1 // `library/{std,core}/src/primitive_docs.rs` should have the same contents.
2 // These are different files so that relative links work properly without
3 // having to have `CARGO_PKG_NAME` set, but conceptually they should always be the same.
4 #[doc(primitive = "bool")]
6 #[doc(alias = "false")]
9 /// The `bool` represents a value, which could only be either [`true`] or [`false`]. If you cast
10 /// a `bool` into an integer, [`true`] will be 1 and [`false`] will be 0.
14 /// `bool` implements various traits, such as [`BitAnd`], [`BitOr`], [`Not`], etc.,
15 /// which allow us to perform boolean operations using `&`, `|` and `!`.
17 /// [`if`] requires a `bool` value as its conditional. [`assert!`], which is an
18 /// important macro in testing, checks whether an expression is [`true`] and panics
22 /// let bool_val = true & false | false;
23 /// assert!(!bool_val);
26 /// [`true`]: ../std/keyword.true.html
27 /// [`false`]: ../std/keyword.false.html
28 /// [`BitAnd`]: ops::BitAnd
29 /// [`BitOr`]: ops::BitOr
31 /// [`if`]: ../std/keyword.if.html
35 /// A trivial example of the usage of `bool`:
38 /// let praise_the_borrow_checker = true;
40 /// // using the `if` conditional
41 /// if praise_the_borrow_checker {
42 /// println!("oh, yeah!");
44 /// println!("what?!!");
47 /// // ... or, a match pattern
48 /// match praise_the_borrow_checker {
49 /// true => println!("keep praising!"),
50 /// false => println!("you should praise!"),
54 /// Also, since `bool` implements the [`Copy`] trait, we don't
55 /// have to worry about the move semantics (just like the integer and float primitives).
57 /// Now an example of `bool` cast to integer type:
60 /// assert_eq!(true as i32, 1);
61 /// assert_eq!(false as i32, 0);
63 #[stable(feature = "rust1", since = "1.0.0")]
66 #[doc(primitive = "never")]
69 /// The `!` type, also called "never".
71 /// `!` represents the type of computations which never resolve to any value at all. For example,
72 /// the [`exit`] function `fn exit(code: i32) -> !` exits the process without ever returning, and
75 /// `break`, `continue` and `return` expressions also have type `!`. For example we are allowed to
79 /// #![feature(never_type)]
80 /// # fn foo() -> u32 {
87 /// Although the `let` is pointless here, it illustrates the meaning of `!`. Since `x` is never
88 /// assigned a value (because `return` returns from the entire function), `x` can be given type
89 /// `!`. We could also replace `return 123` with a `panic!` or a never-ending `loop` and this code
90 /// would still be valid.
92 /// A more realistic usage of `!` is in this code:
95 /// # fn get_a_number() -> Option<u32> { None }
97 /// let num: u32 = match get_a_number() {
104 /// Both match arms must produce values of type [`u32`], but since `break` never produces a value
105 /// at all we know it can never produce a value which isn't a [`u32`]. This illustrates another
106 /// behaviour of the `!` type - expressions with type `!` will coerce into any other type.
108 /// [`u32`]: prim@u32
109 #[doc = concat!("[`exit`]: ", include_str!("../primitive_docs/process_exit.md"))]
111 /// # `!` and generics
113 /// ## Infallible errors
115 /// The main place you'll see `!` used explicitly is in generic code. Consider the [`FromStr`]
119 /// trait FromStr: Sized {
121 /// fn from_str(s: &str) -> Result<Self, Self::Err>;
125 /// When implementing this trait for [`String`] we need to pick a type for [`Err`]. And since
126 /// converting a string into a string will never result in an error, the appropriate type is `!`.
127 /// (Currently the type actually used is an enum with no variants, though this is only because `!`
128 /// was added to Rust at a later date and it may change in the future.) With an [`Err`] type of
129 /// `!`, if we have to call [`String::from_str`] for some reason the result will be a
130 /// [`Result<String, !>`] which we can unpack like this:
133 /// #![feature(exhaustive_patterns)]
134 /// use std::str::FromStr;
135 /// let Ok(s) = String::from_str("hello");
138 /// Since the [`Err`] variant contains a `!`, it can never occur. If the `exhaustive_patterns`
139 /// feature is present this means we can exhaustively match on [`Result<T, !>`] by just taking the
140 /// [`Ok`] variant. This illustrates another behaviour of `!` - it can be used to "delete" certain
141 /// enum variants from generic types like `Result`.
143 /// ## Infinite loops
145 /// While [`Result<T, !>`] is very useful for removing errors, `!` can also be used to remove
146 /// successes as well. If we think of [`Result<T, !>`] as "if this function returns, it has not
147 /// errored," we get a very intuitive idea of [`Result<!, E>`] as well: if the function returns, it
150 /// For example, consider the case of a simple web server, which can be simplified to:
152 /// ```ignore (hypothetical-example)
154 /// let (client, request) = get_request().expect("disconnected");
155 /// let response = request.process();
156 /// response.send(client);
160 /// Currently, this isn't ideal, because we simply panic whenever we fail to get a new connection.
161 /// Instead, we'd like to keep track of this error, like this:
163 /// ```ignore (hypothetical-example)
165 /// match get_request() {
166 /// Err(err) => break err,
167 /// Ok((client, request)) => {
168 /// let response = request.process();
169 /// response.send(client);
175 /// Now, when the server disconnects, we exit the loop with an error instead of panicking. While it
176 /// might be intuitive to simply return the error, we might want to wrap it in a [`Result<!, E>`]
179 /// ```ignore (hypothetical-example)
180 /// fn server_loop() -> Result<!, ConnectionError> {
182 /// let (client, request) = get_request()?;
183 /// let response = request.process();
184 /// response.send(client);
189 /// Now, we can use `?` instead of `match`, and the return type makes a lot more sense: if the loop
190 /// ever stops, it means that an error occurred. We don't even have to wrap the loop in an `Ok`
191 /// because `!` coerces to `Result<!, ConnectionError>` automatically.
193 /// [`String::from_str`]: str::FromStr::from_str
194 #[doc = concat!("[`String`]: ", include_str!("../primitive_docs/string_string.md"))]
195 /// [`FromStr`]: str::FromStr
199 /// When writing your own traits, `!` should have an `impl` whenever there is an obvious `impl`
200 /// which doesn't `panic!`. The reason is that functions returning an `impl Trait` where `!`
201 /// does not have an `impl` of `Trait` cannot diverge as their only possible code path. In other
202 /// words, they can't return `!` from every code path. As an example, this code doesn't compile:
205 /// use std::ops::Add;
207 /// fn foo() -> impl Add<u32> {
212 /// But this code does:
215 /// use std::ops::Add;
217 /// fn foo() -> impl Add<u32> {
226 /// The reason is that, in the first example, there are many possible types that `!` could coerce
227 /// to, because many types implement `Add<u32>`. However, in the second example,
228 /// the `else` branch returns a `0`, which the compiler infers from the return type to be of type
229 /// `u32`. Since `u32` is a concrete type, `!` can and will be coerced to it. See issue [#36375]
230 /// for more information on this quirk of `!`.
232 /// [#36375]: https://github.com/rust-lang/rust/issues/36375
234 /// As it turns out, though, most traits can have an `impl` for `!`. Take [`Debug`]
238 /// #![feature(never_type)]
241 /// # fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result;
243 /// impl Debug for ! {
244 /// fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
250 /// Once again we're using `!`'s ability to coerce into any other type, in this case
251 /// [`fmt::Result`]. Since this method takes a `&!` as an argument we know that it can never be
252 /// called (because there is no value of type `!` for it to be called with). Writing `*self`
253 /// essentially tells the compiler "We know that this code can never be run, so just treat the
254 /// entire function body as having type [`fmt::Result`]". This pattern can be used a lot when
255 /// implementing traits for `!`. Generally, any trait which only has methods which take a `self`
256 /// parameter should have such an impl.
258 /// On the other hand, one trait which would not be appropriate to implement is [`Default`]:
262 /// fn default() -> Self;
266 /// Since `!` has no values, it has no default value either. It's true that we could write an
267 /// `impl` for this which simply panics, but the same is true for any type (we could `impl
268 /// Default` for (eg.) [`File`] by just making [`default()`] panic.)
270 #[doc = concat!("[`File`]: ", include_str!("../primitive_docs/fs_file.md"))]
271 /// [`Debug`]: fmt::Debug
272 /// [`default()`]: Default::default
274 #[unstable(feature = "never_type", issue = "35121")]
277 #[doc(primitive = "char")]
278 #[allow(rustdoc::invalid_rust_codeblocks)]
279 /// A character type.
281 /// The `char` type represents a single character. More specifically, since
282 /// 'character' isn't a well-defined concept in Unicode, `char` is a '[Unicode
285 /// This documentation describes a number of methods and trait implementations on the
286 /// `char` type. For technical reasons, there is additional, separate
287 /// documentation in [the `std::char` module](char/index.html) as well.
291 /// A `char` is a '[Unicode scalar value]', which is any '[Unicode code point]'
292 /// other than a [surrogate code point]. This has a fixed numerical definition:
293 /// code points are in the range 0 to 0x10FFFF, inclusive.
294 /// Surrogate code points, used by UTF-16, are in the range 0xD800 to 0xDFFF.
296 /// No `char` may be constructed, whether as a literal or at runtime, that is not a
297 /// Unicode scalar value:
300 /// // Each of these is a compiler error
301 /// ['\u{D800}', '\u{DFFF}', '\u{110000}'];
305 /// // Panics; from_u32 returns None.
306 /// char::from_u32(0xDE01).unwrap();
310 /// // Undefined behaviour
311 /// unsafe { char::from_u32_unchecked(0x110000) };
314 /// USVs are also the exact set of values that may be encoded in UTF-8. Because
315 /// `char` values are USVs and `str` values are valid UTF-8, it is safe to store
316 /// any `char` in a `str` or read any character from a `str` as a `char`.
318 /// The gap in valid `char` values is understood by the compiler, so in the
319 /// below example the two ranges are understood to cover the whole range of
320 /// possible `char` values and there is no error for a [non-exhaustive match].
323 /// let c: char = 'a';
325 /// '\0' ..= '\u{D7FF}' => false,
326 /// '\u{E000}' ..= '\u{10FFFF}' => true,
330 /// All USVs are valid `char` values, but not all of them represent a real
331 /// character. Many USVs are not currently assigned to a character, but may be
332 /// in the future ("reserved"); some will never be a character
333 /// ("noncharacters"); and some may be given different meanings by different
334 /// users ("private use").
336 /// [Unicode code point]: https://www.unicode.org/glossary/#code_point
337 /// [Unicode scalar value]: https://www.unicode.org/glossary/#unicode_scalar_value
338 /// [non-exhaustive match]: ../book/ch06-02-match.html#matches-are-exhaustive
339 /// [surrogate code point]: https://www.unicode.org/glossary/#surrogate_code_point
343 /// `char` is always four bytes in size. This is a different representation than
344 /// a given character would have as part of a [`String`]. For example:
347 /// let v = vec!['h', 'e', 'l', 'l', 'o'];
349 /// // five elements times four bytes for each element
350 /// assert_eq!(20, v.len() * std::mem::size_of::<char>());
352 /// let s = String::from("hello");
354 /// // five elements times one byte per element
355 /// assert_eq!(5, s.len() * std::mem::size_of::<u8>());
358 #[doc = concat!("[`String`]: ", include_str!("../primitive_docs/string_string.md"))]
360 /// As always, remember that a human intuition for 'character' might not map to
361 /// Unicode's definitions. For example, despite looking similar, the 'é'
362 /// character is one Unicode code point while 'é' is two Unicode code points:
365 /// let mut chars = "é".chars();
366 /// // U+00e9: 'latin small letter e with acute'
367 /// assert_eq!(Some('\u{00e9}'), chars.next());
368 /// assert_eq!(None, chars.next());
370 /// let mut chars = "é".chars();
371 /// // U+0065: 'latin small letter e'
372 /// assert_eq!(Some('\u{0065}'), chars.next());
373 /// // U+0301: 'combining acute accent'
374 /// assert_eq!(Some('\u{0301}'), chars.next());
375 /// assert_eq!(None, chars.next());
378 /// This means that the contents of the first string above _will_ fit into a
379 /// `char` while the contents of the second string _will not_. Trying to create
380 /// a `char` literal with the contents of the second string gives an error:
383 /// error: character literal may only contain one codepoint: 'é'
388 /// Another implication of the 4-byte fixed size of a `char` is that
389 /// per-`char` processing can end up using a lot more memory:
392 /// let s = String::from("love: ❤️");
393 /// let v: Vec<char> = s.chars().collect();
395 /// assert_eq!(12, std::mem::size_of_val(&s[..]));
396 /// assert_eq!(32, std::mem::size_of_val(&v[..]));
398 #[stable(feature = "rust1", since = "1.0.0")]
401 #[doc(primitive = "unit")]
406 /// The `()` type, also called "unit".
408 /// The `()` type has exactly one value `()`, and is used when there
409 /// is no other meaningful value that could be returned. `()` is most
410 /// commonly seen implicitly: functions without a `-> ...` implicitly
411 /// have return type `()`, that is, these are equivalent:
414 /// fn long() -> () {}
419 /// The semicolon `;` can be used to discard the result of an
420 /// expression at the end of a block, making the expression (and thus
421 /// the block) evaluate to `()`. For example,
424 /// fn returns_i64() -> i64 {
427 /// fn returns_unit() {
439 #[stable(feature = "rust1", since = "1.0.0")]
442 // Required to make auto trait impls render.
443 // See /src/librustdoc/passes/collect_trait_impls.rs:collect_trait_impls
447 #[doc(primitive = "pointer")]
448 #[doc(alias = "ptr")]
450 #[doc(alias = "*const")]
451 #[doc(alias = "*mut")]
453 /// Raw, unsafe pointers, `*const T`, and `*mut T`.
455 /// *[See also the `std::ptr` module](ptr).*
457 /// Working with raw pointers in Rust is uncommon, typically limited to a few patterns.
458 /// Raw pointers can be unaligned or [`null`]. However, when a raw pointer is
459 /// dereferenced (using the `*` operator), it must be non-null and aligned.
461 /// Storing through a raw pointer using `*ptr = data` calls `drop` on the old value, so
462 /// [`write`] must be used if the type has drop glue and memory is not already
463 /// initialized - otherwise `drop` would be called on the uninitialized memory.
465 /// Use the [`null`] and [`null_mut`] functions to create null pointers, and the
466 /// [`is_null`] method of the `*const T` and `*mut T` types to check for null.
467 /// The `*const T` and `*mut T` types also define the [`offset`] method, for
470 /// # Common ways to create raw pointers
472 /// ## 1. Coerce a reference (`&T`) or mutable reference (`&mut T`).
475 /// let my_num: i32 = 10;
476 /// let my_num_ptr: *const i32 = &my_num;
477 /// let mut my_speed: i32 = 88;
478 /// let my_speed_ptr: *mut i32 = &mut my_speed;
481 /// To get a pointer to a boxed value, dereference the box:
484 /// let my_num: Box<i32> = Box::new(10);
485 /// let my_num_ptr: *const i32 = &*my_num;
486 /// let mut my_speed: Box<i32> = Box::new(88);
487 /// let my_speed_ptr: *mut i32 = &mut *my_speed;
490 /// This does not take ownership of the original allocation
491 /// and requires no resource management later,
492 /// but you must not use the pointer after its lifetime.
494 /// ## 2. Consume a box (`Box<T>`).
496 /// The [`into_raw`] function consumes a box and returns
497 /// the raw pointer. It doesn't destroy `T` or deallocate any memory.
500 /// let my_speed: Box<i32> = Box::new(88);
501 /// let my_speed: *mut i32 = Box::into_raw(my_speed);
503 /// // By taking ownership of the original `Box<T>` though
504 /// // we are obligated to put it together later to be destroyed.
506 /// drop(Box::from_raw(my_speed));
510 /// Note that here the call to [`drop`] is for clarity - it indicates
511 /// that we are done with the given value and it should be destroyed.
513 /// ## 3. Create it using `ptr::addr_of!`
515 /// Instead of coercing a reference to a raw pointer, you can use the macros
516 /// [`ptr::addr_of!`] (for `*const T`) and [`ptr::addr_of_mut!`] (for `*mut T`).
517 /// These macros allow you to create raw pointers to fields to which you cannot
518 /// create a reference (without causing undefined behaviour), such as an
519 /// unaligned field. This might be necessary if packed structs or uninitialized
520 /// memory is involved.
523 /// #[derive(Debug, Default, Copy, Clone)]
524 /// #[repr(C, packed)]
529 /// let s = S::default();
530 /// let p = std::ptr::addr_of!(s.unaligned); // not allowed with coercion
533 /// ## 4. Get it from C.
536 /// # #![feature(rustc_private)]
537 /// extern crate libc;
542 /// let my_num: *mut i32 = libc::malloc(mem::size_of::<i32>()) as *mut i32;
543 /// if my_num.is_null() {
544 /// panic!("failed to allocate memory");
546 /// libc::free(my_num as *mut libc::c_void);
550 /// Usually you wouldn't literally use `malloc` and `free` from Rust,
551 /// but C APIs hand out a lot of pointers generally, so are a common source
552 /// of raw pointers in Rust.
554 /// [`null`]: ptr::null
555 /// [`null_mut`]: ptr::null_mut
556 /// [`is_null`]: pointer::is_null
557 /// [`offset`]: pointer::offset
558 #[doc = concat!("[`into_raw`]: ", include_str!("../primitive_docs/box_into_raw.md"))]
559 /// [`drop`]: mem::drop
560 /// [`write`]: ptr::write
561 #[stable(feature = "rust1", since = "1.0.0")]
564 #[doc(primitive = "array")]
566 #[doc(alias = "[T;N]")] // unfortunately, rustdoc doesn't have fuzzy search for aliases
567 #[doc(alias = "[T; N]")]
568 /// A fixed-size array, denoted `[T; N]`, for the element type, `T`, and the
569 /// non-negative compile-time constant size, `N`.
571 /// There are two syntactic forms for creating an array:
573 /// * A list with each element, i.e., `[x, y, z]`.
574 /// * A repeat expression `[x; N]`, which produces an array with `N` copies of `x`.
575 /// The type of `x` must be [`Copy`].
577 /// Note that `[expr; 0]` is allowed, and produces an empty array.
578 /// This will still evaluate `expr`, however, and immediately drop the resulting value, so
579 /// be mindful of side effects.
581 /// Arrays of *any* size implement the following traits if the element type allows it:
586 /// - [`IntoIterator`] (implemented for `[T; N]`, `&[T; N]` and `&mut [T; N]`)
587 /// - [`PartialEq`], [`PartialOrd`], [`Eq`], [`Ord`]
589 /// - [`AsRef`], [`AsMut`]
590 /// - [`Borrow`], [`BorrowMut`]
592 /// Arrays of sizes from 0 to 32 (inclusive) implement the [`Default`] trait
593 /// if the element type allows it. As a stopgap, trait implementations are
594 /// statically generated up to size 32.
596 /// Arrays coerce to [slices (`[T]`)][slice], so a slice method may be called on
597 /// an array. Indeed, this provides most of the API for working with arrays.
598 /// Slices have a dynamic size and do not coerce to arrays.
600 /// You can move elements out of an array with a [slice pattern]. If you want
601 /// one element, see [`mem::replace`].
606 /// let mut array: [i32; 3] = [0; 3];
611 /// assert_eq!([1, 2], &array[1..]);
613 /// // This loop prints: 0 1 2
619 /// You can also iterate over reference to the array's elements:
622 /// let array: [i32; 3] = [0; 3];
624 /// for x in &array { }
627 /// You can use a [slice pattern] to move elements out of an array:
630 /// fn move_away(_: String) { /* Do interesting things. */ }
632 /// let [john, roa] = ["John".to_string(), "Roa".to_string()];
639 /// Prior to Rust 1.53, arrays did not implement [`IntoIterator`] by value, so the method call
640 /// `array.into_iter()` auto-referenced into a [slice iterator](slice::iter). Right now, the old
641 /// behavior is preserved in the 2015 and 2018 editions of Rust for compatibility, ignoring
642 /// [`IntoIterator`] by value. In the future, the behavior on the 2015 and 2018 edition
643 /// might be made consistent to the behavior of later editions.
645 /// ```rust,edition2018
646 /// // Rust 2015 and 2018:
648 /// # #![allow(array_into_iter)] // override our `deny(warnings)`
649 /// let array: [i32; 3] = [0; 3];
651 /// // This creates a slice iterator, producing references to each value.
652 /// for item in array.into_iter().enumerate() {
653 /// let (i, x): (usize, &i32) = item;
654 /// println!("array[{i}] = {x}");
657 /// // The `array_into_iter` lint suggests this change for future compatibility:
658 /// for item in array.iter().enumerate() {
659 /// let (i, x): (usize, &i32) = item;
660 /// println!("array[{i}] = {x}");
663 /// // You can explicitly iterate an array by value using `IntoIterator::into_iter`
664 /// for item in IntoIterator::into_iter(array).enumerate() {
665 /// let (i, x): (usize, i32) = item;
666 /// println!("array[{i}] = {x}");
670 /// Starting in the 2021 edition, `array.into_iter()` uses `IntoIterator` normally to iterate
671 /// by value, and `iter()` should be used to iterate by reference like previous editions.
673 /// ```rust,edition2021
676 /// let array: [i32; 3] = [0; 3];
678 /// // This iterates by reference:
679 /// for item in array.iter().enumerate() {
680 /// let (i, x): (usize, &i32) = item;
681 /// println!("array[{i}] = {x}");
684 /// // This iterates by value:
685 /// for item in array.into_iter().enumerate() {
686 /// let (i, x): (usize, i32) = item;
687 /// println!("array[{i}] = {x}");
691 /// Future language versions might start treating the `array.into_iter()`
692 /// syntax on editions 2015 and 2018 the same as on edition 2021. So code using
693 /// those older editions should still be written with this change in mind, to
694 /// prevent breakage in the future. The safest way to accomplish this is to
695 /// avoid the `into_iter` syntax on those editions. If an edition update is not
696 /// viable/desired, there are multiple alternatives:
697 /// * use `iter`, equivalent to the old behavior, creating references
698 /// * use [`IntoIterator::into_iter`], equivalent to the post-2021 behavior (Rust 1.53+)
699 /// * replace `for ... in array.into_iter() {` with `for ... in array {`,
700 /// equivalent to the post-2021 behavior (Rust 1.53+)
702 /// ```rust,edition2018
703 /// // Rust 2015 and 2018:
705 /// let array: [i32; 3] = [0; 3];
707 /// // This iterates by reference:
708 /// for item in array.iter() {
709 /// let x: &i32 = item;
713 /// // This iterates by value:
714 /// for item in IntoIterator::into_iter(array) {
715 /// let x: i32 = item;
719 /// // This iterates by value:
720 /// for item in array {
721 /// let x: i32 = item;
725 /// // IntoIter can also start a chain.
726 /// // This iterates by value:
727 /// for item in IntoIterator::into_iter(array).enumerate() {
728 /// let (i, x): (usize, i32) = item;
729 /// println!("array[{i}] = {x}");
733 /// [slice]: prim@slice
734 /// [`Debug`]: fmt::Debug
735 /// [`Hash`]: hash::Hash
736 /// [`Borrow`]: borrow::Borrow
737 /// [`BorrowMut`]: borrow::BorrowMut
738 /// [slice pattern]: ../reference/patterns.html#slice-patterns
739 #[stable(feature = "rust1", since = "1.0.0")]
742 #[doc(primitive = "slice")]
746 /// A dynamically-sized view into a contiguous sequence, `[T]`. Contiguous here
747 /// means that elements are laid out so that every element is the same
748 /// distance from its neighbors.
750 /// *[See also the `std::slice` module](crate::slice).*
752 /// Slices are a view into a block of memory represented as a pointer and a
757 /// let vec = vec![1, 2, 3];
758 /// let int_slice = &vec[..];
759 /// // coercing an array to a slice
760 /// let str_slice: &[&str] = &["one", "two", "three"];
763 /// Slices are either mutable or shared. The shared slice type is `&[T]`,
764 /// while the mutable slice type is `&mut [T]`, where `T` represents the element
765 /// type. For example, you can mutate the block of memory that a mutable slice
769 /// let mut x = [1, 2, 3];
770 /// let x = &mut x[..]; // Take a full slice of `x`.
772 /// assert_eq!(x, &[1, 7, 3]);
775 /// As slices store the length of the sequence they refer to, they have twice
776 /// the size of pointers to [`Sized`](marker/trait.Sized.html) types.
777 /// Also see the reference on
778 /// [dynamically sized types](../reference/dynamically-sized-types.html).
781 /// # use std::rc::Rc;
782 /// let pointer_size = std::mem::size_of::<&u8>();
783 /// assert_eq!(2 * pointer_size, std::mem::size_of::<&[u8]>());
784 /// assert_eq!(2 * pointer_size, std::mem::size_of::<*const [u8]>());
785 /// assert_eq!(2 * pointer_size, std::mem::size_of::<Box<[u8]>>());
786 /// assert_eq!(2 * pointer_size, std::mem::size_of::<Rc<[u8]>>());
788 #[stable(feature = "rust1", since = "1.0.0")]
791 #[doc(primitive = "str")]
795 /// *[See also the `std::str` module](crate::str).*
797 /// The `str` type, also called a 'string slice', is the most primitive string
798 /// type. It is usually seen in its borrowed form, `&str`. It is also the type
799 /// of string literals, `&'static str`.
801 /// String slices are always valid UTF-8.
805 /// String literals are string slices:
808 /// let hello = "Hello, world!";
810 /// // with an explicit type annotation
811 /// let hello: &'static str = "Hello, world!";
814 /// They are `'static` because they're stored directly in the final binary, and
815 /// so will be valid for the `'static` duration.
819 /// A `&str` is made up of two components: a pointer to some bytes, and a
820 /// length. You can look at these with the [`as_ptr`] and [`len`] methods:
826 /// let story = "Once upon a time...";
828 /// let ptr = story.as_ptr();
829 /// let len = story.len();
831 /// // story has nineteen bytes
832 /// assert_eq!(19, len);
834 /// // We can re-build a str out of ptr and len. This is all unsafe because
835 /// // we are responsible for making sure the two components are valid:
837 /// // First, we build a &[u8]...
838 /// let slice = slice::from_raw_parts(ptr, len);
840 /// // ... and then convert that slice into a string slice
841 /// str::from_utf8(slice)
844 /// assert_eq!(s, Ok(story));
847 /// [`as_ptr`]: str::as_ptr
848 /// [`len`]: str::len
850 /// Note: This example shows the internals of `&str`. `unsafe` should not be
851 /// used to get a string slice under normal circumstances. Use `as_str`
853 #[stable(feature = "rust1", since = "1.0.0")]
856 #[doc(primitive = "tuple")]
861 /// A finite heterogeneous sequence, `(T, U, ..)`.
863 /// Let's cover each of those in turn:
865 /// Tuples are *finite*. In other words, a tuple has a length. Here's a tuple
869 /// ("hello", 5, 'c');
872 /// 'Length' is also sometimes called 'arity' here; each tuple of a different
873 /// length is a different, distinct type.
875 /// Tuples are *heterogeneous*. This means that each element of the tuple can
876 /// have a different type. In that tuple above, it has the type:
880 /// (&'static str, i32, char)
881 /// # = ("hello", 5, 'c');
884 /// Tuples are a *sequence*. This means that they can be accessed by position;
885 /// this is called 'tuple indexing', and it looks like this:
888 /// let tuple = ("hello", 5, 'c');
890 /// assert_eq!(tuple.0, "hello");
891 /// assert_eq!(tuple.1, 5);
892 /// assert_eq!(tuple.2, 'c');
895 /// The sequential nature of the tuple applies to its implementations of various
896 /// traits. For example, in [`PartialOrd`] and [`Ord`], the elements are compared
897 /// sequentially until the first non-equal set is found.
899 /// For more about tuples, see [the book](../book/ch03-02-data-types.html#the-tuple-type).
901 /// # Trait implementations
903 /// If every type inside a tuple implements one of the following traits, then a
904 /// tuple itself also implements it.
916 /// [`Debug`]: fmt::Debug
917 /// [`Hash`]: hash::Hash
919 /// Due to a temporary restriction in Rust's type system, these traits are only
920 /// implemented on tuples of arity 12 or less. In the future, this may change.
927 /// let tuple = ("hello", 5, 'c');
929 /// assert_eq!(tuple.0, "hello");
932 /// Tuples are often used as a return type when you want to return more than
936 /// fn calculate_point() -> (i32, i32) {
937 /// // Don't do a calculation, that's not the point of the example
941 /// let point = calculate_point();
943 /// assert_eq!(point.0, 4);
944 /// assert_eq!(point.1, 5);
946 /// // Combining this with patterns can be nicer.
948 /// let (x, y) = calculate_point();
950 /// assert_eq!(x, 4);
951 /// assert_eq!(y, 5);
954 #[stable(feature = "rust1", since = "1.0.0")]
957 // Required to make auto trait impls render.
958 // See /src/librustdoc/passes/collect_trait_impls.rs:collect_trait_impls
962 #[doc(primitive = "f32")]
963 /// A 32-bit floating point type (specifically, the "binary32" type defined in IEEE 754-2008).
965 /// This type can represent a wide range of decimal numbers, like `3.5`, `27`,
966 /// `-113.75`, `0.0078125`, `34359738368`, `0`, `-1`. So unlike integer types
967 /// (such as `i32`), floating point types can represent non-integer numbers,
970 /// However, being able to represent this wide range of numbers comes at the
971 /// cost of precision: floats can only represent some of the real numbers and
972 /// calculation with floats round to a nearby representable number. For example,
973 /// `5.0` and `1.0` can be exactly represented as `f32`, but `1.0 / 5.0` results
974 /// in `0.20000000298023223876953125` since `0.2` cannot be exactly represented
975 /// as `f32`. Note, however, that printing floats with `println` and friends will
976 /// often discard insignificant digits: `println!("{}", 1.0f32 / 5.0f32)` will
979 /// Additionally, `f32` can represent some special values:
981 /// - −0.0: IEEE 754 floating point numbers have a bit that indicates their sign, so −0.0 is a
982 /// possible value. For comparison −0.0 = +0.0, but floating point operations can carry
983 /// the sign bit through arithmetic operations. This means −0.0 × +0.0 produces −0.0 and
984 /// a negative number rounded to a value smaller than a float can represent also produces −0.0.
985 /// - [∞](#associatedconstant.INFINITY) and
986 /// [−∞](#associatedconstant.NEG_INFINITY): these result from calculations
987 /// like `1.0 / 0.0`.
988 /// - [NaN (not a number)](#associatedconstant.NAN): this value results from
989 /// calculations like `(-1.0).sqrt()`. NaN has some potentially unexpected
991 /// - It is unequal to any float, including itself! This is the reason `f32`
992 /// doesn't implement the `Eq` trait.
993 /// - It is also neither smaller nor greater than any float, making it
994 /// impossible to sort by the default comparison operation, which is the
995 /// reason `f32` doesn't implement the `Ord` trait.
996 /// - It is also considered *infectious* as almost all calculations where one
997 /// of the operands is NaN will also result in NaN. The explanations on this
998 /// page only explicitly document behavior on NaN operands if this default
999 /// is deviated from.
1000 /// - Lastly, there are multiple bit patterns that are considered NaN.
1001 /// Rust does not currently guarantee that the bit patterns of NaN are
1002 /// preserved over arithmetic operations, and they are not guaranteed to be
1003 /// portable or even fully deterministic! This means that there may be some
1004 /// surprising results upon inspecting the bit patterns,
1005 /// as the same calculations might produce NaNs with different bit patterns.
1007 /// When the number resulting from a primitive operation (addition,
1008 /// subtraction, multiplication, or division) on this type is not exactly
1009 /// representable as `f32`, it is rounded according to the roundTiesToEven
1010 /// direction defined in IEEE 754-2008. That means:
1012 /// - The result is the representable value closest to the true value, if there
1013 /// is a unique closest representable value.
1014 /// - If the true value is exactly half-way between two representable values,
1015 /// the result is the one with an even least-significant binary digit.
1016 /// - If the true value's magnitude is ≥ `f32::MAX` + 2<sup>(`f32::MAX_EXP` −
1017 /// `f32::MANTISSA_DIGITS` − 1)</sup>, the result is ∞ or −∞ (preserving the
1018 /// true value's sign).
1020 /// For more information on floating point numbers, see [Wikipedia][wikipedia].
1022 /// *[See also the `std::f32::consts` module](crate::f32::consts).*
1024 /// [wikipedia]: https://en.wikipedia.org/wiki/Single-precision_floating-point_format
1025 #[stable(feature = "rust1", since = "1.0.0")]
1028 #[doc(primitive = "f64")]
1029 /// A 64-bit floating point type (specifically, the "binary64" type defined in IEEE 754-2008).
1031 /// This type is very similar to [`f32`], but has increased
1032 /// precision by using twice as many bits. Please see [the documentation for
1033 /// `f32`][`f32`] or [Wikipedia on double precision
1034 /// values][wikipedia] for more information.
1036 /// *[See also the `std::f64::consts` module](crate::f64::consts).*
1038 /// [`f32`]: prim@f32
1039 /// [wikipedia]: https://en.wikipedia.org/wiki/Double-precision_floating-point_format
1040 #[stable(feature = "rust1", since = "1.0.0")]
1043 #[doc(primitive = "i8")]
1045 /// The 8-bit signed integer type.
1046 #[stable(feature = "rust1", since = "1.0.0")]
1049 #[doc(primitive = "i16")]
1051 /// The 16-bit signed integer type.
1052 #[stable(feature = "rust1", since = "1.0.0")]
1055 #[doc(primitive = "i32")]
1057 /// The 32-bit signed integer type.
1058 #[stable(feature = "rust1", since = "1.0.0")]
1061 #[doc(primitive = "i64")]
1063 /// The 64-bit signed integer type.
1064 #[stable(feature = "rust1", since = "1.0.0")]
1067 #[doc(primitive = "i128")]
1069 /// The 128-bit signed integer type.
1070 #[stable(feature = "i128", since = "1.26.0")]
1073 #[doc(primitive = "u8")]
1075 /// The 8-bit unsigned integer type.
1076 #[stable(feature = "rust1", since = "1.0.0")]
1079 #[doc(primitive = "u16")]
1081 /// The 16-bit unsigned integer type.
1082 #[stable(feature = "rust1", since = "1.0.0")]
1085 #[doc(primitive = "u32")]
1087 /// The 32-bit unsigned integer type.
1088 #[stable(feature = "rust1", since = "1.0.0")]
1091 #[doc(primitive = "u64")]
1093 /// The 64-bit unsigned integer type.
1094 #[stable(feature = "rust1", since = "1.0.0")]
1097 #[doc(primitive = "u128")]
1099 /// The 128-bit unsigned integer type.
1100 #[stable(feature = "i128", since = "1.26.0")]
1103 #[doc(primitive = "isize")]
1105 /// The pointer-sized signed integer type.
1107 /// The size of this primitive is how many bytes it takes to reference any
1108 /// location in memory. For example, on a 32 bit target, this is 4 bytes
1109 /// and on a 64 bit target, this is 8 bytes.
1110 #[stable(feature = "rust1", since = "1.0.0")]
1113 #[doc(primitive = "usize")]
1115 /// The pointer-sized unsigned integer type.
1117 /// The size of this primitive is how many bytes it takes to reference any
1118 /// location in memory. For example, on a 32 bit target, this is 4 bytes
1119 /// and on a 64 bit target, this is 8 bytes.
1120 #[stable(feature = "rust1", since = "1.0.0")]
1123 #[doc(primitive = "reference")]
1125 #[doc(alias = "&mut")]
1127 /// References, both shared and mutable.
1129 /// A reference represents a borrow of some owned value. You can get one by using the `&` or `&mut`
1130 /// operators on a value, or by using a [`ref`](../std/keyword.ref.html) or
1131 /// <code>[ref](../std/keyword.ref.html) [mut](../std/keyword.mut.html)</code> pattern.
1133 /// For those familiar with pointers, a reference is just a pointer that is assumed to be
1134 /// aligned, not null, and pointing to memory containing a valid value of `T` - for example,
1135 /// <code>&[bool]</code> can only point to an allocation containing the integer values `1`
1136 /// ([`true`](../std/keyword.true.html)) or `0` ([`false`](../std/keyword.false.html)), but
1137 /// creating a <code>&[bool]</code> that points to an allocation containing
1138 /// the value `3` causes undefined behaviour.
1139 /// In fact, <code>[Option]\<&T></code> has the same memory representation as a
1140 /// nullable but aligned pointer, and can be passed across FFI boundaries as such.
1142 /// In most cases, references can be used much like the original value. Field access, method
1143 /// calling, and indexing work the same (save for mutability rules, of course). In addition, the
1144 /// comparison operators transparently defer to the referent's implementation, allowing references
1145 /// to be compared the same as owned values.
1147 /// References have a lifetime attached to them, which represents the scope for which the borrow is
1148 /// valid. A lifetime is said to "outlive" another one if its representative scope is as long or
1149 /// longer than the other. The `'static` lifetime is the longest lifetime, which represents the
1150 /// total life of the program. For example, string literals have a `'static` lifetime because the
1151 /// text data is embedded into the binary of the program, rather than in an allocation that needs
1152 /// to be dynamically managed.
1154 /// `&mut T` references can be freely coerced into `&T` references with the same referent type, and
1155 /// references with longer lifetimes can be freely coerced into references with shorter ones.
1157 /// Reference equality by address, instead of comparing the values pointed to, is accomplished via
1158 /// implicit reference-pointer coercion and raw pointer equality via [`ptr::eq`], while
1159 /// [`PartialEq`] compares values.
1165 /// let other_five = 5;
1166 /// let five_ref = &five;
1167 /// let same_five_ref = &five;
1168 /// let other_five_ref = &other_five;
1170 /// assert!(five_ref == same_five_ref);
1171 /// assert!(five_ref == other_five_ref);
1173 /// assert!(ptr::eq(five_ref, same_five_ref));
1174 /// assert!(!ptr::eq(five_ref, other_five_ref));
1177 /// For more information on how to use references, see [the book's section on "References and
1178 /// Borrowing"][book-refs].
1180 /// [book-refs]: ../book/ch04-02-references-and-borrowing.html
1182 /// # Trait implementations
1184 /// The following traits are implemented for all `&T`, regardless of the type of its referent:
1187 /// * [`Clone`] \(Note that this will not defer to `T`'s `Clone` implementation if it exists!)
1190 /// * [`fmt::Pointer`]
1192 /// [`Deref`]: ops::Deref
1193 /// [`Borrow`]: borrow::Borrow
1195 /// `&mut T` references get all of the above except `Copy` and `Clone` (to prevent creating
1196 /// multiple simultaneous mutable borrows), plus the following, regardless of the type of its
1202 /// [`DerefMut`]: ops::DerefMut
1203 /// [`BorrowMut`]: borrow::BorrowMut
1204 /// [bool]: prim@bool
1206 /// The following traits are implemented on `&T` references if the underlying `T` also implements
1209 /// * All the traits in [`std::fmt`] except [`fmt::Pointer`] (which is implemented regardless of the type of its referent) and [`fmt::Write`]
1210 /// * [`PartialOrd`]
1215 /// * [`Fn`] \(in addition, `&T` references get [`FnMut`] and [`FnOnce`] if `T: Fn`)
1217 /// * [`ToSocketAddrs`]
1218 /// * [`Send`] \(`&T` references also require <code>T: [Sync]</code>)
1220 /// [`std::fmt`]: fmt
1221 /// [`Hash`]: hash::Hash
1222 #[doc = concat!("[`ToSocketAddrs`]: ", include_str!("../primitive_docs/net_tosocketaddrs.md"))]
1224 /// `&mut T` references get all of the above except `ToSocketAddrs`, plus the following, if `T`
1225 /// implements that trait:
1228 /// * [`FnMut`] \(in addition, `&mut T` references get [`FnOnce`] if `T: FnMut`)
1229 /// * [`fmt::Write`]
1231 /// * [`DoubleEndedIterator`]
1232 /// * [`ExactSizeIterator`]
1233 /// * [`FusedIterator`]
1234 /// * [`TrustedLen`]
1240 /// [`FusedIterator`]: iter::FusedIterator
1241 /// [`TrustedLen`]: iter::TrustedLen
1242 #[doc = concat!("[`Seek`]: ", include_str!("../primitive_docs/io_seek.md"))]
1243 #[doc = concat!("[`BufRead`]: ", include_str!("../primitive_docs/io_bufread.md"))]
1244 #[doc = concat!("[`Read`]: ", include_str!("../primitive_docs/io_read.md"))]
1245 #[doc = concat!("[`io::Write`]: ", include_str!("../primitive_docs/io_write.md"))]
1247 /// Note that due to method call deref coercion, simply calling a trait method will act like they
1248 /// work on references as well as they do on owned values! The implementations described here are
1249 /// meant for generic contexts, where the final type `T` is a type parameter or otherwise not
1251 #[stable(feature = "rust1", since = "1.0.0")]
1254 #[doc(primitive = "fn")]
1256 /// Function pointers, like `fn(usize) -> bool`.
1258 /// *See also the traits [`Fn`], [`FnMut`], and [`FnOnce`].*
1261 /// [`FnMut`]: ops::FnMut
1262 /// [`FnOnce`]: ops::FnOnce
1264 /// Function pointers are pointers that point to *code*, not data. They can be called
1265 /// just like functions. Like references, function pointers are, among other things, assumed to
1266 /// not be null, so if you want to pass a function pointer over FFI and be able to accommodate null
1267 /// pointers, make your type [`Option<fn()>`](core::option#options-and-pointers-nullable-pointers)
1268 /// with your required signature.
1272 /// Plain function pointers are obtained by casting either plain functions, or closures that don't
1273 /// capture an environment:
1276 /// fn add_one(x: usize) -> usize {
1280 /// let ptr: fn(usize) -> usize = add_one;
1281 /// assert_eq!(ptr(5), 6);
1283 /// let clos: fn(usize) -> usize = |x| x + 5;
1284 /// assert_eq!(clos(5), 10);
1287 /// In addition to varying based on their signature, function pointers come in two flavors: safe
1288 /// and unsafe. Plain `fn()` function pointers can only point to safe functions,
1289 /// while `unsafe fn()` function pointers can point to safe or unsafe functions.
1292 /// fn add_one(x: usize) -> usize {
1296 /// unsafe fn add_one_unsafely(x: usize) -> usize {
1300 /// let safe_ptr: fn(usize) -> usize = add_one;
1302 /// //ERROR: mismatched types: expected normal fn, found unsafe fn
1303 /// //let bad_ptr: fn(usize) -> usize = add_one_unsafely;
1305 /// let unsafe_ptr: unsafe fn(usize) -> usize = add_one_unsafely;
1306 /// let really_safe_ptr: unsafe fn(usize) -> usize = add_one;
1311 /// On top of that, function pointers can vary based on what ABI they use. This
1312 /// is achieved by adding the `extern` keyword before the type, followed by the
1313 /// ABI in question. The default ABI is "Rust", i.e., `fn()` is the exact same
1314 /// type as `extern "Rust" fn()`. A pointer to a function with C ABI would have
1315 /// type `extern "C" fn()`.
1317 /// `extern "ABI" { ... }` blocks declare functions with ABI "ABI". The default
1318 /// here is "C", i.e., functions declared in an `extern {...}` block have "C"
1321 /// For more information and a list of supported ABIs, see [the nomicon's
1322 /// section on foreign calling conventions][nomicon-abi].
1324 /// [nomicon-abi]: ../nomicon/ffi.html#foreign-calling-conventions
1326 /// ### Variadic functions
1328 /// Extern function declarations with the "C" or "cdecl" ABIs can also be *variadic*, allowing them
1329 /// to be called with a variable number of arguments. Normal Rust functions, even those with an
1330 /// `extern "ABI"`, cannot be variadic. For more information, see [the nomicon's section on
1331 /// variadic functions][nomicon-variadic].
1333 /// [nomicon-variadic]: ../nomicon/ffi.html#variadic-functions
1335 /// ### Creating function pointers
1337 /// When `bar` is the name of a function, then the expression `bar` is *not* a
1338 /// function pointer. Rather, it denotes a value of an unnameable type that
1339 /// uniquely identifies the function `bar`. The value is zero-sized because the
1340 /// type already identifies the function. This has the advantage that "calling"
1341 /// the value (it implements the `Fn*` traits) does not require dynamic
1344 /// This zero-sized type *coerces* to a regular function pointer. For example:
1349 /// fn bar(x: i32) {}
1351 /// let not_bar_ptr = bar; // `not_bar_ptr` is zero-sized, uniquely identifying `bar`
1352 /// assert_eq!(mem::size_of_val(¬_bar_ptr), 0);
1354 /// let bar_ptr: fn(i32) = not_bar_ptr; // force coercion to function pointer
1355 /// assert_eq!(mem::size_of_val(&bar_ptr), mem::size_of::<usize>());
1357 /// let footgun = &bar; // this is a shared reference to the zero-sized type identifying `bar`
1360 /// The last line shows that `&bar` is not a function pointer either. Rather, it
1361 /// is a reference to the function-specific ZST. `&bar` is basically never what you
1362 /// want when `bar` is a function.
1364 /// ### Casting to and from integers
1366 /// You cast function pointers directly to integers:
1369 /// let fnptr: fn(i32) -> i32 = |x| x+2;
1370 /// let fnptr_addr = fnptr as usize;
1373 /// However, a direct cast back is not possible. You need to use `transmute`:
1376 /// # let fnptr: fn(i32) -> i32 = |x| x+2;
1377 /// # let fnptr_addr = fnptr as usize;
1378 /// let fnptr = fnptr_addr as *const ();
1379 /// let fnptr: fn(i32) -> i32 = unsafe { std::mem::transmute(fnptr) };
1380 /// assert_eq!(fnptr(40), 42);
1383 /// Crucially, we `as`-cast to a raw pointer before `transmute`ing to a function pointer.
1384 /// This avoids an integer-to-pointer `transmute`, which can be problematic.
1385 /// Transmuting between raw pointers and function pointers (i.e., two pointer types) is fine.
1387 /// Note that all of this is not portable to platforms where function pointers and data pointers
1388 /// have different sizes.
1392 /// Function pointers implement the following traits:
1397 /// * [`PartialOrd`]
1403 /// [`Hash`]: hash::Hash
1404 /// [`Pointer`]: fmt::Pointer
1406 /// Due to a temporary restriction in Rust's type system, these traits are only implemented on
1407 /// functions that take 12 arguments or less, with the `"Rust"` and `"C"` ABIs. In the future, this
1410 /// In addition, function pointers of *any* signature, ABI, or safety are [`Copy`], and all *safe*
1411 /// function pointers implement [`Fn`], [`FnMut`], and [`FnOnce`]. This works because these traits
1412 /// are specially known to the compiler.
1413 #[stable(feature = "rust1", since = "1.0.0")]