error.

Known problems

The code may have a condition to stop iteration, but this lint is not clever enough to analyze it.

Example

let infinite_iter = 0..;
[0..].iter().zip(infinite_iter.take_while(|x| *x > 5));
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of std::mem::forget(t) where t is Drop or has a field that implements Drop.

Why restrict this?

std::mem::forget(t) prevents t from running its destructor, possibly causing leaks. It is not possible to detect all means of creating leaks, but it may be desirable to prohibit the simple ones.

Example

mem::forget(Rc::new(55))
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for mem::replace() on an Option with None.

Why is this bad?

Option already has the method take() for taking its current value (Some(..) or None) and replacing it with None.

Example

use std::mem;

let mut an_option = Some(0);
let replaced = mem::replace(&mut an_option, None);

Is better expressed with:

let mut an_option = Some(0);
let taken = an_option.take();
Applicability: MachineApplicable(?)
Added in: 1.31.0

What it does

Checks for std::mem::replace on a value of type T with T::default().

Why is this bad?

std::mem module already has the method take to take the current value and replace it with the default value of that type.

Example

let mut text = String::from("foo");
let replaced = std::mem::replace(&mut text, String::default());

Is better expressed with:

let mut text = String::from("foo");
let taken = std::mem::take(&mut text);

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.42.0

What it does

Checks for mem::replace(&mut _, mem::uninitialized()) and mem::replace(&mut _, mem::zeroed()).

Why is this bad?

This will lead to undefined behavior even if the value is overwritten later, because the uninitialized value may be observed in the case of a panic.

Example

use std::mem;

#[allow(deprecated, invalid_value)]
fn myfunc (v: &mut Vec<i32>) {
    let taken_v = unsafe { mem::replace(v, mem::uninitialized()) };
    let new_v = may_panic(taken_v); // undefined behavior on panic
    mem::forget(mem::replace(v, new_v));
}

The take_mut crate offers a sound solution, at the cost of either lazily creating a replacement value or aborting on panic, to ensure that the uninitialized value cannot be observed.

Applicability: MachineApplicable(?)
Added in: 1.39.0

What it does

Checks for identifiers which consist of a single character (or fewer than the configured threshold).

Note: This lint can be very noisy when enabled; it may be desirable to only enable it temporarily.

Why restrict this?

To improve readability by requiring that every variable has a name more specific than a single letter can be.

Example

for m in movies {
    let title = m.t;
}

Use instead:

for movie in movies {
    let title = movie.title;
}

Configuration

  • allowed-idents-below-min-chars: Allowed names below the minimum allowed characters. The value ".." can be used as part of the list to indicate, that the configured values should be appended to the default configuration of Clippy. By default, any configuration will replace the default value.

    (default: ["i", "j", "x", "y", "z", "w", "n"])

  • min-ident-chars-threshold: Minimum chars an ident can have, anything below or equal to this will be linted.

    (default: 1)

Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Checks for expressions where std::cmp::min and max are used to clamp values, but switched so that the result is constant.

Why is this bad?

This is in all probability not the intended outcome. At the least it hurts readability of the code.

Example

min(0, max(100, x))

// or

x.max(100).min(0)

It will always be equal to 0. Probably the author meant to clamp the value between 0 and 100, but has erroneously swapped min and max.

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

Split into clippy::cast_ptr_alignment and clippy::transmute_ptr_to_ptr.

Applicability: Unspecified(?)
Deprecated in: pre 1.29.0

What it does

Checks for type parameters which are positioned inconsistently between a type definition and impl block. Specifically, a parameter in an impl block which has the same name as a parameter in the type def, but is in a different place.

Why is this bad?

Type parameters are determined by their position rather than name. Naming type parameters inconsistently may cause you to refer to the wrong type parameter.

Limitations

This lint only applies to impl blocks with simple generic params, e.g. A. If there is anything more complicated, such as a tuple, it will be ignored.

Example

struct Foo<A, B> {
    x: A,
    y: B,
}
// inside the impl, B refers to Foo::A
impl<B, A> Foo<B, A> {}

Use instead:

struct Foo<A, B> {
    x: A,
    y: B,
}
impl<A, B> Foo<A, B> {}
Applicability: Unspecified(?)
Added in: 1.63.0

What it does

Checks for getter methods that return a field that doesn’t correspond to the name of the method, when there is a field’s whose name matches that of the method.

Why is this bad?

It is most likely that such a method is a bug caused by a typo or by copy-pasting.

Example

struct A {
    a: String,
    b: String,
}

impl A {
    fn a(&self) -> &str{
        &self.b
    }
}

Use instead:

struct A {
    a: String,
    b: String,
}

impl A {
    fn a(&self) -> &str{
        &self.a
    }
}
Applicability: MaybeIncorrect(?)
Added in: 1.67.0

What it does

Checks for a op= a op b or a op= b op a patterns.

Why is this bad?

Most likely these are bugs where one meant to write a op= b.

Known problems

Clippy cannot know for sure if a op= a op b should have been a = a op a op b or a = a op b/a op= b. Therefore, it suggests both. If a op= a op b is really the correct behavior it should be written as a = a op a op b as it’s less confusing.

Example

let mut a = 5;
let b = 2;
// ...
a += a + b;
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks assertions without a custom panic message.

Why restrict this?

Without a good custom message, it’d be hard to understand what went wrong when the assertion fails. A good custom message should be more about why the failure of the assertion is problematic and not what is failed because the assertion already conveys that.

Although the same reasoning applies to testing functions, this lint ignores them as they would be too noisy. Also, in most cases understanding the test failure would be easier compared to understanding a complex invariant distributed around the codebase.

Known problems

This lint cannot check the quality of the custom panic messages. Hence, you can suppress this lint simply by adding placeholder messages like “assertion failed”. However, we recommend coming up with good messages that provide useful information instead of placeholder messages that don’t provide any extra information.

Example

fn call(service: Service) {
    assert!(service.ready);
}

Use instead:

fn call(service: Service) {
    assert!(service.ready, "`service.poll_ready()` must be called first to ensure that service is ready to receive requests");
}
Applicability: Unspecified(?)
Added in: 1.70.0

What it does

Checks for repeated slice indexing without asserting beforehand that the length is greater than the largest index used to index into the slice.

Why restrict this?

In the general case where the compiler does not have a lot of information about the length of a slice, indexing it repeatedly will generate a bounds check for every single index.

Asserting that the length of the slice is at least as large as the largest value to index beforehand gives the compiler enough information to elide the bounds checks, effectively reducing the number of bounds checks from however many times the slice was indexed to just one (the assert).

Drawbacks

False positives. It is, in general, very difficult to predict how well the optimizer will be able to elide bounds checks and it very much depends on the surrounding code. For example, indexing into the slice yielded by the slice::chunks_exact iterator will likely have all of the bounds checks elided even without an assert if the chunk_size is a constant.

Asserts are not tracked across function calls. Asserting the length of a slice in a different function likely gives the optimizer enough information about the length of a slice, but this lint will not detect that.

Example

fn sum(v: &[u8]) -> u8 {
    // 4 bounds checks
    v[0] + v[1] + v[2] + v[3]
}

Use instead:

fn sum(v: &[u8]) -> u8 {
    assert!(v.len() > 3);
    // no bounds checks
    v[0] + v[1] + v[2] + v[3]
}
Applicability: MachineApplicable(?)
Added in: 1.74.0

What it does

Suggests the use of const in functions and methods where possible.

Why is this bad?

Not having the function const prevents callers of the function from being const as well.

Known problems

Const functions are currently still being worked on, with some features only being available on nightly. This lint does not consider all edge cases currently and the suggestions may be incorrect if you are using this lint on stable.

Also, the lint only runs one pass over the code. Consider these two non-const functions:

fn a() -> i32 {
    0
}
fn b() -> i32 {
    a()
}

When running Clippy, the lint will only suggest to make a const, because b at this time can’t be const as it calls a non-const function. Making a const and running Clippy again, will suggest to make b const, too.

If you are marking a public function with const, removing it again will break API compatibility.

Example

fn new() -> Self {
    Self { random_number: 42 }
}

Could be a const fn:

const fn new() -> Self {
    Self { random_number: 42 }
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.34.0

What it does

Suggests to use const in thread_local! macro if possible.

Why is this bad?

The thread_local! macro wraps static declarations and makes them thread-local. It supports using a const keyword that may be used for declarations that can be evaluated as a constant expression. This can enable a more efficient thread local implementation that can avoid lazy initialization. For types that do not need to be dropped, this can enable an even more efficient implementation that does not need to track any additional state.

https://doc.rust-lang.org/std/macro.thread_local.html

Example

// example code where clippy issues a warning
thread_local! {
    static BUF: String = String::new();
}

Use instead:

// example code which does not raise clippy warning
thread_local! {
    static BUF: String = const { String::new() };
}

Past names

  • thread_local_initializer_can_be_made_const
Applicability: MachineApplicable(?)
Added in: 1.77.0

What it does

Warns if there is missing documentation for any private documentable item.

Why restrict this?

Doc is good. rustc has a MISSING_DOCS allowed-by-default lint for public members, but has no way to enforce documentation of private items. This lint fixes that.

Configuration

  • missing-docs-in-crate-items: Whether to only check for missing documentation in items visible within the current crate. For example, pub(crate) items.

    (default: false)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for imports that do not rename the item as specified in the enforced-import-renames config option.

Note: Even though this lint is warn-by-default, it will only trigger if import renames are defined in the clippy.toml file.

Why is this bad?

Consistency is important; if a project has defined import renames, then they should be followed. More practically, some item names are too vague outside of their defining scope, in which case this can enforce a more meaningful naming.

Example

An example clippy.toml configuration:

enforced-import-renames = [
    { path = "serde_json::Value", rename = "JsonValue" },
]
use serde_json::Value;

Use instead:

use serde_json::Value as JsonValue;

Configuration

  • enforced-import-renames: The list of imports to always rename, a fully qualified path followed by the rename.

    (default: [])

Applicability: MachineApplicable(?)
Added in: 1.55.0

What it does

Checks the doc comments of publicly visible functions that return a Result type and warns if there is no # Errors section.

Why is this bad?

Documenting the type of errors that can be returned from a function can help callers write code to handle the errors appropriately.

Examples

Since the following function returns a Result it has an # Errors section in its doc comment:

/// # Errors
///
/// Will return `Err` if `filename` does not exist or the user does not have
/// permission to read it.
pub fn read(filename: String) -> io::Result<String> {
    unimplemented!();
}

Configuration

  • check-private-items: Whether to also run the listed lints on private items.

    (default: false)

Applicability: Unspecified(?)
Added in: 1.41.0

What it does

Checks for manual core::fmt::Debug implementations that do not use all fields.

Why is this bad?

A common mistake is to forget to update manual Debug implementations when adding a new field to a struct or a new variant to an enum.

At the same time, it also acts as a style lint to suggest using core::fmt::DebugStruct::finish_non_exhaustive for the times when the user intentionally wants to leave out certain fields (e.g. to hide implementation details).

Known problems

This lint works based on the DebugStruct helper types provided by the Formatter, so this won’t detect Debug impls that use the write! macro. Oftentimes there is more logic to a Debug impl if it uses write! macro, so it tries to be on the conservative side and not lint in those cases in an attempt to prevent false positives.

This lint also does not look through function calls, so calling a function does not consider fields used inside of that function as used by the Debug impl.

Lastly, it also ignores tuple structs as their DebugTuple formatter does not have a finish_non_exhaustive method, as well as enums because their exhaustiveness is already checked by the compiler when matching on the enum, making it much less likely to accidentally forget to update the Debug impl when adding a new variant.

Example

use std::fmt;
struct Foo {
    data: String,
    // implementation detail
    hidden_data: i32
}
impl fmt::Debug for Foo {
    fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
        formatter
            .debug_struct("Foo")
            .field("data", &self.data)
            .finish()
    }
}

Use instead:

use std::fmt;
struct Foo {
    data: String,
    // implementation detail
    hidden_data: i32
}
impl fmt::Debug for Foo {
    fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
        formatter
            .debug_struct("Foo")
            .field("data", &self.data)
            .finish_non_exhaustive()
    }
}
Applicability: Unspecified(?)
Added in: 1.70.0

What it does

It lints if an exported function, method, trait method with default impl, or trait method impl is not #[inline].

Why restrict this?

When a function is not marked #[inline], it is not a “small” candidate for automatic inlining, and LTO is not in use, then it is not possible for the function to be inlined into the code of any crate other than the one in which it is defined. Depending on the role of the function and the relationship of the crates, this could significantly reduce performance.

Certain types of crates might intend for most of the methods in their public API to be able to be inlined across crates even when LTO is disabled. This lint allows those crates to require all exported methods to be #[inline] by default, and then opt out for specific methods where this might not make sense.

Example

pub fn foo() {} // missing #[inline]
fn ok() {} // ok
#[inline] pub fn bar() {} // ok
#[inline(always)] pub fn baz() {} // ok

pub trait Bar {
  fn bar(); // ok
  fn def_bar() {} // missing #[inline]
}

struct Baz;
impl Baz {
   fn private() {} // ok
}

impl Bar for Baz {
  fn bar() {} // ok - Baz is not exported
}

pub struct PubBaz;
impl PubBaz {
   fn private() {} // ok
   pub fn not_private() {} // missing #[inline]
}

impl Bar for PubBaz {
   fn bar() {} // missing #[inline]
   fn def_bar() {} // missing #[inline]
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks the doc comments of publicly visible functions that may panic and warns if there is no # Panics section.

Why is this bad?

Documenting the scenarios in which panicking occurs can help callers who do not want to panic to avoid those situations.

Examples

Since the following function may panic it has a # Panics section in its doc comment:

/// # Panics
///
/// Will panic if y is 0
pub fn divide_by(x: i32, y: i32) -> i32 {
    if y == 0 {
        panic!("Cannot divide by 0")
    } else {
        x / y
    }
}

Configuration

  • check-private-items: Whether to also run the listed lints on private items.

    (default: false)

Applicability: Unspecified(?)
Added in: 1.51.0

What it does

Checks for the doc comments of publicly visible unsafe functions and warns if there is no # Safety section.

Why is this bad?

Unsafe functions should document their safety preconditions, so that users can be sure they are using them safely.

Examples

/// This function should really be documented
pub unsafe fn start_apocalypse(u: &mut Universe) {
    unimplemented!();
}

At least write a line about safety:

/// # Safety
///
/// This function should not be called before the horsemen are ready.
pub unsafe fn start_apocalypse(u: &mut Universe) {
    unimplemented!();
}

Configuration

  • check-private-items: Whether to also run the listed lints on private items.

    (default: false)

Applicability: Unspecified(?)
Added in: 1.39.0

What it does

Checks for empty spin loops

Why is this bad?

The loop body should have something like thread::park() or at least std::hint::spin_loop() to avoid needlessly burning cycles and conserve energy. Perhaps even better use an actual lock, if possible.

Known problems

This lint doesn’t currently trigger on while let or loop { match .. { .. } } loops, which would be considered idiomatic in combination with e.g. AtomicBool::compare_exchange_weak.

Example

use core::sync::atomic::{AtomicBool, Ordering};
let b = AtomicBool::new(true);
// give a ref to `b` to another thread,wait for it to become false
while b.load(Ordering::Acquire) {};

Use instead:

while b.load(Ordering::Acquire) {
    std::hint::spin_loop()
}
Applicability: MachineApplicable(?)
Added in: 1.61.0

What it does

Checks if a provided method is used implicitly by a trait implementation.

Why restrict this?

To ensure that a certain implementation implements every method; for example, a wrapper type where every method should delegate to the corresponding method of the inner type’s implementation.

This lint should typically be enabled on a specific trait impl item rather than globally.

Example

trait Trait {
    fn required();

    fn provided() {}
}

#[warn(clippy::missing_trait_methods)]
impl Trait for Type {
    fn required() { /* ... */ }
}

Use instead:

trait Trait {
    fn required();

    fn provided() {}
}

#[warn(clippy::missing_trait_methods)]
impl Trait for Type {
    fn required() { /* ... */ }

    fn provided() { /* ... */ }
}
Applicability: Unspecified(?)
Added in: 1.66.0

What it does

Checks if transmute calls have all generics specified.

Why is this bad?

If not, one or more unexpected types could be used during transmute(), potentially leading to Undefined Behavior or other problems.

This is particularly dangerous in case a seemingly innocent/unrelated change causes type inference to result in a different type. For example, if transmute() is the tail expression of an if-branch, and the else-branch type changes, the compiler may silently infer a different type to be returned by transmute(). That is because the compiler is free to change the inference of a type as long as that inference is technically correct, regardless of the programmer’s unknown expectation.

Both type-parameters, the input- and the output-type, to any transmute() should be given explicitly: Setting the input-type explicitly avoids confusion about what the argument’s type actually is. Setting the output-type explicitly avoids type-inference to infer a technically correct yet unexpected type.

Example

// Avoid "naked" calls to `transmute()`!
let x: i32 = std::mem::transmute([1u16, 2u16]);

// `first_answers` is intended to transmute a slice of bool to a slice of u8.
// But the programmer forgot to index the first element of the outer slice,
// so we are actually transmuting from "pointers to slices" instead of
// transmuting from "a slice of bool", causing a nonsensical result.
let the_answers: &[&[bool]] = &[&[true, false, true]];
let first_answers: &[u8] = std::mem::transmute(the_answers);

Use instead:

let x = std::mem::transmute::<[u16; 2], i32>([1u16, 2u16]);

// The explicit type parameters on `transmute()` makes the intention clear,
// and cause a type-error if the actual types don't match our expectation.
let the_answers: &[&[bool]] = &[&[true, false, true]];
let first_answers: &[u8] = std::mem::transmute::<&[bool], &[u8]>(the_answers[0]);
Applicability: MaybeIncorrect(?)
Added in: 1.79.0

What it does

Warns for mistyped suffix in literals

Why is this bad?

This is most probably a typo

Known problems

  • Does not match on integers too large to fit in the corresponding unsigned type
  • Does not match on _127 since that is a valid grouping for decimal and octal numbers

Example

`2_32` => `2_i32`
`250_8 => `250_u8`
Applicability: MaybeIncorrect(?)
Added in: 1.30.0

What it does

Checks for items that have the same kind of attributes with mixed styles (inner/outer).

Why is this bad?

Having both style of said attributes makes it more complicated to read code.

Known problems

This lint currently has false-negatives when mixing same attributes but they have different path symbols, for example:

#[custom_attribute]
pub fn foo() {
    #![my_crate::custom_attribute]
}

Example

#[cfg(linux)]
pub fn foo() {
    #![cfg(windows)]
}

Use instead:

#[cfg(linux)]
#[cfg(windows)]
pub fn foo() {
}
Applicability: Unspecified(?)
Added in: 1.78.0

What it does

Warns on hexadecimal literals with mixed-case letter digits.

Why is this bad?

It looks confusing.

Example

0x1a9BAcD

Use instead:

0x1A9BACD
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for a read and a write to the same variable where whether the read occurs before or after the write depends on the evaluation order of sub-expressions.

Why restrict this?

While [the evaluation order of sub-expressions] is fully specified in Rust, it still may be confusing to read an expression where the evaluation order affects its behavior.

Known problems

Code which intentionally depends on the evaluation order, or which is correct for any evaluation order.

Example

let mut x = 0;

let a = {
    x = 1;
    1
} + x;
// Unclear whether a is 1 or 2.

Use instead:

let tmp = {
    x = 1;
    1
};
let a = tmp + x;

Past names

  • eval_order_dependence
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks that module layout uses only self named module files; bans mod.rs files.

Why restrict this?

Having multiple module layout styles in a project can be confusing.

Example

src/
  stuff/
    stuff_files.rs
    mod.rs
  lib.rs

Use instead:

src/
  stuff/
    stuff_files.rs
  stuff.rs
  lib.rs
Applicability: Unspecified(?)
Added in: 1.57.0

What it does

Checks for modules that have the same name as their parent module

Why is this bad?

A typical beginner mistake is to have mod foo; and again mod foo { .. } in foo.rs. The expectation is that items inside the inner mod foo { .. } are then available through foo::x, but they are only available through foo::foo::x. If this is done on purpose, it would be better to choose a more representative module name.

Example

// lib.rs
mod foo;
// foo.rs
mod foo {
    ...
}

Configuration

  • allow-private-module-inception: Whether to allow module inception if it’s not public.

    (default: false)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Detects public item names that are prefixed or suffixed by the containing public module’s name.

Why is this bad?

It requires the user to type the module name twice in each usage, especially if they choose to import the module rather than its contents.

Lack of such repetition is also the style used in the Rust standard library; e.g. io::Error and fmt::Error rather than io::IoError and fmt::FmtError; and array::from_ref rather than array::array_from_ref.

Known issues

Glob re-exports are ignored; e.g. this will not warn even though it should:

pub mod foo {
    mod iteration {
        pub struct FooIter {}
    }
    pub use iteration::*; // creates the path `foo::FooIter`
}

Example

mod cake {
    struct BlackForestCake;
}

Use instead:

mod cake {
    struct BlackForest;
}

Past names

  • stutter

Configuration

  • allowed-prefixes: List of prefixes to allow when determining whether an item’s name ends with the module’s name. If the rest of an item’s name is an allowed prefix (e.g. item ToFoo or to_foo in module foo), then don’t emit a warning.

Example

allowed-prefixes = [ "to", "from" ]

Noteworthy

  • By default, the following prefixes are allowed: to, as, into, from, try_into and try_from

  • PascalCase variant is included automatically for each snake_case variant (e.g. if try_into is included, TryInto will also be included)

  • Use ".." as part of the list to indicate that the configured values should be appended to the default configuration of Clippy. By default, any configuration will replace the default value

    (default: ["to", "as", "into", "from", "try_into", "try_from"])

Applicability: Unspecified(?)
Added in: 1.33.0

What it does

Checks for modulo arithmetic.

Why restrict this?

The results of modulo (%) operation might differ depending on the language, when negative numbers are involved. If you interop with different languages it might be beneficial to double check all places that use modulo arithmetic.

For example, in Rust 17 % -3 = 2, but in Python 17 % -3 = -1.

Example

let x = -17 % 3;

Configuration

  • allow-comparison-to-zero: Don’t lint when comparing the result of a modulo operation to zero.

    (default: true)

Applicability: Unspecified(?)
Added in: 1.42.0

What it does

Checks for getting the remainder of integer division by one or minus one.

Why is this bad?

The result for a divisor of one can only ever be zero; for minus one it can cause panic/overflow (if the left operand is the minimal value of the respective integer type) or results in zero. No one will write such code deliberately, unless trying to win an Underhanded Rust Contest. Even for that contest, it’s probably a bad idea. Use something more underhanded.

Example

let a = x % 1;
let a = x % -1;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for nested assignments.

Why is this bad?

While this is in most cases already a type mismatch, the result of an assignment being () can throw off people coming from languages like python or C, where such assignments return a copy of the assigned value.

Example

a = b = 42;

Use instead:

b = 42;
a = b;
Applicability: Unspecified(?)
Added in: 1.65.0

What it does

Check if a generic is defined both in the bound predicate and in the where clause.

Why is this bad?

It can be confusing for developers when seeing bounds for a generic in multiple places.

Example

fn ty<F: std::fmt::Debug>(a: F)
where
    F: Sized,
{}

Use instead:

fn ty<F>(a: F)
where
    F: Sized + std::fmt::Debug,
{}
Applicability: Unspecified(?)
Added in: 1.78.0

What it does

Checks to see if multiple versions of a crate are being used.

Why is this bad?

This bloats the size of targets, and can lead to confusing error messages when structs or traits are used interchangeably between different versions of a crate.

Known problems

Because this can be caused purely by the dependencies themselves, it’s not always possible to fix this issue. In those cases, you can allow that specific crate using the allowed_duplicate_crates configuration option.

Example

[dependencies]
ctrlc = "=3.1.0"
ansi_term = "=0.11.0"

Configuration

  • allowed-duplicate-crates: A list of crate names to allow duplicates of

    (default: [])

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for multiple inherent implementations of a struct

Why restrict this?

Splitting the implementation of a type makes the code harder to navigate.

Example

struct X;
impl X {
    fn one() {}
}
impl X {
    fn other() {}
}

Could be written:

struct X;
impl X {
    fn one() {}
    fn other() {}
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for unsafe blocks that contain more than one unsafe operation.

Why restrict this?

Combined with undocumented_unsafe_blocks, this lint ensures that each unsafe operation must be independently justified. Combined with unused_unsafe, this lint also ensures elimination of unnecessary unsafe blocks through refactoring.

Example

/// Reads a `char` from the given pointer.
///
/// # Safety
///
/// `ptr` must point to four consecutive, initialized bytes which
/// form a valid `char` when interpreted in the native byte order.
fn read_char(ptr: *const u8) -> char {
    // SAFETY: The caller has guaranteed that the value pointed
    // to by `bytes` is a valid `char`.
    unsafe { char::from_u32_unchecked(*ptr.cast::<u32>()) }
}

Use instead:

/// Reads a `char` from the given pointer.
///
/// # Safety
///
/// - `ptr` must be 4-byte aligned, point to four consecutive
///   initialized bytes, and be valid for reads of 4 bytes.
/// - The bytes pointed to by `ptr` must represent a valid
///   `char` when interpreted in the native byte order.
fn read_char(ptr: *const u8) -> char {
    // SAFETY: `ptr` is 4-byte aligned, points to four consecutive
    // initialized bytes, and is valid for reads of 4 bytes.
    let int_value = unsafe { *ptr.cast::<u32>() };

    // SAFETY: The caller has guaranteed that the four bytes
    // pointed to by `bytes` represent a valid `char`.
    unsafe { char::from_u32_unchecked(int_value) }
}
Applicability: Unspecified(?)
Added in: 1.69.0

What it does

Checks for public functions that have no #[must_use] attribute, but return something not already marked must-use, have no mutable arg and mutate no statics.

Why is this bad?

Not bad at all, this lint just shows places where you could add the attribute.

Known problems

The lint only checks the arguments for mutable types without looking if they are actually changed. On the other hand, it also ignores a broad range of potentially interesting side effects, because we cannot decide whether the programmer intends the function to be called for the side effect or the result. Expect many false positives. At least we don’t lint if the result type is unit or already #[must_use].

Examples

// this could be annotated with `#[must_use]`.
pub fn id<T>(t: T) -> T { t }
Applicability: MachineApplicable(?)
Added in: 1.40.0

What it does

Checks for a #[must_use] attribute on unit-returning functions and methods.

Why is this bad?

Unit values are useless. The attribute is likely a remnant of a refactoring that removed the return type.

Examples

#[must_use]
fn useless() { }
Applicability: MachineApplicable(?)
Added in: 1.40.0

What it does

This lint checks for functions that take immutable references and return mutable ones. This will not trigger if no unsafe code exists as there are multiple safe functions which will do this transformation

To be on the conservative side, if there’s at least one mutable reference with the output lifetime, this lint will not trigger.

Why is this bad?

Creating a mutable reference which can be repeatably derived from an immutable reference is unsound as it allows creating multiple live mutable references to the same object.

This error actually lead to an interim Rust release 1.15.1.

Known problems

This pattern is used by memory allocators to allow allocating multiple objects while returning mutable references to each one. So long as different mutable references are returned each time such a function may be safe.

Example

fn foo(&Foo) -> &mut Bar { .. }
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for instances of mut mut references.

Why is this bad?

Multiple muts don’t add anything meaningful to the source. This is either a copy’n’paste error, or it shows a fundamental misunderstanding of references.

Example

let x = &mut &mut y;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for &mut Mutex::lock calls

Why is this bad?

Mutex::lock is less efficient than calling Mutex::get_mut. In addition you also have a statically guarantee that the mutex isn’t locked, instead of just a runtime guarantee.

Example

use std::sync::{Arc, Mutex};

let mut value_rc = Arc::new(Mutex::new(42_u8));
let value_mutex = Arc::get_mut(&mut value_rc).unwrap();

let mut value = value_mutex.lock().unwrap();
*value += 1;

Use instead:

use std::sync::{Arc, Mutex};

let mut value_rc = Arc::new(Mutex::new(42_u8));
let value_mutex = Arc::get_mut(&mut value_rc).unwrap();

let value = value_mutex.get_mut().unwrap();
*value += 1;
Applicability: MaybeIncorrect(?)
Added in: 1.49.0

What it does

Checks for loops with a range bound that is a mutable variable.

Why is this bad?

One might think that modifying the mutable variable changes the loop bounds. It doesn’t.

Known problems

False positive when mutation is followed by a break, but the break is not immediately after the mutation:

let mut x = 5;
for _ in 0..x {
    x += 1; // x is a range bound that is mutated
    ..; // some other expression
    break; // leaves the loop, so mutation is not an issue
}

False positive on nested loops (#6072)

Example

let mut foo = 42;
for i in 0..foo {
    foo -= 1;
    println!("{i}"); // prints numbers from 0 to 41, not 0 to 21
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for sets/maps with mutable key types.

Why is this bad?

All of HashMap, HashSet, BTreeMap and BtreeSet rely on either the hash or the order of keys be unchanging, so having types with interior mutability is a bad idea.

Known problems

False Positives

It’s correct to use a struct that contains interior mutability as a key when its implementation of Hash or Ord doesn’t access any of the interior mutable types. However, this lint is unable to recognize this, so it will often cause false positives in these cases.

False Negatives

This lint does not follow raw pointers (*const T or *mut T) as Hash and Ord apply only to the address of the contained value. This can cause false negatives for custom collections that use raw pointers internally.

Example

use std::cmp::{PartialEq, Eq};
use std::collections::HashSet;
use std::hash::{Hash, Hasher};
use std::sync::atomic::AtomicUsize;

struct Bad(AtomicUsize);
impl PartialEq for Bad {
    fn eq(&self, rhs: &Self) -> bool {
         ..
    }
}

impl Eq for Bad {}

impl Hash for Bad {
    fn hash<H: Hasher>(&self, h: &mut H) {
        ..
    }
}

fn main() {
    let _: HashSet<Bad> = HashSet::new();
}

Configuration

  • ignore-interior-mutability: A list of paths to types that should be treated as if they do not contain interior mutability

    (default: ["bytes::Bytes"])

Applicability: Unspecified(?)
Added in: 1.42.0

What it does

Checks for usage of Mutex<X> where an atomic will do.

Why restrict this?

Using a mutex just to make access to a plain bool or reference sequential is shooting flies with cannons. std::sync::atomic::AtomicBool and std::sync::atomic::AtomicPtr are leaner and faster.

On the other hand, Mutexes are, in general, easier to verify correctness. An atomic does not behave the same as an equivalent mutex. See this issue’s commentary for more details.

Known problems

This lint cannot detect if the mutex is actually used for waiting before a critical section.

Example

let x = Mutex::new(&y);

Use instead:

let x = AtomicBool::new(y);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of Mutex<X> where X is an integral type.

Why is this bad?

Using a mutex just to make access to a plain integer sequential is shooting flies with cannons. std::sync::atomic::AtomicUsize is leaner and faster.

Known problems

This lint cannot detect if the mutex is actually used for waiting before a critical section.

Example

let x = Mutex::new(0usize);

Use instead:

let x = AtomicUsize::new(0usize);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for naive byte counts

Why is this bad?

The bytecount crate has methods to count your bytes faster, especially for large slices.

Known problems

If you have predominantly small slices, the bytecount::count(..) method may actually be slower. However, if you can ensure that less than 2³²-1 matches arise, the naive_count_32(..) can be faster in those cases.

Example

let count = vec.iter().filter(|x| **x == 0u8).count();

Use instead:

let count = bytecount::count(&vec, 0u8);
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

The lint checks for self in fn parameters that specify the Self-type explicitly

Why is this bad?

Increases the amount and decreases the readability of code

Example

enum ValType {
    I32,
    I64,
    F32,
    F64,
}

impl ValType {
    pub fn bytes(self: Self) -> usize {
        match self {
            Self::I32 | Self::F32 => 4,
            Self::I64 | Self::F64 => 8,
        }
    }
}

Could be rewritten as

enum ValType {
    I32,
    I64,
    F32,
    F64,
}

impl ValType {
    pub fn bytes(self) -> usize {
        match self {
            Self::I32 | Self::F32 => 4,
            Self::I64 | Self::F64 => 8,
        }
    }
}
Applicability: MachineApplicable(?)
Added in: 1.47.0

What it does

It detects useless calls to str::as_bytes() before calling len() or is_empty().

Why is this bad?

The len() and is_empty() methods are also directly available on strings, and they return identical results. In particular, len() on a string returns the number of bytes.

Example

let len = "some string".as_bytes().len();
let b = "some string".as_bytes().is_empty();

Use instead:

let len = "some string".len();
let b = "some string".is_empty();
Applicability: MachineApplicable(?)
Added in: 1.84.0

What it does

Checks for usage of bitwise and/or operators between booleans, where performance may be improved by using a lazy and.

Why is this bad?

The bitwise operators do not support short-circuiting, so it may hinder code performance. Additionally, boolean logic “masked” as bitwise logic is not caught by lints like unnecessary_fold

Known problems

This lint evaluates only when the right side is determined to have no side effects. At this time, that determination is quite conservative.

Example

let (x,y) = (true, false);
if x & !y {} // where both x and y are booleans

Use instead:

let (x,y) = (true, false);
if x && !y {}
Applicability: MachineApplicable(?)
Added in: 1.54.0

What it does

Checks for expressions of the form if c { true } else { false } (or vice versa) and suggests using the condition directly.

Why is this bad?

Redundant code.

Known problems

Maybe false positives: Sometimes, the two branches are painstakingly documented (which we, of course, do not detect), so they may have some value. Even then, the documentation can be rewritten to match the shorter code.

Example

if x {
    false
} else {
    true
}

Use instead:

!x
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for expressions of the form if c { x = true } else { x = false } (or vice versa) and suggest assigning the variable directly from the condition.

Why is this bad?

Redundant code.

Example

if must_keep(x, y) {
    skip = false;
} else {
    skip = true;
}

Use instead:

skip = !must_keep(x, y);
Applicability: MachineApplicable(?)
Added in: 1.71.0

What it does

Checks for address of operations (&) that are going to be dereferenced immediately by the compiler.

Why is this bad?

Suggests that the receiver of the expression borrows the expression.

Known problems

The lint cannot tell when the implementation of a trait for &T and T do different things. Removing a borrow in such a case can change the semantics of the code.

Example

fn fun(_a: &i32) {}

let x: &i32 = &&&&&&5;
fun(&x);

Use instead:

let x: &i32 = &5;
fun(x);

Past names

  • ref_in_deref

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for bindings that needlessly destructure a reference and borrow the inner value with &ref.

Why is this bad?

This pattern has no effect in almost all cases.

Example

let mut v = Vec::<String>::new();
v.iter_mut().filter(|&ref a| a.is_empty());

if let &[ref first, ref second] = v.as_slice() {}

Use instead:

let mut v = Vec::<String>::new();
v.iter_mut().filter(|a| a.is_empty());

if let [first, second] = v.as_slice() {}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for borrow operations (&) that are used as a generic argument to a function when the borrowed value could be used.

Why is this bad?

Suggests that the receiver of the expression borrows the expression.

Known problems

The lint cannot tell when the implementation of a trait for &T and T do different things. Removing a borrow in such a case can change the semantics of the code.

Example

fn f(_: impl AsRef<str>) {}

let x = "foo";
f(&x);

Use instead:

fn f(_: impl AsRef<str>) {}

let x = "foo";
f(x);
Applicability: MachineApplicable(?)
Added in: 1.74.0

What it does

Checks if an iterator is used to check if a string is ascii.

Why is this bad?

The str type already implements the is_ascii method.

Example

"foo".chars().all(|c| c.is_ascii());

Use instead:

"foo".is_ascii();
Applicability: MachineApplicable(?)
Added in: 1.81.0

What it does

Checks for functions collecting an iterator when collect is not needed.

Why is this bad?

collect causes the allocation of a new data structure, when this allocation may not be needed.

Example

let len = iterator.collect::<Vec<_>>().len();

Use instead:

let len = iterator.count();
Applicability: MachineApplicable(?)
Added in: 1.30.0

What it does

The lint checks for if-statements appearing in loops that contain a continue statement in either their main blocks or their else-blocks, when omitting the else-block possibly with some rearrangement of code can make the code easier to understand.

Why is this bad?

Having explicit else blocks for if statements containing continue in their THEN branch adds unnecessary branching and nesting to the code. Having an else block containing just continue can also be better written by grouping the statements following the whole if statement within the THEN block and omitting the else block completely.

Example

while condition() {
    update_condition();
    if x {
        // ...
    } else {
        continue;
    }
    println!("Hello, world");
}

Could be rewritten as

while condition() {
    update_condition();
    if x {
        // ...
        println!("Hello, world");
    }
}

As another example, the following code

loop {
    if waiting() {
        continue;
    } else {
        // Do something useful
    }
    # break;
}

Could be rewritten as

loop {
    if waiting() {
        continue;
    }
    // Do something useful
    # break;
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for fn main() { .. } in doctests

Why is this bad?

The test can be shorter (and likely more readable) if the fn main() is left implicit.

Examples

/// An example of a doctest with a `main()` function
///
/// # Examples
///
/// ```
/// fn main() {
///     // this needs not be in an `fn`
/// }
/// ```
fn needless_main() {
    unimplemented!();
}
Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Checks for empty else branches.

Why is this bad?

An empty else branch does nothing and can be removed.

Example

if check() {
    println!("Check successful!");
} else {
}

Use instead:

if check() {
    println!("Check successful!");
}
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Checks for usage of for_each that would be more simply written as a for loop.

Why is this bad?

for_each may be used after applying iterator transformers like filter for better readability and performance. It may also be used to fit a simple operation on one line. But when none of these apply, a simple for loop is more idiomatic.

Example

let v = vec![0, 1, 2];
v.iter().for_each(|elem| {
    println!("{elem}");
})

Use instead:

let v = vec![0, 1, 2];
for elem in &v {
    println!("{elem}");
}

Known Problems

When doing things such as:

let v = vec![0, 1, 2];
v.iter().for_each(|elem| unsafe {
    libc::printf(c"%d\n".as_ptr(), elem);
});

This lint will not trigger.

Applicability: MachineApplicable(?)
Added in: 1.53.0

What it does

Checks for empty if branches with no else branch.

Why is this bad?

It can be entirely omitted, and often the condition too.

Known issues

This will usually only suggest to remove the if statement, not the condition. Other lints such as no_effect will take care of removing the condition if it’s unnecessary.

Example

if really_expensive_condition(&i) {}
if really_expensive_condition_with_side_effects(&mut i) {}

Use instead:

// <omitted>
really_expensive_condition_with_side_effects(&mut i);
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Checks for late initializations that can be replaced by a let statement with an initializer.

Why is this bad?

Assigning in the let statement is less repetitive.

Example

let a;
a = 1;

let b;
match 3 {
    0 => b = "zero",
    1 => b = "one",
    _ => b = "many",
}

let c;
if true {
    c = 1;
} else {
    c = -1;
}

Use instead:

let a = 1;

let b = match 3 {
    0 => "zero",
    1 => "one",
    _ => "many",
};

let c = if true {
    1
} else {
    -1
};
Applicability: MachineApplicable(?)
Added in: 1.59.0

What it does

Checks for lifetime annotations which can be removed by relying on lifetime elision.

Why is this bad?

The additional lifetimes make the code look more complicated, while there is nothing out of the ordinary going on. Removing them leads to more readable code.

Known problems

  • We bail out if the function has a where clause where lifetimes are mentioned due to potential false positives.

Example

// Unnecessary lifetime annotations
fn in_and_out<'a>(x: &'a u8, y: u8) -> &'a u8 {
    x
}

Use instead:

fn elided(x: &u8, y: u8) -> &u8 {
    x
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for unnecessary match or match-like if let returns for Option and Result when function signatures are the same.

Why is this bad?

This match block does nothing and might not be what the coder intended.

Example

fn foo() -> Result<(), i32> {
    match result {
        Ok(val) => Ok(val),
        Err(err) => Err(err),
    }
}

fn bar() -> Option<i32> {
    if let Some(val) = option {
        Some(val)
    } else {
        None
    }
}

Could be replaced as

fn foo() -> Result<(), i32> {
    result
}

fn bar() -> Option<i32> {
    option
}
Applicability: MachineApplicable(?)
Added in: 1.61.0

What it does

Lints ?Sized bounds applied to type parameters that cannot be unsized

Why is this bad?

The ?Sized bound is misleading because it cannot be satisfied by an unsized type

Example

// `T` cannot be unsized because `Clone` requires it to be `Sized`
fn f<T: Clone + ?Sized>(t: &T) {}

Use instead:

fn f<T: Clone>(t: &T) {}

// or choose alternative bounds for `T` so that it can be unsized
Applicability: MaybeIncorrect(?)
Added in: 1.81.0

What it does

Checks for no-op uses of Option::{as_deref, as_deref_mut}, for example, Option<&T>::as_deref() returns the same type.

Why is this bad?

Redundant code and improving readability.

Example

let a = Some(&1);
let b = a.as_deref(); // goes from Option<&i32> to Option<&i32>

Use instead:

let a = Some(&1);
let b = a;
Applicability: MachineApplicable(?)
Added in: 1.57.0

What it does

Checks for calling take function after as_ref.

Why is this bad?

Redundant code. take writes None to its argument. In this case the modification is useless as it’s a temporary that cannot be read from afterwards.

Example

let x = Some(3);
x.as_ref().take();

Use instead:

let x = Some(3);
x.as_ref();
Applicability: MachineApplicable(?)
Added in: 1.62.0

What it does

The lint checks for parenthesis on literals in range statements that are superfluous.

Why is this bad?

Having superfluous parenthesis makes the code less readable overhead when reading.

Example

for i in (0)..10 {
  println!("{i}");
}

Use instead:

for i in 0..10 {
  println!("{i}");
}
Applicability: MachineApplicable(?)
Added in: 1.63.0

What it does

Check if a &mut function argument is actually used mutably.

Be careful if the function is publicly reexported as it would break compatibility with users of this function, when the users pass this function as an argument.

Why is this bad?

Less mut means less fights with the borrow checker. It can also lead to more opportunities for parallelization.

Example

fn foo(y: &mut i32) -> i32 {
    12 + *y
}

Use instead:

fn foo(y: &i32) -> i32 {
    12 + *y
}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: 1.73.0

What it does

Checks for functions taking arguments by value, but not consuming them in its body.

Why is this bad?

Taking arguments by reference is more flexible and can sometimes avoid unnecessary allocations.

Known problems

  • This lint suggests taking an argument by reference, however sometimes it is better to let users decide the argument type (by using Borrow trait, for example), depending on how the function is used.

Example

fn foo(v: Vec<i32>) {
    assert_eq!(v.len(), 42);
}

should be

fn foo(v: &[i32]) {
    assert_eq!(v.len(), 42);
}
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for usage of pub(self) and pub(in self).

Why is this bad?

It’s unnecessary, omitting the pub entirely will give the same results.

Example

pub(self) type OptBox<T> = Option<Box<T>>;

Use instead:

type OptBox<T> = Option<Box<T>>;
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Suggests alternatives for useless applications of ? in terminating expressions

Why is this bad?

There’s no reason to use ? to short-circuit when execution of the body will end there anyway.

Example

struct TO {
    magic: Option<usize>,
}

fn f(to: TO) -> Option<usize> {
    Some(to.magic?)
}

struct TR {
    magic: Result<usize, bool>,
}

fn g(tr: Result<TR, bool>) -> Result<usize, bool> {
    tr.and_then(|t| Ok(t.magic?))
}

Use instead:

struct TO {
    magic: Option<usize>,
}

fn f(to: TO) -> Option<usize> {
   to.magic
}

struct TR {
    magic: Result<usize, bool>,
}

fn g(tr: Result<TR, bool>) -> Result<usize, bool> {
    tr.and_then(|t| t.magic)
}
Applicability: MachineApplicable(?)
Added in: 1.51.0

What it does

Checks for looping over the range of 0..len of some collection just to get the values by index.

Why is this bad?

Just iterating the collection itself makes the intent more clear and is probably faster because it eliminates the bounds check that is done when indexing.

Example

let vec = vec!['a', 'b', 'c'];
for i in 0..vec.len() {
    println!("{}", vec[i]);
}

Use instead:

let vec = vec!['a', 'b', 'c'];
for i in vec {
    println!("{}", i);
}
Applicability: HasPlaceholders(?)
Added in: pre 1.29.0

What it does

Checks for raw string literals with an unnecessary amount of hashes around them.

Why is this bad?

It’s just unnecessary, and makes it look like there’s more escaping needed than is actually necessary.

Example

let r = r###"Hello, "world"!"###;

Use instead:

let r = r#"Hello, "world"!"#;
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Checks for raw string literals where a string literal can be used instead.

Why restrict this?

For consistent style by using simpler string literals whenever possible.

However, there are many cases where using a raw string literal is more idiomatic than a string literal, so it’s opt-in.

Example

let r = r"Hello, world!";

Use instead:

let r = "Hello, world!";
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Checks for return statements at the end of a block.

Why is this bad?

Removing the return and semicolon will make the code more rusty.

Example

fn foo(x: usize) -> usize {
    return x;
}

simplify to

fn foo(x: usize) -> usize {
    x
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for return statements on Err paired with the ? operator.

Why is this bad?

The return is unnecessary.

Returns may be used to add attributes to the return expression. Return statements with attributes are therefore be accepted by this lint.

Example

fn foo(x: usize) -> Result<(), Box<dyn Error>> {
    if x == 0 {
        return Err(...)?;
    }
    Ok(())
}

simplify to

fn foo(x: usize) -> Result<(), Box<dyn Error>> {
    if x == 0 {
        Err(...)?;
    }
    Ok(())
}

if paired with try_err, use instead:

fn foo(x: usize) -> Result<(), Box<dyn Error>> {
    if x == 0 {
        return Err(...);
    }
    Ok(())
}
Applicability: MachineApplicable(?)
Added in: 1.73.0

What it does

Checks for usage of str::splitn (or str::rsplitn) where using str::split would be the same.

Why is this bad?

The function split is simpler and there is no performance difference in these cases, considering that both functions return a lazy iterator.

Example

let str = "key=value=add";
let _ = str.splitn(3, '=').next().unwrap();

Use instead:

let str = "key=value=add";
let _ = str.split('=').next().unwrap();
Applicability: MachineApplicable(?)
Added in: 1.59.0

What it does

Checks for needlessly including a base struct on update when all fields are changed anyway.

This lint is not applied to structs marked with non_exhaustive.

Why is this bad?

This will cost resources (because the base has to be somewhere), and make the code less readable.

Example

Point {
    x: 1,
    y: 1,
    z: 1,
    ..zero_point
};

Use instead:

// Missing field `z`
Point {
    x: 1,
    y: 1,
    ..zero_point
};
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for the usage of negated comparison operators on types which only implement PartialOrd (e.g., f64).

Why is this bad?

These operators make it easy to forget that the underlying types actually allow not only three potential Orderings (Less, Equal, Greater) but also a fourth one (Uncomparable). This is especially easy to miss if the operator based comparison result is negated.

Example

let a = 1.0;
let b = f64::NAN;

let not_less_or_equal = !(a <= b);

Use instead:

use std::cmp::Ordering;

let _not_less_or_equal = match a.partial_cmp(&b) {
    None | Some(Ordering::Greater) => true,
    _ => false,
};
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for multiplication by -1 as a form of negation.

Why is this bad?

It’s more readable to just negate.

Known problems

This only catches integers (for now).

Example

let a = x * -1;

Use instead:

let a = -x;
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for negative feature names with prefix no- or not-

Why is this bad?

Features are supposed to be additive, and negatively-named features violate it.

Example

[features]
default = []
no-abc = []
not-def = []

Use instead:

[features]
default = ["abc", "def"]
abc = []
def = []

Applicability: Unspecified(?)
Added in: 1.57.0

What it does

Checks for loops that will always break, return or continue an outer loop.

Why is this bad?

This loop never loops, all it does is obfuscating the code.

Example

loop {
    ..;
    break;
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for new not returning a type that contains Self.

Why is this bad?

As a convention, new methods are used to make a new instance of a type.

Example

In an impl block:

impl Foo {
    fn new() -> NotAFoo {
    }
}
struct Bar(Foo);
impl Foo {
    // Bad. The type name must contain `Self`
    fn new() -> Bar {
    }
}
impl Foo {
    // Good. Return type contains `Self`
    fn new() -> Result<Foo, FooError> {
    }
}

Or in a trait definition:

pub trait Trait {
    // Bad. The type name must contain `Self`
    fn new();
}
pub trait Trait {
    // Good. Return type contains `Self`
    fn new() -> Self;
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for public types with a pub fn new() -> Self method and no implementation of Default.

Why is this bad?

The user might expect to be able to use Default as the type can be constructed without arguments.

Example

pub struct Foo(Bar);

impl Foo {
    pub fn new() -> Self {
        Foo(Bar::new())
    }
}

To fix the lint, add a Default implementation that delegates to new:

pub struct Foo(Bar);

impl Default for Foo {
    fn default() -> Self {
        Foo::new()
    }
}

Past names

  • new_without_default_derive
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for statements which have no effect.

Why is this bad?

Unlike dead code, these statements are actually executed. However, as they have no effect, all they do is make the code less readable.

Example

0;
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for replace statements which have no effect.

Why is this bad?

It’s either a mistake or confusing.

Example

"1234".replace("12", "12");
"1234".replacen("12", "12", 1);
Applicability: Unspecified(?)
Added in: 1.63.0

What it does

Checks for binding to underscore prefixed variable without side-effects.

Why is this bad?

Unlike dead code, these bindings are actually executed. However, as they have no effect and shouldn’t be used further on, all they do is make the code less readable.

Example

let _i_serve_no_purpose = 1;
Applicability: Unspecified(?)
Added in: 1.58.0

What it does

Checks for Rust ABI functions with the #[no_mangle] attribute.

Why is this bad?

The Rust ABI is not stable, but in many simple cases matches enough with the C ABI that it is possible to forget to add extern "C" to a function called from C. Changes to the Rust ABI can break this at any point.

Example

 #[no_mangle]
 fn example(arg_one: u32, arg_two: usize) {}

Use instead:

 #[no_mangle]
 extern "C" fn example(arg_one: u32, arg_two: usize) {}
Applicability: MaybeIncorrect(?)
Added in: 1.69.0

What it does

Checks for non-ASCII characters in string and char literals.

Why restrict this?

Yeah, we know, the 90’s called and wanted their charset back. Even so, there still are editors and other programs out there that don’t work well with Unicode. So if the code is meant to be used internationally, on multiple operating systems, or has other portability requirements, activating this lint could be useful.

Example

let x = String::from("€");

Use instead:

let x = String::from("\u{20ac}");
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for non-canonical implementations of Clone when Copy is already implemented.

Why is this bad?

If both Clone and Copy are implemented, they must agree. This can done by dereferencing self in Clone’s implementation, which will avoid any possibility of the implementations becoming out of sync.

Example

#[derive(Eq, PartialEq)]
struct A(u32);

impl Clone for A {
    fn clone(&self) -> Self {
        Self(self.0)
    }
}

impl Copy for A {}

Use instead:

#[derive(Eq, PartialEq)]
struct A(u32);

impl Clone for A {
    fn clone(&self) -> Self {
        *self
    }
}

impl Copy for A {}

Past names

  • incorrect_clone_impl_on_copy_type
Applicability: MaybeIncorrect(?)
Added in: 1.72.0

What it does

Checks for non-canonical implementations of PartialOrd when Ord is already implemented.

Why is this bad?

If both PartialOrd and Ord are implemented, they must agree. This is commonly done by wrapping the result of cmp in Some for partial_cmp. Not doing this may silently introduce an error upon refactoring.

Known issues

Code that calls the .into() method instead will be flagged, despite .into() wrapping it in Some.

Example

#[derive(Eq, PartialEq)]
struct A(u32);

impl Ord for A {
    fn cmp(&self, other: &Self) -> Ordering {
        // ...
    }
}

impl PartialOrd for A {
    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
        // ...
    }
}

Use instead:

#[derive(Eq, PartialEq)]
struct A(u32);

impl Ord for A {
    fn cmp(&self, other: &Self) -> Ordering {
        // ...
    }
}

impl PartialOrd for A {
    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
        Some(self.cmp(other))
    }
}

Past names

  • incorrect_partial_ord_impl_on_ord_type
Applicability: Unspecified(?)
Added in: 1.73.0

What it does

Checks for any and all combinators in cfg with only one condition.

Why is this bad?

If there is only one condition, no need to wrap it into any or all combinators.

Example

#[cfg(any(unix))]
pub struct Bar;

Use instead:

#[cfg(unix)]
pub struct Bar;
Applicability: MaybeIncorrect(?)
Added in: 1.71.0

What it does

Checks for non-octal values used to set Unix file permissions.

Why is this bad?

They will be converted into octal, creating potentially unintended file permissions.

Example

use std::fs::OpenOptions;
use std::os::unix::fs::OpenOptionsExt;

let mut options = OpenOptions::new();
options.mode(644);

Use instead:

use std::fs::OpenOptions;
use std::os::unix::fs::OpenOptionsExt;

let mut options = OpenOptions::new();
options.mode(0o644);
Applicability: MachineApplicable(?)
Added in: 1.53.0

What it does

This lint warns about a Send implementation for a type that contains fields that are not safe to be sent across threads. It tries to detect fields that can cause a soundness issue when sent to another thread (e.g., Rc) while allowing !Send fields that are expected to exist in a Send type, such as raw pointers.

Why is this bad?

Sending the struct to another thread effectively sends all of its fields, and the fields that do not implement Send can lead to soundness bugs such as data races when accessed in a thread that is different from the thread that created it.

See:

Known Problems

This lint relies on heuristics to distinguish types that are actually unsafe to be sent across threads and !Send types that are expected to exist in Send type. Its rule can filter out basic cases such as Vec<*const T>, but it’s not perfect. Feel free to create an issue if you have a suggestion on how this heuristic can be improved.

Example

struct ExampleStruct<T> {
    rc_is_not_send: Rc<String>,
    unbounded_generic_field: T,
}

// This impl is unsound because it allows sending `!Send` types through `ExampleStruct`
unsafe impl<T> Send for ExampleStruct<T> {}

Use thread-safe types like std::sync::Arc or specify correct bounds on generic type parameters (T: Send).

Configuration

  • enable-raw-pointer-heuristic-for-send: Whether to apply the raw pointer heuristic to determine if a type is Send.

    (default: true)

Applicability: Unspecified(?)
Added in: 1.57.0

What it does

Checks for conversions from NonZero types to regular integer types, and suggests using NonZero types for the target as well.

Why is this bad?

Converting from NonZero types to regular integer types and then back to NonZero types is less efficient and loses the type-safety guarantees provided by NonZero types. Using NonZero types consistently can lead to more optimized code and prevent certain classes of errors related to zero values.

Example

use std::num::{NonZeroU32, NonZeroU64};

fn example(x: u64, y: NonZeroU32) {
    // Bad: Converting NonZeroU32 to u64 unnecessarily
    let r1 = x / u64::from(y.get());
    let r2 = x % u64::from(y.get());
}

Use instead:

use std::num::{NonZeroU32, NonZeroU64};

fn example(x: u64, y: NonZeroU32) {
    // Good: Preserving the NonZero property
    let r1 = x / NonZeroU64::from(y);
    let r2 = x % NonZeroU64::from(y);
}
Applicability: MachineApplicable(?)
Added in: 1.81.0

What it does

Checks for boolean expressions that can be written more concisely.

Why is this bad?

Readability of boolean expressions suffers from unnecessary duplication.

Known problems

Ignores short circuiting behavior of || and &&. Ignores |, & and ^.

Example

if a && true {}
if !(a == b) {}

Use instead:

if a {}
if a != b {}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for duplicate open options as well as combinations that make no sense.

Why is this bad?

In the best case, the code will be harder to read than necessary. I don’t know the worst case.

Example

use std::fs::OpenOptions;

OpenOptions::new().read(true).truncate(true);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks that common macros are used with consistent bracing.

Why is this bad?

This is mostly a consistency lint although using () or [] doesn’t give you a semicolon in item position, which can be unexpected.

Example

vec!{1, 2, 3};

Use instead:

vec![1, 2, 3];

Configuration

  • standard-macro-braces: Enforce the named macros always use the braces specified.

A MacroMatcher can be added like so { name = "macro_name", brace = "(" }. If the macro could be used with a full path two MacroMatchers have to be added one with the full path crate_name::macro_name and one with just the macro name.

(default: [])

Applicability: MachineApplicable(?)
Added in: 1.55.0

What it does

Checks for public functions that dereference raw pointer arguments but are not marked unsafe.

Why is this bad?

The function should almost definitely be marked unsafe, since for an arbitrary raw pointer, there is no way of telling for sure if it is valid.

In general, this lint should never be disabled unless it is definitely a false positive (please submit an issue if so) since it breaks Rust’s soundness guarantees, directly exposing API users to potentially dangerous program behavior. This is also true for internal APIs, as it is easy to leak unsoundness.

Context

In Rust, an unsafe {...} block is used to indicate that the code in that section has been verified in some way that the compiler can not. For a function that accepts a raw pointer then accesses the pointer’s data, this is generally impossible as the incoming pointer could point anywhere, valid or not. So, the signature should be marked unsafe fn: this indicates that the function’s caller must provide some verification that the arguments it sends are valid (and then call the function within an unsafe block).

Known problems

  • It does not check functions recursively so if the pointer is passed to a private non-unsafe function which does the dereferencing, the lint won’t trigger (false negative).
  • It only checks for arguments whose type are raw pointers, not raw pointers got from an argument in some other way (fn foo(bar: &[*const u8]) or some_argument.get_raw_ptr()) (false negative).

Example

pub fn foo(x: *const u8) {
    println!("{}", unsafe { *x });
}

// this call "looks" safe but will segfault or worse!
// foo(invalid_ptr);

Use instead:

pub unsafe fn foo(x: *const u8) {
    println!("{}", unsafe { *x });
}

// this would cause a compiler error for calling without `unsafe`
// foo(invalid_ptr);

// sound call if the caller knows the pointer is valid
unsafe { foo(valid_ptr); }
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of .then_some(..).unwrap_or(..)

Why is this bad?

This can be written more clearly with if .. else ..

Limitations

This lint currently only looks for usages of .then_some(..).unwrap_or(..), but will be expanded to account for similar patterns.

Example

let x = true;
x.then_some("a").unwrap_or("b");

Use instead:

let x = true;
if x { "a" } else { "b" };
Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

Checks for \0 escapes in string and byte literals that look like octal character escapes in C.

Why is this bad?

C and other languages support octal character escapes in strings, where a backslash is followed by up to three octal digits. For example, \033 stands for the ASCII character 27 (ESC). Rust does not support this notation, but has the escape code \0 which stands for a null byte/character, and any following digits do not form part of the escape sequence. Therefore, \033 is not a compiler error but the result may be surprising.

Known problems

The actual meaning can be the intended one. \x00 can be used in these cases to be unambiguous.

The lint does not trigger for format strings in print!(), write!() and friends since the string is already preprocessed when Clippy lints can see it.

Example

let one = "\033[1m Bold? \033[0m";  // \033 intended as escape
let two = "\033\0";                 // \033 intended as null-3-3

Use instead:

let one = "\x1b[1mWill this be bold?\x1b[0m";
let two = "\x0033\x00";
Applicability: MaybeIncorrect(?)
Added in: 1.59.0

What it does

Checks for usage of ok().expect(..).

Why is this bad?

Because you usually call expect() on the Result directly to get a better error message.

Known problems

The error type needs to implement Debug

Example

x.ok().expect("why did I do this again?");

Use instead:

x.expect("why did I do this again?");
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for arguments that are only used in recursion with no side-effects.

Why is this bad?

It could contain a useless calculation and can make function simpler.

The arguments can be involved in calculations and assignments but as long as the calculations have no side-effects (function calls or mutating dereference) and the assigned variables are also only in recursion, it is useless.

Known problems

Too many code paths in the linting code are currently untested and prone to produce false positives or are prone to have performance implications.

In some cases, this would not catch all useless arguments.

fn foo(a: usize, b: usize) -> usize {
    let f = |x| x + 1;

    if a == 0 {
        1
    } else {
        foo(a - 1, f(b))
    }
}

For example, the argument b is only used in recursion, but the lint would not catch it.

List of some examples that can not be caught:

  • binary operation of non-primitive types
  • closure usage
  • some break relative operations
  • struct pattern binding

Also, when you recurse the function name with path segments, it is not possible to detect.

Example

fn f(a: usize, b: usize) -> usize {
    if a == 0 {
        1
    } else {
        f(a - 1, b + 1)
    }
}

Use instead:

fn f(a: usize) -> usize {
    if a == 0 {
        1
    } else {
        f(a - 1)
    }
}
Applicability: MaybeIncorrect(?)
Added in: 1.61.0

What it does

Checks for arguments to == which have their address taken to satisfy a bound and suggests to dereference the other argument instead

Why is this bad?

It is more idiomatic to dereference the other argument.

Example

&x == y

Use instead:

x == *y
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of .as_ref().cloned() and .as_mut().cloned() on Options

Why is this bad?

This can be written more concisely by cloning the Option directly.

Example

fn foo(bar: &Option<Vec<u8>>) -> Option<Vec<u8>> {
    bar.as_ref().cloned()
}

Use instead:

fn foo(bar: &Option<Vec<u8>>) -> Option<Vec<u8>> {
    bar.clone()
}
Applicability: MachineApplicable(?)
Added in: 1.77.0

What it does

Checks for usage of _.as_ref().map(Deref::deref) or its aliases (such as String::as_str).

Why is this bad?

Readability, this can be written more concisely as _.as_deref().

Example

opt.as_ref().map(String::as_str)

Can be written as

opt.as_deref()

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.42.0

What it does

Checks for usage of option_env!(...).unwrap() and suggests usage of the env! macro.

Why is this bad?

Unwrapping the result of option_env! will panic at run-time if the environment variable doesn’t exist, whereas env! catches it at compile-time.

Example

let _ = option_env!("HOME").unwrap();

Is better expressed as:

let _ = env!("HOME");
Applicability: Unspecified(?)
Added in: 1.43.0

What it does

Checks for iterators of Options using .filter(Option::is_some).map(Option::unwrap) that may be replaced with a .flatten() call.

Why is this bad?

Option is like a collection of 0-1 things, so flatten automatically does this without suspicious-looking unwrap calls.

Example

let _ = std::iter::empty::<Option<i32>>().filter(Option::is_some).map(Option::unwrap);

Use instead:

let _ = std::iter::empty::<Option<i32>>().flatten();
Applicability: MachineApplicable(?)
Added in: 1.53.0

What it does

Lints usage of if let Some(v) = ... { y } else { x } and match .. { Some(v) => y, None/_ => x } which are more idiomatically done with Option::map_or (if the else bit is a pure expression) or Option::map_or_else (if the else bit is an impure expression).

Why is this bad?

Using the dedicated functions of the Option type is clearer and more concise than an if let expression.

Notes

This lint uses a deliberately conservative metric for checking if the inside of either body contains loop control expressions break or continue (which cannot be used within closures). If these are found, this lint will not be raised.

Example

let _ = if let Some(foo) = optional {
    foo
} else {
    5
};
let _ = match optional {
    Some(val) => val + 1,
    None => 5
};
let _ = if let Some(foo) = optional {
    foo
} else {
    let y = do_complicated_function();
    y*y
};

should be

let _ = optional.map_or(5, |foo| foo);
let _ = optional.map_or(5, |val| val + 1);
let _ = optional.map_or_else(||{
    let y = do_complicated_function();
    y*y
}, |foo| foo);
Applicability: MaybeIncorrect(?)
Added in: 1.47.0

What it does

Checks for usage of _.map_or(Err(_), Ok).

Why is this bad?

Readability, this can be written more concisely as _.ok_or(_).

Example

opt.map_or(Err("error"), Ok);

Use instead:

opt.ok_or("error");
Applicability: MachineApplicable(?)
Added in: 1.76.0

What it does

Checks for usage of _.map_or(None, _).

Why is this bad?

Readability, this can be written more concisely as _.and_then(_).

Known problems

The order of the arguments is not in execution order.

Example

opt.map_or(None, |a| Some(a + 1));

Use instead:

opt.and_then(|a| Some(a + 1));
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of option.map(f) where f is a function or closure that returns the unit type ().

Why is this bad?

Readability, this can be written more clearly with an if let statement

Example

let x: Option<String> = do_stuff();
x.map(log_err_msg);
x.map(|msg| log_err_msg(format_msg(msg)));

The correct use would be:

let x: Option<String> = do_stuff();
if let Some(msg) = x {
    log_err_msg(msg);
}

if let Some(msg) = x {
    log_err_msg(format_msg(msg));
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of Option<Option<_>> in function signatures and type definitions

Why is this bad?

Option<_> represents an optional value. Option<Option<_>> represents an optional value which itself wraps an optional. This is logically the same thing as an optional value but has an unneeded extra level of wrapping.

If you have a case where Some(Some(_)), Some(None) and None are distinct cases, consider a custom enum instead, with clear names for each case.

Example

fn get_data() -> Option<Option<u32>> {
    None
}

Better:

pub enum Contents {
    Data(Vec<u8>), // Was Some(Some(Vec<u8>))
    NotYetFetched, // Was Some(None)
    None,          // Was None
}

fn get_data() -> Contents {
    Contents::None
}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for calls to .or(foo(..)), .unwrap_or(foo(..)), .or_insert(foo(..)) etc., and suggests to use .or_else(|| foo(..)), .unwrap_or_else(|| foo(..)), .unwrap_or_default() or .or_default() etc. instead.

Why is this bad?

The function will always be called. This is only bad if it allocates or does some non-trivial amount of work.

Known problems

If the function has side-effects, not calling it will change the semantic of the program, but you shouldn’t rely on that.

The lint also cannot figure out whether the function you call is actually expensive to call or not.

Example

foo.unwrap_or(String::from("empty"));

Use instead:

foo.unwrap_or_else(|| String::from("empty"));
Applicability: HasPlaceholders(?)
Added in: pre 1.29.0

What it does

Checks for .or(…).unwrap() calls to Options and Results.

Why is this bad?

You should use .unwrap_or(…) instead for clarity.

Example

// Result
let value = result.or::<Error>(Ok(fallback)).unwrap();

// Option
let value = option.or(Some(fallback)).unwrap();

Use instead:

// Result
let value = result.unwrap_or(fallback);

// Option
let value = option.unwrap_or(fallback);
Applicability: MachineApplicable(?)
Added in: 1.61.0

What it does

Checks for out of bounds array indexing with a constant index.

Why is this bad?

This will always panic at runtime.

Example

let x = [1, 2, 3, 4];

x[9];
&x[2..9];

Use instead:

// Index within bounds

x[0];
x[3];
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for boolean expressions that contain terminals that can be eliminated.

Why is this bad?

This is most likely a logic bug.

Known problems

Ignores short circuiting behavior.

Example

// The `b` is unnecessary, the expression is equivalent to `if a`.
if a && b || a { ... }

Use instead:

if a {}

Past names

  • logic_bug
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of panic!.

Why restrict this?

This macro, or panics in general, may be unwanted in production code.

Example

panic!("even with a good reason");

Configuration

  • allow-panic-in-tests: Whether panic should be allowed in test functions or #[cfg(test)]

    (default: false)

Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Checks for usage of panic! or assertions in a function whose return type is Result.

Why restrict this?

For some codebases, it is desirable for functions of type result to return an error instead of crashing. Hence panicking macros should be avoided.

Known problems

Functions called from a function returning a Result may invoke a panicking macro. This is not checked.

Example

fn result_with_panic() -> Result<bool, String>
{
    panic!("error");
}

Use instead:

fn result_without_panic() -> Result<bool, String> {
    Err(String::from("error"))
}
Applicability: Unspecified(?)
Added in: 1.48.0

What it does

Detects C-style underflow/overflow checks.

Why is this bad?

These checks will, by default, panic in debug builds rather than check whether the operation caused an overflow.

Example

if a + b < a {
    // handle overflow
}

Use instead:

if a.checked_add(b).is_none() {
    // handle overflow
}

Or:

if a.overflowing_add(b).1 {
    // handle overflow
}

Past names

  • overflow_check_conditional
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for calls of unwrap[_err]() that will always fail.

Why is this bad?

If panicking is desired, an explicit panic!() should be used.

Known problems

This lint only checks if conditions not assignments. So something like let x: Option<()> = None; x.unwrap(); will not be recognized.

Example

if option.is_none() {
    do_something_with(option.unwrap())
}

This code will always panic. The if condition should probably be inverted.

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks whether some but not all fields of a struct are public.

Either make all fields of a type public, or make none of them public

Why restrict this?

Most types should either be:

  • Abstract data types: complex objects with opaque implementation which guard interior invariants and expose intentionally limited API to the outside world.
  • Data: relatively simple objects which group a bunch of related attributes together, but have no invariants.

Example

pub struct Color {
    pub r: u8,
    pub g: u8,
    b: u8,
}

Use instead:

pub struct Color {
    pub r: u8,
    pub g: u8,
    pub b: u8,
}
Applicability: Unspecified(?)
Added in: 1.66.0

What it does

Checks for manual re-implementations of PartialEq::ne.

Why is this bad?

PartialEq::ne is required to always return the negated result of PartialEq::eq, which is exactly what the default implementation does. Therefore, there should never be any need to re-implement it.

Example

struct Foo;

impl PartialEq for Foo {
   fn eq(&self, other: &Foo) -> bool { true }
   fn ne(&self, other: &Foo) -> bool { !(self == other) }
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for binary comparisons to a literal Option::None.

Why is this bad?

A programmer checking if some foo is None via a comparison foo == None is usually inspired from other programming languages (e.g. foo is None in Python). Checking if a value of type Option<T> is (not) equal to None in that way relies on T: PartialEq to do the comparison, which is unneeded.

Example

fn foo(f: Option<u32>) -> &'static str {
    if f != None { "yay" } else { "nay" }
}

Use instead:

fn foo(f: Option<u32>) -> &'static str {
    if f.is_some() { "yay" } else { "nay" }
}
Applicability: MachineApplicable(?)
Added in: 1.65.0

What it does

  • Checks for push calls on PathBuf that can cause overwrites.

Why is this bad?

Calling push with a root path at the start can overwrite the previous defined path.

Example

use std::path::PathBuf;

let mut x = PathBuf::from("/foo");
x.push("/bar");
assert_eq!(x, PathBuf::from("/bar"));

Could be written:

use std::path::PathBuf;

let mut x = PathBuf::from("/foo");
x.push("bar");
assert_eq!(x, PathBuf::from("/foo/bar"));
Applicability: MachineApplicable(?)
Added in: 1.36.0

What it does

Looks for calls to Path::ends_with calls where the argument looks like a file extension.

By default, Clippy has a short list of known filenames that start with a dot but aren’t necessarily file extensions (e.g. the .git folder), which are allowed by default. The allowed-dotfiles configuration can be used to allow additional file extensions that Clippy should not lint.

Why is this bad?

This doesn’t actually compare file extensions. Rather, ends_with compares the given argument to the last component of the path and checks if it matches exactly.

Known issues

File extensions are often at most three characters long, so this only lints in those cases in an attempt to avoid false positives. Any extension names longer than that are assumed to likely be real path components and are therefore ignored.

Example

fn is_markdown(path: &Path) -> bool {
    path.ends_with(".md")
}

Use instead:

fn is_markdown(path: &Path) -> bool {
    path.extension().is_some_and(|ext| ext == "md")
}

Configuration

  • allowed-dotfiles: Additional dotfiles (files or directories starting with a dot) to allow

    (default: [])

Applicability: MaybeIncorrect(?)
Added in: 1.74.0

What it does

Checks for calls to push immediately after creating a new PathBuf.

Why is this bad?

Multiple .join() calls are usually easier to read than multiple .push calls across multiple statements. It might also be possible to use PathBuf::from instead.

Known problems

.join() introduces an implicit clone(). PathBuf::from can alternatively be used when the PathBuf is newly constructed. This will avoid the implicit clone.

Example

let mut path_buf = PathBuf::new();
path_buf.push("foo");

Use instead:

let path_buf = PathBuf::from("foo");
// or
let path_buf = PathBuf::new().join("foo");
Applicability: HasPlaceholders(?)
Added in: 1.82.0

What it does

Checks for patterns that aren’t exact representations of the types they are applied to.

To satisfy this lint, you will have to adjust either the expression that is matched against or the pattern itself, as well as the bindings that are introduced by the adjusted patterns. For matching you will have to either dereference the expression with the * operator, or amend the patterns to explicitly match against &<pattern> or &mut <pattern> depending on the reference mutability. For the bindings you need to use the inverse. You can leave them as plain bindings if you wish for the value to be copied, but you must use ref mut <variable> or ref <variable> to construct a reference into the matched structure.

If you are looking for a way to learn about ownership semantics in more detail, it is recommended to look at IDE options available to you to highlight types, lifetimes and reference semantics in your code. The available tooling would expose these things in a general way even outside of the various pattern matching mechanics. Of course this lint can still be used to highlight areas of interest and ensure a good understanding of ownership semantics.

Why restrict this?

It increases ownership hints in the code, and will guard against some changes in ownership.

Example

This example shows the basic adjustments necessary to satisfy the lint. Note how the matched expression is explicitly dereferenced with * and the inner variable is bound to a shared borrow via ref inner.

// Bad
let value = &Some(Box::new(23));
match value {
    Some(inner) => println!("{}", inner),
    None => println!("none"),
}

// Good
let value = &Some(Box::new(23));
match *value {
    Some(ref inner) => println!("{}", inner),
    None => println!("none"),
}

The following example demonstrates one of the advantages of the more verbose style. Note how the second version uses ref mut a to explicitly declare a a shared mutable borrow, while b is simply taken by value. This ensures that the loop body cannot accidentally modify the wrong part of the structure.

// Bad
let mut values = vec![(2, 3), (3, 4)];
for (a, b) in &mut values {
    *a += *b;
}

// Good
let mut values = vec![(2, 3), (3, 4)];
for &mut (ref mut a, b) in &mut values {
    *a += b;
}
Applicability: Unspecified(?)
Added in: 1.47.0

What it does

Checks for calls to std::fs::Permissions.set_readonly with argument false.

Why is this bad?

On Unix platforms this results in the file being world writable, equivalent to chmod a+w <file>.

Example

use std::fs::File;
let f = File::create("foo.txt").unwrap();
let metadata = f.metadata().unwrap();
let mut permissions = metadata.permissions();
permissions.set_readonly(false);
Applicability: Unspecified(?)
Added in: 1.68.0

What it does

Checks if any pointer is being passed to an asm! block with nomem option.

Why is this bad?

nomem forbids any reads or writes to memory and passing a pointer suggests that either of those will happen.

Example

fn f(p: *mut u32) {
    unsafe { core::arch::asm!("mov [{p}], 42", p = in(reg) p, options(nomem, nostack)); }
}

Use instead:

fn f(p: *mut u32) {
    unsafe { core::arch::asm!("mov [{p}], 42", p = in(reg) p, options(nostack)); }
}
Applicability: Unspecified(?)
Added in: 1.81.0

What it does

Checks for possible missing comma in an array. It lints if an array element is a binary operator expression and it lies on two lines.

Why is this bad?

This could lead to unexpected results.

Example

let a = &[
    -1, -2, -3 // <= no comma here
    -4, -5, -6
];
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for operations where precedence may be unclear and suggests to add parentheses. Currently it catches the following:

  • mixed usage of arithmetic and bit shifting/combining operators without parentheses

Why is this bad?

Not everyone knows the precedence of those operators by heart, so expressions like these may trip others trying to reason about the code.

Example

  • 1 << 2 + 3 equals 32, while (1 << 2) + 3 equals 7
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

This lint warns when you use println!("") to print a newline.

Why is this bad?

You should use println!(), which is simpler.

Example

println!("");

Use instead:

println!();
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

This lint checks for function arguments of type &String, &Vec, &PathBuf, and Cow<_>. It will also suggest you replace .clone() calls with the appropriate .to_owned()/to_string() calls.

Why is this bad?

Requiring the argument to be of the specific type makes the function less useful for no benefit; slices in the form of &[T] or &str usually suffice and can be obtained from other types, too.

Known problems

There may be fn(&Vec)-typed references pointing to your function. If you have them, you will get a compiler error after applying this lint’s suggestions. You then have the choice to undo your changes or change the type of the reference.

Note that if the function is part of your public interface, there may be other crates referencing it, of which you may not be aware. Carefully deprecate the function before applying the lint suggestions in this case.

Example

fn foo(&Vec<u32>) { .. }

Use instead:

fn foo(&[u32]) { .. }
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for as casts between raw pointers that don’t change their constness, namely *const T to *const U and *mut T to *mut U.

Why is this bad?

Though as casts between raw pointers are not terrible, pointer::cast is safer because it cannot accidentally change the pointer’s mutability, nor cast the pointer to other types like usize.

Example

let ptr: *const u32 = &42_u32;
let mut_ptr: *mut u32 = &mut 42_u32;
let _ = ptr as *const i32;
let _ = mut_ptr as *mut i32;

Use instead:

let ptr: *const u32 = &42_u32;
let mut_ptr: *mut u32 = &mut 42_u32;
let _ = ptr.cast::<i32>();
let _ = mut_ptr.cast::<i32>();

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.51.0

What it does

Checks for as casts between raw pointers that change their constness, namely *const T to *mut T and *mut T to *const T.

Why is this bad?

Though as casts between raw pointers are not terrible, pointer::cast_mut and pointer::cast_const are safer because they cannot accidentally cast the pointer to another type. Or, when null pointers are involved, null() and null_mut() can be used directly.

Example

let ptr: *const u32 = &42_u32;
let mut_ptr = ptr as *mut u32;
let ptr = mut_ptr as *const u32;
let ptr1 = std::ptr::null::<u32>() as *mut u32;
let ptr2 = std::ptr::null_mut::<u32>() as *const u32;
let ptr3 = std::ptr::null::<u32>().cast_mut();
let ptr4 = std::ptr::null_mut::<u32>().cast_const();

Use instead:

let ptr: *const u32 = &42_u32;
let mut_ptr = ptr.cast_mut();
let ptr = mut_ptr.cast_const();
let ptr1 = std::ptr::null_mut::<u32>();
let ptr2 = std::ptr::null::<u32>();
let ptr3 = std::ptr::null_mut::<u32>();
let ptr4 = std::ptr::null::<u32>();
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Use std::ptr::eq when applicable

Why is this bad?

ptr::eq can be used to compare &T references (which coerce to *const T implicitly) by their address rather than comparing the values they point to.

Example

let a = &[1, 2, 3];
let b = &[1, 2, 3];

assert!(a as *const _ as usize == b as *const _ as usize);

Use instead:

let a = &[1, 2, 3];
let b = &[1, 2, 3];

assert!(std::ptr::eq(a, b));
Applicability: MachineApplicable(?)
Added in: 1.49.0

What it does

Checks for usage of the offset pointer method with a usize casted to an isize.

Why is this bad?

If we’re always increasing the pointer address, we can avoid the numeric cast by using the add method instead.

Example

let vec = vec![b'a', b'b', b'c'];
let ptr = vec.as_ptr();
let offset = 1_usize;

unsafe {
    ptr.offset(offset as isize);
}

Could be written:

let vec = vec![b'a', b'b', b'c'];
let ptr = vec.as_ptr();
let offset = 1_usize;

unsafe {
    ptr.add(offset);
}
Applicability: MachineApplicable(?)
Added in: 1.30.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

clippy::enum_variant_names now covers this case via the avoid-breaking-exported-api config.

Applicability: Unspecified(?)
Deprecated in: 1.54.0

What it does

Checks whether any field of the struct is prefixed with an _ (underscore) and also marked pub (public)

Why is this bad?

Fields prefixed with an _ are inferred as unused, which suggests it should not be marked as pub, because marking it as pub infers it will be used.

Example

struct FileHandle {
    pub _descriptor: usize,
}

Use instead:

struct FileHandle {
    _descriptor: usize,
}

OR

struct FileHandle {
    pub descriptor: usize,
}

Configuration

  • pub-underscore-fields-behavior: Lint “public” fields in a struct that are prefixed with an underscore based on their exported visibility, or whether they are marked as “pub”.

    (default: "PubliclyExported")

Applicability: Unspecified(?)
Added in: 1.77.0

What it does

Restricts the usage of pub use ...

Why restrict this?

A project may wish to limit pub use instances to prevent unintentional exports, or to encourage placing exported items directly in public modules.

Example

pub mod outer {
    mod inner {
        pub struct Test {}
    }
    pub use inner::Test;
}

use outer::Test;

Use instead:

pub mod outer {
    pub struct Test {}
}

use outer::Test;
Applicability: Unspecified(?)
Added in: 1.62.0

What it does

Checks for usage of pub(<loc>) with in.

Why restrict this?

Consistency. Use it or don’t, just be consistent about it.

Also see the pub_without_shorthand lint for an alternative.

Example

pub(super) type OptBox<T> = Option<Box<T>>;

Use instead:

pub(in super) type OptBox<T> = Option<Box<T>>;
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Checks for usage of pub(<loc>) without in.

Note: As you cannot write a module’s path in pub(<loc>), this will only trigger on pub(super) and the like.

Why restrict this?

Consistency. Use it or don’t, just be consistent about it.

Also see the pub_with_shorthand lint for an alternative.

Example

pub(in super) type OptBox<T> = Option<Box<T>>;

Use instead:

pub(super) type OptBox<T> = Option<Box<T>>;
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Checks for expressions that could be replaced by the question mark operator.

Why is this bad?

Question mark usage is more idiomatic.

Example

if option.is_none() {
    return None;
}

Could be written:

option?;
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for expressions that use the question mark operator and rejects them.

Why restrict this?

Sometimes code wants to avoid the question mark operator because for instance a local block requires a macro to re-throw errors to attach additional information to the error.

Example

let result = expr?;

Could be written:

utility_macro!(expr);
Applicability: Unspecified(?)
Added in: 1.69.0

What it does

Checks for inclusive ranges where 1 is subtracted from the upper bound, e.g., x..=(y-1).

Why is this bad?

The code is more readable with an exclusive range like x..y.

Known problems

This will cause a warning that cannot be fixed if the consumer of the range only accepts a specific range type, instead of the generic RangeBounds trait (#3307).

Example

for i in x..=(y-1) {
    // ..
}

Use instead:

for i in x..y {
    // ..
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for exclusive ranges where 1 is added to the upper bound, e.g., x..(y+1).

Why is this bad?

The code is more readable with an inclusive range like x..=y.

Known problems

Will add unnecessary pair of parentheses when the expression is not wrapped in a pair but starts with an opening parenthesis and ends with a closing one. I.e., let _ = (f()+1)..(f()+1) results in let _ = ((f()+1)..=f()).

Also in many cases, inclusive ranges are still slower to run than exclusive ranges, because they essentially add an extra branch that LLVM may fail to hoist out of the loop.

This will cause a warning that cannot be fixed if the consumer of the range only accepts a specific range type, instead of the generic RangeBounds trait (#3307).

Example

for i in x..(y+1) {
    // ..
}

Use instead:

for i in x..=y {
    // ..
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

Iterator::step_by(0) now panics and is no longer an infinite iterator.

Applicability: Unspecified(?)
Deprecated in: pre 1.29.0

What it does

Checks for zipping a collection with the range of 0.._.len().

Why is this bad?

The code is better expressed with .enumerate().

Example

let _ = x.iter().zip(0..x.len());

Use instead:

let _ = x.iter().enumerate();
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for Rc<T> and Arc<T> when T is a mutable buffer type such as String or Vec.

Why restrict this?

Expressions such as Rc<String> usually have no advantage over Rc<str>, since it is larger and involves an extra level of indirection, and doesn’t implement Borrow<str>.

While mutating a buffer type would still be possible with Rc::get_mut(), it only works if there are no additional references yet, which usually defeats the purpose of enclosing it in a shared ownership type. Instead, additionally wrapping the inner type with an interior mutable container (such as RefCell or Mutex) would normally be used.

Known problems

This pattern can be desirable to avoid the overhead of a RefCell or Mutex for cases where mutation only happens before there are any additional references.

Example

fn foo(interned: Rc<String>) { ... }

Better:

fn foo(interned: Rc<str>) { ... }

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: 1.48.0

What it does

Checks for reference-counted pointers (Arc, Rc, rc::Weak, and sync::Weak) in vec![elem; len]

Why is this bad?

This will create elem once and clone it len times - doing so with Arc/Rc/Weak is a bit misleading, as it will create references to the same pointer, rather than different instances.

Example

let v = vec![std::sync::Arc::new("some data".to_string()); 100];
// or
let v = vec![std::rc::Rc::new("some data".to_string()); 100];

Use instead:

// Initialize each value separately:
let mut data = Vec::with_capacity(100);
for _ in 0..100 {
    data.push(std::rc::Rc::new("some data".to_string()));
}

// Or if you want clones of the same reference,
// Create the reference beforehand to clarify that
// it should be cloned for each value
let data = std::rc::Rc::new("some data".to_string());
let v = vec![data; 100];
Applicability: HasPlaceholders(?)
Added in: 1.63.0

What it does

Checks for Rc<Mutex<T>>.

Why restrict this?

Rc is used in single thread and Mutex is used in multi thread. Consider using Rc<RefCell<T>> in single thread or Arc<Mutex<T>> in multi thread.

Known problems

Sometimes combining generic types can lead to the requirement that a type use Rc in conjunction with Mutex. We must consider those cases false positives, but alas they are quite hard to rule out. Luckily they are also rare.

Example

use std::rc::Rc;
use std::sync::Mutex;
fn foo(interned: Rc<Mutex<i32>>) { ... }

Better:

use std::rc::Rc;
use std::cell::RefCell
fn foo(interned: Rc<RefCell<i32>>) { ... }

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: 1.55.0

What it does

Looks for calls to [Stdin::read_line] to read a line from the standard input into a string, then later attempting to use that string for an operation that will never work for strings with a trailing newline character in it (e.g. parsing into a i32).

Why is this bad?

The operation will always fail at runtime no matter what the user enters, thus making it a useless operation.

Example

let mut input = String::new();
std::io::stdin().read_line(&mut input).expect("Failed to read a line");
let num: i32 = input.parse().expect("Not a number!");
assert_eq!(num, 42); // we never even get here!

Use instead:

let mut input = String::new();
std::io::stdin().read_line(&mut input).expect("Failed to read a line");
let num: i32 = input.trim_end().parse().expect("Not a number!");
//                  ^^^^^^^^^^^ remove the trailing newline
assert_eq!(num, 42);
Applicability: MachineApplicable(?)
Added in: 1.73.0

What it does

This lint catches reads into a zero-length Vec. Especially in the case of a call to with_capacity, this lint warns that read gets the number of bytes from the Vec’s length, not its capacity.

Why is this bad?

Reading zero bytes is almost certainly not the intended behavior.

Known problems

In theory, a very unusual read implementation could assign some semantic meaning to zero-byte reads. But it seems exceptionally unlikely that code intending to do a zero-byte read would allocate a Vec for it.

Example

use std::io;
fn foo<F: io::Read>(mut f: F) {
    let mut data = Vec::with_capacity(100);
    f.read(&mut data).unwrap();
}

Use instead:

use std::io;
fn foo<F: io::Read>(mut f: F) {
    let mut data = Vec::with_capacity(100);
    data.resize(100, 0);
    f.read(&mut data).unwrap();
}
Applicability: MaybeIncorrect(?)
Added in: 1.63.0

What it does

Looks for calls to RwLock::write where the lock is only used for reading.

Why is this bad?

The write portion of RwLock is exclusive, meaning that no other thread can access the lock while this writer is active.

Example

use std::sync::RwLock;
fn assert_is_zero(lock: &RwLock<i32>) {
    let num = lock.write().unwrap();
    assert_eq!(*num, 0);
}

Use instead:

use std::sync::RwLock;
fn assert_is_zero(lock: &RwLock<i32>) {
    let num = lock.read().unwrap();
    assert_eq!(*num, 0);
}
Applicability: MaybeIncorrect(?)
Added in: 1.73.0

What it does

Checks for format trait implementations (e.g. Display) with a recursive call to itself which uses self as a parameter. This is typically done indirectly with the write! macro or with to_string().

Why is this bad?

This will lead to infinite recursion and a stack overflow.

Example

use std::fmt;

struct Structure(i32);
impl fmt::Display for Structure {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        write!(f, "{}", self.to_string())
    }
}

Use instead:

use std::fmt;

struct Structure(i32);
impl fmt::Display for Structure {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        write!(f, "{}", self.0)
    }
}

Past names

  • to_string_in_display
Applicability: Unspecified(?)
Added in: 1.48.0

What it does

Checks for usage of redundant allocations anywhere in the code.

Why is this bad?

Expressions such as Rc<&T>, Rc<Rc<T>>, Rc<Arc<T>>, Rc<Box<T>>, Arc<&T>, Arc<Rc<T>>, Arc<Arc<T>>, Arc<Box<T>>, Box<&T>, Box<Rc<T>>, Box<Arc<T>>, Box<Box<T>>, add an unnecessary level of indirection.

Example

fn foo(bar: Rc<&usize>) {}

Better:

fn foo(bar: &usize) {}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: MaybeIncorrect(?)
Added in: 1.44.0

What it does

Checks for usage of as_str() on a String chained with a method available on the String itself.

Why is this bad?

The as_str() conversion is pointless and can be removed for simplicity and cleanliness.

Example

let owned_string = "This is a string".to_owned();
owned_string.as_str().as_bytes()

Use instead:

let owned_string = "This is a string".to_owned();
owned_string.as_bytes()
Applicability: MachineApplicable(?)
Added in: 1.74.0

What it does

Checks for async block that only returns await on a future.

Why is this bad?

It is simpler and more efficient to use the future directly.

Example

let f = async {
   1 + 2
};
let fut = async {
    f.await
};

Use instead:

let f = async {
   1 + 2
};
let fut = f;
Applicability: MachineApplicable(?)
Added in: 1.70.0

What it does

Checks for [all @ ..] patterns.

Why is this bad?

In all cases, all works fine and can often make code simpler, as you possibly won’t need to convert from say a Vec to a slice by dereferencing.

Example

if let [all @ ..] = &*v {
    // NOTE: Type is a slice here
    println!("all elements: {all:#?}");
}

Use instead:

if let all = v {
    // NOTE: Type is a `Vec` here
    println!("all elements: {all:#?}");
}
// or
println!("all elements: {v:#?}");
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Checks for a redundant clone() (and its relatives) which clones an owned value that is going to be dropped without further use.

Why is this bad?

It is not always possible for the compiler to eliminate useless allocations and deallocations generated by redundant clone()s.

Known problems

False-negatives: analysis performed by this lint is conservative and limited.

Example

{
    let x = Foo::new();
    call(x.clone());
    call(x.clone()); // this can just pass `x`
}

["lorem", "ipsum"].join(" ").to_string();

Path::new("/a/b").join("c").to_path_buf();
Applicability: MachineApplicable(?)
Added in: 1.32.0

What it does

Checks for closures which just call another function where the function can be called directly. unsafe functions, calls where types get adjusted or where the callee is marked #[track_caller] are ignored.

Why is this bad?

Needlessly creating a closure adds code for no benefit and gives the optimizer more work.

Known problems

If creating the closure inside the closure has a side- effect then moving the closure creation out will change when that side- effect runs. See #1439 for more details.

Example

xs.map(|x| foo(x))

Use instead:

// where `foo(_)` is a plain function that takes the exact argument type of `x`.
xs.map(foo)
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Detects closures called in the same expression where they are defined.

Why is this bad?

It is unnecessarily adding to the expression’s complexity.

Example

let a = (|| 42)();

Use instead:

let a = 42;
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for closures which only invoke a method on the closure argument and can be replaced by referencing the method directly.

Why is this bad?

It’s unnecessary to create the closure.

Example

Some('a').map(|s| s.to_uppercase());

may be rewritten as

Some('a').map(char::to_uppercase);
Applicability: MachineApplicable(?)
Added in: 1.35.0

What it does

Checks for ineffective double comparisons against constants.

Why is this bad?

Only one of the comparisons has any effect on the result, the programmer probably intended to flip one of the comparison operators, or compare a different value entirely.

Example

if status_code <= 400 && status_code < 500 {}
Applicability: Unspecified(?)
Added in: 1.73.0

What it does

Checks for else blocks that can be removed without changing semantics.

Why is this bad?

The else block adds unnecessary indentation and verbosity.

Known problems

Some may prefer to keep the else block for clarity.

Example

fn my_func(count: u32) {
    if count == 0 {
        print!("Nothing to do");
        return;
    } else {
        print!("Moving on...");
    }
}

Use instead:

fn my_func(count: u32) {
    if count == 0 {
        print!("Nothing to do");
        return;
    }
    print!("Moving on...");
}
Applicability: Unspecified(?)
Added in: 1.50.0

What it does

Checks for feature names with prefix use-, with- or suffix -support

Why is this bad?

These prefixes and suffixes have no significant meaning.

Example

[features]
default = ["use-abc", "with-def", "ghi-support"]
use-abc = []  // redundant
with-def = []   // redundant
ghi-support = []   // redundant

Use instead:

[features]
default = ["abc", "def", "ghi"]
abc = []
def = []
ghi = []
Applicability: Unspecified(?)
Added in: 1.57.0

What it does

Checks for fields in struct literals where shorthands could be used.

Why is this bad?

If the field and variable names are the same, the field name is redundant.

Example

let bar: u8 = 123;

struct Foo {
    bar: u8,
}

let foo = Foo { bar: bar };

the last line can be simplified to

let foo = Foo { bar };

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for unnecessary guards in match expressions.

Why is this bad?

It’s more complex and much less readable. Making it part of the pattern can improve exhaustiveness checking as well.

Example

match x {
    Some(x) if matches!(x, Some(1)) => ..,
    Some(x) if x == Some(2) => ..,
    _ => todo!(),
}

Use instead:

match x {
    Some(Some(1)) => ..,
    Some(Some(2)) => ..,
    _ => todo!(),
}
Applicability: MaybeIncorrect(?)
Added in: 1.73.0

What it does

Checks for redundant redefinitions of local bindings.

Why is this bad?

Redundant redefinitions of local bindings do not change behavior and are likely to be unintended.

Note that although these bindings do not affect your code’s meaning, they may affect rustc’s stack allocation.

Example

let a = 0;
let a = a;

fn foo(b: i32) {
   let b = b;
}

Use instead:

let a = 0;
// no redefinition with the same name

fn foo(b: i32) {
  // no redefinition with the same name
}
Applicability: Unspecified(?)
Added in: 1.73.0

What it does

Checks for patterns in the form name @ _.

Why is this bad?

It’s almost always more readable to just use direct bindings.

Example

match v {
    Some(x) => (),
    y @ _ => (),
}

Use instead:

match v {
    Some(x) => (),
    y => (),
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Lint for redundant pattern matching over Result, Option, std::task::Poll, std::net::IpAddr or bools

Why is this bad?

It’s more concise and clear to just use the proper utility function or using the condition directly

Known problems

For suggestions involving bindings in patterns, this will change the drop order for the matched type. Both if let and while let will drop the value at the end of the block, both if and while will drop the value before entering the block. For most types this change will not matter, but for a few types this will not be an acceptable change (e.g. locks). See the reference for more about drop order.

Example

if let Ok(_) = Ok::<i32, i32>(42) {}
if let Err(_) = Err::<i32, i32>(42) {}
if let None = None::<()> {}
if let Some(_) = Some(42) {}
if let Poll::Pending = Poll::Pending::<()> {}
if let Poll::Ready(_) = Poll::Ready(42) {}
if let IpAddr::V4(_) = IpAddr::V4(Ipv4Addr::LOCALHOST) {}
if let IpAddr::V6(_) = IpAddr::V6(Ipv6Addr::LOCALHOST) {}
match Ok::<i32, i32>(42) {
    Ok(_) => true,
    Err(_) => false,
};

let cond = true;
if let true = cond {}
matches!(cond, true);

The more idiomatic use would be:

if Ok::<i32, i32>(42).is_ok() {}
if Err::<i32, i32>(42).is_err() {}
if None::<()>.is_none() {}
if Some(42).is_some() {}
if Poll::Pending::<()>.is_pending() {}
if Poll::Ready(42).is_ready() {}
if IpAddr::V4(Ipv4Addr::LOCALHOST).is_ipv4() {}
if IpAddr::V6(Ipv6Addr::LOCALHOST).is_ipv6() {}
Ok::<i32, i32>(42).is_ok();

let cond = true;
if cond {}
cond;

Past names

  • if_let_redundant_pattern_matching
Applicability: MachineApplicable(?)
Added in: 1.31.0

What it does

Checks for items declared pub(crate) that are not crate visible because they are inside a private module.

Why is this bad?

Writing pub(crate) is misleading when it’s redundant due to the parent module’s visibility.

Example

mod internal {
    pub(crate) fn internal_fn() { }
}

This function is not visible outside the module and it can be declared with pub or private visibility

mod internal {
    pub fn internal_fn() { }
}
Applicability: MachineApplicable(?)
Added in: 1.44.0

What it does

Checks for redundant slicing expressions which use the full range, and do not change the type.

Why is this bad?

It unnecessarily adds complexity to the expression.

Known problems

If the type being sliced has an implementation of Index<RangeFull> that actually changes anything then it can’t be removed. However, this would be surprising to people reading the code and should have a note with it.

Example

fn get_slice(x: &[u32]) -> &[u32] {
    &x[..]
}

Use instead:

fn get_slice(x: &[u32]) -> &[u32] {
    x
}
Applicability: MachineApplicable(?)
Added in: 1.51.0

What it does

Checks for constants and statics with an explicit 'static lifetime.

Why is this bad?

Adding 'static to every reference can create very complicated types.

Example

const FOO: &'static [(&'static str, &'static str, fn(&Bar) -> bool)] =
&[...]
static FOO: &'static [(&'static str, &'static str, fn(&Bar) -> bool)] =
&[...]

This code can be rewritten as

 const FOO: &[(&str, &str, fn(&Bar) -> bool)] = &[...]
 static FOO: &[(&str, &str, fn(&Bar) -> bool)] = &[...]

Past names

  • const_static_lifetime

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.37.0

What it does

Warns about needless / redundant type annotations.

Why restrict this?

Code without type annotations is shorter and in most cases more idiomatic and easier to modify.

Limitations

This lint doesn’t support:

  • Generics
  • Refs returned from anything else than a MethodCall
  • Complex types (tuples, arrays, etc…)
  • Path to anything else than a primitive type.

Example

let foo: String = String::new();

Use instead:

let foo = String::new();
Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Checks for casts of references to pointer using as and suggests std::ptr::from_ref and std::ptr::from_mut instead.

Why is this bad?

Using as casts may result in silently changing mutability or type.

Example

let a_ref = &1;
let a_ptr = a_ref as *const _;

Use instead:

let a_ref = &1;
let a_ptr = std::ptr::from_ref(a_ref);
Applicability: MachineApplicable(?)
Added in: 1.78.0

What it does

Checks for ref bindings which create a reference to a reference.

Why is this bad?

The address-of operator at the use site is clearer about the need for a reference.

Example

let x = Some("");
if let Some(ref x) = x {
    // use `x` here
}

Use instead:

let x = Some("");
if let Some(x) = x {
    // use `&x` here
}
Applicability: MachineApplicable(?)
Added in: 1.54.0

What it does

Warns when a function signature uses &Option<T> instead of Option<&T>.

Why is this bad?

More flexibility, better memory optimization, and more idiomatic Rust code.

&Option<T> in a function signature breaks encapsulation because the caller must own T and move it into an Option to call with it. When returned, the owner must internally store it as Option<T> in order to return it. At a lower level, &Option<T> points to memory with the presence bit flag plus the T value, whereas Option<&T> is usually optimized to a single pointer, so it may be more optimal.

See this YouTube video by Logan Smith for an in-depth explanation of why this is important.

Known problems

This lint recommends changing the function signatures, but it cannot automatically change the function calls or the function implementations.

Example

// caller uses  foo(&opt)
fn foo(a: &Option<String>) {}
fn bar(&self) -> &Option<String> { &None }

Use instead:

// caller should use  `foo1(opt.as_ref())`
fn foo1(a: Option<&String>) {}
// better yet, use string slice  `foo2(opt.as_deref())`
fn foo2(a: Option<&str>) {}
fn bar(&self) -> Option<&String> { None }

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: 1.83.0

What it does

Checks for usage of &Option<&T>.

Why is this bad?

Since & is Copy, it’s useless to have a reference on Option<&T>.

Known problems

It may be irrelevant to use this lint on public API code as it will make a breaking change to apply it.

Example

let x: &Option<&u32> = &Some(&0u32);

Use instead:

let x: Option<&u32> = Some(&0u32);
Applicability: MaybeIncorrect(?)
Added in: 1.49.0

What it does

Checks for usages of the ref keyword.

Why restrict this?

The ref keyword can be confusing for people unfamiliar with it, and often it is more concise to use & instead.

Example

let opt = Some(5);
if let Some(ref foo) = opt {}

Use instead:

let opt = Some(5);
if let Some(foo) = &opt {}
Applicability: Unspecified(?)
Added in: 1.71.0

What it does

Checks for regex compilation inside a loop with a literal.

Why is this bad?

Compiling a regex is a much more expensive operation than using one, and a compiled regex can be used multiple times. This is documented as an antipattern on the regex documentation

Example

for haystack in haystacks {
    let regex = regex::Regex::new(MY_REGEX).unwrap();
    if regex.is_match(haystack) {
        // Perform operation
    }
}

can be replaced with

let regex = regex::Regex::new(MY_REGEX).unwrap();
for haystack in haystacks {
    if regex.is_match(haystack) {
        // Perform operation
    }
}
Applicability: Unspecified(?)
Added in: 1.83.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

The regex! macro was removed from the regex crate in 2018.

Applicability: Unspecified(?)
Deprecated in: 1.47.0

What it does

Lints when the name of function parameters from trait impl is different than its default implementation.

Why restrict this?

Using the default name for parameters of a trait method is more consistent.

Example

struct A(u32);

impl PartialEq for A {
    fn eq(&self, b: &Self) -> bool {
        self.0 == b.0
    }
}

Use instead:

struct A(u32);

impl PartialEq for A {
    fn eq(&self, other: &Self) -> bool {
        self.0 == other.0
    }
}

Configuration

  • allow-renamed-params-for: List of trait paths to ignore when checking renamed function parameters.

Example

allow-renamed-params-for = [ "std::convert::From" ]

Noteworthy

  • By default, the following traits are ignored: From, TryFrom, FromStr

  • ".." can be used as part of the list to indicate that the configured values should be appended to the default configuration of Clippy. By default, any configuration will replace the default value.

    (default: ["core::convert::From", "core::convert::TryFrom", "core::str::FromStr"])

Applicability: Unspecified(?)
Added in: 1.80.0

What it does

Checks for usage of .repeat(1) and suggest the following method for each types.

  • .to_string() for str
  • .clone() for String
  • .to_vec() for slice

The lint will evaluate constant expressions and values as arguments of .repeat(..) and emit a message if they are equivalent to 1. (Related discussion in rust-clippy#7306)

Why is this bad?

For example, String.repeat(1) is equivalent to .clone(). If cloning the string is the intention behind this, clone() should be used.

Example

fn main() {
    let x = String::from("hello world").repeat(1);
}

Use instead:

fn main() {
    let x = String::from("hello world").clone();
}
Applicability: MachineApplicable(?)
Added in: 1.47.0

What it does

Looks for patterns such as vec![Vec::with_capacity(x); n] or iter::repeat(Vec::with_capacity(x)).

Why is this bad?

These constructs work by cloning the element, but cloning a Vec<_> does not respect the old vector’s capacity and effectively discards it.

This makes iter::repeat(Vec::with_capacity(x)) especially suspicious because the user most certainly expected that the yielded Vec<_> will have the requested capacity, otherwise one can simply write iter::repeat(Vec::new()) instead and it will have the same effect.

Similarly for vec![x; n], the element x is cloned to fill the vec. Unlike iter::repeat however, the vec repeat macro does not have to clone the value n times but just n - 1 times, because it can reuse the passed value for the last slot. That means that the last Vec<_> gets the requested capacity but all other ones do not.

Example


let _: Vec<Vec<u8>> = vec![Vec::with_capacity(42); 123];
let _: Vec<Vec<u8>> = iter::repeat(Vec::with_capacity(42)).take(123).collect();

Use instead:


let _: Vec<Vec<u8>> = iter::repeat_with(|| Vec::with_capacity(42)).take(123).collect();
//                                      ^^^ this closure executes 123 times
//                                          and the vecs will have the expected capacity
Applicability: MaybeIncorrect(?)
Added in: 1.76.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

min_value and max_value are now deprecated.

Applicability: Unspecified(?)
Deprecated in: 1.44.0

What it does

Informs the user about a more concise way to create a vector with a known capacity.

Why is this bad?

The Vec::with_capacity constructor is less complex.

Example

let mut v: Vec<usize> = vec![];
v.reserve(10);

Use instead:

let mut v: Vec<usize> = Vec::with_capacity(10);
Applicability: HasPlaceholders(?)
Added in: 1.74.0

What it does

Checks for unnecessary ‘..’ pattern binding on struct when all fields are explicitly matched.

Why restrict this?

Correctness and readability. It’s like having a wildcard pattern after matching all enum variants explicitly.

Example

let a = A { a: 5 };

match a {
    A { a: 5, .. } => {},
    _ => {},
}

Use instead:

match a {
    A { a: 5 } => {},
    _ => {},
}
Applicability: Unspecified(?)
Added in: 1.43.0

What it does

Checks for iterators of Results using .filter(Result::is_ok).map(Result::unwrap) that may be replaced with a .flatten() call.

Why is this bad?

Result implements IntoIterator<Item = T>. This means that Result can be flattened automatically without suspicious-looking unwrap calls.

Example

let _ = std::iter::empty::<Result<i32, ()>>().filter(Result::is_ok).map(Result::unwrap);

Use instead:

let _ = std::iter::empty::<Result<i32, ()>>().flatten();
Applicability: MachineApplicable(?)
Added in: 1.77.0

What it does

Checks for functions that return Result with an unusually large Err-variant.

Why is this bad?

A Result is at least as large as the Err-variant. While we expect that variant to be seldom used, the compiler needs to reserve and move that much memory every single time. Furthermore, errors are often simply passed up the call-stack, making use of the ?-operator and its type-conversion mechanics. If the Err-variant further up the call-stack stores the Err-variant in question (as library code often does), it itself needs to be at least as large, propagating the problem.

Known problems

The size determined by Clippy is platform-dependent.

Examples

pub enum ParseError {
    UnparsedBytes([u8; 512]),
    UnexpectedEof,
}

// The `Result` has at least 512 bytes, even in the `Ok`-case
pub fn parse() -> Result<(), ParseError> {
    Ok(())
}

should be

pub enum ParseError {
    UnparsedBytes(Box<[u8; 512]>),
    UnexpectedEof,
}

// The `Result` is slightly larger than a pointer
pub fn parse() -> Result<(), ParseError> {
    Ok(())
}

Configuration

  • large-error-threshold: The maximum size of the Err-variant in a Result returned from a function

    (default: 128)

Applicability: Unspecified(?)
Added in: 1.65.0

What it does

Checks for usage of _.map_or(None, Some).

Why is this bad?

Readability, this can be written more concisely as _.ok().

Example

assert_eq!(Some(1), r.map_or(None, Some));

Use instead:

assert_eq!(Some(1), r.ok());
Applicability: MachineApplicable(?)
Added in: 1.44.0

What it does

Checks for usage of result.map(f) where f is a function or closure that returns the unit type ().

Why is this bad?

Readability, this can be written more clearly with an if let statement

Example

let x: Result<String, String> = do_stuff();
x.map(log_err_msg);
x.map(|msg| log_err_msg(format_msg(msg)));

The correct use would be:

let x: Result<String, String> = do_stuff();
if let Ok(msg) = x {
    log_err_msg(msg);
};
if let Ok(msg) = x {
    log_err_msg(format_msg(msg));
};
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for public functions that return a Result with an Err type of (). It suggests using a custom type that implements std::error::Error.

Why is this bad?

Unit does not implement Error and carries no further information about what went wrong.

Known problems

Of course, this lint assumes that Result is used for a fallible operation (which is after all the intended use). However code may opt to (mis)use it as a basic two-variant-enum. In that case, the suggestion is misguided, and the code should use a custom enum instead.

Examples

pub fn read_u8() -> Result<u8, ()> { Err(()) }

should become

use std::fmt;

#[derive(Debug)]
pub struct EndOfStream;

impl fmt::Display for EndOfStream {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        write!(f, "End of Stream")
    }
}

impl std::error::Error for EndOfStream { }

pub fn read_u8() -> Result<u8, EndOfStream> { Err(EndOfStream) }

Note that there are crates that simplify creating the error type, e.g. thiserror.

Applicability: Unspecified(?)
Added in: 1.49.0

What it does

This lint warns when a method returning Self doesn’t have the #[must_use] attribute.

Why is this bad?

Methods returning Self often create new values, having the #[must_use] attribute prevents users from “forgetting” to use the newly created value.

The #[must_use] attribute can be added to the type itself to ensure that instances are never forgotten. Functions returning a type marked with #[must_use] will not be linted, as the usage is already enforced by the type attribute.

Limitations

This lint is only applied on methods taking a self argument. It would be mostly noise if it was added on constructors for example.

Example

pub struct Bar;
impl Bar {
    // Missing attribute
    pub fn bar(&self) -> Self {
        Self
    }
}

Use instead:

// It's better to have the `#[must_use]` attribute on the method like this:
pub struct Bar;
impl Bar {
    #[must_use]
    pub fn bar(&self) -> Self {
        Self
    }
}

// Or on the type definition like this:
#[must_use]
pub struct Bar;
impl Bar {
    pub fn bar(&self) -> Self {
        Self
    }
}
Applicability: Unspecified(?)
Added in: 1.59.0

What it does

Checks for range expressions x..y where both x and y are constant and x is greater to y. Also triggers if x is equal to y when they are conditions to a for loop.

Why is this bad?

Empty ranges yield no values so iterating them is a no-op. Moreover, trying to use a reversed range to index a slice will panic at run-time.

Example

fn main() {
    (10..=0).for_each(|x| println!("{}", x));

    let arr = [1, 2, 3, 4, 5];
    let sub = &arr[3..1];
}

Use instead:

fn main() {
    (0..=10).rev().for_each(|x| println!("{}", x));

    let arr = [1, 2, 3, 4, 5];
    let sub = &arr[1..3];
}

Past names

  • reverse_range_loop
Applicability: MaybeIncorrect(?)
Added in: 1.45.0

What it does

Checks for consecutive ifs with the same function call.

Why is this bad?

This is probably a copy & paste error. Despite the fact that function can have side effects and if works as intended, such an approach is implicit and can be considered a “code smell”.

Example

if foo() == bar {
    …
} else if foo() == bar {
    …
}

This probably should be:

if foo() == bar {
    …
} else if foo() == baz {
    …
}

or if the original code was not a typo and called function mutates a state, consider move the mutation out of the if condition to avoid similarity to a copy & paste error:

let first = foo();
if first == bar {
    …
} else {
    let second = foo();
    if second == bar {
    …
    }
}
Applicability: Unspecified(?)
Added in: 1.41.0

What it does

Checks whether a for loop is being used to push a constant value into a Vec.

Why is this bad?

This kind of operation can be expressed more succinctly with vec![item; SIZE] or vec.resize(NEW_SIZE, item) and using these alternatives may also have better performance.

Example

let item1 = 2;
let item2 = 3;
let mut vec: Vec<u8> = Vec::new();
for _ in 0..20 {
   vec.push(item1);
}
for _ in 0..30 {
    vec.push(item2);
}

Use instead:

let item1 = 2;
let item2 = 3;
let mut vec: Vec<u8> = vec![item1; 20];
vec.resize(20 + 30, item2);
Applicability: Unspecified(?)
Added in: 1.47.0

What it does

It lints if a struct has two methods with the same name: one from a trait, another not from a trait.

Why restrict this?

Confusing.

Example

trait T {
    fn foo(&self) {}
}

struct S;

impl T for S {
    fn foo(&self) {}
}

impl S {
    fn foo(&self) {}
}
Applicability: Unspecified(?)
Added in: 1.57.0

What it does

Checks for an iterator or string search (such as find(), position(), or rposition()) followed by a call to is_some() or is_none().

Why is this bad?

Readability, this can be written more concisely as:

  • _.any(_), or _.contains(_) for is_some(),
  • !_.any(_), or !_.contains(_) for is_none().

Example

let vec = vec![1];
vec.iter().find(|x| **x == 0).is_some();

"hello world".find("world").is_none();

Use instead:

let vec = vec![1];
vec.iter().any(|x| *x == 0);

!"hello world".contains("world");
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks if the seek method of the Seek trait is called with SeekFrom::Current(0), and if it is, suggests using stream_position instead.

Why is this bad?

Readability. Use dedicated method.

Example

use std::fs::File;
use std::io::{self, Write, Seek, SeekFrom};

fn main() -> io::Result<()> {
    let mut f = File::create("foo.txt")?;
    f.write_all(b"Hello")?;
    eprintln!("Written {} bytes", f.seek(SeekFrom::Current(0))?);

    Ok(())
}

Use instead:

use std::fs::File;
use std::io::{self, Write, Seek, SeekFrom};

fn main() -> io::Result<()> {
    let mut f = File::create("foo.txt")?;
    f.write_all(b"Hello")?;
    eprintln!("Written {} bytes", f.stream_position()?);

    Ok(())
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.67.0

What it does

Checks for jumps to the start of a stream that implements Seek and uses the seek method providing Start as parameter.

Why is this bad?

Readability. There is a specific method that was implemented for this exact scenario.

Example

fn foo<T: io::Seek>(t: &mut T) {
    t.seek(io::SeekFrom::Start(0));
}

Use instead:

fn foo<T: io::Seek>(t: &mut T) {
    t.rewind();
}
Applicability: MachineApplicable(?)
Added in: 1.67.0

What it does

Checks for explicit self-assignments.

Why is this bad?

Self-assignments are redundant and unlikely to be intentional.

Known problems

If expression contains any deref coercions or indexing operations they are assumed not to have any side effects.

Example

struct Event {
    x: i32,
}

fn copy_position(a: &mut Event, b: &Event) {
    a.x = a.x;
}

Should be:

struct Event {
    x: i32,
}

fn copy_position(a: &mut Event, b: &Event) {
    a.x = b.x;
}
Applicability: Unspecified(?)
Added in: 1.48.0

What it does

Warns when constructors have the same name as their types.

Why is this bad?

Repeating the name of the type is redundant.

Example

struct Foo {}

impl Foo {
    pub fn foo() -> Foo {
        Foo {}
    }
}

Use instead:

struct Foo {}

impl Foo {
    pub fn new() -> Foo {
        Foo {}
    }
}
Applicability: Unspecified(?)
Added in: 1.55.0

What it does

Checks that module layout uses only mod.rs files.

Why restrict this?

Having multiple module layout styles in a project can be confusing.

Example

src/
  stuff/
    stuff_files.rs
  stuff.rs
  lib.rs

Use instead:

src/
  stuff/
    stuff_files.rs
    mod.rs
  lib.rs
Applicability: Unspecified(?)
Added in: 1.57.0

What it does

Looks for blocks of expressions and fires if the last expression returns () but is not followed by a semicolon.

Why is this bad?

The semicolon might be optional but when extending the block with new code, it doesn’t require a change in previous last line.

Example

fn main() {
    println!("Hello world")
}

Use instead:

fn main() {
    println!("Hello world");
}
Applicability: MachineApplicable(?)
Added in: 1.52.0

What it does

Suggests moving the semicolon after a block to the inside of the block, after its last expression.

Why restrict this?

For consistency it’s best to have the semicolon inside/outside the block. Either way is fine and this lint suggests inside the block. Take a look at semicolon_outside_block for the other alternative.

Example

unsafe { f(x) };

Use instead:

unsafe { f(x); }

Configuration

  • semicolon-inside-block-ignore-singleline: Whether to lint only if it’s multiline.

    (default: false)

Applicability: MachineApplicable(?)
Added in: 1.68.0

What it does

Suggests moving the semicolon from a block’s final expression outside of the block.

Why restrict this?

For consistency it’s best to have the semicolon inside/outside the block. Either way is fine and this lint suggests outside the block. Take a look at semicolon_inside_block for the other alternative.

Example

unsafe { f(x); }

Use instead:

unsafe { f(x) };

Configuration

  • semicolon-outside-block-ignore-multiline: Whether to lint only if it’s singleline.

    (default: false)

Applicability: MachineApplicable(?)
Added in: 1.68.0

What it does

Warns if literal suffixes are separated by an underscore. To enforce separated literal suffix style, see the unseparated_literal_suffix lint.

Why restrict this?

Suffix style should be consistent.

Example

123832_i32

Use instead:

123832i32
Applicability: MachineApplicable(?)
Added in: 1.58.0

What it does

Checks for misuses of the serde API.

Why is this bad?

Serde is very finicky about how its API should be used, but the type system can’t be used to enforce it (yet?).

Example

Implementing Visitor::visit_string but not Visitor::visit_str.

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of contains to see if a value is not present in a set like HashSet or BTreeSet, followed by an insert.

Why is this bad?

Using just insert and checking the returned bool is more efficient.

Known problems

In case the value that wants to be inserted is borrowed and also expensive or impossible to clone. In such a scenario, the developer might want to check with contains before inserting, to avoid the clone. In this case, it will report a false positive.

Example

use std::collections::HashSet;
let mut set = HashSet::new();
let value = 5;
if !set.contains(&value) {
    set.insert(value);
    println!("inserted {value:?}");
}

Use instead:

use std::collections::HashSet;
let mut set = HashSet::new();
let value = 5;
if set.insert(&value) {
    println!("inserted {value:?}");
}
Applicability: Unspecified(?)
Added in: 1.81.0

What it does

Checks for bindings that shadow other bindings already in scope, while reusing the original value.

Why restrict this?

Some argue that name shadowing like this hurts readability, because a value may be bound to different things depending on position in the code.

See also shadow_same and shadow_unrelated for other restrictions on shadowing.

Example

let x = 2;
let x = x + 1;

use different variable name:

let x = 2;
let y = x + 1;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for bindings that shadow other bindings already in scope, while just changing reference level or mutability.

Why restrict this?

To require that what are formally distinct variables be given distinct names.

See also shadow_reuse and shadow_unrelated for other restrictions on shadowing.

Example

let x = &x;

Use instead:

let y = &x; // use different variable name
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for bindings that shadow other bindings already in scope, either without an initialization or with one that does not even use the original value.

Why restrict this?

Shadowing a binding with a closely related one is part of idiomatic Rust, but shadowing a binding by accident with an unrelated one may indicate a mistake.

Additionally, name shadowing in general can hurt readability, especially in large code bases, because it is easy to lose track of the active binding at any place in the code. If linting against all shadowing is desired, you may wish to use the shadow_same and shadow_reuse lints as well.

Example

let x = y;
let x = z; // shadows the earlier binding

Use instead:

let x = y;
let w = z; // use different variable name
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for the use of short circuit boolean conditions as a statement.

Why is this bad?

Using a short circuit boolean condition as a statement may hide the fact that the second part is executed or not depending on the outcome of the first part.

Example

f() && g(); // We should write `if f() { g(); }`.
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

assert!(a == b) can now print the values the same way `assert_eq!(a, b) can.

Applicability: Unspecified(?)
Deprecated in: pre 1.29.0

What it does

Checks for methods that should live in a trait implementation of a std trait (see llogiq’s blog post for further information) instead of an inherent implementation.

Why is this bad?

Implementing the traits improve ergonomics for users of the code, often with very little cost. Also people seeing a mul(...) method may expect * to work equally, so you should have good reason to disappoint them.

Example

struct X;
impl X {
    fn add(&self, other: &X) -> X {
        // ..
    }
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for #[should_panic] attributes without specifying the expected panic message.

Why is this bad?

The expected panic message should be specified to ensure that the test is actually panicking with the expected message, and not another unrelated panic.

Example

fn random() -> i32 { 0 }

#[should_panic]
#[test]
fn my_test() {
    let _ = 1 / random();
}

Use instead:

fn random() -> i32 { 0 }

#[should_panic = "attempt to divide by zero"]
#[test]
fn my_test() {
    let _ = 1 / random();
}
Applicability: HasPlaceholders(?)
Added in: 1.74.0

What it does

Checks for temporaries returned from function calls in a match scrutinee that have the clippy::has_significant_drop attribute.

Why is this bad?

The clippy::has_significant_drop attribute can be added to types whose Drop impls have an important side-effect, such as unlocking a mutex, making it important for users to be able to accurately understand their lifetimes. When a temporary is returned in a function call in a match scrutinee, its lifetime lasts until the end of the match block, which may be surprising.

For Mutexes this can lead to a deadlock. This happens when the match scrutinee uses a function call that returns a MutexGuard and then tries to lock again in one of the match arms. In that case the MutexGuard in the scrutinee will not be dropped until the end of the match block and thus will not unlock.

Example

let mutex = Mutex::new(State {});

match mutex.lock().unwrap().foo() {
    true => {
        mutex.lock().unwrap().bar(); // Deadlock!
    }
    false => {}
};

println!("All done!");

Use instead:

let mutex = Mutex::new(State {});

let is_foo = mutex.lock().unwrap().foo();
match is_foo {
    true => {
        mutex.lock().unwrap().bar();
    }
    false => {}
};

println!("All done!");
Applicability: MaybeIncorrect(?)
Added in: 1.60.0

What it does

Searches for elements marked with #[clippy::has_significant_drop] that could be early dropped but are in fact dropped at the end of their scopes. In other words, enforces the “tightening” of their possible lifetimes.

Why is this bad?

Elements marked with #[clippy::has_significant_drop] are generally synchronizing primitives that manage shared resources, as such, it is desired to release them as soon as possible to avoid unnecessary resource contention.

Example

fn main() {
  let lock = some_sync_resource.lock();
  let owned_rslt = lock.do_stuff_with_resource();
  // Only `owned_rslt` is needed but `lock` is still held.
  do_heavy_computation_that_takes_time(owned_rslt);
}

Use instead:

fn main() {
    let owned_rslt = some_sync_resource.lock().do_stuff_with_resource();
    do_heavy_computation_that_takes_time(owned_rslt);
}
Applicability: MaybeIncorrect(?)
Added in: 1.69.0

What it does

Checks for names that are very similar and thus confusing.

Note: this lint looks for similar names throughout each scope. To allow it, you need to allow it on the scope level, not on the name that is reported.

Why is this bad?

It’s hard to distinguish between names that differ only by a single character.

Example

let checked_exp = something;
let checked_expr = something_else;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for functions that are only used once. Does not lint tests.

Why restrict this?

If a function is only used once (perhaps because it used to be used more widely), then the code could be simplified by moving that function’s code into its caller.

However, there are reasons not to do this everywhere:

  • Splitting a large function into multiple parts often improves readability by giving names to its parts.
  • A function’s signature might serve a necessary purpose, such as constraining the type of a closure passed to it.
  • Generic functions might call non-generic functions to reduce duplication in the produced machine code.

If this lint is used, prepare to #[allow] it a lot.

Example

pub fn a<T>(t: &T)
where
    T: AsRef<str>,
{
    a_inner(t.as_ref())
}

fn a_inner(t: &str) {
    /* snip */
}

Use instead:

pub fn a<T>(t: &T)
where
    T: AsRef<str>,
{
    let t = t.as_ref();
    /* snip */
}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Warns when using push_str/insert_str with a single-character string literal where push/insert with a char would work fine.

Why is this bad?

It’s less clear that we are pushing a single character.

Example

string.insert_str(0, "R");
string.push_str("R");

Use instead:

string.insert(0, 'R');
string.push('R');

Past names

  • single_char_push_str
Applicability: MachineApplicable(?)
Added in: 1.49.0

What it does

Checks for lifetimes with names which are one character long.

Why restrict this?

A single character is likely not enough to express the purpose of a lifetime. Using a longer name can make code easier to understand.

Known problems

Rust programmers and learning resources tend to use single character lifetimes, so this lint is at odds with the ecosystem at large. In addition, the lifetime’s purpose may be obvious or, rarely, expressible in one character.

Example

struct DiagnosticCtx<'a> {
    source: &'a str,
}

Use instead:

struct DiagnosticCtx<'src> {
    source: &'src str,
}
Applicability: Unspecified(?)
Added in: 1.60.0

What it does

Checks for string methods that receive a single-character str as an argument, e.g., _.split("x").

Why is this bad?

While this can make a perf difference on some systems, benchmarks have proven inconclusive. But at least using a char literal makes it clear that we are looking at a single character.

Known problems

Does not catch multi-byte unicode characters. This is by design, on many machines, splitting by a non-ascii char is actually slower. Please do your own measurements instead of relying solely on the results of this lint.

Example

_.split("x");

Use instead:

_.split('x');
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checking for imports with single component use path.

Why is this bad?

Import with single component use path such as use cratename; is not necessary, and thus should be removed.

Example

use regex;

fn main() {
    regex::Regex::new(r"^\d{4}-\d{2}-\d{2}$").unwrap();
}

Better as

fn main() {
    regex::Regex::new(r"^\d{4}-\d{2}-\d{2}$").unwrap();
}
Applicability: MachineApplicable(?)
Added in: 1.43.0

What it does

Checks whether a for loop has a single element.

Why is this bad?

There is no reason to have a loop of a single element.

Example

let item1 = 2;
for item in &[item1] {
    println!("{}", item);
}

Use instead:

let item1 = 2;
let item = &item1;
println!("{}", item);
Applicability: MachineApplicable(?)
Added in: 1.49.0

What it does

Checks for matches with a single arm where an if let will usually suffice.

This intentionally does not lint if there are comments inside of the other arm, so as to allow the user to document why having another explicit pattern with an empty body is necessary, or because the comments need to be preserved for other reasons.

Why is this bad?

Just readability – if let nests less than a match.

Example

match x {
    Some(ref foo) => bar(foo),
    _ => (),
}

Use instead:

if let Some(ref foo) = x {
    bar(foo);
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for matches with two arms where an if let else will usually suffice.

Why is this bad?

Just readability – if let nests less than a match.

Known problems

Personal style preferences may differ.

Example

Using match:

match x {
    Some(ref foo) => bar(foo),
    _ => bar(&other_ref),
}

Using if let with else:

if let Some(ref foo) = x {
    bar(foo);
} else {
    bar(&other_ref);
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for Vec or array initializations that contain only one range.

Why is this bad?

This is almost always incorrect, as it will result in a Vec that has only one element. Almost always, the programmer intended for it to include all elements in the range or for the end of the range to be the length instead.

Example

let x = [0..200];

Use instead:

// If it was intended to include every element in the range...
let x = (0..200).collect::<Vec<i32>>();
// ...Or if 200 was meant to be the len
let x = [0; 200];
Applicability: MaybeIncorrect(?)
Added in: 1.72.0

What it does

Detects expressions where size_of::<T> or size_of_val::<T> is used as a count of elements of type T

Why is this bad?

These functions expect a count of T and not a number of bytes

Example

const SIZE: usize = 128;
let x = [2u8; SIZE];
let mut y = [2u8; SIZE];
unsafe { copy_nonoverlapping(x.as_ptr(), y.as_mut_ptr(), size_of::<u8>() * SIZE) };
Applicability: Unspecified(?)
Added in: 1.50.0

What it does

Checks for calls to std::mem::size_of_val() where the argument is a reference to a reference.

Why is this bad?

Calling size_of_val() with a reference to a reference as the argument yields the size of the reference-type, not the size of the value behind the reference.

Example

struct Foo {
    buffer: [u8],
}

impl Foo {
    fn size(&self) -> usize {
        // Note that `&self` as an argument is a `&&Foo`: Because `self`
        // is already a reference, `&self` is a double-reference.
        // The return value of `size_of_val()` therefore is the
        // size of the reference-type, not the size of `self`.
        std::mem::size_of_val(&self)
    }
}

Use instead:

struct Foo {
    buffer: [u8],
}

impl Foo {
    fn size(&self) -> usize {
        // Correct
        std::mem::size_of_val(self)
    }
}
Applicability: Unspecified(?)
Added in: 1.68.0

What it does

Checks for usage of _.skip_while(condition).next().

Why is this bad?

Readability, this can be written more concisely as _.find(!condition).

Example

vec.iter().skip_while(|x| **x == 0).next();

Use instead:

vec.iter().find(|x| **x != 0);
Applicability: Unspecified(?)
Added in: 1.42.0

What it does

Checks slow zero-filled vector initialization

Why is this bad?

These structures are non-idiomatic and less efficient than simply using vec![0; len].

Specifically, for vec![0; len], the compiler can use a specialized type of allocation that also zero-initializes the allocated memory in the same call (see: alloc_zeroed).

Writing Vec::new() followed by vec.resize(len, 0) is suboptimal because, while it does do the same number of allocations, it involves two operations for allocating and initializing. The resize call first allocates memory (since Vec::new() did not), and only then zero-initializes it.

Example

let mut vec1 = Vec::new();
vec1.resize(len, 0);

let mut vec2 = Vec::with_capacity(len);
vec2.resize(len, 0);

let mut vec3 = Vec::with_capacity(len);
vec3.extend(repeat(0).take(len));

Use instead:

let mut vec1 = vec![0; len];
let mut vec2 = vec![0; len];
let mut vec3 = vec![0; len];
Applicability: Unspecified(?)
Added in: 1.32.0

What it does

When sorting primitive values (integers, bools, chars, as well as arrays, slices, and tuples of such items), it is typically better to use an unstable sort than a stable sort.

Why is this bad?

Typically, using a stable sort consumes more memory and cpu cycles. Because values which compare equal are identical, preserving their relative order (the guarantee that a stable sort provides) means nothing, while the extra costs still apply.

Known problems

As pointed out in issue #8241, a stable sort can instead be significantly faster for certain scenarios (eg. when a sorted vector is extended with new data and resorted).

For more information and benchmarking results, please refer to the issue linked above.

Example

let mut vec = vec![2, 1, 3];
vec.sort();

Use instead:

let mut vec = vec![2, 1, 3];
vec.sort_unstable();
Applicability: MachineApplicable(?)
Added in: 1.47.0

What it does

Finds items imported through std when available through alloc.

Why restrict this?

Crates which have no_std compatibility and require alloc may wish to ensure types are imported from alloc to ensure disabling std does not cause the crate to fail to compile. This lint is also useful for crates migrating to become no_std compatible.

Example

use std::vec::Vec;

Use instead:

use alloc::vec::Vec;
Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

Finds items imported through std when available through core.

Why restrict this?

Crates which have no_std compatibility may wish to ensure types are imported from core to ensure disabling std does not cause the crate to fail to compile. This lint is also useful for crates migrating to become no_std compatible.

Example

use std::hash::Hasher;

Use instead:

use core::hash::Hasher;
Applicability: MachineApplicable(?)
Added in: 1.64.0

What it does

Checks for usages of str.trim().split("\n") and str.trim().split("\r\n").

Why is this bad?

Hard-coding the line endings makes the code less compatible. str.lines should be used instead.

Example

"some\ntext\nwith\nnewlines\n".trim().split('\n');

Use instead:

"some\ntext\nwith\nnewlines\n".lines();

Known Problems

This lint cannot detect if the split is intentionally restricted to a single type of newline ("\n" or "\r\n"), for example during the parsing of a specific file format in which precisely one newline type is valid.

Applicability: MaybeIncorrect(?)
Added in: 1.77.0

What it does

This lint checks for .to_string() method calls on values of type &str.

Why restrict this?

The to_string method is also used on other types to convert them to a string. When called on a &str it turns the &str into the owned variant String, which can be more specifically expressed with .to_owned().

Example

// example code where clippy issues a warning
let _ = "str".to_string();

Use instead:

// example code which does not raise clippy warning
let _ = "str".to_owned();
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for all instances of x + _ where x is of type String, but only if string_add_assign does not match.

Why restrict this?

This particular Add implementation is asymmetric (the other operand need not be String, but x does), while addition as mathematically defined is symmetric, and the String::push_str(_) function is a perfectly good replacement. Therefore, some dislike it and wish not to have it in their code.

That said, other people think that string addition, having a long tradition in other languages is actually fine, which is why we decided to make this particular lint allow by default.

Example

let x = "Hello".to_owned();
x + ", World";

Use instead:

let mut x = "Hello".to_owned();
x.push_str(", World");
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for string appends of the form x = x + y (without let!).

Why is this bad?

It’s not really bad, but some people think that the .push_str(_) method is more readable.

Example

let mut x = "Hello".to_owned();
x = x + ", World";

// More readable
x += ", World";
x.push_str(", World");
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for the use of .extend(s.chars()) where s is a &str or String.

Why is this bad?

.push_str(s) is clearer

Example

let abc = "abc";
let def = String::from("def");
let mut s = String::new();
s.extend(abc.chars());
s.extend(def.chars());

The correct use would be:

let abc = "abc";
let def = String::from("def");
let mut s = String::new();
s.push_str(abc);
s.push_str(&def);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Check if the string is transformed to byte array and casted back to string.

Why is this bad?

It’s unnecessary, the string can be used directly.

Example

std::str::from_utf8(&"Hello World!".as_bytes()[6..11]).unwrap();

Use instead:

&"Hello World!"[6..11];
Applicability: MachineApplicable(?)
Added in: 1.50.0

What it does

Checks for the as_bytes method called on string literals that contain only ASCII characters.

Why is this bad?

Byte string literals (e.g., b"foo") can be used instead. They are shorter but less discoverable than as_bytes().

Known problems

"str".as_bytes() and the suggested replacement of b"str" are not equivalent because they have different types. The former is &[u8] while the latter is &[u8; 3]. That means in general they will have a different set of methods and different trait implementations.

fn f(v: Vec<u8>) {}

f("...".as_bytes().to_owned()); // works
f(b"...".to_owned()); // does not work, because arg is [u8; 3] not Vec<u8>

fn g(r: impl std::io::Read) {}

g("...".as_bytes()); // works
g(b"..."); // does not work

The actual equivalent of "str".as_bytes() with the same type is not b"str" but &b"str"[..], which is a great deal of punctuation and not more readable than a function call.

Example

let bstr = "a byte string".as_bytes();

Use instead:

let bstr = b"a byte string";
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for <string_lit>.chars().any(|i| i == c).

Why is this bad?

It’s significantly slower than using a pattern instead, like matches!(c, '\\' | '.' | '+').

Despite this being faster, this is not perf as this is pretty common, and is a rather nice way to check if a char is any in a set. In any case, this restriction lint is available for situations where that additional performance is absolutely necessary.

Example

"\\.+*?()|[]{}^$#&-~".chars().any(|x| x == c);

Use instead:

matches!(c, '\\' | '.' | '+' | '*' | '(' | ')' | '|' | '[' | ']' | '{' | '}' | '^' | '$' | '#' | '&' | '-' | '~');
Applicability: MachineApplicable(?)
Added in: 1.73.0

What it does

Checks for slice operations on strings

Why restrict this?

UTF-8 characters span multiple bytes, and it is easy to inadvertently confuse character counts and string indices. This may lead to panics, and should warrant some test cases containing wide UTF-8 characters. This lint is most useful in code that should avoid panics at all costs.

Known problems

Probably lots of false positives. If an index comes from a known valid position (e.g. obtained via char_indices over the same string), it is totally OK.

Example

&"Ölkanne"[1..];
Applicability: Unspecified(?)
Added in: 1.58.0

What it does

This lint checks for .to_string() method calls on values of type String.

Why restrict this?

The to_string method is also used on other types to convert them to a string. When called on a String it only clones the String, which can be more specifically expressed with .clone().

Example

// example code where clippy issues a warning
let msg = String::from("Hello World");
let _ = msg.to_string();

Use instead:

// example code which does not raise clippy warning
let msg = String::from("Hello World");
let _ = msg.clone();
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of libc::strlen on a CString or CStr value, and suggest calling as_bytes().len() or to_bytes().len() respectively instead.

Why is this bad?

This avoids calling an unsafe libc function. Currently, it also avoids calculating the length.

Example

use std::ffi::CString;
let cstring = CString::new("foo").expect("CString::new failed");
let len = unsafe { libc::strlen(cstring.as_ptr()) };

Use instead:

use std::ffi::CString;
let cstring = CString::new("foo").expect("CString::new failed");
let len = cstring.as_bytes().len();
Applicability: MachineApplicable(?)
Added in: 1.55.0

What it does

Checks for excessive use of bools in structs.

Why is this bad?

Excessive bools in a struct is often a sign that it’s used as a state machine, which is much better implemented as an enum. If it’s not the case, excessive bools usually benefit from refactoring into two-variant enums for better readability and API.

Example

struct S {
    is_pending: bool,
    is_processing: bool,
    is_finished: bool,
}

Use instead:

enum S {
    Pending,
    Processing,
    Finished,
}

Configuration

  • max-struct-bools: The maximum number of bool fields a struct can have

    (default: 3)

Applicability: Unspecified(?)
Added in: 1.43.0

What it does

Detects struct fields that are prefixed or suffixed by the same characters or the name of the struct itself.

Why is this bad?

Information common to all struct fields is better represented in the struct name.

Limitations

Characters with no casing will be considered when comparing prefixes/suffixes This applies to numbers and non-ascii characters without casing e.g. foo1 and foo2 is considered to have different prefixes (the prefixes are foo1 and foo2 respectively), as also bar螃, bar蟹

Example

struct Cake {
    cake_sugar: u8,
    cake_flour: u8,
    cake_eggs: u8
}

Use instead:

struct Cake {
    sugar: u8,
    flour: u8,
    eggs: u8
}

Configuration

  • struct-field-name-threshold: The minimum number of struct fields for the lints about field names to trigger

    (default: 3)

Applicability: Unspecified(?)
Added in: 1.75.0

What it does

Looks for floating-point expressions that can be expressed using built-in methods to improve both accuracy and performance.

Why is this bad?

Negatively impacts accuracy and performance.

Example

use std::f32::consts::E;

let a = 3f32;
let _ = (2f32).powf(a);
let _ = E.powf(a);
let _ = a.powf(1.0 / 2.0);
let _ = a.log(2.0);
let _ = a.log(10.0);
let _ = a.log(E);
let _ = a.powf(2.0);
let _ = a * 2.0 + 4.0;
let _ = if a < 0.0 {
    -a
} else {
    a
};
let _ = if a < 0.0 {
    a
} else {
    -a
};

is better expressed as

use std::f32::consts::E;

let a = 3f32;
let _ = a.exp2();
let _ = a.exp();
let _ = a.sqrt();
let _ = a.log2();
let _ = a.log10();
let _ = a.ln();
let _ = a.powi(2);
let _ = a.mul_add(2.0, 4.0);
let _ = a.abs();
let _ = -a.abs();
Applicability: MachineApplicable(?)
Added in: 1.43.0

What it does

Lints for suspicious operations in impls of arithmetic operators, e.g. subtracting elements in an Add impl.

Why is this bad?

This is probably a typo or copy-and-paste error and not intended.

Example

impl Add for Foo {
    type Output = Foo;

    fn add(self, other: Foo) -> Foo {
        Foo(self.0 - other.0)
    }
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of the non-existent =*, =! and =- operators.

Why is this bad?

This is either a typo of *=, != or -= or confusing.

Example

a =- 42; // confusing, should it be `a -= 42` or `a = -42`?
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for Command::arg() invocations that look like they should be multiple arguments instead, such as arg("-t ext2").

Why is this bad?

Command::arg() does not split arguments by space. An argument like arg("-t ext2") will be passed as a single argument to the command, which is likely not what was intended.

Example

std::process::Command::new("echo").arg("-n hello").spawn().unwrap();

Use instead:

std::process::Command::new("echo").args(["-n", "hello"]).spawn().unwrap();
Applicability: MaybeIncorrect(?)
Added in: 1.69.0

What it does

Detects the use of outer doc comments (///, /**) followed by a bang (!): ///!

Why is this bad?

Triple-slash comments (known as “outer doc comments”) apply to items that follow it. An outer doc comment followed by a bang (i.e. ///!) has no specific meaning.

The user most likely meant to write an inner doc comment (//!, /*!), which applies to the parent item (i.e. the item that the comment is contained in, usually a module or crate).

Known problems

Inner doc comments can only appear before items, so there are certain cases where the suggestion made by this lint is not valid code. For example:

fn foo() {}
///!
fn bar() {}

This lint detects the doc comment and suggests changing it to //!, but an inner doc comment is not valid at that position.

Example

In this example, the doc comment is attached to the function, rather than the module.

pub mod util {
    ///! This module contains utility functions.

    pub fn dummy() {}
}

Use instead:

pub mod util {
    //! This module contains utility functions.

    pub fn dummy() {}
}
Applicability: MaybeIncorrect(?)
Added in: 1.70.0

What it does

Checks for formatting of else. It lints if the else is followed immediately by a newline or the else seems to be missing.

Why is this bad?

This is probably some refactoring remnant, even if the code is correct, it might look confusing.

Example

if foo {
} { // looks like an `else` is missing here
}

if foo {
} if bar { // looks like an `else` is missing here
}

if foo {
} else

{ // this is the `else` block of the previous `if`, but should it be?
}

if foo {
} else

if bar { // this is the `else` block of the previous `if`, but should it be?
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for calls to map followed by a count.

Why is this bad?

It looks suspicious. Maybe map was confused with filter. If the map call is intentional, this should be rewritten using inspect. Or, if you intend to drive the iterator to completion, you can just use for_each instead.

Example

let _ = (0..3).map(|x| x + 2).count();
Applicability: Unspecified(?)
Added in: 1.39.0

What it does

Lints for suspicious operations in impls of OpAssign, e.g. subtracting elements in an AddAssign impl.

Why is this bad?

This is probably a typo or copy-and-paste error and not intended.

Example

impl AddAssign for Foo {
    fn add_assign(&mut self, other: Foo) {
        *self = *self - other;
    }
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for the suspicious use of OpenOptions::create() without an explicit OpenOptions::truncate().

Why is this bad?

create() alone will either create a new file or open an existing file. If the file already exists, it will be overwritten when written to, but the file will not be truncated by default. If less data is written to the file than it already contains, the remainder of the file will remain unchanged, and the end of the file will contain old data. In most cases, one should either use create_new to ensure the file is created from scratch, or ensure truncate is called so that the truncation behaviour is explicit. truncate(true) will ensure the file is entirely overwritten with new data, whereas truncate(false) will explicitly keep the default behavior.

Example

use std::fs::OpenOptions;

OpenOptions::new().create(true);

Use instead:

use std::fs::OpenOptions;

OpenOptions::new().create(true).truncate(true);
Applicability: MaybeIncorrect(?)
Added in: 1.77.0

What it does

Checks for unlikely usages of binary operators that are almost certainly typos and/or copy/paste errors, given the other usages of binary operators nearby.

Why is this bad?

They are probably bugs and if they aren’t then they look like bugs and you should add a comment explaining why you are doing such an odd set of operations.

Known problems

There may be some false positives if you are trying to do something unusual that happens to look like a typo.

Example

struct Vec3 {
    x: f64,
    y: f64,
    z: f64,
}

impl Eq for Vec3 {}

impl PartialEq for Vec3 {
    fn eq(&self, other: &Self) -> bool {
        // This should trigger the lint because `self.x` is compared to `other.y`
        self.x == other.y && self.y == other.y && self.z == other.z
    }
}

Use instead:

// same as above except:
impl PartialEq for Vec3 {
    fn eq(&self, other: &Self) -> bool {
        // Note we now compare other.x to self.x
        self.x == other.x && self.y == other.y && self.z == other.z
    }
}
Applicability: MachineApplicable(?)
Added in: 1.50.0

What it does

Checks for calls to [splitn] (https://doc.rust-lang.org/std/primitive.str.html#method.splitn) and related functions with either zero or one splits.

Why is this bad?

These calls don’t actually split the value and are likely to be intended as a different number.

Example

for x in s.splitn(1, ":") {
    // ..
}

Use instead:

for x in s.splitn(2, ":") {
    // ..
}
Applicability: Unspecified(?)
Added in: 1.54.0

What it does

Checks for the usage of _.to_owned(), on a Cow<'_, _>.

Why is this bad?

Calling to_owned() on a Cow creates a clone of the Cow itself, without taking ownership of the Cow contents (i.e. it’s equivalent to calling Cow::clone). The similarly named into_owned method, on the other hand, clones the Cow contents, effectively turning any Cow::Borrowed into a Cow::Owned.

Given the potential ambiguity, consider replacing to_owned with clone for better readability or, if getting a Cow::Owned was the original intent, using into_owned instead.

Example

let s = "Hello world!";
let cow = Cow::Borrowed(s);

let data = cow.to_owned();
assert!(matches!(data, Cow::Borrowed(_)))

Use instead:

let s = "Hello world!";
let cow = Cow::Borrowed(s);

let data = cow.clone();
assert!(matches!(data, Cow::Borrowed(_)))

or

let s = "Hello world!";
let cow = Cow::Borrowed(s);

let _data: String = cow.into_owned();
Applicability: MaybeIncorrect(?)
Added in: 1.65.0

What it does

Checks the formatting of a unary operator on the right hand side of a binary operator. It lints if there is no space between the binary and unary operators, but there is a space between the unary and its operand.

Why is this bad?

This is either a typo in the binary operator or confusing.

Example

// &&! looks like a different operator
if foo &&! bar {}

Use instead:

if foo && !bar {}
Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Warns for a Bitwise XOR (^) operator being probably confused as a powering. It will not trigger if any of the numbers are not in decimal.

Why restrict this?

It’s most probably a typo and may lead to unexpected behaviours.

Example

let x = 3_i32 ^ 4_i32;

Use instead:

let x = 3_i32.pow(4);
Applicability: MaybeIncorrect(?)
Added in: 1.67.0

What it does

Checks for calls to core::mem::swap where either parameter is derived from a pointer

Why is this bad?

When at least one parameter to swap is derived from a pointer it may overlap with the other. This would then lead to undefined behavior.

Example

unsafe fn swap(x: &[*mut u32], y: &[*mut u32]) {
    for (&x, &y) in x.iter().zip(y) {
        core::mem::swap(&mut *x, &mut *y);
    }
}

Use instead:

unsafe fn swap(x: &[*mut u32], y: &[*mut u32]) {
    for (&x, &y) in x.iter().zip(y) {
        core::ptr::swap(x, y);
    }
}
Applicability: MachineApplicable(?)
Added in: 1.63.0

What it does

Checks doc comments for usage of tab characters.

Why is this bad?

The rust style-guide promotes spaces instead of tabs for indentation. To keep a consistent view on the source, also doc comments should not have tabs. Also, explaining ascii-diagrams containing tabs can get displayed incorrectly when the display settings of the author and reader differ.

Example

///
/// Struct to hold two strings:
/// 	- first		one
/// 	- second	one
pub struct DoubleString {
   ///
   /// 	- First String:
   /// 		- needs to be inside here
   first_string: String,
   ///
   /// 	- Second String:
   /// 		- needs to be inside here
   second_string: String,
}

Will be converted to:

///
/// Struct to hold two strings:
///     - first        one
///     - second    one
pub struct DoubleString {
   ///
   ///     - First String:
   ///         - needs to be inside here
   first_string: String,
   ///
   ///     - Second String:
   ///         - needs to be inside here
   second_string: String,
}
Applicability: MaybeIncorrect(?)
Added in: 1.41.0

What it does

Checks for construction of a structure or tuple just to assign a value in it.

Why is this bad?

Readability. If the structure is only created to be updated, why not write the structure you want in the first place?

Example

(0, 0).0 = 1
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for #[test] in doctests unless they are marked with either ignore, no_run or compile_fail.

Why is this bad?

Code in examples marked as #[test] will somewhat surprisingly not be run by cargo test. If you really want to show how to test stuff in an example, mark it no_run to make the intent clear.

Examples

/// An example of a doctest with a `main()` function
///
/// # Examples
///
/// ```
/// #[test]
/// fn equality_works() {
///     assert_eq!(1_u8, 1);
/// }
/// ```
fn test_attr_in_doctest() {
    unimplemented!();
}
Applicability: Unspecified(?)
Added in: 1.76.0

What it does

Triggers when a testing function (marked with the #[test] attribute) isn’t inside a testing module (marked with #[cfg(test)]).

Why restrict this?

The idiomatic (and more performant) way of writing tests is inside a testing module (flagged with #[cfg(test)]), having test functions outside of this module is confusing and may lead to them being “hidden”.

Example

#[test]
fn my_cool_test() {
    // [...]
}

#[cfg(test)]
mod tests {
    // [...]
}

Use instead:

#[cfg(test)]
mod tests {
    #[test]
    fn my_cool_test() {
        // [...]
    }
}
Applicability: Unspecified(?)
Added in: 1.70.0

What it does

Checks for .to_digit(..).is_some() on chars.

Why is this bad?

This is a convoluted way of checking if a char is a digit. It’s more straight forward to use the dedicated is_digit method.

Example

let is_digit = c.to_digit(radix).is_some();

can be written as:

let is_digit = c.is_digit(radix);
Applicability: MachineApplicable(?)
Added in: 1.41.0

What it does

Checks for ToString::to_string applied to a type that implements Display in a macro that does formatting.

Why is this bad?

Since the type implements Display, the use of to_string is unnecessary.

Example

println!("error: something failed at {}", Location::caller().to_string());

Use instead:

println!("error: something failed at {}", Location::caller());
Applicability: MachineApplicable(?)
Added in: 1.58.0

What it does

Checks for direct implementations of ToString.

Why is this bad?

This trait is automatically implemented for any type which implements the Display trait. As such, ToString shouldn’t be implemented directly: Display should be implemented instead, and you get the ToString implementation for free.

Example

struct Point {
  x: usize,
  y: usize,
}

impl ToString for Point {
  fn to_string(&self) -> String {
    format!("({}, {})", self.x, self.y)
  }
}

Use instead:

struct Point {
  x: usize,
  y: usize,
}

impl std::fmt::Display for Point {
  fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
    write!(f, "({}, {})", self.x, self.y)
  }
}
Applicability: Unspecified(?)
Added in: 1.78.0

What it does

Checks for usage of todo!.

Why restrict this?

The todo! macro indicates the presence of unfinished code, so it should not be present in production code.

Example

todo!();

Finish the implementation, or consider marking it as explicitly unimplemented.

unimplemented!();
Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Checks if the first paragraph in the documentation of items listed in the module page is too long.

Why is this bad?

Documentation will show the first paragraph of the docstring in the summary page of a module. Having a nice, short summary in the first paragraph is part of writing good docs.

Example

/// A very short summary.
/// A much longer explanation that goes into a lot more detail about
/// how the thing works, possibly with doclinks and so one,
/// and probably spanning a many rows.
struct Foo {}

Use instead:

/// A very short summary.
///
/// A much longer explanation that goes into a lot more detail about
/// how the thing works, possibly with doclinks and so one,
/// and probably spanning a many rows.
struct Foo {}
Applicability: MachineApplicable(?)
Added in: 1.82.0

What it does

Checks for functions with too many parameters.

Why is this bad?

Functions with lots of parameters are considered bad style and reduce readability (“what does the 5th parameter mean?”). Consider grouping some parameters into a new type.

Example

fn foo(x: u32, y: u32, name: &str, c: Color, w: f32, h: f32, a: f32, b: f32) {
    // ..
}

Configuration

  • too-many-arguments-threshold: The maximum number of argument a function or method can have

    (default: 7)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for functions with a large amount of lines.

Why is this bad?

Functions with a lot of lines are harder to understand due to having to look at a larger amount of code to understand what the function is doing. Consider splitting the body of the function into multiple functions.

Example

fn im_too_long() {
    println!("");
    // ... 100 more LoC
    println!("");
}

Configuration

  • too-many-lines-threshold: The maximum number of lines a function or method can have

    (default: 100)

Applicability: Unspecified(?)
Added in: 1.34.0

What it does

Checks for function arguments and let bindings denoted as ref.

Why is this bad?

The ref declaration makes the function take an owned value, but turns the argument into a reference (which means that the value is destroyed when exiting the function). This adds not much value: either take a reference type, or take an owned value and create references in the body.

For let bindings, let x = &foo; is preferred over let ref x = foo. The type of x is more obvious with the former.

Known problems

If the argument is dereferenced within the function, removing the ref will lead to errors. This can be fixed by removing the dereferences, e.g., changing *x to x within the function.

Example

fn foo(ref _x: u8) {}

Use instead:

fn foo(_x: &u8) {}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Displays a warning when a struct with a trailing zero-sized array is declared without a repr attribute.

Why is this bad?

Zero-sized arrays aren’t very useful in Rust itself, so such a struct is likely being created to pass to C code or in some other situation where control over memory layout matters (for example, in conjunction with manual allocation to make it easy to compute the offset of the array). Either way, #[repr(C)] (or another repr attribute) is needed.

Example

struct RarelyUseful {
    some_field: u32,
    last: [u32; 0],
}

Use instead:

#[repr(C)]
struct MoreOftenUseful {
    some_field: usize,
    last: [u32; 0],
}
Applicability: Unspecified(?)
Added in: 1.58.0

What it does

Checks for cases where generics or trait objects are being used and multiple syntax specifications for trait bounds are used simultaneously.

Why is this bad?

Duplicate bounds makes the code less readable than specifying them only once.

Example

fn func<T: Clone + Default>(arg: T) where T: Clone + Default {}

Use instead:

fn func<T: Clone + Default>(arg: T) {}

// or

fn func<T>(arg: T) where T: Clone + Default {}
fn foo<T: Default + Default>(bar: T) {}

Use instead:

fn foo<T: Default>(bar: T) {}
fn foo<T>(bar: T) where T: Default + Default {}

Use instead:

fn foo<T>(bar: T) where T: Default {}
Applicability: MachineApplicable(?)
Added in: 1.47.0

What it does

Checks for transmutes from a &[u8] to a &str.

Why is this bad?

Not every byte slice is a valid UTF-8 string.

Known problems

  • from_utf8 which this lint suggests using is slower than transmute as it needs to validate the input. If you are certain that the input is always a valid UTF-8, use from_utf8_unchecked which is as fast as transmute but has a semantically meaningful name.
  • You might want to handle errors returned from from_utf8 instead of calling unwrap.

Example

let b: &[u8] = &[1_u8, 2_u8];
unsafe {
    let _: &str = std::mem::transmute(b); // where b: &[u8]
}

// should be:
let _ = std::str::from_utf8(b).unwrap();
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for transmutes from a float to an integer.

Why is this bad?

Transmutes are dangerous and error-prone, whereas to_bits is intuitive and safe.

Example

unsafe {
    let _: u32 = std::mem::transmute(1f32);
}

// should be:
let _: u32 = 1f32.to_bits();
Applicability: Unspecified(?)
Added in: 1.41.0

What it does

Checks for transmutes from an integer to a bool.

Why is this bad?

This might result in an invalid in-memory representation of a bool.

Example

let x = 1_u8;
unsafe {
    let _: bool = std::mem::transmute(x); // where x: u8
}

// should be:
let _: bool = x != 0;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for transmutes from an integer to a char.

Why is this bad?

Not every integer is a Unicode scalar value.

Known problems

  • from_u32 which this lint suggests using is slower than transmute as it needs to validate the input. If you are certain that the input is always a valid Unicode scalar value, use from_u32_unchecked which is as fast as transmute but has a semantically meaningful name.
  • You might want to handle None returned from from_u32 instead of calling unwrap.

Example

let x = 1_u32;
unsafe {
    let _: char = std::mem::transmute(x); // where x: u32
}

// should be:
let _ = std::char::from_u32(x).unwrap();
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for transmutes from an integer to a float.

Why is this bad?

Transmutes are dangerous and error-prone, whereas from_bits is intuitive and safe.

Example

unsafe {
    let _: f32 = std::mem::transmute(1_u32); // where x: u32
}

// should be:
let _: f32 = f32::from_bits(1_u32);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for transmutes from T to NonZero<T>, and suggests the new_unchecked method instead.

Why is this bad?

Transmutes work on any types and thus might cause unsoundness when those types change elsewhere. new_unchecked only works for the appropriate types instead.

Example

let _: NonZero<u32> = unsafe { std::mem::transmute(123) };

Use instead:

let _: NonZero<u32> = unsafe { NonZero::new_unchecked(123) };
Applicability: Unspecified(?)
Added in: 1.69.0

What it does

Checks for null function pointer creation through transmute.

Why is this bad?

Creating a null function pointer is undefined behavior.

More info: https://doc.rust-lang.org/nomicon/ffi.html#the-nullable-pointer-optimization

Known problems

Not all cases can be detected at the moment of this writing. For example, variables which hold a null pointer and are then fed to a transmute call, aren’t detectable yet.

Example

let null_fn: fn() = unsafe { std::mem::transmute( std::ptr::null::<()>() ) };

Use instead:

let null_fn: Option<fn()> = None;
Applicability: Unspecified(?)
Added in: 1.68.0

What it does

Checks for transmutes from a number to an array of u8

Why this is bad?

Transmutes are dangerous and error-prone, whereas to_ne_bytes is intuitive and safe.

Example

unsafe {
    let x: [u8; 8] = std::mem::transmute(1i64);
}

// should be
let x: [u8; 8] = 0i64.to_ne_bytes();
Applicability: Unspecified(?)
Added in: 1.58.0

What it does

Checks for transmutes from a pointer to a pointer, or from a reference to a reference.

Why is this bad?

Transmutes are dangerous, and these can instead be written as casts.

Example

let ptr = &1u32 as *const u32;
unsafe {
    // pointer-to-pointer transmute
    let _: *const f32 = std::mem::transmute(ptr);
    // ref-ref transmute
    let _: &f32 = std::mem::transmute(&1u32);
}
// These can be respectively written:
let _ = ptr as *const f32;
let _ = unsafe{ &*(&1u32 as *const u32 as *const f32) };
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for transmutes from a pointer to a reference.

Why is this bad?

This can always be rewritten with & and *.

Known problems

  • mem::transmute in statics and constants is stable from Rust 1.46.0, while dereferencing raw pointer is not stable yet. If you need to do this in those places, you would have to use transmute instead.

Example

unsafe {
    let _: &T = std::mem::transmute(p); // where p: *const T
}

// can be written:
let _: &T = &*p;

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for transmutes between types which do not have a representation defined relative to each other.

Why is this bad?

The results of such a transmute are not defined.

Known problems

This lint has had multiple problems in the past and was moved to nursery. See issue #8496 for more details.

Example

struct Foo<T>(u32, T);
let _ = unsafe { core::mem::transmute::<Foo<u32>, Foo<i32>>(Foo(0u32, 0u32)) };

Use instead:

#[repr(C)]
struct Foo<T>(u32, T);
let _ = unsafe { core::mem::transmute::<Foo<u32>, Foo<i32>>(Foo(0u32, 0u32)) };
Applicability: Unspecified(?)
Added in: 1.60.0

What it does

Checks for transmutes that could be a pointer cast.

Why is this bad?

Readability. The code tricks people into thinking that something complex is going on.

Example

unsafe { std::mem::transmute::<*const [i32], *const [u16]>(p) };

Use instead:

p as *const [u16];
Applicability: MachineApplicable(?)
Added in: 1.47.0

What it does

Checks for transmute calls which would receive a null pointer.

Why is this bad?

Transmuting a null pointer is undefined behavior.

Known problems

Not all cases can be detected at the moment of this writing. For example, variables which hold a null pointer and are then fed to a transmute call, aren’t detectable yet.

Example

let null_ref: &u64 = unsafe { std::mem::transmute(0 as *const u64) };
Applicability: Unspecified(?)
Added in: 1.35.0

What it does

Warns about calling str::trim (or variants) before str::split_whitespace.

Why is this bad?

split_whitespace already ignores leading and trailing whitespace.

Example

" A B C ".trim().split_whitespace();

Use instead:

" A B C ".split_whitespace();
Applicability: MachineApplicable(?)
Added in: 1.62.0

What it does

Checks for trivial regex creation (with Regex::new, RegexBuilder::new, or RegexSet::new).

Why is this bad?

Matching the regex can likely be replaced by == or str::starts_with, str::ends_with or std::contains or other str methods.

Known problems

If the same regex is going to be applied to multiple inputs, the precomputations done by Regex construction can give significantly better performance than any of the str-based methods.

Example

Regex::new("^foobar")
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for functions taking arguments by reference, where the argument type is Copy and small enough to be more efficient to always pass by value.

Why is this bad?

In many calling conventions instances of structs will be passed through registers if they fit into two or less general purpose registers.

Known problems

This lint is target register size dependent, it is limited to 32-bit to try and reduce portability problems between 32 and 64-bit, but if you are compiling for 8 or 16-bit targets then the limit will be different.

The configuration option trivial_copy_size_limit can be set to override this limit for a project.

This lint attempts to allow passing arguments by reference if a reference to that argument is returned. This is implemented by comparing the lifetime of the argument and return value for equality. However, this can cause false positives in cases involving multiple lifetimes that are bounded by each other.

Also, it does not take account of other similar cases where getting memory addresses matters; namely, returning the pointer to the argument in question, and passing the argument, as both references and pointers, to a function that needs the memory address. For further details, refer to this issue that explains a real case in which this false positive led to an undefined behavior introduced with unsafe code.

Example

fn foo(v: &u32) {}

Use instead:

fn foo(v: u32) {}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

  • trivial-copy-size-limit: The maximum size (in bytes) to consider a Copy type for passing by value instead of by reference. By default there is no limit

    (default: target_pointer_width * 2)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of Err(x)?.

Why restrict this?

The ? operator is designed to allow calls that can fail to be easily chained. For example, foo()?.bar() or foo(bar()?). Because Err(x)? can’t be used that way (it will always return), it is more clear to write return Err(x).

Example

fn foo(fail: bool) -> Result<i32, String> {
    if fail {
      Err("failed")?;
    }
    Ok(0)
}

Could be written:

fn foo(fail: bool) -> Result<i32, String> {
    if fail {
      return Err("failed".into());
    }
    Ok(0)
}
Applicability: MachineApplicable(?)
Added in: 1.38.0

What it does

Checks for tuple<=>array conversions that are not done with .into().

Why is this bad?

It may be unnecessary complexity. .into() works for converting tuples<=> arrays of up to 12 elements and conveys the intent more clearly, while also leaving less room for hard to spot bugs!

Known issues

The suggested code may hide potential asymmetry in some cases. See #11085 for more info.

Example

let t1 = &[(1, 2), (3, 4)];
let v1: Vec<[u32; 2]> = t1.iter().map(|&(a, b)| [a, b]).collect();

Use instead:

let t1 = &[(1, 2), (3, 4)];
let v1: Vec<[u32; 2]> = t1.iter().map(|&t| t.into()).collect();

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: Unspecified(?)
Added in: 1.72.0

What it does

Checks for types used in structs, parameters and let declarations above a certain complexity threshold.

Why is this bad?

Too complex types make the code less readable. Consider using a type definition to simplify them.

Example

struct PointMatrixContainer {
    matrix: Rc<Vec<Vec<Box<(u32, u32, u32, u32)>>>>,
}

fn main() {
    let point_matrix: Vec<Vec<Box<(u32, u32, u32, u32)>>> = vec![
        vec![
            Box::new((1, 2, 3, 4)),
            Box::new((5, 6, 7, 8)),
        ],
        vec![
            Box::new((9, 10, 11, 12)),
        ],
    ];

    let shared_point_matrix: Rc<Vec<Vec<Box<(u32, u32, u32, u32)>>>> = Rc::new(point_matrix);

    let container = PointMatrixContainer {
        matrix: shared_point_matrix,
    };

    // ...
}

Use instead:

Example

type PointMatrix = Vec<Vec<Box<(u32, u32, u32, u32)>>>;
type SharedPointMatrix = Rc<PointMatrix>;

struct PointMatrixContainer {
    matrix: SharedPointMatrix,
}

fn main() {
    let point_matrix: PointMatrix = vec![
        vec![
            Box::new((1, 2, 3, 4)),
            Box::new((5, 6, 7, 8)),
        ],
        vec![
            Box::new((9, 10, 11, 12)),
        ],
    ];

    let shared_point_matrix: SharedPointMatrix = Rc::new(point_matrix);

    let container = PointMatrixContainer {
        matrix: shared_point_matrix,
    };

    // ...
}

Configuration

  • type-complexity-threshold: The maximum complexity a type can have

    (default: 250)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Looks for calls to .type_id() on a Box<dyn _>.

Why is this bad?

This almost certainly does not do what the user expects and can lead to subtle bugs. Calling .type_id() on a Box<dyn Trait> returns a fixed TypeId of the Box itself, rather than returning the TypeId of the underlying type behind the trait object.

For Box<dyn Any> specifically (and trait objects that have Any as its supertrait), this lint will provide a suggestion, which is to dereference the receiver explicitly to go from Box<dyn Any> to dyn Any. This makes sure that .type_id() resolves to a dynamic call on the trait object and not on the box.

If the fixed TypeId of the Box is the intended behavior, it’s better to be explicit about it and write TypeId::of::<Box<dyn Trait>>(): this makes it clear that a fixed TypeId is returned and not the TypeId of the implementor.

Example

use std::any::{Any, TypeId};

let any_box: Box<dyn Any> = Box::new(42_i32);
assert_eq!(any_box.type_id(), TypeId::of::<i32>()); // ⚠️ this fails!

Use instead:

use std::any::{Any, TypeId};

let any_box: Box<dyn Any> = Box::new(42_i32);
assert_eq!((*any_box).type_id(), TypeId::of::<i32>());
//          ^ dereference first, to call `type_id` on `dyn Any`
Applicability: MaybeIncorrect(?)
Added in: 1.73.0

What it does

This lint warns about unnecessary type repetitions in trait bounds

Why is this bad?

Repeating the type for every bound makes the code less readable than combining the bounds

Example

pub fn foo<T>(t: T) where T: Copy, T: Clone {}

Use instead:

pub fn foo<T>(t: T) where T: Copy + Clone {}

Configuration

  • max-trait-bounds: The maximum number of bounds a trait can have to be linted

    (default: 3)

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: Unspecified(?)
Added in: 1.38.0

What it does

Lints subtraction between an Instant and a Duration.

Why is this bad?

Unchecked subtraction could cause underflow on certain platforms, leading to unintentional panics.

Example

let time_passed = Instant::now() - Duration::from_secs(5);

Use instead:

let time_passed = Instant::now().checked_sub(Duration::from_secs(5));

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.67.0

What it does

Checks that there isn’t an infinite recursion in trait implementations.

Why is this bad?

This is a hard to find infinite recursion that will crash any code using it.

Example

enum Foo {
    A,
    B,
}

impl PartialEq for Foo {
    fn eq(&self, other: &Self) -> bool {
        self == other // bad!
    }
}

Use instead:

In such cases, either use #[derive(PartialEq)] or don’t implement it.

Applicability: Unspecified(?)
Added in: 1.77.0

What it does

Checks for unsafe blocks and impls without a // SAFETY: comment explaining why the unsafe operations performed inside the block are safe.

Note the comment must appear on the line(s) preceding the unsafe block with nothing appearing in between. The following is ok:

foo(
    // SAFETY:
    // This is a valid safety comment
    unsafe { *x }
)

But neither of these are:

// SAFETY:
// This is not a valid safety comment
foo(
    /* SAFETY: Neither is this */ unsafe { *x },
);

Why restrict this?

Undocumented unsafe blocks and impls can make it difficult to read and maintain code. Writing out the safety justification may help in discovering unsoundness or bugs.

Example

use std::ptr::NonNull;
let a = &mut 42;

let ptr = unsafe { NonNull::new_unchecked(a) };

Use instead:

use std::ptr::NonNull;
let a = &mut 42;

// SAFETY: references are guaranteed to be non-null.
let ptr = unsafe { NonNull::new_unchecked(a) };

Configuration

  • accept-comment-above-attributes: Whether to accept a safety comment to be placed above the attributes for the unsafe block

    (default: true)

  • accept-comment-above-statement: Whether to accept a safety comment to be placed above the statement containing the unsafe block

    (default: true)

Applicability: Unspecified(?)
Added in: 1.58.0

What it does

Checks for string literals that contain Unicode in a form that is not equal to its NFC-recomposition.

Why is this bad?

If such a string is compared to another, the results may be surprising.

Example

You may not see it, but “à”“ and “à”“ aren’t the same string. The former when escaped is actually "a\u{300}" while the latter is "\u{e0}".

Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of unimplemented!.

Why restrict this?

This macro, or panics in general, may be unwanted in production code.

Example

unimplemented!();
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

It detects references to uninhabited types, such as ! and warns when those are either dereferenced or returned from a function.

Why is this bad?

Dereferencing a reference to an uninhabited type would create an instance of such a type, which cannot exist. This constitutes undefined behaviour. Such a reference could have been created by unsafe code.

Example

The following function can return a reference to an uninhabited type (Infallible) because it uses unsafe code to create it. However, the user of such a function could dereference the return value and trigger an undefined behavior from safe code.

fn create_ref() -> &'static std::convert::Infallible {
    unsafe { std::mem::transmute(&()) }
}
Applicability: Unspecified(?)
Added in: 1.76.0

What it does

Checks for MaybeUninit::uninit().assume_init().

Why is this bad?

For most types, this is undefined behavior.

Known problems

For now, we accept empty tuples and tuples / arrays of MaybeUninit. There may be other types that allow uninitialized data, but those are not yet rigorously defined.

Example

// Beware the UB
use std::mem::MaybeUninit;

let _: usize = unsafe { MaybeUninit::uninit().assume_init() };

Note that the following is OK:

use std::mem::MaybeUninit;

let _: [MaybeUninit<bool>; 5] = unsafe {
    MaybeUninit::uninit().assume_init()
};
Applicability: Unspecified(?)
Added in: 1.39.0

What it does

Checks for set_len() call that creates Vec with uninitialized elements. This is commonly caused by calling set_len() right after allocating or reserving a buffer with new(), default(), with_capacity(), or reserve().

Why is this bad?

It creates a Vec with uninitialized data, which leads to undefined behavior with most safe operations. Notably, uninitialized Vec<u8> must not be used with generic Read.

Moreover, calling set_len() on a Vec created with new() or default() creates out-of-bound values that lead to heap memory corruption when used.

Known Problems

This lint only checks directly adjacent statements.

Example

let mut vec: Vec<u8> = Vec::with_capacity(1000);
unsafe { vec.set_len(1000); }
reader.read(&mut vec); // undefined behavior!

How to fix?

  1. Use an initialized buffer:
    let mut vec: Vec<u8> = vec![0; 1000];
    reader.read(&mut vec);
    
  2. Wrap the content in MaybeUninit:
    let mut vec: Vec<MaybeUninit<T>> = Vec::with_capacity(1000);
    vec.set_len(1000);  // `MaybeUninit` can be uninitialized
    
  3. If you are on 1.60.0 or later, Vec::spare_capacity_mut() is available:
    let mut vec: Vec<u8> = Vec::with_capacity(1000);
    let remaining = vec.spare_capacity_mut();  // `&mut [MaybeUninit<u8>]`
    // perform initialization with `remaining`
    vec.set_len(...);  // Safe to call `set_len()` on initialized part
    
Applicability: Unspecified(?)
Added in: 1.58.0

What it does

Detect when a variable is not inlined in a format string, and suggests to inline it.

Why is this bad?

Non-inlined code is slightly more difficult to read and understand, as it requires arguments to be matched against the format string. The inlined syntax, where allowed, is simpler.

Example

format!("{}", var);
format!("{v:?}", v = var);
format!("{0} {0}", var);
format!("{0:1$}", var, width);
format!("{:.*}", prec, var);

Use instead:

format!("{var}");
format!("{var:?}");
format!("{var} {var}");
format!("{var:width$}");
format!("{var:.prec$}");

If allow-mixed-uninlined-format-args is set to false in clippy.toml, the following code will also trigger the lint:

format!("{} {}", var, 1+2);

Use instead:

format!("{var} {}", 1+2);

Known Problems

If a format string contains a numbered argument that cannot be inlined nothing will be suggested, e.g. println!("{0}={1}", var, 1+2).

Configuration

  • allow-mixed-uninlined-format-args: Whether to allow mixed uninlined format args, e.g. format!("{} {}", a, foo.bar)

    (default: true)

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.66.0

What it does

Checks for passing a unit value as an argument to a function without using a unit literal (()).

Why is this bad?

This is likely the result of an accidental semicolon.

Example

foo({
    let a = bar();
    baz(a);
})
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for comparisons to unit. This includes all binary comparisons (like == and <) and asserts.

Why is this bad?

Unit is always equal to itself, and thus is just a clumsily written constant. Mostly this happens when someone accidentally adds semicolons at the end of the operands.

Example

if {
    foo();
} == {
    bar();
} {
    baz();
}

is equal to

{
    foo();
    bar();
    baz();
}

For asserts:

assert_eq!({ foo(); }, { bar(); });

will always succeed

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Detects ().hash(_).

Why is this bad?

Hashing a unit value doesn’t do anything as the implementation of Hash for () is a no-op.

Example

match my_enum {
	Empty => ().hash(&mut state),
	WithValue(x) => x.hash(&mut state),
}

Use instead:

match my_enum {
	Empty => 0_u8.hash(&mut state),
	WithValue(x) => x.hash(&mut state),
}
Applicability: MaybeIncorrect(?)
Added in: 1.58.0

What it does

Checks for functions that expect closures of type Fn(…) -> Ord where the implemented closure returns the unit type. The lint also suggests to remove the semi-colon at the end of the statement if present.

Why is this bad?

Likely, returning the unit type is unintentional, and could simply be caused by an extra semi-colon. Since () implements Ord it doesn’t cause a compilation error. This is the same reasoning behind the unit_cmp lint.

Known problems

If returning unit is intentional, then there is no way of specifying this without triggering needless_return lint

Example

let mut twins = vec![(1, 1), (2, 2)];
twins.sort_by_key(|x| { x.1; });
Applicability: Unspecified(?)
Added in: 1.47.0

What it does

Checks for a return type containing a Box<T> where T implements Sized

The lint ignores Box<T> where T is larger than unnecessary_box_size, as returning a large T directly may be detrimental to performance.

Why is this bad?

It’s better to just return T in these cases. The caller may not need the value to be boxed, and it’s expensive to free the memory once the Box<T> been dropped.

Example

fn foo() -> Box<String> {
    Box::new(String::from("Hello, world!"))
}

Use instead:

fn foo() -> String {
    String::from("Hello, world!")
}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

  • unnecessary-box-size: The byte size a T in Box<T> can have, below which it triggers the clippy::unnecessary_box lint

    (default: 128)

Applicability: Unspecified(?)
Added in: 1.70.0

What it does

Checks for casts to the same type, casts of int literals to integer types, casts of float literals to float types, and casts between raw pointers that don’t change type or constness.

Why is this bad?

It’s just unnecessary.

Known problems

When the expression on the left is a function call, the lint considers the return type to be a type alias if it’s aliased through a use statement (like use std::io::Result as IoResult). It will not lint such cases.

This check will only work on primitive types without any intermediate references: raw pointers and trait objects may or may not work.

Example

let _ = 2i32 as i32;
let _ = 0.5 as f32;

Better:

let _ = 2_i32;
let _ = 0.5_f32;
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for #[cfg_attr(clippy, allow(clippy::lint))] and suggests to replace it with #[allow(clippy::lint)].

Why is this bad?

There is no reason to put clippy attributes behind a clippy cfg as they are not run by anything else than clippy.

Example

#![cfg_attr(clippy, allow(clippy::deprecated_cfg_attr))]

Use instead:

#![allow(clippy::deprecated_cfg_attr)]
Applicability: MachineApplicable(?)
Added in: 1.78.0

What it does

Checks for calls to TryInto::try_into and TryFrom::try_from when their infallible counterparts could be used.

Why is this bad?

In those cases, the TryInto and TryFrom trait implementation is a blanket impl that forwards to Into or From, which always succeeds. The returned Result<_, Infallible> requires error handling to get the contained value even though the conversion can never fail.

Example

let _: Result<i64, _> = 1i32.try_into();
let _: Result<i64, _> = <_>::try_from(1i32);

Use from/into instead:

let _: i64 = 1i32.into();
let _: i64 = <_>::from(1i32);
Applicability: MachineApplicable(?)
Added in: 1.75.0

What it does

Checks for filter_map calls that could be replaced by filter or map. More specifically it checks if the closure provided is only performing one of the filter or map operations and suggests the appropriate option.

Why is this bad?

Complexity. The intent is also clearer if only a single operation is being performed.

Example

let _ = (0..3).filter_map(|x| if x > 2 { Some(x) } else { None });

// As there is no transformation of the argument this could be written as:
let _ = (0..3).filter(|&x| x > 2);
let _ = (0..4).filter_map(|x| Some(x + 1));

// As there is no conditional check on the argument this could be written as:
let _ = (0..4).map(|x| x + 1);
Applicability: MaybeIncorrect(?)
Added in: 1.31.0

What it does

Checks for find_map calls that could be replaced by find or map. More specifically it checks if the closure provided is only performing one of the find or map operations and suggests the appropriate option.

Why is this bad?

Complexity. The intent is also clearer if only a single operation is being performed.

Example

let _ = (0..3).find_map(|x| if x > 2 { Some(x) } else { None });

// As there is no transformation of the argument this could be written as:
let _ = (0..3).find(|&x| x > 2);
let _ = (0..4).find_map(|x| Some(x + 1));

// As there is no conditional check on the argument this could be written as:
let _ = (0..4).map(|x| x + 1).next();
Applicability: MaybeIncorrect(?)
Added in: 1.61.0

What it does

Checks the usage of .first().is_some() or .first().is_none() to check if a slice is empty.

Why is this bad?

Using .is_empty() is shorter and better communicates the intention.

Example

let v = vec![1, 2, 3];
if v.first().is_none() {
    // The vector is empty...
}

Use instead:

let v = vec![1, 2, 3];
if v.is_empty() {
    // The vector is empty...
}
Applicability: MachineApplicable(?)
Added in: 1.83.0

What it does

Checks for usage of fold when a more succinct alternative exists. Specifically, this checks for folds which could be replaced by any, all, sum or product.

Why is this bad?

Readability.

Example

(0..3).fold(false, |acc, x| acc || x > 2);

Use instead:

(0..3).any(|x| x > 2);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks the usage of .get().is_some() or .get().is_none() on std map types.

Why is this bad?

It can be done in one call with .contains()/.contains_key().

Example

let s: HashSet<String> = HashSet::new();
if s.get("a").is_some() {
    // code
}

Use instead:

let s: HashSet<String> = HashSet::new();
if s.contains("a") {
    // code
}
Applicability: MaybeIncorrect(?)
Added in: 1.78.0

What it does

Checks for usage of .collect::<Vec<String>>().join("") on iterators.

Why is this bad?

.collect::<String>() is more concise and might be more performant

Example

let vector = vec!["hello",  "world"];
let output = vector.iter().map(|item| item.to_uppercase()).collect::<Vec<String>>().join("");
println!("{}", output);

The correct use would be:

let vector = vec!["hello",  "world"];
let output = vector.iter().map(|item| item.to_uppercase()).collect::<String>();
println!("{}", output);

Known problems

While .collect::<String>() is sometimes more performant, there are cases where using .collect::<String>() over .collect::<Vec<String>>().join("") will prevent loop unrolling and will result in a negative performance impact.

Additionally, differences have been observed between aarch64 and x86_64 assembly output, with aarch64 tending to producing faster assembly in more cases when using .collect::<String>()

Applicability: MachineApplicable(?)
Added in: 1.61.0

What it does

As the counterpart to or_fun_call, this lint looks for unnecessary lazily evaluated closures on Option and Result.

This lint suggests changing the following functions, when eager evaluation results in simpler code:

  • unwrap_or_else to unwrap_or
  • and_then to and
  • or_else to or
  • get_or_insert_with to get_or_insert
  • ok_or_else to ok_or
  • then to then_some (for msrv >= 1.62.0)

Why is this bad?

Using eager evaluation is shorter and simpler in some cases.

Known problems

It is possible, but not recommended for Deref and Index to have side effects. Eagerly evaluating them can change the semantics of the program.

Example

// example code where clippy issues a warning
let opt: Option<u32> = None;

opt.unwrap_or_else(|| 42);

Use instead:

let opt: Option<u32> = None;

opt.unwrap_or(42);

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.48.0

What it does

Detects functions that are written to return &str that could return &'static str but instead return a &'a str.

Why is this bad?

This leaves the caller unable to use the &str as &'static str, causing unneccessary allocations or confusion. This is also most likely what you meant to write.

Example

impl MyType {
    fn returns_literal(&self) -> &str {
        "Literal"
    }
}

Use instead:

impl MyType {
    fn returns_literal(&self) -> &'static str {
        "Literal"
    }
}

Or, in case you may return a non-literal str in future:

impl MyType {
    fn returns_literal<'a>(&'a self) -> &'a str {
        "Literal"
    }
}
Applicability: MachineApplicable(?)
Added in: 1.83.0

What it does

Checks for .unwrap() related calls on Results and Options that are constructed.

Why is this bad?

It is better to write the value directly without the indirection.

Examples

let val1 = Some(1).unwrap();
let val2 = Ok::<_, ()>(1).unwrap();
let val3 = Err::<(), _>(1).unwrap_err();

Use instead:

let val1 = 1;
let val2 = 1;
let val3 = 1;
Applicability: MachineApplicable(?)
Added in: 1.72.0

What it does

Suggests removing the use of a map() (or map_err()) method when an Option or Result is being constructed.

Why is this bad?

It introduces unnecessary complexity. Instead, the function can be called before constructing the Option or Result from its return value.

Example

Some(4).map(i32::swap_bytes)

Use instead:

Some(i32::swap_bytes(4))
Applicability: MachineApplicable(?)
Added in: 1.74.0

What it does

Converts some constructs mapping an Enum value for equality comparison.

Why is this bad?

Calls such as opt.map_or(false, |val| val == 5) are needlessly long and cumbersome, and can be reduced to, for example, opt == Some(5) assuming opt implements PartialEq. This lint offers readability and conciseness improvements.

Example

pub fn a(x: Option<i32>) -> bool {
    x.map_or(false, |n| n == 5)
}

Use instead:

pub fn a(x: Option<i32>) -> bool {
    x == Some(5)
}
Applicability: MaybeIncorrect(?)
Added in: 1.75.0

What it does

Checks for unnecessary calls to min() or max() in the following cases

  • Either both side is constant
  • One side is clearly larger than the other, like i32::MIN and an i32 variable

Why is this bad?

In the aformentioned cases it is not necessary to call min() or max() to compare values, it may even cause confusion.

Example

let _ = 0.min(7_u32);

Use instead:

let _ = 0;
Applicability: MachineApplicable(?)
Added in: 1.81.0

What it does

Detects passing a mutable reference to a function that only requires an immutable reference.

Why is this bad?

The mutable reference rules out all other references to the value. Also the code misleads about the intent of the call site.

Example

vec.push(&mut value);

Use instead:

vec.push(&value);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for expression statements that can be reduced to a sub-expression.

Why is this bad?

Expressions by themselves often have no side-effects. Having such expressions reduces readability.

Example

compute_array()[0];
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Detects cases of owned empty strings being passed as an argument to a function expecting &str

Why is this bad?

This results in longer and less readable code

Example

vec!["1", "2", "3"].join(&String::new());

Use instead:

vec!["1", "2", "3"].join("");
Applicability: MachineApplicable(?)
Added in: 1.62.0

What it does

Checks for usage of .map_or_else() “map closure” for Result type.

Why is this bad?

This can be written more concisely by using unwrap_or_else().

Example

let x: Result<u32, ()> = Ok(0);
let y = x.map_or_else(|err| handle_error(err), |n| n);

Use instead:

let x: Result<u32, ()> = Ok(0);
let y = x.unwrap_or_else(|err| handle_error(err));
Applicability: MachineApplicable(?)
Added in: 1.78.0

What it does

Checks for // SAFETY: comments on safe code.

Why restrict this?

Safe code has no safety requirements, so there is no need to describe safety invariants.

Example

use std::ptr::NonNull;
let a = &mut 42;

// SAFETY: references are guaranteed to be non-null.
let ptr = NonNull::new(a).unwrap();

Use instead:

use std::ptr::NonNull;
let a = &mut 42;

let ptr = NonNull::new(a).unwrap();
Applicability: Unspecified(?)
Added in: 1.67.0

What it does

Checks for the doc comments of publicly visible safe functions and traits and warns if there is a # Safety section.

Why restrict this?

Safe functions and traits are safe to implement and therefore do not need to describe safety preconditions that users are required to uphold.

Examples

/// # Safety
///
/// This function should not be called before the horsemen are ready.
pub fn start_apocalypse_but_safely(u: &mut Universe) {
    unimplemented!();
}

The function is safe, so there shouldn’t be any preconditions that have to be explained for safety reasons.

/// This function should really be documented
pub fn start_apocalypse(u: &mut Universe) {
    unimplemented!();
}

Configuration

  • check-private-items: Whether to also run the listed lints on private items.

    (default: false)

Applicability: Unspecified(?)
Added in: 1.67.0

What it does

Checks for imports ending in ::{self}.

Why restrict this?

In most cases, this can be written much more cleanly by omitting ::{self}.

Known problems

Removing ::{self} will cause any non-module items at the same path to also be imported. This might cause a naming conflict (https://github.com/rust-lang/rustfmt/issues/3568). This lint makes no attempt to detect this scenario and that is why it is a restriction lint.

Example

use std::io::{self};

Use instead:

use std::io;
Applicability: MaybeIncorrect(?)
Added in: 1.53.0

What it does

Checks for usage of Vec::sort_by passing in a closure which compares the two arguments, either directly or indirectly.

Why is this bad?

It is more clear to use Vec::sort_by_key (or Vec::sort if possible) than to use Vec::sort_by and a more complicated closure.

Known problems

If the suggested Vec::sort_by_key uses Reverse and it isn’t already imported by a use statement, then it will need to be added manually.

Example

vec.sort_by(|a, b| a.foo().cmp(&b.foo()));

Use instead:

vec.sort_by_key(|a| a.foo());
Applicability: MachineApplicable(?)
Added in: 1.46.0

What it does

Checks for initialization of an identical struct from another instance of the type, either by copying a base without setting any field or by moving all fields individually.

Why is this bad?

Readability suffers from unnecessary struct building.

Example

struct S { s: String }

let a = S { s: String::from("Hello, world!") };
let b = S { ..a };

Use instead:

struct S { s: String }

let a = S { s: String::from("Hello, world!") };
let b = a;

The struct literal S { ..a } in the assignment to b could be replaced with just a.

Known Problems

Has false positives when the base is a place expression that cannot be moved out of, see #10547.

Empty structs are ignored by the lint.

Applicability: MachineApplicable(?)
Added in: 1.70.0

What it does

Checks for unnecessary calls to ToOwned::to_owned and other to_owned-like functions.

Why is this bad?

The unnecessary calls result in useless allocations.

Known problems

unnecessary_to_owned can falsely trigger if IntoIterator::into_iter is applied to an owned copy of a resource and the resource is later used mutably. See #8148.

Example

let path = std::path::Path::new("x");
foo(&path.to_string_lossy().to_string());
fn foo(s: &str) {}

Use instead:

let path = std::path::Path::new("x");
foo(&path.to_string_lossy());
fn foo(s: &str) {}
Applicability: MachineApplicable(?)
Added in: 1.59.0

What it does

Checks for calls of unwrap[_err]() that cannot fail.

Why is this bad?

Using if let or match is more idiomatic.

Example

if option.is_some() {
    do_something_with(option.unwrap())
}

Could be written:

if let Some(value) = option {
    do_something_with(value)
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for private functions that only return Ok or Some.

Why is this bad?

It is not meaningful to wrap values when no None or Err is returned.

Known problems

There can be false positives if the function signature is designed to fit some external requirement.

Example

fn get_cool_number(a: bool, b: bool) -> Option<i32> {
    if a && b {
        return Some(50);
    }
    if a {
        Some(0)
    } else {
        Some(10)
    }
}

Use instead:

fn get_cool_number(a: bool, b: bool) -> i32 {
    if a && b {
        return 50;
    }
    if a {
        0
    } else {
        10
    }
}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: MaybeIncorrect(?)
Added in: 1.50.0

What it does

Checks for structure field patterns bound to wildcards.

Why restrict this?

Using .. instead is shorter and leaves the focus on the fields that are actually bound.

Example

let f = Foo { a: 0, b: 0, c: 0 };

match f {
    Foo { a: _, b: 0, .. } => {},
    Foo { a: _, b: _, c: _ } => {},
}

Use instead:

let f = Foo { a: 0, b: 0, c: 0 };

match f {
    Foo { b: 0, .. } => {},
    Foo { .. } => {},
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for tuple patterns with a wildcard pattern (_) is next to a rest pattern (..).

NOTE: While _, .. means there is at least one element left, .. means there are 0 or more elements left. This can make a difference when refactoring, but shouldn’t result in errors in the refactored code, since the wildcard pattern isn’t used anyway.

Why is this bad?

The wildcard pattern is unneeded as the rest pattern can match that element as well.

Example

match t {
    TupleStruct(0, .., _) => (),
    _ => (),
}

Use instead:

match t {
    TupleStruct(0, ..) => (),
    _ => (),
}
Applicability: MachineApplicable(?)
Added in: 1.40.0

What it does

Checks for unnested or-patterns, e.g., Some(0) | Some(2) and suggests replacing the pattern with a nested one, Some(0 | 2).

Another way to think of this is that it rewrites patterns in disjunctive normal form (DNF) into conjunctive normal form (CNF).

Why is this bad?

In the example above, Some is repeated, which unnecessarily complicates the pattern.

Example

fn main() {
    if let Some(0) | Some(2) = Some(0) {}
}

Use instead:

fn main() {
    if let Some(0 | 2) = Some(0) {}
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.46.0

What it does

Checks for usage of unreachable!.

Why restrict this?

This macro, or panics in general, may be unwanted in production code.

Example

unreachable!();
Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Warns if a long integral or floating-point constant does not contain underscores.

Why is this bad?

Reading long numbers is difficult without separators.

Example

61864918973511

Use instead:

61_864_918_973_511

Configuration

  • unreadable-literal-lint-fractions: Should the fraction of a decimal be linted to include separators.

    (default: true)

Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for deriving serde::Deserialize on a type that has methods using unsafe.

Why is this bad?

Deriving serde::Deserialize will create a constructor that may violate invariants held by another constructor.

Example

use serde::Deserialize;

#[derive(Deserialize)]
pub struct Foo {
    // ..
}

impl Foo {
    pub fn new() -> Self {
        // setup here ..
    }

    pub unsafe fn parts() -> (&str, &str) {
        // assumes invariants hold
    }
}
Applicability: Unspecified(?)
Added in: 1.45.0

What it does

Checks for imports that remove “unsafe” from an item’s name.

Why is this bad?

Renaming makes it less clear which traits and structures are unsafe.

Example

use std::cell::{UnsafeCell as TotallySafeCell};

extern crate crossbeam;
use crossbeam::{spawn_unsafe as spawn};
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

The suggested alternative could be substantially slower.

Applicability: Unspecified(?)
Deprecated in: pre 1.29.0

What it does

Warns if literal suffixes are not separated by an underscore. To enforce unseparated literal suffix style, see the separated_literal_suffix lint.

Why restrict this?

Suffix style should be consistent.

Example

123832i32

Use instead:

123832_i32
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for transmutes between collections whose types have different ABI, size or alignment.

Why is this bad?

This is undefined behavior.

Known problems

Currently, we cannot know whether a type is a collection, so we just lint the ones that come with std.

Example

// different size, therefore likely out-of-bounds memory access
// You absolutely do not want this in your code!
unsafe {
    std::mem::transmute::<_, Vec<u32>>(vec![2_u16])
};

You must always iterate, map and collect the values:

vec![2_u16].into_iter().map(u32::from).collect::<Vec<_>>();
Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

Vec::as_mut_slice is now stable.

Applicability: Unspecified(?)
Deprecated in: pre 1.29.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

Vec::as_slice is now stable.

Applicability: Unspecified(?)
Deprecated in: pre 1.29.0

What it does

Checks for functions that are declared async but have no .awaits inside of them.

Why is this bad?

Async functions with no async code create overhead, both mentally and computationally. Callers of async methods either need to be calling from an async function themselves or run it on an executor, both of which causes runtime overhead and hassle for the caller.

Example

async fn get_random_number() -> i64 {
    4 // Chosen by fair dice roll. Guaranteed to be random.
}
let number_future = get_random_number();

Use instead:

fn get_random_number_improved() -> i64 {
    4 // Chosen by fair dice roll. Guaranteed to be random.
}
let number_future = async { get_random_number_improved() };
Applicability: Unspecified(?)
Added in: 1.54.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

Iterator::collect is now marked as #[must_use].

Applicability: Unspecified(?)
Deprecated in: 1.39.0

What it does

Checks for uses of the enumerate method where the index is unused (_)

Why is this bad?

The index from .enumerate() is immediately dropped.

Example

let v = vec![1, 2, 3, 4];
for (_, x) in v.iter().enumerate() {
    println!("{x}");
}

Use instead:

let v = vec![1, 2, 3, 4];
for x in v.iter() {
    println!("{x}");
}
Applicability: MachineApplicable(?)
Added in: 1.75.0

What it does

Detects formatting parameters that have no effect on the output of format!(), println!() or similar macros.

Why is this bad?

Shorter format specifiers are easier to read, it may also indicate that an expected formatting operation such as adding padding isn’t happening.

Example

println!("{:.}", 1.0);

println!("not padded: {:5}", format_args!("..."));

Use instead:

println!("{}", 1.0);

println!("not padded: {}", format_args!("..."));
// OR
println!("padded: {:5}", format!("..."));
Applicability: MaybeIncorrect(?)
Added in: 1.66.0

What it does

Checks for unused written/read amount.

Why is this bad?

io::Write::write(_vectored) and io::Read::read(_vectored) are not guaranteed to process the entire buffer. They return how many bytes were processed, which might be smaller than a given buffer’s length. If you don’t need to deal with partial-write/read, use write_all/read_exact instead.

When working with asynchronous code (either with the futures crate or with tokio), a similar issue exists for AsyncWriteExt::write() and AsyncReadExt::read() : these functions are also not guaranteed to process the entire buffer. Your code should either handle partial-writes/reads, or call the write_all/read_exact methods on those traits instead.

Known problems

Detects only common patterns.

Examples

use std::io;
fn foo<W: io::Write>(w: &mut W) -> io::Result<()> {
    w.write(b"foo")?;
    Ok(())
}

Use instead:

use std::io;
fn foo<W: io::Write>(w: &mut W) -> io::Result<()> {
    w.write_all(b"foo")?;
    Ok(())
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for the creation of a peekable iterator that is never .peek()ed

Why is this bad?

Creating a peekable iterator without using any of its methods is likely a mistake, or just a leftover after a refactor.

Example

let collection = vec![1, 2, 3];
let iter = collection.iter().peekable();

for item in iter {
    // ...
}

Use instead:

let collection = vec![1, 2, 3];
let iter = collection.iter();

for item in iter {
    // ...
}
Applicability: Unspecified(?)
Added in: 1.65.0

What it does

Checks for calls to Result::ok() without using the returned Option.

Why is this bad?

Using Result::ok() may look like the result is checked like unwrap or expect would do but it only silences the warning caused by #[must_use] on the Result.

Example

some_function().ok();

Use instead:

let _ = some_function();
Applicability: MaybeIncorrect(?)
Added in: 1.82.0

What it does

Detects cases where a whole-number literal float is being rounded, using the floor, ceil, or round methods.

Why is this bad?

This is unnecessary and confusing to the reader. Doing this is probably a mistake.

Example

let x = 1f32.ceil();

Use instead:

let x = 1f32;
Applicability: MachineApplicable(?)
Added in: 1.63.0

What it does

Checks methods that contain a self argument but don’t use it

Why is this bad?

It may be clearer to define the method as an associated function instead of an instance method if it doesn’t require self.

Example

struct A;
impl A {
    fn method(&self) {}
}

Could be written:

struct A;
impl A {
    fn method() {}
}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: 1.40.0

What it does

Checks for use Trait where the Trait is only used for its methods and not referenced by a path directly.

Why is this bad?

Traits imported that aren’t used directly can be imported anonymously with use Trait as _. It is more explicit, avoids polluting the current scope with unused names and can be useful to show which imports are required for traits.

Example

use std::fmt::Write;

fn main() {
    let mut s = String::new();
    let _ = write!(s, "hello, world!");
    println!("{s}");
}

Use instead:

use std::fmt::Write as _;

fn main() {
    let mut s = String::new();
    let _ = write!(s, "hello, world!");
    println!("{s}");
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: 1.83.0

What it does

Checks for unit (()) expressions that can be removed.

Why is this bad?

Such expressions add no value, but can make the code less readable. Depending on formatting they can make a break or return statement look like a function call.

Example

fn return_unit() -> () {
    ()
}

is equivalent to

fn return_unit() {}
Applicability: MachineApplicable(?)
Added in: 1.31.0

What it does

Warns if hexadecimal or binary literals are not grouped by nibble or byte.

Why is this bad?

Negatively impacts readability.

Example

let x: u32 = 0xFFF_FFF;
let y: u8 = 0b01_011_101;
Applicability: MaybeIncorrect(?)
Added in: 1.49.0

What it does

Checks for functions of type Result that contain expect() or unwrap()

Why restrict this?

These functions promote recoverable errors to non-recoverable errors, which may be undesirable in code bases which wish to avoid panics, or be a bug in the specific function.

Known problems

This can cause false positives in functions that handle both recoverable and non recoverable errors.

Example

Before:

fn divisible_by_3(i_str: String) -> Result<(), String> {
    let i = i_str
        .parse::<i32>()
        .expect("cannot divide the input by three");

    if i % 3 != 0 {
        Err("Number is not divisible by 3")?
    }

    Ok(())
}

After:

fn divisible_by_3(i_str: String) -> Result<(), String> {
    let i = i_str
        .parse::<i32>()
        .map_err(|e| format!("cannot divide the input by three: {}", e))?;

    if i % 3 != 0 {
        Err("Number is not divisible by 3")?
    }

    Ok(())
}
Applicability: Unspecified(?)
Added in: 1.48.0

What it does

Checks for usages of the following functions with an argument that constructs a default value (e.g., Default::default or String::new):

  • unwrap_or
  • unwrap_or_else
  • or_insert
  • or_insert_with

Why is this bad?

Readability. Using unwrap_or_default in place of unwrap_or/unwrap_or_else, or or_default in place of or_insert/or_insert_with, is simpler and more concise.

Known problems

In some cases, the argument of unwrap_or, etc. is needed for type inference. The lint uses a heuristic to try to identify such cases. However, the heuristic can produce false negatives.

Examples

x.unwrap_or(Default::default());
map.entry(42).or_insert_with(String::new);

Use instead:

x.unwrap_or_default();
map.entry(42).or_default();

Past names

  • unwrap_or_else_default
Applicability: MachineApplicable(?)
Added in: 1.56.0

What it does

Checks for .unwrap() or .unwrap_err() calls on Results and .unwrap() call on Options.

Why restrict this?

It is better to handle the None or Err case, or at least call .expect(_) with a more helpful message. Still, for a lot of quick-and-dirty code, unwrap is a good choice, which is why this lint is Allow by default.

result.unwrap() will let the thread panic on Err values. Normally, you want to implement more sophisticated error handling, and propagate errors upwards with ? operator.

Even if you want to panic on errors, not all Errors implement good messages on display. Therefore, it may be beneficial to look at the places where they may get displayed. Activate this lint to do just that.

Examples

option.unwrap();
result.unwrap();

Use instead:

option.expect("more helpful message");
result.expect("more helpful message");

If expect_used is enabled, instead:

option?;

// or

result?;

Past names

  • option_unwrap_used
  • result_unwrap_used

Configuration

  • allow-unwrap-in-tests: Whether unwrap should be allowed in test functions or #[cfg(test)]

    (default: false)

Applicability: Unspecified(?)
Added in: 1.45.0

What it does

Checks for fully capitalized names and optionally names containing a capitalized acronym.

Why is this bad?

In CamelCase, acronyms count as one word. See naming conventions for more.

By default, the lint only triggers on fully-capitalized names. You can use the upper-case-acronyms-aggressive: true config option to enable linting on all camel case names

Known problems

When two acronyms are contiguous, the lint can’t tell where the first acronym ends and the second starts, so it suggests to lowercase all of the letters in the second acronym.

Example

struct HTTPResponse;

Use instead:

struct HttpResponse;

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

  • upper-case-acronyms-aggressive: Enables verbose mode. Triggers if there is more than one uppercase char next to each other

    (default: false)

Applicability: MaybeIncorrect(?)
Added in: 1.51.0

What it does

Checks for usage of Debug formatting. The purpose of this lint is to catch debugging remnants.

Why restrict this?

The purpose of the Debug trait is to facilitate debugging Rust code, and no guarantees are made about its output. It should not be used in user-facing output.

Example

println!("{:?}", foo);
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for unnecessary repetition of structure name when a replacement with Self is applicable.

Why is this bad?

Unnecessary repetition. Mixed use of Self and struct name feels inconsistent.

Known problems

  • Unaddressed false negative in fn bodies of trait implementations

Example

struct Foo;
impl Foo {
    fn new() -> Foo {
        Foo {}
    }
}

could be

struct Foo;
impl Foo {
    fn new() -> Self {
        Self {}
    }
}

Configuration

  • msrv: The minimum rust version that the project supports. Defaults to the rust-version field in Cargo.toml

    (default: current version)

Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for the use of bindings with a single leading underscore.

Why is this bad?

A single leading underscore is usually used to indicate that a binding will not be used. Using such a binding breaks this expectation.

Known problems

The lint does not work properly with desugaring and macro, it has been allowed in the meantime.

Example

let _x = 0;
let y = _x + 1; // Here we are using `_x`, even though it has a leading
                // underscore. We should rename `_x` to `x`
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for the use of item with a single leading underscore.

Why is this bad?

A single leading underscore is usually used to indicate that a item will not be used. Using such a item breaks this expectation.

Example

fn _foo() {}

struct _FooStruct {}

fn main() {
    _foo();
    let _ = _FooStruct{};
}

Use instead:

fn foo() {}

struct FooStruct {}

fn main() {
    foo();
    let _ = FooStruct{};
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of .as_ref() or .as_mut() where the types before and after the call are the same.

Why is this bad?

The call is unnecessary.

Example

let x: &[i32] = &[1, 2, 3, 4, 5];
do_stuff(x.as_ref());

The correct use would be:

let x: &[i32] = &[1, 2, 3, 4, 5];
do_stuff(x);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for extern crate and use items annotated with lint attributes.

This lint permits lint attributes for lints emitted on the items themself. For use items these lints are:

  • ambiguous_glob_reexports
  • dead_code
  • deprecated
  • hidden_glob_reexports
  • unreachable_pub
  • unused
  • unused_braces
  • unused_import_braces
  • clippy::disallowed_types
  • clippy::enum_glob_use
  • clippy::macro_use_imports
  • clippy::module_name_repetitions
  • clippy::redundant_pub_crate
  • clippy::single_component_path_imports
  • clippy::unsafe_removed_from_name
  • clippy::wildcard_imports

For extern crate items these lints are:

  • unused_imports on items with #[macro_use]

Why is this bad?

Lint attributes have no effect on crate imports. Most likely a ! was forgotten.

Example

#[deny(dead_code)]
extern crate foo;
#[forbid(dead_code)]
use foo::bar;

Use instead:

#[allow(unused_imports)]
use foo::baz;
#[allow(unused_imports)]
#[macro_use]
extern crate baz;
Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for Into, TryInto, From, TryFrom, or IntoIter calls which uselessly convert to the same type.

Why is this bad?

Redundant code.

Example

// format!() returns a `String`
let s: String = format!("hello").into();

Use instead:

let s: String = format!("hello");

Past names

  • identity_conversion
Applicability: MachineApplicable(?)
Added in: 1.45.0

What it does

Checks for the use of format!("string literal with no argument") and format!("{}", foo) where foo is a string.

Why is this bad?

There is no point of doing that. format!("foo") can be replaced by "foo".to_owned() if you really need a String. The even worse &format!("foo") is often encountered in the wild. format!("{}", foo) can be replaced by foo.clone() if foo: String or foo.to_owned() if foo: &str.

Examples

let foo = "foo";
format!("{}", foo);

Use instead:

let foo = "foo";
foo.to_owned();
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for variable declarations immediately followed by a conditional affectation.

Why is this bad?

This is not idiomatic Rust.

Example

let foo;

if bar() {
    foo = 42;
} else {
    foo = 0;
}

let mut baz = None;

if bar() {
    baz = Some(42);
}

should be written

let foo = if bar() {
    42
} else {
    0
};

let baz = if bar() {
    Some(42)
} else {
    None
};
Applicability: HasPlaceholders(?)
Added in: pre 1.29.0

What it does

Checks for transmutes to the original type of the object and transmutes that could be a cast.

Why is this bad?

Readability. The code tricks people into thinking that something complex is going on.

Example

core::intrinsics::transmute(t); // where the result type is the same as `t`'s
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for usage of vec![..] when using [..] would be possible.

Why is this bad?

This is less efficient.

Example

fn foo(_x: &[u8]) {}

foo(&vec![1, 2]);

Use instead:

foo(&[1, 2]);

Configuration

  • allow-useless-vec-in-tests: Whether useless_vec should ignore test functions or #[cfg(test)]

    (default: false)

  • too-large-for-stack: The maximum size of objects (in bytes) that will be linted. Larger objects are ok on the heap

    (default: 200)

Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for usage of Vec<Box<T>> where T: Sized anywhere in the code. Check the Box documentation for more information.

Why is this bad?

Vec already keeps its contents in a separate area on the heap. So if you Box its contents, you just add another level of indirection.

Known problems

Vec<Box<T: Sized>> makes sense if T is a large type (see #3530, 1st comment).

Example

struct X {
    values: Vec<Box<i32>>,
}

Better:

struct X {
    values: Vec<i32>,
}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

  • vec-box-size-threshold: The size of the boxed type in bytes, where boxing in a Vec is allowed

    (default: 4096)

Applicability: Unspecified(?)
Added in: 1.33.0

What it does

Checks for calls to push immediately after creating a new Vec.

If the Vec is created using with_capacity this will only lint if the capacity is a constant and the number of pushes is greater than or equal to the initial capacity.

If the Vec is extended after the initial sequence of pushes and it was default initialized then this will only lint after there were at least four pushes. This number may change in the future.

Why is this bad?

The vec![] macro is both more performant and easier to read than multiple push calls.

Example

let mut v = Vec::new();
v.push(0);
v.push(1);
v.push(2);

Use instead:

let v = vec![0, 1, 2];
Applicability: HasPlaceholders(?)
Added in: 1.51.0

What it does

Finds occurrences of Vec::resize(0, an_int)

Why is this bad?

This is probably an argument inversion mistake.

Example

vec![1, 2, 3, 4, 5].resize(0, 5)

Use instead:

vec![1, 2, 3, 4, 5].clear()
Applicability: MaybeIncorrect(?)
Added in: 1.46.0

What it does

Checks for bit masks that can be replaced by a call to trailing_zeros

Why is this bad?

x.trailing_zeros() > 4 is much clearer than x & 15 == 0

Known problems

llvm generates better code for x & 15 == 0 on x86

Example

if x & 0b1111 == 0 { }

Use instead:

if x.trailing_zeros() > 4 { }

Configuration

  • verbose-bit-mask-threshold: The maximum allowed size of a bit mask before suggesting to use ‘trailing_zeros’

    (default: 1)

Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Checks for usage of File::read_to_end and File::read_to_string.

Why restrict this?

fs::{read, read_to_string} provide the same functionality when buf is empty with fewer imports and no intermediate values. See also: fs::read docs, fs::read_to_string docs

Example

let mut f = File::open("foo.txt").unwrap();
let mut bytes = Vec::new();
f.read_to_end(&mut bytes).unwrap();

Can be written more concisely as

let mut bytes = fs::read("foo.txt").unwrap();
Applicability: Unspecified(?)
Added in: 1.44.0

What it does

Checks for usage of waker.clone().wake()

Why is this bad?

Cloning the waker is not necessary, wake_by_ref() enables the same operation without extra cloning/dropping.

Example

waker.clone().wake();

Should be written

waker.wake_by_ref();
Applicability: MachineApplicable(?)
Added in: 1.75.0

What it does

Checks for while loops comparing floating point values.

Why is this bad?

If you increment floating point values, errors can compound, so, use integers instead if possible.

Known problems

The lint will catch all while loops comparing floating point values without regarding the increment.

Example

let mut x = 0.0;
while x < 42.0 {
    x += 1.0;
}

Use instead:

let mut x = 0;
while x < 42 {
    x += 1;
}
Applicability: Unspecified(?)
Added in: 1.80.0

What it does

Checks whether variables used within while loop condition can be (and are) mutated in the body.

Why is this bad?

If the condition is unchanged, entering the body of the loop will lead to an infinite loop.

Known problems

If the while-loop is in a closure, the check for mutation of the condition variables in the body can cause false negatives. For example when only Upvar a is in the condition and only Upvar b gets mutated in the body, the lint will not trigger.

Example

let i = 0;
while i > 10 {
    println!("let me loop forever!");
}
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Detects loop + match combinations that are easier written as a while let loop.

Why is this bad?

The while let loop is usually shorter and more readable.

Example

let y = Some(1);
loop {
    let x = match y {
        Some(x) => x,
        None => break,
    };
    // ..
}

Use instead:

let y = Some(1);
while let Some(x) = y {
    // ..
};
Applicability: HasPlaceholders(?)
Added in: pre 1.29.0

What it does

Checks for while let expressions on iterators.

Why is this bad?

Readability. A simple for loop is shorter and conveys the intent better.

Example

while let Some(val) = iter.next() {
    ..
}

Use instead:

for val in &mut iter {
    ..
}
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for wildcard dependencies in the Cargo.toml.

Why is this bad?

As the edition guide says, it is highly unlikely that you work with any possible version of your dependency, and wildcard dependencies would cause unnecessary breakage in the ecosystem.

Example

[dependencies]
regex = "*"
Applicability: Unspecified(?)
Added in: 1.32.0

What it does

Checks for wildcard enum matches using _.

Why restrict this?

New enum variants added by library updates can be missed.

Known problems

Suggested replacements may be incorrect if guards exhaustively cover some variants, and also may not use correct path to enum if it’s not present in the current scope.

Example

match x {
    Foo::A(_) => {},
    _ => {},
}

Use instead:

match x {
    Foo::A(_) => {},
    Foo::B(_) => {},
}
Applicability: MaybeIncorrect(?)
Added in: 1.34.0

What it does

Checks for wildcard imports use _::*.

Why is this bad?

wildcard imports can pollute the namespace. This is especially bad if you try to import something through a wildcard, that already has been imported by name from a different source:

use crate1::foo; // Imports a function named foo
use crate2::*; // Has a function named foo

foo(); // Calls crate1::foo

This can lead to confusing error messages at best and to unexpected behavior at worst.

Exceptions

Wildcard imports are allowed from modules that their name contains prelude. Many crates (including the standard library) provide modules named “prelude” specifically designed for wildcard import.

use super::* is allowed in test modules. This is defined as any module with “test” in the name.

These exceptions can be disabled using the warn-on-all-wildcard-imports configuration flag.

Known problems

If macros are imported through the wildcard, this macro is not included by the suggestion and has to be added by hand.

Applying the suggestion when explicit imports of the things imported with a glob import exist, may result in unused_imports warnings.

Example

use crate1::*;

foo();

Use instead:

use crate1::foo;

foo();

Configuration

  • allowed-wildcard-imports: List of path segments allowed to have wildcard imports.

Example

allowed-wildcard-imports = [ "utils", "common" ]

Noteworthy

  1. This configuration has no effects if used with warn_on_all_wildcard_imports = true.
  2. Paths with any segment that containing the word ‘prelude’ are already allowed by default.

(default: [])

  • warn-on-all-wildcard-imports: Whether to allow certain wildcard imports (prelude, super in tests).

    (default: false)

Applicability: MachineApplicable(?)
Added in: 1.43.0

What it does

Checks for wildcard pattern used with others patterns in same match arm.

Why is this bad?

Wildcard pattern already covers any other pattern as it will match anyway. It makes the code less readable, especially to spot wildcard pattern use in match arm.

Example

match s {
    "a" => {},
    "bar" | _ => {},
}

Use instead:

match s {
    "a" => {},
    _ => {},
}
Applicability: Unspecified(?)
Added in: 1.42.0

What it does

This lint warns about the use of literals as write!/writeln! args.

Why is this bad?

Using literals as writeln! args is inefficient (c.f., https://github.com/matthiaskrgr/rust-str-bench) and unnecessary (i.e., just put the literal in the format string)

Example

writeln!(buf, "{}", "foo");

Use instead:

writeln!(buf, "foo");
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

This lint warns when you use write!() with a format string that ends in a newline.

Why is this bad?

You should use writeln!() instead, which appends the newline.

Example

write!(buf, "Hello {}!\n", name);

Use instead:

writeln!(buf, "Hello {}!", name);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

This lint warns when you use writeln!(buf, "") to print a newline.

Why is this bad?

You should use writeln!(buf), which is simpler.

Example

writeln!(buf, "");

Use instead:

writeln!(buf);
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Nothing. This lint has been deprecated

Deprecation reason

clippy::wrong_self_convention now covers this case via the avoid-breaking-exported-api config.

Applicability: Unspecified(?)
Deprecated in: 1.54.0

What it does

Checks for methods with certain name prefixes or suffixes, and which do not adhere to standard conventions regarding how self is taken. The actual rules are:

PrefixPostfixself takenself type
as_none&self or &mut selfany
from_nonenoneany
into_noneselfany
is_none&mut self or &self or noneany
to__mut&mut selfany
to_not _mutselfCopy
to_not _mut&selfnot Copy

Note: Clippy doesn’t trigger methods with to_ prefix in:

  • Traits definition. Clippy can not tell if a type that implements a trait is Copy or not.
  • Traits implementation, when &self is taken. The method signature is controlled by the trait and often &self is required for all types that implement the trait (see e.g. the std::string::ToString trait).

Clippy allows Pin<&Self> and Pin<&mut Self> if &self and &mut self is required.

Please find more info here: https://rust-lang.github.io/api-guidelines/naming.html#ad-hoc-conversions-follow-as_-to_-into_-conventions-c-conv

Why is this bad?

Consistency breeds readability. If you follow the conventions, your users won’t be surprised that they, e.g., need to supply a mutable reference to a as_.. function.

Example

impl X {
    fn as_str(self) -> &'static str {
        // ..
    }
}

Use instead:

impl X {
    fn as_str(&self) -> &'static str {
        // ..
    }
}

Configuration

  • avoid-breaking-exported-api: Suppress lints whenever the suggested change would cause breakage for other crates.

    (default: true)

Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for transmutes that can’t ever be correct on any architecture.

Why is this bad?

It’s basically guaranteed to be undefined behavior.

Known problems

When accessing C, users might want to store pointer sized objects in extradata arguments to save an allocation.

Example

let ptr: *const T = core::intrinsics::transmute('x')
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Checks for 0.0 / 0.0.

Why is this bad?

It’s less readable than f32::NAN or f64::NAN.

Example

let nan = 0.0f32 / 0.0;

Use instead:

let nan = f32::NAN;
Applicability: Unspecified(?)
Added in: pre 1.29.0

What it does

Warns if an integral constant literal starts with 0.

Why is this bad?

In some languages (including the infamous C language and most of its family), this marks an octal constant. In Rust however, this is a decimal constant. This could be confusing for both the writer and a reader of the constant.

Example

In Rust:

fn main() {
    let a = 0123;
    println!("{}", a);
}

prints 123, while in C:

#include <stdio.h>

int main() {
    int a = 0123;
    printf("%d\n", a);
}

prints 83 (as 83 == 0o123 while 123 == 0o173).

Applicability: MaybeIncorrect(?)
Added in: pre 1.29.0

What it does

Catch casts from 0 to some pointer type

Why is this bad?

This generally means null and is better expressed as {std, core}::ptr::{null, null_mut}.

Example

let a = 0 as *const u32;

Use instead:

let a = std::ptr::null::<u32>();
Applicability: MachineApplicable(?)
Added in: pre 1.29.0

What it does

Checks for array or vec initializations which call a function or method, but which have a repeat count of zero.

Why is this bad?

Such an initialization, despite having a repeat length of 0, will still call the inner function. This may not be obvious and as such there may be unintended side effects in code.

Example

fn side_effect() -> i32 {
    println!("side effect");
    10
}
let a = [side_effect(); 0];

Use instead:

fn side_effect() -> i32 {
    println!("side effect");
    10
}
side_effect();
let a: [i32; 0] = [];
Applicability: Unspecified(?)
Added in: 1.79.0

What it does

Checks for maps with zero-sized value types anywhere in the code.

Why is this bad?

Since there is only a single value for a zero-sized type, a map containing zero sized values is effectively a set. Using a set in that case improves readability and communicates intent more clearly.

Known problems

  • A zero-sized type cannot be recovered later if it contains private fields.
  • This lints the signature of public items

Example

fn unique_words(text: &str) -> HashMap<&str, ()> {
    todo!();
}

Use instead:

fn unique_words(text: &str) -> HashSet<&str> {
    todo!();
}
Applicability: Unspecified(?)
Added in: 1.50.0

What it does

Looks for code that spawns a process but never calls wait() on the child.

Why is this bad?

As explained in the standard library documentation, calling wait() is necessary on Unix platforms to properly release all OS resources associated with the process. Not doing so will effectively leak process IDs and/or other limited global resources, which can eventually lead to resource exhaustion, so it’s recommended to call wait() in long-running applications. Such processes are called “zombie processes”.

Example

use std::process::Command;

let _child = Command::new("ls").spawn().expect("failed to execute child");

Use instead:

use std::process::Command;

let mut child = Command::new("ls").spawn().expect("failed to execute child");
child.wait().expect("failed to wait on child");
Applicability: MaybeIncorrect(?)
Added in: 1.74.0

What it does

Checks for offset(_), wrapping_{add, sub}, etc. on raw pointers to zero-sized types

Why is this bad?

This is a no-op, and likely unintended

Example

unsafe { (&() as *const ()).offset(1) };
Applicability: Unspecified(?)
Added in: 1.41.0