5 This document is the primary reference for the Rust programming language. It
6 provides three kinds of material:
8 - Chapters that informally describe each language construct and their use.
9 - Chapters that informally describe the memory model, concurrency model,
10 runtime services, linkage model and debugging facilities.
11 - Appendix chapters providing rationale and references to languages that
12 influenced the design.
14 This document does not serve as an introduction to the language. Background
15 familiarity with the language is assumed. A separate [book] is available to
16 help acquire such background familiarity.
18 This document also does not serve as a reference to the [standard] library
19 included in the language distribution. Those libraries are documented
20 separately by extracting documentation attributes from their source code. Many
21 of the features that one might expect to be language features are library
22 features in Rust, so what you're looking for may be there, not here.
24 You may also be interested in the [grammar].
26 [book]: book/index.html
27 [standard]: std/index.html
28 [grammar]: grammar.html
32 ## Unicode productions
34 A few productions in Rust's grammar permit Unicode code points outside the
35 ASCII range. We define these productions in terms of character properties
36 specified in the Unicode standard, rather than in terms of ASCII-range code
37 points. The grammar has a [Special Unicode Productions][unicodeproductions]
38 section that lists these productions.
40 [unicodeproductions]: grammar.html#special-unicode-productions
42 ## String table productions
44 Some rules in the grammar — notably [unary
45 operators](#unary-operator-expressions), [binary
46 operators](#binary-operator-expressions), and [keywords][keywords] — are
47 given in a simplified form: as a listing of a table of unquoted, printable
48 whitespace-separated strings. These cases form a subset of the rules regarding
49 the [token](#tokens) rule, and are assumed to be the result of a
50 lexical-analysis phase feeding the parser, driven by a DFA, operating over the
51 disjunction of all such string table entries.
53 [keywords]: grammar.html#keywords
55 When such a string enclosed in double-quotes (`"`) occurs inside the grammar,
56 it is an implicit reference to a single member of such a string table
57 production. See [tokens](#tokens) for more information.
63 Rust input is interpreted as a sequence of Unicode code points encoded in UTF-8.
64 Most Rust grammar rules are defined in terms of printable ASCII-range
65 code points, but a small number are defined in terms of Unicode properties or
66 explicit code point lists. [^inputformat]
68 [^inputformat]: Substitute definitions for the special Unicode productions are
69 provided to the grammar verifier, restricted to ASCII range, when verifying the
70 grammar in this document.
74 An identifier is any nonempty Unicode[^non_ascii_idents] string of the following form:
76 [^non_ascii_idents]: Non-ASCII characters in identifiers are currently feature
77 gated. This is expected to improve soon.
81 * The first character has property `XID_start`
82 * The remaining characters have property `XID_continue`
86 * The first character is `_`
87 * The identifier is more than one character, `_` alone is not an identifier
88 * The remaining characters have property `XID_continue`
90 that does _not_ occur in the set of [keywords][keywords].
92 > **Note**: `XID_start` and `XID_continue` as character properties cover the
93 > character ranges used to form the more familiar C and Java language-family
98 Comments in Rust code follow the general C++ style of line (`//`) and
99 block (`/* ... */`) comment forms. Nested block comments are supported.
101 Line comments beginning with exactly _three_ slashes (`///`), and block
102 comments (`/** ... */`), are interpreted as a special syntax for `doc`
103 [attributes](#attributes). That is, they are equivalent to writing
104 `#[doc="..."]` around the body of the comment, i.e., `/// Foo` turns into
107 Line comments beginning with `//!` and block comments `/*! ... */` are
108 doc comments that apply to the parent of the comment, rather than the item
109 that follows. That is, they are equivalent to writing `#![doc="..."]` around
110 the body of the comment. `//!` comments are usually used to document
111 modules that occupy a source file.
113 Non-doc comments are interpreted as a form of whitespace.
117 Whitespace is any non-empty string containing only the following characters:
119 - `U+0020` (space, `' '`)
120 - `U+0009` (tab, `'\t'`)
121 - `U+000A` (LF, `'\n'`)
122 - `U+000D` (CR, `'\r'`)
124 Rust is a "free-form" language, meaning that all forms of whitespace serve only
125 to separate _tokens_ in the grammar, and have no semantic significance.
127 A Rust program has identical meaning if each whitespace element is replaced
128 with any other legal whitespace element, such as a single space character.
132 Tokens are primitive productions in the grammar defined by regular
133 (non-recursive) languages. "Simple" tokens are given in [string table
134 production](#string-table-productions) form, and occur in the rest of the
135 grammar as double-quoted strings. Other tokens have exact rules given.
139 A literal is an expression consisting of a single token, rather than a sequence
140 of tokens, that immediately and directly denotes the value it evaluates to,
141 rather than referring to it by name or some other evaluation rule. A literal is
142 a form of constant expression, so is evaluated (primarily) at compile time.
146 ##### Characters and strings
148 | | Example | `#` sets | Characters | Escapes |
149 |----------------------------------------------|-----------------|------------|-------------|---------------------|
150 | [Character](#character-literals) | `'H'` | `N/A` | All Unicode | [Quote](#quote-escapes) & [Byte](#byte-escapes) & [Unicode](#unicode-escapes) |
151 | [String](#string-literals) | `"hello"` | `N/A` | All Unicode | [Quote](#quote-escapes) & [Byte](#byte-escapes) & [Unicode](#unicode-escapes) |
152 | [Raw](#raw-string-literals) | `r#"hello"#` | `0...` | All Unicode | `N/A` |
153 | [Byte](#byte-literals) | `b'H'` | `N/A` | All ASCII | [Quote](#quote-escapes) & [Byte](#byte-escapes) |
154 | [Byte string](#byte-string-literals) | `b"hello"` | `N/A` | All ASCII | [Quote](#quote-escapes) & [Byte](#byte-escapes) |
155 | [Raw byte string](#raw-byte-string-literals) | `br#"hello"#` | `0...` | All ASCII | `N/A` |
161 | `\x7F` | 8-bit character code (exactly 2 digits) |
163 | `\r` | Carriage return |
168 ##### Unicode escapes
171 | `\u{7FFF}` | 24-bit Unicode character code (up to 6 digits) |
176 | `\'` | Single quote |
177 | `\"` | Double quote |
181 | [Number literals](#number-literals)`*` | Example | Exponentiation | Suffixes |
182 |----------------------------------------|---------|----------------|----------|
183 | Decimal integer | `98_222` | `N/A` | Integer suffixes |
184 | Hex integer | `0xff` | `N/A` | Integer suffixes |
185 | Octal integer | `0o77` | `N/A` | Integer suffixes |
186 | Binary integer | `0b1111_0000` | `N/A` | Integer suffixes |
187 | Floating-point | `123.0E+77` | `Optional` | Floating-point suffixes |
189 `*` All number literals allow `_` as a visual separator: `1_234.0E+18f64`
192 | Integer | Floating-point |
193 |---------|----------------|
194 | `u8`, `i8`, `u16`, `i16`, `u32`, `i32`, `u64`, `i64`, `isize`, `usize` | `f32`, `f64` |
196 #### Character and string literals
198 ##### Character literals
200 A _character literal_ is a single Unicode character enclosed within two
201 `U+0027` (single-quote) characters, with the exception of `U+0027` itself,
202 which must be _escaped_ by a preceding `U+005C` character (`\`).
204 ##### String literals
206 A _string literal_ is a sequence of any Unicode characters enclosed within two
207 `U+0022` (double-quote) characters, with the exception of `U+0022` itself,
208 which must be _escaped_ by a preceding `U+005C` character (`\`).
210 Line-break characters are allowed in string literals. Normally they represent
211 themselves (i.e. no translation), but as a special exception, when an unescaped
212 `U+005C` character (`\`) occurs immediately before the newline (`U+000A`), the
213 `U+005C` character, the newline, and all whitespace at the beginning of the
214 next line are ignored. Thus `a` and `b` are equal:
224 ##### Character escapes
226 Some additional _escapes_ are available in either character or non-raw string
227 literals. An escape starts with a `U+005C` (`\`) and continues with one of the
230 * An _8-bit code point escape_ starts with `U+0078` (`x`) and is
231 followed by exactly two _hex digits_. It denotes the Unicode code point
232 equal to the provided hex value.
233 * A _24-bit code point escape_ starts with `U+0075` (`u`) and is followed
234 by up to six _hex digits_ surrounded by braces `U+007B` (`{`) and `U+007D`
235 (`}`). It denotes the Unicode code point equal to the provided hex value.
236 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
237 (`r`), or `U+0074` (`t`), denoting the Unicode values `U+000A` (LF),
238 `U+000D` (CR) or `U+0009` (HT) respectively.
239 * The _null escape_ is the character `U+0030` (`0`) and denotes the Unicode
240 value `U+0000` (NUL).
241 * The _backslash escape_ is the character `U+005C` (`\`) which must be
242 escaped in order to denote *itself*.
244 ##### Raw string literals
246 Raw string literals do not process any escapes. They start with the character
247 `U+0072` (`r`), followed by zero or more of the character `U+0023` (`#`) and a
248 `U+0022` (double-quote) character. The _raw string body_ can contain any sequence
249 of Unicode characters and is terminated only by another `U+0022` (double-quote)
250 character, followed by the same number of `U+0023` (`#`) characters that preceded
251 the opening `U+0022` (double-quote) character.
253 All Unicode characters contained in the raw string body represent themselves,
254 the characters `U+0022` (double-quote) (except when followed by at least as
255 many `U+0023` (`#`) characters as were used to start the raw string literal) or
256 `U+005C` (`\`) do not have any special meaning.
258 Examples for string literals:
261 "foo"; r"foo"; // foo
262 "\"foo\""; r#""foo""#; // "foo"
265 r##"foo #"# bar"##; // foo #"# bar
267 "\x52"; "R"; r"R"; // R
268 "\\x52"; r"\x52"; // \x52
271 #### Byte and byte string literals
275 A _byte literal_ is a single ASCII character (in the `U+0000` to `U+007F`
276 range) or a single _escape_ preceded by the characters `U+0062` (`b`) and
277 `U+0027` (single-quote), and followed by the character `U+0027`. If the character
278 `U+0027` is present within the literal, it must be _escaped_ by a preceding
279 `U+005C` (`\`) character. It is equivalent to a `u8` unsigned 8-bit integer
282 ##### Byte string literals
284 A non-raw _byte string literal_ is a sequence of ASCII characters and _escapes_,
285 preceded by the characters `U+0062` (`b`) and `U+0022` (double-quote), and
286 followed by the character `U+0022`. If the character `U+0022` is present within
287 the literal, it must be _escaped_ by a preceding `U+005C` (`\`) character.
288 Alternatively, a byte string literal can be a _raw byte string literal_, defined
289 below. A byte string literal of length `n` is equivalent to a `&'static [u8; n]` borrowed fixed-sized array
290 of unsigned 8-bit integers.
292 Some additional _escapes_ are available in either byte or non-raw byte string
293 literals. An escape starts with a `U+005C` (`\`) and continues with one of the
296 * A _byte escape_ escape starts with `U+0078` (`x`) and is
297 followed by exactly two _hex digits_. It denotes the byte
298 equal to the provided hex value.
299 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
300 (`r`), or `U+0074` (`t`), denoting the bytes values `0x0A` (ASCII LF),
301 `0x0D` (ASCII CR) or `0x09` (ASCII HT) respectively.
302 * The _null escape_ is the character `U+0030` (`0`) and denotes the byte
303 value `0x00` (ASCII NUL).
304 * The _backslash escape_ is the character `U+005C` (`\`) which must be
305 escaped in order to denote its ASCII encoding `0x5C`.
307 ##### Raw byte string literals
309 Raw byte string literals do not process any escapes. They start with the
310 character `U+0062` (`b`), followed by `U+0072` (`r`), followed by zero or more
311 of the character `U+0023` (`#`), and a `U+0022` (double-quote) character. The
312 _raw string body_ can contain any sequence of ASCII characters and is terminated
313 only by another `U+0022` (double-quote) character, followed by the same number of
314 `U+0023` (`#`) characters that preceded the opening `U+0022` (double-quote)
315 character. A raw byte string literal can not contain any non-ASCII byte.
317 All characters contained in the raw string body represent their ASCII encoding,
318 the characters `U+0022` (double-quote) (except when followed by at least as
319 many `U+0023` (`#`) characters as were used to start the raw string literal) or
320 `U+005C` (`\`) do not have any special meaning.
322 Examples for byte string literals:
325 b"foo"; br"foo"; // foo
326 b"\"foo\""; br#""foo""#; // "foo"
329 br##"foo #"# bar"##; // foo #"# bar
331 b"\x52"; b"R"; br"R"; // R
332 b"\\x52"; br"\x52"; // \x52
337 A _number literal_ is either an _integer literal_ or a _floating-point
338 literal_. The grammar for recognizing the two kinds of literals is mixed.
340 ##### Integer literals
342 An _integer literal_ has one of four forms:
344 * A _decimal literal_ starts with a *decimal digit* and continues with any
345 mixture of *decimal digits* and _underscores_.
346 * A _hex literal_ starts with the character sequence `U+0030` `U+0078`
347 (`0x`) and continues as any mixture of hex digits and underscores.
348 * An _octal literal_ starts with the character sequence `U+0030` `U+006F`
349 (`0o`) and continues as any mixture of octal digits and underscores.
350 * A _binary literal_ starts with the character sequence `U+0030` `U+0062`
351 (`0b`) and continues as any mixture of binary digits and underscores.
353 Like any literal, an integer literal may be followed (immediately,
354 without any spaces) by an _integer suffix_, which forcibly sets the
355 type of the literal. The integer suffix must be the name of one of the
356 integral types: `u8`, `i8`, `u16`, `i16`, `u32`, `i32`, `u64`, `i64`,
359 The type of an _unsuffixed_ integer literal is determined by type inference:
361 * If an integer type can be _uniquely_ determined from the surrounding
362 program context, the unsuffixed integer literal has that type.
364 * If the program context under-constrains the type, it defaults to the
365 signed 32-bit integer `i32`.
367 * If the program context over-constrains the type, it is considered a
370 Examples of integer literals of various forms:
377 0o70_i16; // type i16
378 0b1111_1111_1001_0000_i32; // type i32
379 0usize; // type usize
382 Note that the Rust syntax considers `-1i8` as an application of the [unary minus
383 operator](#unary-operator-expressions) to an integer literal `1i8`, rather than
384 a single integer literal.
386 ##### Floating-point literals
388 A _floating-point literal_ has one of two forms:
390 * A _decimal literal_ followed by a period character `U+002E` (`.`). This is
391 optionally followed by another decimal literal, with an optional _exponent_.
392 * A single _decimal literal_ followed by an _exponent_.
394 Like integer literals, a floating-point literal may be followed by a
395 suffix, so long as the pre-suffix part does not end with `U+002E` (`.`).
396 The suffix forcibly sets the type of the literal. There are two valid
397 _floating-point suffixes_, `f32` and `f64` (the 32-bit and 64-bit floating point
398 types), which explicitly determine the type of the literal.
400 The type of an _unsuffixed_ floating-point literal is determined by
403 * If a floating-point type can be _uniquely_ determined from the
404 surrounding program context, the unsuffixed floating-point literal
407 * If the program context under-constrains the type, it defaults to `f64`.
409 * If the program context over-constrains the type, it is considered a
412 Examples of floating-point literals of various forms:
415 123.0f64; // type f64
418 12E+99_f64; // type f64
419 let x: f64 = 2.; // type f64
422 This last example is different because it is not possible to use the suffix
423 syntax with a floating point literal ending in a period. `2.f64` would attempt
424 to call a method named `f64` on `2`.
426 The representation semantics of floating-point numbers are described in
427 ["Machine Types"](#machine-types).
429 #### Boolean literals
431 The two values of the boolean type are written `true` and `false`.
435 Symbols are a general class of printable [tokens](#tokens) that play structural
436 roles in a variety of grammar productions. They are a
437 set of remaining miscellaneous printable tokens that do not
438 otherwise appear as [unary operators](#unary-operator-expressions), [binary
439 operators](#binary-operator-expressions), or [keywords][keywords].
440 They are catalogued in [the Symbols section][symbols] of the Grammar document.
442 [symbols]: grammar.html#symbols
447 A _path_ is a sequence of one or more path components _logically_ separated by
448 a namespace qualifier (`::`). If a path consists of only one component, it may
449 refer to either an [item](#items) or a [variable](#variables) in a local control
450 scope. If a path has multiple components, it refers to an item.
452 Every item has a _canonical path_ within its crate, but the path naming an item
453 is only meaningful within a given crate. There is no global namespace across
454 crates; an item's canonical path merely identifies it within the crate.
456 Two examples of simple paths consisting of only identifier components:
463 Path components are usually [identifiers](#identifiers), but they may
464 also include angle-bracket-enclosed lists of type arguments. In
465 [expression](#expressions) context, the type argument list is given
466 after a `::` namespace qualifier in order to disambiguate it from a
467 relational expression involving the less-than symbol (`<`). In type
468 expression context, the final namespace qualifier is omitted.
470 Two examples of paths with type arguments:
473 # struct HashMap<K, V>(K,V);
475 # fn id<T>(t: T) -> T { t }
476 type T = HashMap<i32,String>; // Type arguments used in a type expression
477 let x = id::<i32>(10); // Type arguments used in a call expression
481 Paths can be denoted with various leading qualifiers to change the meaning of
484 * Paths starting with `::` are considered to be global paths where the
485 components of the path start being resolved from the crate root. Each
486 identifier in the path must resolve to an item.
494 ::a::foo(); // call a's foo function
500 * Paths starting with the keyword `super` begin resolution relative to the
501 parent module. Each further identifier must resolve to an item.
509 super::a::foo(); // call a's foo function
515 * Paths starting with the keyword `self` begin resolution relative to the
516 current module. Each further identifier must resolve to an item.
526 Additionally keyword `super` may be repeated several times after the first
527 `super` or `self` to refer to ancestor modules.
536 super::super::foo(); // call a's foo function
537 self::super::super::foo(); // call a's foo function
547 A number of minor features of Rust are not central enough to have their own
548 syntax, and yet are not implementable as functions. Instead, they are given
549 names, and invoked through a consistent syntax: `some_extension!(...)`.
551 Users of `rustc` can define new syntax extensions in two ways:
553 * [Compiler plugins][plugin] can include arbitrary Rust code that
554 manipulates syntax trees at compile time. Note that the interface
555 for compiler plugins is considered highly unstable.
557 * [Macros](book/macros.html) define new syntax in a higher-level,
562 `macro_rules` allows users to define syntax extension in a declarative way. We
563 call such extensions "macros by example" or simply "macros" — to be distinguished
564 from the "procedural macros" defined in [compiler plugins][plugin].
566 Currently, macros can expand to expressions, statements, items, or patterns.
568 (A `sep_token` is any token other than `*` and `+`. A `non_special_token` is
569 any token other than a delimiter or `$`.)
571 The macro expander looks up macro invocations by name, and tries each macro
572 rule in turn. It transcribes the first successful match. Matching and
573 transcription are closely related to each other, and we will describe them
578 The macro expander matches and transcribes every token that does not begin with
579 a `$` literally, including delimiters. For parsing reasons, delimiters must be
580 balanced, but they are otherwise not special.
582 In the matcher, `$` _name_ `:` _designator_ matches the nonterminal in the Rust
583 syntax named by _designator_. Valid designators are:
585 * `item`: an [item](#items)
586 * `block`: a [block](#block-expressions)
587 * `stmt`: a [statement](#statements)
588 * `pat`: a [pattern](#match-expressions)
589 * `expr`: an [expression](#expressions)
590 * `ty`: a [type](#types)
591 * `ident`: an [identifier](#identifiers)
592 * `path`: a [path](#paths)
593 * `tt`: either side of the `=>` in macro rules
594 * `meta`: the contents of an [attribute](#attributes)
596 In the transcriber, the
597 designator is already known, and so only the name of a matched nonterminal comes
598 after the dollar sign.
600 In both the matcher and transcriber, the Kleene star-like operator indicates
601 repetition. The Kleene star operator consists of `$` and parentheses, optionally
602 followed by a separator token, followed by `*` or `+`. `*` means zero or more
603 repetitions, `+` means at least one repetition. The parentheses are not matched or
604 transcribed. On the matcher side, a name is bound to _all_ of the names it
605 matches, in a structure that mimics the structure of the repetition encountered
606 on a successful match. The job of the transcriber is to sort that structure
609 The rules for transcription of these repetitions are called "Macro By Example".
610 Essentially, one "layer" of repetition is discharged at a time, and all of them
611 must be discharged by the time a name is transcribed. Therefore, `( $( $i:ident
612 ),* ) => ( $i )` is an invalid macro, but `( $( $i:ident ),* ) => ( $( $i:ident
613 ),* )` is acceptable (if trivial).
615 When Macro By Example encounters a repetition, it examines all of the `$`
616 _name_ s that occur in its body. At the "current layer", they all must repeat
617 the same number of times, so ` ( $( $i:ident ),* ; $( $j:ident ),* ) => ( $(
618 ($i,$j) ),* )` is valid if given the argument `(a,b,c ; d,e,f)`, but not
619 `(a,b,c ; d,e)`. The repetition walks through the choices at that layer in
620 lockstep, so the former input transcribes to `(a,d), (b,e), (c,f)`.
622 Nested repetitions are allowed.
624 ### Parsing limitations
626 The parser used by the macro system is reasonably powerful, but the parsing of
627 Rust syntax is restricted in two ways:
629 1. Macro definitions are required to include suitable separators after parsing
630 expressions and other bits of the Rust grammar. This implies that
631 a macro definition like `$i:expr [ , ]` is not legal, because `[` could be part
632 of an expression. A macro definition like `$i:expr,` or `$i:expr;` would be legal,
633 however, because `,` and `;` are legal separators. See [RFC 550] for more information.
634 2. The parser must have eliminated all ambiguity by the time it reaches a `$`
635 _name_ `:` _designator_. This requirement most often affects name-designator
636 pairs when they occur at the beginning of, or immediately after, a `$(...)*`;
637 requiring a distinctive token in front can solve the problem.
639 [RFC 550]: https://github.com/rust-lang/rfcs/blob/master/text/0550-macro-future-proofing.md
641 # Crates and source files
643 Although Rust, like any other language, can be implemented by an interpreter as
644 well as a compiler, the only existing implementation is a compiler,
646 always been designed to be compiled. For these reasons, this section assumes a
649 Rust's semantics obey a *phase distinction* between compile-time and
650 run-time.[^phase-distinction] Semantic rules that have a *static
651 interpretation* govern the success or failure of compilation, while
653 that have a *dynamic interpretation* govern the behavior of the program at
656 [^phase-distinction]: This distinction would also exist in an interpreter.
657 Static checks like syntactic analysis, type checking, and lints should
658 happen before the program is executed regardless of when it is executed.
660 The compilation model centers on artifacts called _crates_. Each compilation
661 processes a single crate in source form, and if successful, produces a single
662 crate in binary form: either an executable or some sort of
663 library.[^cratesourcefile]
665 [^cratesourcefile]: A crate is somewhat analogous to an *assembly* in the
666 ECMA-335 CLI model, a *library* in the SML/NJ Compilation Manager, a *unit*
667 in the Owens and Flatt module system, or a *configuration* in Mesa.
669 A _crate_ is a unit of compilation and linking, as well as versioning,
670 distribution and runtime loading. A crate contains a _tree_ of nested
671 [module](#modules) scopes. The top level of this tree is a module that is
672 anonymous (from the point of view of paths within the module) and any item
673 within a crate has a canonical [module path](#paths) denoting its location
674 within the crate's module tree.
676 The Rust compiler is always invoked with a single source file as input, and
677 always produces a single output crate. The processing of that source file may
678 result in other source files being loaded as modules. Source files have the
681 A Rust source file describes a module, the name and location of which —
682 in the module tree of the current crate — are defined from outside the
683 source file: either by an explicit `mod_item` in a referencing source file, or
684 by the name of the crate itself. Every source file is a module, but not every
685 module needs its own source file: [module definitions](#modules) can be nested
688 Each source file contains a sequence of zero or more `item` definitions, and
689 may optionally begin with any number of [attributes](#items-and-attributes)
690 that apply to the containing module, most of which influence the behavior of
691 the compiler. The anonymous crate module can have additional attributes that
692 apply to the crate as a whole.
695 // Specify the crate name.
696 #![crate_name = "projx"]
698 // Specify the type of output artifact.
699 #![crate_type = "lib"]
701 // Turn on a warning.
702 // This can be done in any module, not just the anonymous crate module.
703 #![warn(non_camel_case_types)]
706 A crate that contains a `main` function can be compiled to an executable. If a
707 `main` function is present, its return type must be `()`
708 ("[unit](#tuple-types)") and it must take no arguments.
710 # Items and attributes
712 Crates contain [items](#items), each of which may have some number of
713 [attributes](#attributes) attached to it.
717 An _item_ is a component of a crate. Items are organized within a crate by a
718 nested set of [modules](#modules). Every crate has a single "outermost"
719 anonymous module; all further items within the crate have [paths](#paths)
720 within the module tree of the crate.
722 Items are entirely determined at compile-time, generally remain fixed during
723 execution, and may reside in read-only memory.
725 There are several kinds of item:
727 * [`extern crate` declarations](#extern-crate-declarations)
728 * [`use` declarations](#use-declarations)
729 * [modules](#modules)
730 * [functions](#functions)
731 * [type definitions](grammar.html#type-definitions)
732 * [structs](#structs)
733 * [enumerations](#enumerations)
734 * [constant items](#constant-items)
735 * [static items](#static-items)
737 * [implementations](#implementations)
739 Some items form an implicit scope for the declaration of sub-items. In other
740 words, within a function or module, declarations of items can (in many cases)
741 be mixed with the statements, control blocks, and similar artifacts that
742 otherwise compose the item body. The meaning of these scoped items is the same
743 as if the item was declared outside the scope — it is still a static item
744 — except that the item's *path name* within the module namespace is
745 qualified by the name of the enclosing item, or is private to the enclosing
746 item (in the case of functions). The grammar specifies the exact locations in
747 which sub-item declarations may appear.
751 All items except modules, constants and statics may be *parameterized* by type.
752 Type parameters are given as a comma-separated list of identifiers enclosed in
753 angle brackets (`<...>`), after the name of the item and before its definition.
754 The type parameters of an item are considered "part of the name", not part of
755 the type of the item. A referencing [path](#paths) must (in principle) provide
756 type arguments as a list of comma-separated types enclosed within angle
757 brackets, in order to refer to the type-parameterized item. In practice, the
758 type-inference system can usually infer such argument types from context. There
759 are no general type-parametric types, only type-parametric items. That is, Rust
760 has no notion of type abstraction: there are no higher-ranked (or "forall") types
761 abstracted over other types, though higher-ranked types do exist for lifetimes.
765 A module is a container for zero or more [items](#items).
767 A _module item_ is a module, surrounded in braces, named, and prefixed with the
768 keyword `mod`. A module item introduces a new, named module into the tree of
769 modules making up a crate. Modules can nest arbitrarily.
771 An example of a module:
775 type Complex = (f64, f64);
776 fn sin(f: f64) -> f64 {
780 fn cos(f: f64) -> f64 {
784 fn tan(f: f64) -> f64 {
791 Modules and types share the same namespace. Declaring a named type with
792 the same name as a module in scope is forbidden: that is, a type definition,
793 trait, struct, enumeration, or type parameter can't shadow the name of a module
794 in scope, or vice versa.
796 A module without a body is loaded from an external file, by default with the
797 same name as the module, plus the `.rs` extension. When a nested submodule is
798 loaded from an external file, it is loaded from a subdirectory path that
799 mirrors the module hierarchy.
802 // Load the `vec` module from `vec.rs`
806 // Load the `local_data` module from `thread/local_data.rs`
807 // or `thread/local_data/mod.rs`.
812 The directories and files used for loading external file modules can be
813 influenced with the `path` attribute.
816 #[path = "thread_files"]
818 // Load the `local_data` module from `thread_files/tls.rs`
824 #### Extern crate declarations
826 An _`extern crate` declaration_ specifies a dependency on an external crate.
827 The external crate is then bound into the declaring scope as the `ident`
828 provided in the `extern_crate_decl`.
830 The external crate is resolved to a specific `soname` at compile time, and a
831 runtime linkage requirement to that `soname` is passed to the linker for
832 loading at runtime. The `soname` is resolved at compile time by scanning the
833 compiler's library path and matching the optional `crateid` provided against
834 the `crateid` attributes that were declared on the external crate when it was
835 compiled. If no `crateid` is provided, a default `name` attribute is assumed,
836 equal to the `ident` given in the `extern_crate_decl`.
838 Three examples of `extern crate` declarations:
843 extern crate std; // equivalent to: extern crate std as std;
845 extern crate std as ruststd; // linking to 'std' under another name
848 #### Use declarations
850 A _use declaration_ creates one or more local name bindings synonymous with
851 some other [path](#paths). Usually a `use` declaration is used to shorten the
852 path required to refer to a module item. These declarations may appear in
853 [modules](#modules) and [blocks](grammar.html#block-expressions), usually at the top.
855 > **Note**: Unlike in many languages,
856 > `use` declarations in Rust do *not* declare linkage dependency with external crates.
857 > Rather, [`extern crate` declarations](#extern-crate-declarations) declare linkage dependencies.
859 Use declarations support a number of convenient shortcuts:
861 * Rebinding the target name as a new local name, using the syntax `use p::q::r as x;`
862 * Simultaneously binding a list of paths differing only in their final element,
863 using the glob-like brace syntax `use a::b::{c,d,e,f};`
864 * Binding all paths matching a given prefix, using the asterisk wildcard syntax
866 * Simultaneously binding a list of paths differing only in their final element
867 and their immediate parent module, using the `self` keyword, such as
868 `use a::b::{self, c, d};`
870 An example of `use` declarations:
873 use std::option::Option::{Some, None};
874 use std::collections::hash_map::{self, HashMap};
877 fn bar(map1: HashMap<String, usize>, map2: hash_map::HashMap<String, usize>){}
880 // Equivalent to 'foo(vec![std::option::Option::Some(1.0f64),
881 // std::option::Option::None]);'
882 foo(vec![Some(1.0f64), None]);
884 // Both `hash_map` and `HashMap` are in scope.
885 let map1 = HashMap::new();
886 let map2 = hash_map::HashMap::new();
891 Like items, `use` declarations are private to the containing module, by
892 default. Also like items, a `use` declaration can be public, if qualified by
893 the `pub` keyword. Such a `use` declaration serves to _re-export_ a name. A
894 public `use` declaration can therefore _redirect_ some public name to a
895 different target definition: even a definition with a private canonical path,
896 inside a different module. If a sequence of such redirections form a cycle or
897 cannot be resolved unambiguously, they represent a compile-time error.
899 An example of re-exporting:
904 pub use quux::foo::{bar, baz};
913 In this example, the module `quux` re-exports two public names defined in
916 Also note that the paths contained in `use` items are relative to the crate
917 root. So, in the previous example, the `use` refers to `quux::foo::{bar,
918 baz}`, and not simply to `foo::{bar, baz}`. This also means that top-level
919 module declarations should be at the crate root if direct usage of the declared
920 modules within `use` items is desired. It is also possible to use `self` and
921 `super` at the beginning of a `use` item to refer to the current and direct
922 parent modules respectively. All rules regarding accessing declared modules in
923 `use` declarations apply to both module declarations and `extern crate`
926 An example of what will and will not work for `use` items:
929 # #![allow(unused_imports)]
930 use foo::baz::foobaz; // good: foo is at the root of the crate
938 use foo::example::iter; // good: foo is at crate root
939 // use example::iter; // bad: example is not at the crate root
940 use self::baz::foobaz; // good: self refers to module 'foo'
941 use foo::bar::foobar; // good: foo is at crate root
948 use super::bar::foobar; // good: super refers to module 'foo'
958 A _function item_ defines a sequence of [statements](#statements) and a
959 final [expression](#expressions), along with a name and a set of
960 parameters. Other than a name, all these are optional.
961 Functions are declared with the keyword `fn`. Functions may declare a
962 set of *input* [*variables*](#variables) as parameters, through which the caller
963 passes arguments into the function, and the *output* [*type*](#types)
964 of the value the function will return to its caller on completion.
966 A function may also be copied into a first-class *value*, in which case the
967 value has the corresponding [*function type*](#function-types), and can be used
968 otherwise exactly as a function item (with a minor additional cost of calling
969 the function indirectly).
971 Every control path in a function logically ends with a `return` expression or a
972 diverging expression. If the outermost block of a function has a
973 value-producing expression in its final-expression position, that expression is
974 interpreted as an implicit `return` expression applied to the final-expression.
976 An example of a function:
979 fn add(x: i32, y: i32) -> i32 {
984 As with `let` bindings, function arguments are irrefutable patterns, so any
985 pattern that is valid in a let binding is also valid as an argument.
988 fn first((value, _): (i32, i32)) -> i32 { value }
992 #### Generic functions
994 A _generic function_ allows one or more _parameterized types_ to appear in its
995 signature. Each type parameter must be explicitly declared in an
996 angle-bracket-enclosed and comma-separated list, following the function name.
999 // foo is generic over A and B
1001 fn foo<A, B>(x: A, y: B) {
1004 Inside the function signature and body, the name of the type parameter can be
1005 used as a type name. [Trait](#traits) bounds can be specified for type parameters
1006 to allow methods with that trait to be called on values of that type. This is
1007 specified using the `where` syntax:
1010 fn foo<T>(x: T) where T: Debug {
1013 When a generic function is referenced, its type is instantiated based on the
1014 context of the reference. For example, calling the `foo` function here:
1017 use std::fmt::Debug;
1019 fn foo<T>(x: &[T]) where T: Debug {
1027 will instantiate type parameter `T` with `i32`.
1029 The type parameters can also be explicitly supplied in a trailing
1030 [path](#paths) component after the function name. This might be necessary if
1031 there is not sufficient context to determine the type parameters. For example,
1032 `mem::size_of::<u32>() == 4`.
1034 #### Diverging functions
1036 A special kind of function can be declared with a `!` character where the
1037 output type would normally be. For example:
1040 fn my_err(s: &str) -> ! {
1046 We call such functions "diverging" because they never return a value to the
1047 caller. Every control path in a diverging function must end with a `panic!()` or
1048 a call to another diverging function on every control path. The `!` annotation
1049 does *not* denote a type.
1051 It might be necessary to declare a diverging function because as mentioned
1052 previously, the typechecker checks that every control path in a function ends
1053 with a [`return`](#return-expressions) or diverging expression. So, if `my_err`
1054 were declared without the `!` annotation, the following code would not
1058 # fn my_err(s: &str) -> ! { panic!() }
1060 fn f(i: i32) -> i32 {
1065 my_err("Bad number!");
1070 This will not compile without the `!` annotation on `my_err`, since the `else`
1071 branch of the conditional in `f` does not return an `i32`, as required by the
1072 signature of `f`. Adding the `!` annotation to `my_err` informs the
1073 typechecker that, should control ever enter `my_err`, no further type judgments
1074 about `f` need to hold, since control will never resume in any context that
1075 relies on those judgments. Thus the return type on `f` only needs to reflect
1076 the `if` branch of the conditional.
1078 #### Extern functions
1080 Extern functions are part of Rust's foreign function interface, providing the
1081 opposite functionality to [external blocks](#external-blocks). Whereas
1082 external blocks allow Rust code to call foreign code, extern functions with
1083 bodies defined in Rust code _can be called by foreign code_. They are defined
1084 in the same way as any other Rust function, except that they have the `extern`
1088 // Declares an extern fn, the ABI defaults to "C"
1089 extern fn new_i32() -> i32 { 0 }
1091 // Declares an extern fn with "stdcall" ABI
1092 extern "stdcall" fn new_i32_stdcall() -> i32 { 0 }
1095 Unlike normal functions, extern fns have type `extern "ABI" fn()`. This is the
1096 same type as the functions declared in an extern block.
1099 # extern fn new_i32() -> i32 { 0 }
1100 let fptr: extern "C" fn() -> i32 = new_i32;
1103 Extern functions may be called directly from Rust code as Rust uses large,
1104 contiguous stack segments like C.
1108 A _type alias_ defines a new name for an existing [type](#types). Type
1109 aliases are declared with the keyword `type`. Every value has a single,
1110 specific type, but may implement several different traits, or be compatible with
1111 several different type constraints.
1113 For example, the following defines the type `Point` as a synonym for the type
1114 `(u8, u8)`, the type of pairs of unsigned 8 bit integers:
1117 type Point = (u8, u8);
1118 let p: Point = (41, 68);
1121 Currently a type alias to an enum type cannot be used to qualify the
1127 let _: F = E::A; // OK
1128 // let _: F = F::A; // Doesn't work
1133 A _struct_ is a nominal [struct type](#struct-types) defined with the
1136 An example of a `struct` item and its use:
1139 struct Point {x: i32, y: i32}
1140 let p = Point {x: 10, y: 11};
1144 A _tuple struct_ is a nominal [tuple type](#tuple-types), also defined with
1145 the keyword `struct`. For example:
1148 struct Point(i32, i32);
1149 let p = Point(10, 11);
1150 let px: i32 = match p { Point(x, _) => x };
1153 A _unit-like struct_ is a struct without any fields, defined by leaving off
1154 the list of fields entirely. Such a struct implicitly defines a constant of
1155 its type with the same name. For example:
1159 let c = [Cookie, Cookie {}, Cookie, Cookie {}];
1166 const Cookie: Cookie = Cookie {};
1167 let c = [Cookie, Cookie {}, Cookie, Cookie {}];
1170 The precise memory layout of a struct is not specified. One can specify a
1171 particular layout using the [`repr` attribute](#ffi-attributes).
1175 An _enumeration_ is a simultaneous definition of a nominal [enumerated
1176 type](#enumerated-types) as well as a set of *constructors*, that can be used
1177 to create or pattern-match values of the corresponding enumerated type.
1179 Enumerations are declared with the keyword `enum`.
1181 An example of an `enum` item and its use:
1189 let mut a: Animal = Animal::Dog;
1193 Enumeration constructors can have either named or unnamed fields:
1198 Cat { name: String, weight: f64 },
1201 let mut a: Animal = Animal::Dog("Cocoa".to_string(), 37.2);
1202 a = Animal::Cat { name: "Spotty".to_string(), weight: 2.7 };
1205 In this example, `Cat` is a _struct-like enum variant_,
1206 whereas `Dog` is simply called an enum variant.
1208 Each enum value has a _discriminant_ which is an integer associated to it. You
1209 can specify it explicitly:
1217 The right hand side of the specification is interpreted as an `isize` value,
1218 but the compiler is allowed to use a smaller type in the actual memory layout.
1219 The [`repr` attribute](#ffi-attributes) can be added in order to change
1220 the type of the right hand side and specify the memory layout.
1222 If a discriminant isn't specified, they start at zero, and add one for each
1225 You can cast an enum to get its discriminant:
1228 # enum Foo { Bar = 123 }
1229 let x = Foo::Bar as u32; // x is now 123u32
1232 This only works as long as none of the variants have data attached. If
1233 it were `Bar(i32)`, this is disallowed.
1237 A *constant item* is a named _constant value_ which is not associated with a
1238 specific memory location in the program. Constants are essentially inlined
1239 wherever they are used, meaning that they are copied directly into the relevant
1240 context when used. References to the same constant are not necessarily
1241 guaranteed to refer to the same memory address.
1243 Constant values must not have destructors, and otherwise permit most forms of
1244 data. Constants may refer to the address of other constants, in which case the
1245 address will have the `static` lifetime. The compiler is, however, still at
1246 liberty to translate the constant many times, so the address referred to may not
1249 Constants must be explicitly typed. The type may be `bool`, `char`, a number, or
1250 a type derived from those primitive types. The derived types are references with
1251 the `static` lifetime, fixed-size arrays, tuples, enum variants, and structs.
1254 const BIT1: u32 = 1 << 0;
1255 const BIT2: u32 = 1 << 1;
1257 const BITS: [u32; 2] = [BIT1, BIT2];
1258 const STRING: &'static str = "bitstring";
1260 struct BitsNStrings<'a> {
1265 const BITS_N_STRINGS: BitsNStrings<'static> = BitsNStrings {
1273 A *static item* is similar to a *constant*, except that it represents a precise
1274 memory location in the program. A static is never "inlined" at the usage site,
1275 and all references to it refer to the same memory location. Static items have
1276 the `static` lifetime, which outlives all other lifetimes in a Rust program.
1277 Static items may be placed in read-only memory if they do not contain any
1278 interior mutability.
1280 Statics may contain interior mutability through the `UnsafeCell` language item.
1281 All access to a static is safe, but there are a number of restrictions on
1284 * Statics may not contain any destructors.
1285 * The types of static values must ascribe to `Sync` to allow thread-safe access.
1286 * Statics may not refer to other statics by value, only by reference.
1287 * Constants cannot refer to statics.
1289 Constants should in general be preferred over statics, unless large amounts of
1290 data are being stored, or single-address and mutability properties are required.
1292 #### Mutable statics
1294 If a static item is declared with the `mut` keyword, then it is allowed to
1295 be modified by the program. One of Rust's goals is to make concurrency bugs
1296 hard to run into, and this is obviously a very large source of race conditions
1297 or other bugs. For this reason, an `unsafe` block is required when either
1298 reading or writing a mutable static variable. Care should be taken to ensure
1299 that modifications to a mutable static are safe with respect to other threads
1300 running in the same process.
1302 Mutable statics are still very useful, however. They can be used with C
1303 libraries and can also be bound from C libraries (in an `extern` block).
1306 # fn atomic_add(_: &mut u32, _: u32) -> u32 { 2 }
1308 static mut LEVELS: u32 = 0;
1310 // This violates the idea of no shared state, and this doesn't internally
1311 // protect against races, so this function is `unsafe`
1312 unsafe fn bump_levels_unsafe1() -> u32 {
1318 // Assuming that we have an atomic_add function which returns the old value,
1319 // this function is "safe" but the meaning of the return value may not be what
1320 // callers expect, so it's still marked as `unsafe`
1321 unsafe fn bump_levels_unsafe2() -> u32 {
1322 return atomic_add(&mut LEVELS, 1);
1326 Mutable statics have the same restrictions as normal statics, except that the
1327 type of the value is not required to ascribe to `Sync`.
1331 A _trait_ describes an abstract interface that types can
1332 implement. This interface consists of associated items, which come in
1339 Associated functions whose first parameter is named `self` are called
1340 methods and may be invoked using `.` notation (e.g., `x.foo()`).
1342 All traits define an implicit type parameter `Self` that refers to
1343 "the type that is implementing this interface". Traits may also
1344 contain additional type parameters. These type parameters (including
1345 `Self`) may be constrained by other traits and so forth as usual.
1347 Trait bounds on `Self` are considered "supertraits". These are
1348 required to be acyclic. Supertraits are somewhat different from other
1349 constraints in that they affect what methods are available in the
1350 vtable when the trait is used as a [trait object](#trait-objects).
1352 Traits are implemented for specific types through separate
1353 [implementations](#implementations).
1355 Consider the following trait:
1358 # type Surface = i32;
1359 # type BoundingBox = i32;
1361 fn draw(&self, Surface);
1362 fn bounding_box(&self) -> BoundingBox;
1366 This defines a trait with two methods. All values that have
1367 [implementations](#implementations) of this trait in scope can have their
1368 `draw` and `bounding_box` methods called, using `value.bounding_box()`
1369 [syntax](#method-call-expressions).
1371 Traits can include default implementations of methods, as in:
1376 fn baz(&self) { println!("We called baz."); }
1380 Here the `baz` method has a default implementation, so types that implement
1381 `Foo` need only implement `bar`. It is also possible for implementing types
1382 to override a method that has a default implementation.
1384 Type parameters can be specified for a trait to make it generic. These appear
1385 after the trait name, using the same syntax used in [generic
1386 functions](#generic-functions).
1390 fn len(&self) -> u32;
1391 fn elt_at(&self, n: u32) -> T;
1392 fn iter<F>(&self, F) where F: Fn(T);
1396 It is also possible to define associated types for a trait. Consider the
1397 following example of a `Container` trait. Notice how the type is available
1398 for use in the method signatures:
1404 fn insert(&mut self, Self::E);
1408 In order for a type to implement this trait, it must not only provide
1409 implementations for every method, but it must specify the type `E`. Here's
1410 an implementation of `Container` for the standard library type `Vec`:
1415 # fn empty() -> Self;
1416 # fn insert(&mut self, Self::E);
1418 impl<T> Container for Vec<T> {
1420 fn empty() -> Vec<T> { Vec::new() }
1421 fn insert(&mut self, x: T) { self.push(x); }
1425 Generic functions may use traits as _bounds_ on their type parameters. This
1426 will have two effects:
1428 - Only types that have the trait may instantiate the parameter.
1429 - Within the generic function, the methods of the trait can be
1430 called on values that have the parameter's type.
1435 # type Surface = i32;
1436 # trait Shape { fn draw(&self, Surface); }
1437 fn draw_twice<T: Shape>(surface: Surface, sh: T) {
1443 Traits also define a [trait object](#trait-objects) with the same
1444 name as the trait. Values of this type are created by coercing from a
1445 pointer of some specific type to a pointer of trait type. For example,
1446 `&T` could be coerced to `&Shape` if `T: Shape` holds (and similarly
1447 for `Box<T>`). This coercion can either be implicit or
1448 [explicit](#type-cast-expressions). Here is an example of an explicit
1453 impl Shape for i32 { }
1454 let mycircle = 0i32;
1455 let myshape: Box<Shape> = Box::new(mycircle) as Box<Shape>;
1458 The resulting value is a box containing the value that was cast, along with
1459 information that identifies the methods of the implementation that was used.
1460 Values with a trait type can have [methods called](#method-call-expressions) on
1461 them, for any method in the trait, and can be used to instantiate type
1462 parameters that are bounded by the trait.
1464 Trait methods may be static, which means that they lack a `self` argument.
1465 This means that they can only be called with function call syntax (`f(x)`) and
1466 not method call syntax (`obj.f()`). The way to refer to the name of a static
1467 method is to qualify it with the trait name, treating the trait name like a
1468 module. For example:
1472 fn from_i32(n: i32) -> Self;
1475 fn from_i32(n: i32) -> f64 { n as f64 }
1477 let x: f64 = Num::from_i32(42);
1480 Traits may inherit from other traits. Consider the following example:
1483 trait Shape { fn area(&self) -> f64; }
1484 trait Circle : Shape { fn radius(&self) -> f64; }
1487 The syntax `Circle : Shape` means that types that implement `Circle` must also
1488 have an implementation for `Shape`. Multiple supertraits are separated by `+`,
1489 `trait Circle : Shape + PartialEq { }`. In an implementation of `Circle` for a
1490 given type `T`, methods can refer to `Shape` methods, since the typechecker
1491 checks that any type with an implementation of `Circle` also has an
1492 implementation of `Shape`:
1497 trait Shape { fn area(&self) -> f64; }
1498 trait Circle : Shape { fn radius(&self) -> f64; }
1499 impl Shape for Foo {
1500 fn area(&self) -> f64 {
1504 impl Circle for Foo {
1505 fn radius(&self) -> f64 {
1506 println!("calling area: {}", self.area());
1516 In type-parameterized functions, methods of the supertrait may be called on
1517 values of subtrait-bound type parameters. Referring to the previous example of
1518 `trait Circle : Shape`:
1521 # trait Shape { fn area(&self) -> f64; }
1522 # trait Circle : Shape { fn radius(&self) -> f64; }
1523 fn radius_times_area<T: Circle>(c: T) -> f64 {
1524 // `c` is both a Circle and a Shape
1525 c.radius() * c.area()
1529 Likewise, supertrait methods may also be called on trait objects.
1532 # trait Shape { fn area(&self) -> f64; }
1533 # trait Circle : Shape { fn radius(&self) -> f64; }
1534 # impl Shape for i32 { fn area(&self) -> f64 { 0.0 } }
1535 # impl Circle for i32 { fn radius(&self) -> f64 { 0.0 } }
1536 # let mycircle = 0i32;
1537 let mycircle = Box::new(mycircle) as Box<Circle>;
1538 let nonsense = mycircle.radius() * mycircle.area();
1543 An _implementation_ is an item that implements a [trait](#traits) for a
1546 Implementations are defined with the keyword `impl`.
1549 # #[derive(Copy, Clone)]
1550 # struct Point {x: f64, y: f64};
1551 # type Surface = i32;
1552 # struct BoundingBox {x: f64, y: f64, width: f64, height: f64};
1553 # trait Shape { fn draw(&self, Surface); fn bounding_box(&self) -> BoundingBox; }
1554 # fn do_draw_circle(s: Surface, c: Circle) { }
1560 impl Copy for Circle {}
1562 impl Clone for Circle {
1563 fn clone(&self) -> Circle { *self }
1566 impl Shape for Circle {
1567 fn draw(&self, s: Surface) { do_draw_circle(s, *self); }
1568 fn bounding_box(&self) -> BoundingBox {
1569 let r = self.radius;
1571 x: self.center.x - r,
1572 y: self.center.y - r,
1580 It is possible to define an implementation without referring to a trait. The
1581 methods in such an implementation can only be used as direct calls on the values
1582 of the type that the implementation targets. In such an implementation, the
1583 trait type and `for` after `impl` are omitted. Such implementations are limited
1584 to nominal types (enums, structs, trait objects), and the implementation must
1585 appear in the same crate as the `self` type:
1588 struct Point {x: i32, y: i32}
1592 println!("Point is at ({}, {})", self.x, self.y);
1596 let my_point = Point {x: 10, y:11};
1600 When a trait _is_ specified in an `impl`, all methods declared as part of the
1601 trait must be implemented, with matching types and type parameter counts.
1603 An implementation can take type parameters, which can be different from the
1604 type parameters taken by the trait it implements. Implementation parameters
1605 are written after the `impl` keyword.
1608 # trait Seq<T> { fn dummy(&self, _: T) { } }
1609 impl<T> Seq<T> for Vec<T> {
1612 impl Seq<bool> for u32 {
1613 /* Treat the integer as a sequence of bits */
1619 External blocks form the basis for Rust's foreign function interface.
1620 Declarations in an external block describe symbols in external, non-Rust
1623 Functions within external blocks are declared in the same way as other Rust
1624 functions, with the exception that they may not have a body and are instead
1625 terminated by a semicolon.
1627 Functions within external blocks may be called by Rust code, just like
1628 functions defined in Rust. The Rust compiler automatically translates between
1629 the Rust ABI and the foreign ABI.
1631 A number of [attributes](#attributes) control the behavior of external blocks.
1633 By default external blocks assume that the library they are calling uses the
1634 standard C "cdecl" ABI. Other ABIs may be specified using an `abi` string, as
1638 // Interface to the Windows API
1639 extern "stdcall" { }
1642 The `link` attribute allows the name of the library to be specified. When
1643 specified the compiler will attempt to link against the native library of the
1647 #[link(name = "crypto")]
1651 The type of a function declared in an extern block is `extern "abi" fn(A1, ...,
1652 An) -> R`, where `A1...An` are the declared types of its arguments and `R` is
1653 the declared return type.
1655 It is valid to add the `link` attribute on an empty extern block. You can use
1656 this to satisfy the linking requirements of extern blocks elsewhere in your code
1657 (including upstream crates) instead of adding the attribute to each extern block.
1659 ## Visibility and Privacy
1661 These two terms are often used interchangeably, and what they are attempting to
1662 convey is the answer to the question "Can this item be used at this location?"
1664 Rust's name resolution operates on a global hierarchy of namespaces. Each level
1665 in the hierarchy can be thought of as some item. The items are one of those
1666 mentioned above, but also include external crates. Declaring or defining a new
1667 module can be thought of as inserting a new tree into the hierarchy at the
1668 location of the definition.
1670 To control whether interfaces can be used across modules, Rust checks each use
1671 of an item to see whether it should be allowed or not. This is where privacy
1672 warnings are generated, or otherwise "you used a private item of another module
1673 and weren't allowed to."
1675 By default, everything in Rust is *private*, with one exception. Enum variants
1676 in a `pub` enum are also public by default. When an item is declared as `pub`,
1677 it can be thought of as being accessible to the outside world. For example:
1681 // Declare a private struct
1684 // Declare a public struct with a private field
1689 // Declare a public enum with two public variants
1691 PubliclyAccessibleState,
1692 PubliclyAccessibleState2,
1696 With the notion of an item being either public or private, Rust allows item
1697 accesses in two cases:
1699 1. If an item is public, then it can be used externally through any of its
1701 2. If an item is private, it may be accessed by the current module and its
1704 These two cases are surprisingly powerful for creating module hierarchies
1705 exposing public APIs while hiding internal implementation details. To help
1706 explain, here's a few use cases and what they would entail:
1708 * A library developer needs to expose functionality to crates which link
1709 against their library. As a consequence of the first case, this means that
1710 anything which is usable externally must be `pub` from the root down to the
1711 destination item. Any private item in the chain will disallow external
1714 * A crate needs a global available "helper module" to itself, but it doesn't
1715 want to expose the helper module as a public API. To accomplish this, the
1716 root of the crate's hierarchy would have a private module which then
1717 internally has a "public API". Because the entire crate is a descendant of
1718 the root, then the entire local crate can access this private module through
1721 * When writing unit tests for a module, it's often a common idiom to have an
1722 immediate child of the module to-be-tested named `mod test`. This module
1723 could access any items of the parent module through the second case, meaning
1724 that internal implementation details could also be seamlessly tested from the
1727 In the second case, it mentions that a private item "can be accessed" by the
1728 current module and its descendants, but the exact meaning of accessing an item
1729 depends on what the item is. Accessing a module, for example, would mean
1730 looking inside of it (to import more items). On the other hand, accessing a
1731 function would mean that it is invoked. Additionally, path expressions and
1732 import statements are considered to access an item in the sense that the
1733 import/expression is only valid if the destination is in the current visibility
1736 Here's an example of a program which exemplifies the three cases outlined
1740 // This module is private, meaning that no external crate can access this
1741 // module. Because it is private at the root of this current crate, however, any
1742 // module in the crate may access any publicly visible item in this module.
1743 mod crate_helper_module {
1745 // This function can be used by anything in the current crate
1746 pub fn crate_helper() {}
1748 // This function *cannot* be used by anything else in the crate. It is not
1749 // publicly visible outside of the `crate_helper_module`, so only this
1750 // current module and its descendants may access it.
1751 fn implementation_detail() {}
1754 // This function is "public to the root" meaning that it's available to external
1755 // crates linking against this one.
1756 pub fn public_api() {}
1758 // Similarly to 'public_api', this module is public so external crates may look
1761 use crate_helper_module;
1763 pub fn my_method() {
1764 // Any item in the local crate may invoke the helper module's public
1765 // interface through a combination of the two rules above.
1766 crate_helper_module::crate_helper();
1769 // This function is hidden to any module which is not a descendant of
1771 fn my_implementation() {}
1777 fn test_my_implementation() {
1778 // Because this module is a descendant of `submodule`, it's allowed
1779 // to access private items inside of `submodule` without a privacy
1781 super::my_implementation();
1789 For a Rust program to pass the privacy checking pass, all paths must be valid
1790 accesses given the two rules above. This includes all use statements,
1791 expressions, types, etc.
1793 ### Re-exporting and Visibility
1795 Rust allows publicly re-exporting items through a `pub use` directive. Because
1796 this is a public directive, this allows the item to be used in the current
1797 module through the rules above. It essentially allows public access into the
1798 re-exported item. For example, this program is valid:
1801 pub use self::implementation::api;
1803 mod implementation {
1812 This means that any external crate referencing `implementation::api::f` would
1813 receive a privacy violation, while the path `api::f` would be allowed.
1815 When re-exporting a private item, it can be thought of as allowing the "privacy
1816 chain" being short-circuited through the reexport instead of passing through
1817 the namespace hierarchy as it normally would.
1821 Any item declaration may have an _attribute_ applied to it. Attributes in Rust
1822 are modeled on Attributes in ECMA-335, with the syntax coming from ECMA-334
1823 (C#). An attribute is a general, free-form metadatum that is interpreted
1824 according to name, convention, and language and compiler version. Attributes
1825 may appear as any of:
1827 * A single identifier, the attribute name
1828 * An identifier followed by the equals sign '=' and a literal, providing a
1830 * An identifier followed by a parenthesized list of sub-attribute arguments
1832 Attributes with a bang ("!") after the hash ("#") apply to the item that the
1833 attribute is declared within. Attributes that do not have a bang after the hash
1834 apply to the item that follows the attribute.
1836 An example of attributes:
1839 // General metadata applied to the enclosing module or crate.
1840 #![crate_type = "lib"]
1842 // A function marked as a unit test
1848 // A conditionally-compiled module
1849 #[cfg(target_os="linux")]
1854 // A lint attribute used to suppress a warning/error
1855 #[allow(non_camel_case_types)]
1859 > **Note:** At some point in the future, the compiler will distinguish between
1860 > language-reserved and user-available attributes. Until then, there is
1861 > effectively no difference between an attribute handled by a loadable syntax
1862 > extension and the compiler.
1864 ### Crate-only attributes
1866 - `crate_name` - specify the crate's crate name.
1867 - `crate_type` - see [linkage](#linkage).
1868 - `feature` - see [compiler features](#compiler-features).
1869 - `no_builtins` - disable optimizing certain code patterns to invocations of
1870 library functions that are assumed to exist
1871 - `no_main` - disable emitting the `main` symbol. Useful when some other
1872 object being linked to defines `main`.
1873 - `no_start` - disable linking to the `native` crate, which specifies the
1874 "start" language item.
1875 - `no_std` - disable linking to the `std` crate.
1876 - `plugin` - load a list of named crates as compiler plugins, e.g.
1877 `#![plugin(foo, bar)]`. Optional arguments for each plugin,
1878 i.e. `#![plugin(foo(... args ...))]`, are provided to the plugin's
1879 registrar function. The `plugin` feature gate is required to use
1881 - `recursion_limit` - Sets the maximum depth for potentially
1882 infinitely-recursive compile-time operations like
1883 auto-dereference or macro expansion. The default is
1884 `#![recursion_limit="64"]`.
1886 ### Module-only attributes
1888 - `no_implicit_prelude` - disable injecting `use std::prelude::*` in this
1890 - `path` - specifies the file to load the module from. `#[path="foo.rs"] mod
1891 bar;` is equivalent to `mod bar { /* contents of foo.rs */ }`. The path is
1892 taken relative to the directory that the current module is in.
1894 ### Function-only attributes
1896 - `main` - indicates that this function should be passed to the entry point,
1897 rather than the function in the crate root named `main`.
1898 - `plugin_registrar` - mark this function as the registration point for
1899 [compiler plugins][plugin], such as loadable syntax extensions.
1900 - `start` - indicates that this function should be used as the entry point,
1901 overriding the "start" language item. See the "start" [language
1902 item](#language-items) for more details.
1903 - `test` - indicates that this function is a test function, to only be compiled
1904 in case of `--test`.
1905 - `should_panic` - indicates that this test function should panic, inverting the success condition.
1906 - `cold` - The function is unlikely to be executed, so optimize it (and calls
1908 - `naked` - The function utilizes a custom ABI or custom inline ASM that requires
1909 epilogue and prologue to be skipped.
1911 ### Static-only attributes
1913 - `thread_local` - on a `static mut`, this signals that the value of this
1914 static may change depending on the current thread. The exact consequences of
1915 this are implementation-defined.
1919 On an `extern` block, the following attributes are interpreted:
1921 - `link_args` - specify arguments to the linker, rather than just the library
1922 name and type. This is feature gated and the exact behavior is
1923 implementation-defined (due to variety of linker invocation syntax).
1924 - `link` - indicate that a native library should be linked to for the
1925 declarations in this block to be linked correctly. `link` supports an optional
1926 `kind` key with three possible values: `dylib`, `static`, and `framework`. See
1927 [external blocks](#external-blocks) for more about external blocks. Two
1928 examples: `#[link(name = "readline")]` and
1929 `#[link(name = "CoreFoundation", kind = "framework")]`.
1930 - `linked_from` - indicates what native library this block of FFI items is
1931 coming from. This attribute is of the form `#[linked_from = "foo"]` where
1932 `foo` is the name of a library in either `#[link]` or a `-l` flag. This
1933 attribute is currently required to export symbols from a Rust dynamic library
1934 on Windows, and it is feature gated behind the `linked_from` feature.
1936 On declarations inside an `extern` block, the following attributes are
1939 - `link_name` - the name of the symbol that this function or static should be
1941 - `linkage` - on a static, this specifies the [linkage
1942 type](http://llvm.org/docs/LangRef.html#linkage-types).
1946 - `repr` - on C-like enums, this sets the underlying type used for
1947 representation. Takes one argument, which is the primitive
1948 type this enum should be represented for, or `C`, which specifies that it
1949 should be the default `enum` size of the C ABI for that platform. Note that
1950 enum representation in C is undefined, and this may be incorrect when the C
1951 code is compiled with certain flags.
1955 - `repr` - specifies the representation to use for this struct. Takes a list
1956 of options. The currently accepted ones are `C` and `packed`, which may be
1957 combined. `C` will use a C ABI compatible struct layout, and `packed` will
1958 remove any padding between fields (note that this is very fragile and may
1959 break platforms which require aligned access).
1961 ### Macro-related attributes
1963 - `macro_use` on a `mod` — macros defined in this module will be visible in the
1964 module's parent, after this module has been included.
1966 - `macro_use` on an `extern crate` — load macros from this crate. An optional
1967 list of names `#[macro_use(foo, bar)]` restricts the import to just those
1968 macros named. The `extern crate` must appear at the crate root, not inside
1969 `mod`, which ensures proper function of the [`$crate` macro
1970 variable](book/macros.html#the-variable-crate).
1972 - `macro_reexport` on an `extern crate` — re-export the named macros.
1974 - `macro_export` - export a macro for cross-crate usage.
1976 - `no_link` on an `extern crate` — even if we load this crate for macros, don't
1977 link it into the output.
1979 See the [macros section of the
1980 book](book/macros.html#scoping-and-macro-importexport) for more information on
1984 ### Miscellaneous attributes
1986 - `export_name` - on statics and functions, this determines the name of the
1988 - `link_section` - on statics and functions, this specifies the section of the
1989 object file that this item's contents will be placed into.
1990 - `no_mangle` - on any item, do not apply the standard name mangling. Set the
1991 symbol for this item to its identifier.
1992 - `simd` - on certain tuple structs, derive the arithmetic operators, which
1993 lower to the target's SIMD instructions, if any; the `simd` feature gate
1994 is necessary to use this attribute.
1995 - `unsafe_destructor_blind_to_params` - on `Drop::drop` method, asserts that the
1996 destructor code (and all potential specializations of that code) will
1997 never attempt to read from nor write to any references with lifetimes
1998 that come in via generic parameters. This is a constraint we cannot
1999 currently express via the type system, and therefore we rely on the
2000 programmer to assert that it holds. Adding this to a Drop impl causes
2001 the associated destructor to be considered "uninteresting" by the
2002 Drop-Check rule, and thus it can help sidestep data ordering
2003 constraints that would otherwise be introduced by the Drop-Check
2004 rule. Such sidestepping of the constraints, if done incorrectly, can
2005 lead to undefined behavior (in the form of reading or writing to data
2006 outside of its dynamic extent), and thus this attribute has the word
2007 "unsafe" in its name. To use this, the
2008 `unsafe_destructor_blind_to_params` feature gate must be enabled.
2009 - `unsafe_no_drop_flag` - on structs, remove the flag that prevents
2010 destructors from being run twice. Destructors might be run multiple times on
2011 the same object with this attribute. To use this, the `unsafe_no_drop_flag` feature
2012 gate must be enabled.
2013 - `doc` - Doc comments such as `/// foo` are equivalent to `#[doc = "foo"]`.
2014 - `rustc_on_unimplemented` - Write a custom note to be shown along with the error
2015 when the trait is found to be unimplemented on a type.
2016 You may use format arguments like `{T}`, `{A}` to correspond to the
2017 types at the point of use corresponding to the type parameters of the
2018 trait of the same name. `{Self}` will be replaced with the type that is supposed
2019 to implement the trait but doesn't. To use this, the `on_unimplemented` feature gate
2022 ### Conditional compilation
2024 Sometimes one wants to have different compiler outputs from the same code,
2025 depending on build target, such as targeted operating system, or to enable
2028 There are two kinds of configuration options, one that is either defined or not
2029 (`#[cfg(foo)]`), and the other that contains a string that can be checked
2030 against (`#[cfg(bar = "baz")]`). Currently, only compiler-defined configuration
2031 options can have the latter form.
2034 // The function is only included in the build when compiling for OSX
2035 #[cfg(target_os = "macos")]
2040 // This function is only included when either foo or bar is defined
2041 #[cfg(any(foo, bar))]
2042 fn needs_foo_or_bar() {
2046 // This function is only included when compiling for a unixish OS with a 32-bit
2048 #[cfg(all(unix, target_pointer_width = "32"))]
2049 fn on_32bit_unix() {
2053 // This function is only included when foo is not defined
2055 fn needs_not_foo() {
2060 This illustrates some conditional compilation can be achieved using the
2061 `#[cfg(...)]` attribute. `any`, `all` and `not` can be used to assemble
2062 arbitrarily complex configurations through nesting.
2064 The following configurations must be defined by the implementation:
2066 * `debug_assertions` - Enabled by default when compiling without optimizations.
2067 This can be used to enable extra debugging code in development but not in
2068 production. For example, it controls the behavior of the standard library's
2069 `debug_assert!` macro.
2070 * `target_arch = "..."` - Target CPU architecture, such as `"x86"`, `"x86_64"`
2071 `"mips"`, `"powerpc"`, `"powerpc64"`, `"arm"`, or `"aarch64"`.
2072 * `target_endian = "..."` - Endianness of the target CPU, either `"little"` or
2074 * `target_env = ".."` - An option provided by the compiler by default
2075 describing the runtime environment of the target platform. Some examples of
2076 this are `musl` for builds targeting the MUSL libc implementation, `msvc` for
2077 Windows builds targeting MSVC, and `gnu` frequently the rest of the time. This
2078 option may also be blank on some platforms.
2079 * `target_family = "..."` - Operating system family of the target, e. g.
2080 `"unix"` or `"windows"`. The value of this configuration option is defined
2081 as a configuration itself, like `unix` or `windows`.
2082 * `target_os = "..."` - Operating system of the target, examples include
2083 `"windows"`, `"macos"`, `"ios"`, `"linux"`, `"android"`, `"freebsd"`, `"dragonfly"`,
2084 `"bitrig"` , `"openbsd"` or `"netbsd"`.
2085 * `target_pointer_width = "..."` - Target pointer width in bits. This is set
2086 to `"32"` for targets with 32-bit pointers, and likewise set to `"64"` for
2088 * `target_vendor = "..."` - Vendor of the target, for example `apple`, `pc`, or
2090 * `test` - Enabled when compiling the test harness (using the `--test` flag).
2091 * `unix` - See `target_family`.
2092 * `windows` - See `target_family`.
2094 You can also set another attribute based on a `cfg` variable with `cfg_attr`:
2100 Will be the same as `#[b]` if `a` is set by `cfg`, and nothing otherwise.
2102 ### Lint check attributes
2104 A lint check names a potentially undesirable coding pattern, such as
2105 unreachable code or omitted documentation, for the static entity to which the
2108 For any lint check `C`:
2110 * `allow(C)` overrides the check for `C` so that violations will go
2112 * `deny(C)` signals an error after encountering a violation of `C`,
2113 * `forbid(C)` is the same as `deny(C)`, but also forbids changing the lint
2115 * `warn(C)` warns about violations of `C` but continues compilation.
2117 The lint checks supported by the compiler can be found via `rustc -W help`,
2118 along with their default settings. [Compiler
2119 plugins](book/compiler-plugins.html#lint-plugins) can provide additional lint checks.
2123 // Missing documentation is ignored here
2124 #[allow(missing_docs)]
2125 pub fn undocumented_one() -> i32 { 1 }
2127 // Missing documentation signals a warning here
2128 #[warn(missing_docs)]
2129 pub fn undocumented_too() -> i32 { 2 }
2131 // Missing documentation signals an error here
2132 #[deny(missing_docs)]
2133 pub fn undocumented_end() -> i32 { 3 }
2137 This example shows how one can use `allow` and `warn` to toggle a particular
2141 #[warn(missing_docs)]
2143 #[allow(missing_docs)]
2145 // Missing documentation is ignored here
2146 pub fn undocumented_one() -> i32 { 1 }
2148 // Missing documentation signals a warning here,
2149 // despite the allow above.
2150 #[warn(missing_docs)]
2151 pub fn undocumented_two() -> i32 { 2 }
2154 // Missing documentation signals a warning here
2155 pub fn undocumented_too() -> i32 { 3 }
2159 This example shows how one can use `forbid` to disallow uses of `allow` for
2163 #[forbid(missing_docs)]
2165 // Attempting to toggle warning signals an error here
2166 #[allow(missing_docs)]
2168 pub fn undocumented_too() -> i32 { 2 }
2174 Some primitive Rust operations are defined in Rust code, rather than being
2175 implemented directly in C or assembly language. The definitions of these
2176 operations have to be easy for the compiler to find. The `lang` attribute
2177 makes it possible to declare these operations. For example, the `str` module
2178 in the Rust standard library defines the string equality function:
2182 pub fn eq_slice(a: &str, b: &str) -> bool {
2187 The name `str_eq` has a special meaning to the Rust compiler, and the presence
2188 of this definition means that it will use this definition when generating calls
2189 to the string equality function.
2191 The set of language items is currently considered unstable. A complete
2192 list of the built-in language items will be added in the future.
2194 ### Inline attributes
2196 The inline attribute suggests that the compiler should place a copy of
2197 the function or static in the caller, rather than generating code to
2198 call the function or access the static where it is defined.
2200 The compiler automatically inlines functions based on internal heuristics.
2201 Incorrectly inlining functions can actually make the program slower, so it
2202 should be used with care.
2204 `#[inline]` and `#[inline(always)]` always cause the function to be serialized
2205 into the crate metadata to allow cross-crate inlining.
2207 There are three different types of inline attributes:
2209 * `#[inline]` hints the compiler to perform an inline expansion.
2210 * `#[inline(always)]` asks the compiler to always perform an inline expansion.
2211 * `#[inline(never)]` asks the compiler to never perform an inline expansion.
2215 The `derive` attribute allows certain traits to be automatically implemented
2216 for data structures. For example, the following will create an `impl` for the
2217 `PartialEq` and `Clone` traits for `Foo`, the type parameter `T` will be given
2218 the `PartialEq` or `Clone` constraints for the appropriate `impl`:
2221 #[derive(PartialEq, Clone)]
2228 The generated `impl` for `PartialEq` is equivalent to
2231 # struct Foo<T> { a: i32, b: T }
2232 impl<T: PartialEq> PartialEq for Foo<T> {
2233 fn eq(&self, other: &Foo<T>) -> bool {
2234 self.a == other.a && self.b == other.b
2237 fn ne(&self, other: &Foo<T>) -> bool {
2238 self.a != other.a || self.b != other.b
2243 ### Compiler Features
2245 Certain aspects of Rust may be implemented in the compiler, but they're not
2246 necessarily ready for every-day use. These features are often of "prototype
2247 quality" or "almost production ready", but may not be stable enough to be
2248 considered a full-fledged language feature.
2250 For this reason, Rust recognizes a special crate-level attribute of the form:
2253 #![feature(feature1, feature2, feature3)]
2256 This directive informs the compiler that the feature list: `feature1`,
2257 `feature2`, and `feature3` should all be enabled. This is only recognized at a
2258 crate-level, not at a module-level. Without this directive, all features are
2259 considered off, and using the features will result in a compiler error.
2261 The currently implemented features of the reference compiler are:
2263 * `advanced_slice_patterns` - See the [match expressions](#match-expressions)
2264 section for discussion; the exact semantics of
2265 slice patterns are subject to change, so some types
2268 * `slice_patterns` - OK, actually, slice patterns are just scary and
2269 completely unstable.
2271 * `asm` - The `asm!` macro provides a means for inline assembly. This is often
2272 useful, but the exact syntax for this feature along with its
2273 semantics are likely to change, so this macro usage must be opted
2276 * `associated_consts` - Allows constants to be defined in `impl` and `trait`
2277 blocks, so that they can be associated with a type or
2278 trait in a similar manner to methods and associated
2281 * `box_patterns` - Allows `box` patterns, the exact semantics of which
2282 is subject to change.
2284 * `box_syntax` - Allows use of `box` expressions, the exact semantics of which
2285 is subject to change.
2287 * `cfg_target_vendor` - Allows conditional compilation using the `target_vendor`
2288 matcher which is subject to change.
2290 * `concat_idents` - Allows use of the `concat_idents` macro, which is in many
2291 ways insufficient for concatenating identifiers, and may be
2292 removed entirely for something more wholesome.
2294 * `custom_attribute` - Allows the usage of attributes unknown to the compiler
2295 so that new attributes can be added in a backwards compatible
2298 * `custom_derive` - Allows the use of `#[derive(Foo,Bar)]` as sugar for
2299 `#[derive_Foo] #[derive_Bar]`, which can be user-defined syntax
2302 * `inclusive_range_syntax` - Allows use of the `a...b` and `...b` syntax for inclusive ranges.
2304 * `inclusive_range` - Allows use of the types that represent desugared inclusive ranges.
2306 * `intrinsics` - Allows use of the "rust-intrinsics" ABI. Compiler intrinsics
2307 are inherently unstable and no promise about them is made.
2309 * `lang_items` - Allows use of the `#[lang]` attribute. Like `intrinsics`,
2310 lang items are inherently unstable and no promise about them
2313 * `link_args` - This attribute is used to specify custom flags to the linker,
2314 but usage is strongly discouraged. The compiler's usage of the
2315 system linker is not guaranteed to continue in the future, and
2316 if the system linker is not used then specifying custom flags
2317 doesn't have much meaning.
2319 * `link_llvm_intrinsics` – Allows linking to LLVM intrinsics via
2320 `#[link_name="llvm.*"]`.
2322 * `linkage` - Allows use of the `linkage` attribute, which is not portable.
2324 * `log_syntax` - Allows use of the `log_syntax` macro attribute, which is a
2325 nasty hack that will certainly be removed.
2327 * `main` - Allows use of the `#[main]` attribute, which changes the entry point
2328 into a Rust program. This capability is subject to change.
2330 * `macro_reexport` - Allows macros to be re-exported from one crate after being imported
2331 from another. This feature was originally designed with the sole
2332 use case of the Rust standard library in mind, and is subject to
2335 * `non_ascii_idents` - The compiler supports the use of non-ascii identifiers,
2336 but the implementation is a little rough around the
2337 edges, so this can be seen as an experimental feature
2338 for now until the specification of identifiers is fully
2341 * `no_std` - Allows the `#![no_std]` crate attribute, which disables the implicit
2342 `extern crate std`. This typically requires use of the unstable APIs
2343 behind the libstd "facade", such as libcore and libcollections. It
2344 may also cause problems when using syntax extensions, including
2347 * `on_unimplemented` - Allows the `#[rustc_on_unimplemented]` attribute, which allows
2348 trait definitions to add specialized notes to error messages
2349 when an implementation was expected but not found.
2351 * `optin_builtin_traits` - Allows the definition of default and negative trait
2352 implementations. Experimental.
2354 * `plugin` - Usage of [compiler plugins][plugin] for custom lints or syntax extensions.
2355 These depend on compiler internals and are subject to change.
2357 * `plugin_registrar` - Indicates that a crate provides [compiler plugins][plugin].
2359 * `quote` - Allows use of the `quote_*!` family of macros, which are
2360 implemented very poorly and will likely change significantly
2361 with a proper implementation.
2363 * `rustc_attrs` - Gates internal `#[rustc_*]` attributes which may be
2364 for internal use only or have meaning added to them in the future.
2366 * `rustc_diagnostic_macros`- A mysterious feature, used in the implementation
2367 of rustc, not meant for mortals.
2369 * `simd` - Allows use of the `#[simd]` attribute, which is overly simple and
2370 not the SIMD interface we want to expose in the long term.
2372 * `simd_ffi` - Allows use of SIMD vectors in signatures for foreign functions.
2373 The SIMD interface is subject to change.
2375 * `start` - Allows use of the `#[start]` attribute, which changes the entry point
2376 into a Rust program. This capability, especially the signature for the
2377 annotated function, is subject to change.
2379 * `thread_local` - The usage of the `#[thread_local]` attribute is experimental
2380 and should be seen as unstable. This attribute is used to
2381 declare a `static` as being unique per-thread leveraging
2382 LLVM's implementation which works in concert with the kernel
2383 loader and dynamic linker. This is not necessarily available
2384 on all platforms, and usage of it is discouraged.
2386 * `trace_macros` - Allows use of the `trace_macros` macro, which is a nasty
2387 hack that will certainly be removed.
2389 * `unboxed_closures` - Rust's new closure design, which is currently a work in
2390 progress feature with many known bugs.
2392 * `unsafe_no_drop_flag` - Allows use of the `#[unsafe_no_drop_flag]` attribute,
2393 which removes hidden flag added to a type that
2394 implements the `Drop` trait. The design for the
2395 `Drop` flag is subject to change, and this feature
2396 may be removed in the future.
2398 * `unmarked_api` - Allows use of items within a `#![staged_api]` crate
2399 which have not been marked with a stability marker.
2400 Such items should not be allowed by the compiler to exist,
2401 so if you need this there probably is a compiler bug.
2403 * `allow_internal_unstable` - Allows `macro_rules!` macros to be tagged with the
2404 `#[allow_internal_unstable]` attribute, designed
2405 to allow `std` macros to call
2406 `#[unstable]`/feature-gated functionality
2407 internally without imposing on callers
2408 (i.e. making them behave like function calls in
2409 terms of encapsulation).
2410 * - `default_type_parameter_fallback` - Allows type parameter defaults to
2411 influence type inference.
2413 * - `stmt_expr_attributes` - Allows attributes on expressions and
2414 non-item statements.
2416 * - `deprecated` - Allows using the `#[deprecated]` attribute.
2418 * - `type_ascription` - Allows type ascription expressions `expr: Type`.
2420 * - `abi_vectorcall` - Allows the usage of the vectorcall calling convention
2421 (e.g. `extern "vectorcall" func fn_();`)
2423 If a feature is promoted to a language feature, then all existing programs will
2424 start to receive compilation warnings about `#![feature]` directives which enabled
2425 the new feature (because the directive is no longer necessary). However, if a
2426 feature is decided to be removed from the language, errors will be issued (if
2427 there isn't a parser error first). The directive in this case is no longer
2428 necessary, and it's likely that existing code will break if the feature isn't
2431 If an unknown feature is found in a directive, it results in a compiler error.
2432 An unknown feature is one which has never been recognized by the compiler.
2434 # Statements and expressions
2436 Rust is _primarily_ an expression language. This means that most forms of
2437 value-producing or effect-causing evaluation are directed by the uniform syntax
2438 category of _expressions_. Each kind of expression can typically _nest_ within
2439 each other kind of expression, and rules for evaluation of expressions involve
2440 specifying both the value produced by the expression and the order in which its
2441 sub-expressions are themselves evaluated.
2443 In contrast, statements in Rust serve _mostly_ to contain and explicitly
2444 sequence expression evaluation.
2448 A _statement_ is a component of a block, which is in turn a component of an
2449 outer [expression](#expressions) or [function](#functions).
2451 Rust has two kinds of statement: [declaration
2452 statements](#declaration-statements) and [expression
2453 statements](#expression-statements).
2455 ### Declaration statements
2457 A _declaration statement_ is one that introduces one or more *names* into the
2458 enclosing statement block. The declared names may denote new variables or new
2461 #### Item declarations
2463 An _item declaration statement_ has a syntactic form identical to an
2464 [item](#items) declaration within a module. Declaring an item — a
2465 function, enumeration, struct, type, static, trait, implementation or module
2466 — locally within a statement block is simply a way of restricting its
2467 scope to a narrow region containing all of its uses; it is otherwise identical
2468 in meaning to declaring the item outside the statement block.
2470 > **Note**: there is no implicit capture of the function's dynamic environment when
2471 > declaring a function-local item.
2473 #### `let` statements
2475 A _`let` statement_ introduces a new set of variables, given by a pattern. The
2476 pattern may be followed by a type annotation, and/or an initializer expression.
2477 When no type annotation is given, the compiler will infer the type, or signal
2478 an error if insufficient type information is available for definite inference.
2479 Any variables introduced by a variable declaration are visible from the point of
2480 declaration until the end of the enclosing block scope.
2482 ### Expression statements
2484 An _expression statement_ is one that evaluates an [expression](#expressions)
2485 and ignores its result. The type of an expression statement `e;` is always
2486 `()`, regardless of the type of `e`. As a rule, an expression statement's
2487 purpose is to trigger the effects of evaluating its expression.
2491 An expression may have two roles: it always produces a *value*, and it may have
2492 *effects* (otherwise known as "side effects"). An expression *evaluates to* a
2493 value, and has effects during *evaluation*. Many expressions contain
2494 sub-expressions (operands). The meaning of each kind of expression dictates
2497 * Whether or not to evaluate the sub-expressions when evaluating the expression
2498 * The order in which to evaluate the sub-expressions
2499 * How to combine the sub-expressions' values to obtain the value of the expression
2501 In this way, the structure of expressions dictates the structure of execution.
2502 Blocks are just another kind of expression, so blocks, statements, expressions,
2503 and blocks again can recursively nest inside each other to an arbitrary depth.
2505 #### Lvalues, rvalues and temporaries
2507 Expressions are divided into two main categories: _lvalues_ and _rvalues_.
2508 Likewise within each expression, sub-expressions may occur in _lvalue context_
2509 or _rvalue context_. The evaluation of an expression depends both on its own
2510 category and the context it occurs within.
2512 An lvalue is an expression that represents a memory location. These expressions
2513 are [paths](#path-expressions) (which refer to local variables, function and
2514 method arguments, or static variables), dereferences (`*expr`), [indexing
2515 expressions](#index-expressions) (`expr[expr]`), and [field
2516 references](#field-expressions) (`expr.f`). All other expressions are rvalues.
2518 The left operand of an [assignment](#assignment-expressions) or
2519 [compound-assignment](#compound-assignment-expressions) expression is
2520 an lvalue context, as is the single operand of a unary
2521 [borrow](#unary-operator-expressions). The discriminant or subject of
2522 a [match expression](#match-expressions) may be an lvalue context, if
2523 ref bindings are made, but is otherwise an rvalue context. All other
2524 expression contexts are rvalue contexts.
2526 When an lvalue is evaluated in an _lvalue context_, it denotes a memory
2527 location; when evaluated in an _rvalue context_, it denotes the value held _in_
2528 that memory location.
2530 ##### Temporary lifetimes
2532 When an rvalue is used in an lvalue context, a temporary un-named
2533 lvalue is created and used instead. The lifetime of temporary values
2534 is typically the innermost enclosing statement; the tail expression of
2535 a block is considered part of the statement that encloses the block.
2537 When a temporary rvalue is being created that is assigned into a `let`
2538 declaration, however, the temporary is created with the lifetime of
2539 the enclosing block instead, as using the enclosing statement (the
2540 `let` declaration) would be a guaranteed error (since a pointer to the
2541 temporary would be stored into a variable, but the temporary would be
2542 freed before the variable could be used). The compiler uses simple
2543 syntactic rules to decide which values are being assigned into a `let`
2544 binding, and therefore deserve a longer temporary lifetime.
2546 Here are some examples:
2548 - `let x = foo(&temp())`. The expression `temp()` is an rvalue. As it
2549 is being borrowed, a temporary is created which will be freed after
2550 the innermost enclosing statement (the `let` declaration, in this case).
2551 - `let x = temp().foo()`. This is the same as the previous example,
2552 except that the value of `temp()` is being borrowed via autoref on a
2553 method-call. Here we are assuming that `foo()` is an `&self` method
2554 defined in some trait, say `Foo`. In other words, the expression
2555 `temp().foo()` is equivalent to `Foo::foo(&temp())`.
2556 - `let x = &temp()`. Here, the same temporary is being assigned into
2557 `x`, rather than being passed as a parameter, and hence the
2558 temporary's lifetime is considered to be the enclosing block.
2559 - `let x = SomeStruct { foo: &temp() }`. As in the previous case, the
2560 temporary is assigned into a struct which is then assigned into a
2561 binding, and hence it is given the lifetime of the enclosing block.
2562 - `let x = [ &temp() ]`. As in the previous case, the
2563 temporary is assigned into an array which is then assigned into a
2564 binding, and hence it is given the lifetime of the enclosing block.
2565 - `let ref x = temp()`. In this case, the temporary is created using a ref binding,
2566 but the result is the same: the lifetime is extended to the enclosing block.
2568 #### Moved and copied types
2570 When a [local variable](#variables) is used as an
2571 [rvalue](#lvalues-rvalues-and-temporaries), the variable will be copied
2572 if its type implements `Copy`. All others are moved.
2574 ### Literal expressions
2576 A _literal expression_ consists of one of the [literal](#literals) forms
2577 described earlier. It directly describes a number, character, string, boolean
2578 value, or the unit value.
2582 "hello"; // string type
2583 '5'; // character type
2587 ### Path expressions
2589 A [path](#paths) used as an expression context denotes either a local variable
2590 or an item. Path expressions are [lvalues](#lvalues-rvalues-and-temporaries).
2592 ### Tuple expressions
2594 Tuples are written by enclosing zero or more comma-separated expressions in
2595 parentheses. They are used to create [tuple-typed](#tuple-types) values.
2599 ("a", 4usize, true);
2602 You can disambiguate a single-element tuple from a value in parentheses with a
2606 (0,); // single-element tuple
2607 (0); // zero in parentheses
2610 ### Struct expressions
2612 There are several forms of struct expressions. A _struct expression_
2613 consists of the [path](#paths) of a [struct item](#structs), followed by
2614 a brace-enclosed list of one or more comma-separated name-value pairs,
2615 providing the field values of a new instance of the struct. A field name
2616 can be any identifier, and is separated from its value expression by a colon.
2617 The location denoted by a struct field is mutable if and only if the
2618 enclosing struct is mutable.
2620 A _tuple struct expression_ consists of the [path](#paths) of a [struct
2621 item](#structs), followed by a parenthesized list of one or more
2622 comma-separated expressions (in other words, the path of a struct item
2623 followed by a tuple expression). The struct item must be a tuple struct
2626 A _unit-like struct expression_ consists only of the [path](#paths) of a
2627 [struct item](#structs).
2629 The following are examples of struct expressions:
2632 # struct Point { x: f64, y: f64 }
2633 # struct TuplePoint(f64, f64);
2634 # mod game { pub struct User<'a> { pub name: &'a str, pub age: u32, pub score: usize } }
2635 # struct Cookie; fn some_fn<T>(t: T) {}
2636 Point {x: 10.0, y: 20.0};
2637 TuplePoint(10.0, 20.0);
2638 let u = game::User {name: "Joe", age: 35, score: 100_000};
2639 some_fn::<Cookie>(Cookie);
2642 A struct expression forms a new value of the named struct type. Note
2643 that for a given *unit-like* struct type, this will always be the same
2646 A struct expression can terminate with the syntax `..` followed by an
2647 expression to denote a functional update. The expression following `..` (the
2648 base) must have the same struct type as the new struct type being formed.
2649 The entire expression denotes the result of constructing a new struct (with
2650 the same type as the base expression) with the given values for the fields that
2651 were explicitly specified and the values in the base expression for all other
2655 # struct Point3d { x: i32, y: i32, z: i32 }
2656 let base = Point3d {x: 1, y: 2, z: 3};
2657 Point3d {y: 0, z: 10, .. base};
2660 ### Block expressions
2662 A _block expression_ is similar to a module in terms of the declarations that
2663 are possible. Each block conceptually introduces a new namespace scope. Use
2664 items can bring new names into scopes and declared items are in scope for only
2667 A block will execute each statement sequentially, and then execute the
2668 expression (if given). If the block ends in a statement, its value is `()`:
2671 let x: () = { println!("Hello."); };
2674 If it ends in an expression, its value and type are that of the expression:
2677 let x: i32 = { println!("Hello."); 5 };
2682 ### Method-call expressions
2684 A _method call_ consists of an expression followed by a single dot, an
2685 identifier, and a parenthesized expression-list. Method calls are resolved to
2686 methods on specific traits, either statically dispatching to a method if the
2687 exact `self`-type of the left-hand-side is known, or dynamically dispatching if
2688 the left-hand-side expression is an indirect [trait object](#trait-objects).
2690 ### Field expressions
2692 A _field expression_ consists of an expression followed by a single dot and an
2693 identifier, when not immediately followed by a parenthesized expression-list
2694 (the latter is a [method call expression](#method-call-expressions)). A field
2695 expression denotes a field of a [struct](#struct-types).
2700 (Struct {a: 10, b: 20}).a;
2703 A field access is an [lvalue](#lvalues-rvalues-and-temporaries) referring to
2704 the value of that field. When the type providing the field inherits mutability,
2705 it can be [assigned](#assignment-expressions) to.
2707 Also, if the type of the expression to the left of the dot is a
2708 pointer, it is automatically dereferenced as many times as necessary
2709 to make the field access possible. In cases of ambiguity, we prefer
2710 fewer autoderefs to more.
2712 ### Array expressions
2714 An [array](#array-and-slice-types) _expression_ is written by enclosing zero
2715 or more comma-separated expressions of uniform type in square brackets.
2717 In the `[expr ';' expr]` form, the expression after the `';'` must be a
2718 constant expression that can be evaluated at compile time, such as a
2719 [literal](#literals) or a [static item](#static-items).
2723 ["a", "b", "c", "d"];
2724 [0; 128]; // array with 128 zeros
2725 [0u8, 0u8, 0u8, 0u8];
2728 ### Index expressions
2730 [Array](#array-and-slice-types)-typed expressions can be indexed by
2731 writing a square-bracket-enclosed expression (the index) after them. When the
2732 array is mutable, the resulting [lvalue](#lvalues-rvalues-and-temporaries) can
2735 Indices are zero-based, and may be of any integral type. Vector access is
2736 bounds-checked at compile-time for constant arrays being accessed with a constant index value.
2737 Otherwise a check will be performed at run-time that will put the thread in a _panicked state_ if it fails.
2742 let x = (["a", "b"])[10]; // compiler error: const index-expr is out of bounds
2745 let y = (["a", "b"])[n]; // panics
2747 let arr = ["a", "b"];
2751 Also, if the type of the expression to the left of the brackets is a
2752 pointer, it is automatically dereferenced as many times as necessary
2753 to make the indexing possible. In cases of ambiguity, we prefer fewer
2756 ### Range expressions
2758 The `..` operator will construct an object of one of the `std::ops::Range` variants.
2761 1..2; // std::ops::Range
2762 3..; // std::ops::RangeFrom
2763 ..4; // std::ops::RangeTo
2764 ..; // std::ops::RangeFull
2767 The following expressions are equivalent.
2770 let x = std::ops::Range {start: 0, end: 10};
2776 Similarly, the `...` operator will construct an object of one of the
2777 `std::ops::RangeInclusive` variants.
2780 # #![feature(inclusive_range_syntax)]
2781 1...2; // std::ops::RangeInclusive
2782 ...4; // std::ops::RangeToInclusive
2785 The following expressions are equivalent.
2788 # #![feature(inclusive_range_syntax, inclusive_range)]
2789 let x = std::ops::RangeInclusive::NonEmpty {start: 0, end: 10};
2795 ### Unary operator expressions
2797 Rust defines the following unary operators. They are all written as prefix operators,
2798 before the expression they apply to.
2801 : Negation. Signed integer types and floating-point types support negation. It
2802 is an error to apply negation to unsigned types; for example, the compiler
2805 : Dereference. When applied to a [pointer](#pointer-types) it denotes the
2806 pointed-to location. For pointers to mutable locations, the resulting
2807 [lvalue](#lvalues-rvalues-and-temporaries) can be assigned to.
2808 On non-pointer types, it calls the `deref` method of the `std::ops::Deref`
2809 trait, or the `deref_mut` method of the `std::ops::DerefMut` trait (if
2810 implemented by the type and required for an outer expression that will or
2811 could mutate the dereference), and produces the result of dereferencing the
2812 `&` or `&mut` borrowed pointer returned from the overload method.
2814 : Logical negation. On the boolean type, this flips between `true` and
2815 `false`. On integer types, this inverts the individual bits in the
2816 two's complement representation of the value.
2818 : Borrowing. When applied to an lvalue, these operators produce a
2819 reference (pointer) to the lvalue. The lvalue is also placed into
2820 a borrowed state for the duration of the reference. For a shared
2821 borrow (`&`), this implies that the lvalue may not be mutated, but
2822 it may be read or shared again. For a mutable borrow (`&mut`), the
2823 lvalue may not be accessed in any way until the borrow expires.
2824 If the `&` or `&mut` operators are applied to an rvalue, a
2825 temporary value is created; the lifetime of this temporary value
2826 is defined by [syntactic rules](#temporary-lifetimes).
2828 ### Binary operator expressions
2830 Binary operators expressions are given in terms of [operator
2831 precedence](#operator-precedence).
2833 #### Arithmetic operators
2835 Binary arithmetic expressions are syntactic sugar for calls to built-in traits,
2836 defined in the `std::ops` module of the `std` library. This means that
2837 arithmetic operators can be overridden for user-defined types. The default
2838 meaning of the operators on standard types is given here.
2841 : Addition and array/string concatenation.
2842 Calls the `add` method on the `std::ops::Add` trait.
2845 Calls the `sub` method on the `std::ops::Sub` trait.
2848 Calls the `mul` method on the `std::ops::Mul` trait.
2851 Calls the `div` method on the `std::ops::Div` trait.
2854 Calls the `rem` method on the `std::ops::Rem` trait.
2856 #### Bitwise operators
2858 Like the [arithmetic operators](#arithmetic-operators), bitwise operators are
2859 syntactic sugar for calls to methods of built-in traits. This means that
2860 bitwise operators can be overridden for user-defined types. The default
2861 meaning of the operators on standard types is given here. Bitwise `&`, `|` and
2862 `^` applied to boolean arguments are equivalent to logical `&&`, `||` and `!=`
2863 evaluated in non-lazy fashion.
2867 Calls the `bitand` method of the `std::ops::BitAnd` trait.
2869 : Bitwise inclusive OR.
2870 Calls the `bitor` method of the `std::ops::BitOr` trait.
2872 : Bitwise exclusive OR.
2873 Calls the `bitxor` method of the `std::ops::BitXor` trait.
2876 Calls the `shl` method of the `std::ops::Shl` trait.
2878 : Right shift (arithmetic).
2879 Calls the `shr` method of the `std::ops::Shr` trait.
2881 #### Lazy boolean operators
2883 The operators `||` and `&&` may be applied to operands of boolean type. The
2884 `||` operator denotes logical 'or', and the `&&` operator denotes logical
2885 'and'. They differ from `|` and `&` in that the right-hand operand is only
2886 evaluated when the left-hand operand does not already determine the result of
2887 the expression. That is, `||` only evaluates its right-hand operand when the
2888 left-hand operand evaluates to `false`, and `&&` only when it evaluates to
2891 #### Comparison operators
2893 Comparison operators are, like the [arithmetic
2894 operators](#arithmetic-operators), and [bitwise operators](#bitwise-operators),
2895 syntactic sugar for calls to built-in traits. This means that comparison
2896 operators can be overridden for user-defined types. The default meaning of the
2897 operators on standard types is given here.
2901 Calls the `eq` method on the `std::cmp::PartialEq` trait.
2904 Calls the `ne` method on the `std::cmp::PartialEq` trait.
2907 Calls the `lt` method on the `std::cmp::PartialOrd` trait.
2910 Calls the `gt` method on the `std::cmp::PartialOrd` trait.
2912 : Less than or equal.
2913 Calls the `le` method on the `std::cmp::PartialOrd` trait.
2915 : Greater than or equal.
2916 Calls the `ge` method on the `std::cmp::PartialOrd` trait.
2918 #### Type cast expressions
2920 A type cast expression is denoted with the binary operator `as`.
2922 Executing an `as` expression casts the value on the left-hand side to the type
2923 on the right-hand side.
2925 An example of an `as` expression:
2928 # fn sum(values: &[f64]) -> f64 { 0.0 }
2929 # fn len(values: &[f64]) -> i32 { 0 }
2931 fn average(values: &[f64]) -> f64 {
2932 let sum: f64 = sum(values);
2933 let size: f64 = len(values) as f64;
2938 Some of the conversions which can be done through the `as` operator
2939 can also be done implicitly at various points in the program, such as
2940 argument passing and assignment to a `let` binding with an explicit
2941 type. Implicit conversions are limited to "harmless" conversions that
2942 do not lose information and which have minimal or no risk of
2943 surprising side-effects on the dynamic execution semantics.
2945 #### Assignment expressions
2947 An _assignment expression_ consists of an
2948 [lvalue](#lvalues-rvalues-and-temporaries) expression followed by an equals
2949 sign (`=`) and an [rvalue](#lvalues-rvalues-and-temporaries) expression.
2951 Evaluating an assignment expression [either copies or
2952 moves](#moved-and-copied-types) its right-hand operand to its left-hand
2961 #### Compound assignment expressions
2963 The `+`, `-`, `*`, `/`, `%`, `&`, `|`, `^`, `<<`, and `>>` operators may be
2964 composed with the `=` operator. The expression `lval OP= val` is equivalent to
2965 `lval = lval OP val`. For example, `x = x + 1` may be written as `x += 1`.
2967 Any such expression always has the [`unit`](#tuple-types) type.
2969 #### Operator precedence
2971 The precedence of Rust binary operators is ordered as follows, going from
2974 ```{.text .precedence}
2988 Operators at the same precedence level are evaluated left-to-right. [Unary
2989 operators](#unary-operator-expressions) have the same precedence level and are
2990 stronger than any of the binary operators.
2992 ### Grouped expressions
2994 An expression enclosed in parentheses evaluates to the result of the enclosed
2995 expression. Parentheses can be used to explicitly specify evaluation order
2996 within an expression.
2998 An example of a parenthesized expression:
3001 let x: i32 = (2 + 3) * 4;
3005 ### Call expressions
3007 A _call expression_ invokes a function, providing zero or more input variables
3008 and an optional location to move the function's output into. If the function
3009 eventually returns, then the expression completes.
3011 Some examples of call expressions:
3014 # fn add(x: i32, y: i32) -> i32 { 0 }
3016 let x: i32 = add(1i32, 2i32);
3017 let pi: Result<f32, _> = "3.14".parse();
3020 ### Lambda expressions
3022 A _lambda expression_ (sometimes called an "anonymous function expression")
3023 defines a function and denotes it as a value, in a single expression. A lambda
3024 expression is a pipe-symbol-delimited (`|`) list of identifiers followed by an
3027 A lambda expression denotes a function that maps a list of parameters
3028 (`ident_list`) onto the expression that follows the `ident_list`. The
3029 identifiers in the `ident_list` are the parameters to the function. These
3030 parameters' types need not be specified, as the compiler infers them from
3033 Lambda expressions are most useful when passing functions as arguments to other
3034 functions, as an abbreviation for defining and capturing a separate function.
3036 Significantly, lambda expressions _capture their environment_, which regular
3037 [function definitions](#functions) do not. The exact type of capture depends
3038 on the [function type](#function-types) inferred for the lambda expression. In
3039 the simplest and least-expensive form (analogous to a ```|| { }``` expression),
3040 the lambda expression captures its environment by reference, effectively
3041 borrowing pointers to all outer variables mentioned inside the function.
3042 Alternately, the compiler may infer that a lambda expression should copy or
3043 move values (depending on their type) from the environment into the lambda
3044 expression's captured environment.
3046 In this example, we define a function `ten_times` that takes a higher-order
3047 function argument, and we then call it with a lambda expression as an argument:
3050 fn ten_times<F>(f: F) where F: Fn(i32) {
3051 for index in 0..10 {
3056 ten_times(|j| println!("hello, {}", j));
3061 A `loop` expression denotes an infinite loop.
3063 A `loop` expression may optionally have a _label_. The label is written as
3064 a lifetime preceding the loop expression, as in `'foo: loop{ }`. If a
3065 label is present, then labeled `break` and `continue` expressions nested
3066 within this loop may exit out of this loop or return control to its head.
3067 See [break expressions](#break-expressions) and [continue
3068 expressions](#continue-expressions).
3070 ### `break` expressions
3072 A `break` expression has an optional _label_. If the label is absent, then
3073 executing a `break` expression immediately terminates the innermost loop
3074 enclosing it. It is only permitted in the body of a loop. If the label is
3075 present, then `break 'foo` terminates the loop with label `'foo`, which need not
3076 be the innermost label enclosing the `break` expression, but must enclose it.
3078 ### `continue` expressions
3080 A `continue` expression has an optional _label_. If the label is absent, then
3081 executing a `continue` expression immediately terminates the current iteration
3082 of the innermost loop enclosing it, returning control to the loop *head*. In
3083 the case of a `while` loop, the head is the conditional expression controlling
3084 the loop. In the case of a `for` loop, the head is the call-expression
3085 controlling the loop. If the label is present, then `continue 'foo` returns
3086 control to the head of the loop with label `'foo`, which need not be the
3087 innermost label enclosing the `continue` expression, but must enclose it.
3089 A `continue` expression is only permitted in the body of a loop.
3093 A `while` loop begins by evaluating the boolean loop conditional expression.
3094 If the loop conditional expression evaluates to `true`, the loop body block
3095 executes and control returns to the loop conditional expression. If the loop
3096 conditional expression evaluates to `false`, the `while` expression completes.
3109 Like `loop` expressions, `while` loops can be controlled with `break` or
3110 `continue`, and may optionally have a _label_. See [infinite
3111 loops](#infinite-loops), [break expressions](#break-expressions), and
3112 [continue expressions](#continue-expressions) for more information.
3114 ### `for` expressions
3116 A `for` expression is a syntactic construct for looping over elements provided
3117 by an implementation of `std::iter::IntoIterator`.
3119 An example of a `for` loop over the contents of an array:
3123 # fn bar(f: &Foo) { }
3128 let v: &[Foo] = &[a, b, c];
3135 An example of a for loop over a series of integers:
3138 # fn bar(b:usize) { }
3144 Like `loop` expressions, `for` loops can be controlled with `break` or
3145 `continue`, and may optionally have a _label_. See [infinite
3146 loops](#infinite-loops), [break expressions](#break-expressions), and
3147 [continue expressions](#continue-expressions) for more information.
3149 ### `if` expressions
3151 An `if` expression is a conditional branch in program control. The form of an
3152 `if` expression is a condition expression, followed by a consequent block, any
3153 number of `else if` conditions and blocks, and an optional trailing `else`
3154 block. The condition expressions must have type `bool`. If a condition
3155 expression evaluates to `true`, the consequent block is executed and any
3156 subsequent `else if` or `else` block is skipped. If a condition expression
3157 evaluates to `false`, the consequent block is skipped and any subsequent `else
3158 if` condition is evaluated. If all `if` and `else if` conditions evaluate to
3159 `false` then any `else` block is executed.
3161 ### `match` expressions
3163 A `match` expression branches on a *pattern*. The exact form of matching that
3164 occurs depends on the pattern. Patterns consist of some combination of
3165 literals, destructured arrays or enum constructors, structs and tuples,
3166 variable binding specifications, wildcards (`..`), and placeholders (`_`). A
3167 `match` expression has a *head expression*, which is the value to compare to
3168 the patterns. The type of the patterns must equal the type of the head
3171 In a pattern whose head expression has an `enum` type, a placeholder (`_`)
3172 stands for a *single* data field, whereas a wildcard `..` stands for *all* the
3173 fields of a particular variant.
3175 A `match` behaves differently depending on whether or not the head expression
3176 is an [lvalue or an rvalue](#lvalues-rvalues-and-temporaries). If the head
3177 expression is an rvalue, it is first evaluated into a temporary location, and
3178 the resulting value is sequentially compared to the patterns in the arms until
3179 a match is found. The first arm with a matching pattern is chosen as the branch
3180 target of the `match`, any variables bound by the pattern are assigned to local
3181 variables in the arm's block, and control enters the block.
3183 When the head expression is an lvalue, the match does not allocate a temporary
3184 location (however, a by-value binding may copy or move from the lvalue). When
3185 possible, it is preferable to match on lvalues, as the lifetime of these
3186 matches inherits the lifetime of the lvalue, rather than being restricted to
3187 the inside of the match.
3189 An example of a `match` expression:
3195 1 => println!("one"),
3196 2 => println!("two"),
3197 3 => println!("three"),
3198 4 => println!("four"),
3199 5 => println!("five"),
3200 _ => println!("something else"),
3204 Patterns that bind variables default to binding to a copy or move of the
3205 matched value (depending on the matched value's type). This can be changed to
3206 bind to a reference by using the `ref` keyword, or to a mutable reference using
3209 Subpatterns can also be bound to variables by the use of the syntax `variable @
3210 subpattern`. For example:
3216 e @ 1 ... 5 => println!("got a range element {}", e),
3217 _ => println!("anything"),
3221 Patterns can also dereference pointers by using the `&`, `&mut` and `box`
3222 symbols, as appropriate. For example, these two matches on `x: &i32` are
3227 let y = match *x { 0 => "zero", _ => "some" };
3228 let z = match x { &0 => "zero", _ => "some" };
3233 Multiple match patterns may be joined with the `|` operator. A range of values
3234 may be specified with `...`. For example:
3239 let message = match x {
3240 0 | 1 => "not many",
3246 Range patterns only work on scalar types (like integers and characters; not
3247 like arrays and structs, which have sub-components). A range pattern may not
3248 be a sub-range of another range pattern inside the same `match`.
3250 Finally, match patterns can accept *pattern guards* to further refine the
3251 criteria for matching a case. Pattern guards appear after the pattern and
3252 consist of a bool-typed expression following the `if` keyword. A pattern guard
3253 may refer to the variables bound within the pattern they follow.
3256 # let maybe_digit = Some(0);
3257 # fn process_digit(i: i32) { }
3258 # fn process_other(i: i32) { }
3260 let message = match maybe_digit {
3261 Some(x) if x < 10 => process_digit(x),
3262 Some(x) => process_other(x),
3267 ### `if let` expressions
3269 An `if let` expression is semantically identical to an `if` expression but in
3270 place of a condition expression it expects a `let` statement with a refutable
3271 pattern. If the value of the expression on the right hand side of the `let`
3272 statement matches the pattern, the corresponding block will execute, otherwise
3273 flow proceeds to the first `else` block that follows.
3276 let dish = ("Ham", "Eggs");
3278 // this body will be skipped because the pattern is refuted
3279 if let ("Bacon", b) = dish {
3280 println!("Bacon is served with {}", b);
3283 // this body will execute
3284 if let ("Ham", b) = dish {
3285 println!("Ham is served with {}", b);
3289 ### `while let` loops
3291 A `while let` loop is semantically identical to a `while` loop but in place of
3292 a condition expression it expects `let` statement with a refutable pattern. If
3293 the value of the expression on the right hand side of the `let` statement
3294 matches the pattern, the loop body block executes and control returns to the
3295 pattern matching statement. Otherwise, the while expression completes.
3297 ### `return` expressions
3299 Return expressions are denoted with the keyword `return`. Evaluating a `return`
3300 expression moves its argument into the designated output location for the
3301 current function call, destroys the current function activation frame, and
3302 transfers control to the caller frame.
3304 An example of a `return` expression:
3307 fn max(a: i32, b: i32) -> i32 {
3319 Every variable, item and value in a Rust program has a type. The _type_ of a
3320 *value* defines the interpretation of the memory holding it.
3322 Built-in types and type-constructors are tightly integrated into the language,
3323 in nontrivial ways that are not possible to emulate in user-defined types.
3324 User-defined types have limited capabilities.
3328 The primitive types are the following:
3330 * The boolean type `bool` with values `true` and `false`.
3331 * The machine types (integer and floating-point).
3332 * The machine-dependent integer types.
3340 The machine types are the following:
3342 * The unsigned word types `u8`, `u16`, `u32` and `u64`, with values drawn from
3343 the integer intervals [0, 2^8 - 1], [0, 2^16 - 1], [0, 2^32 - 1] and
3344 [0, 2^64 - 1] respectively.
3346 * The signed two's complement word types `i8`, `i16`, `i32` and `i64`, with
3347 values drawn from the integer intervals [-(2^(7)), 2^7 - 1],
3348 [-(2^(15)), 2^15 - 1], [-(2^(31)), 2^31 - 1], [-(2^(63)), 2^63 - 1]
3351 * The IEEE 754-2008 `binary32` and `binary64` floating-point types: `f32` and
3352 `f64`, respectively.
3354 #### Machine-dependent integer types
3356 The `usize` type is an unsigned integer type with the same number of bits as the
3357 platform's pointer type. It can represent every memory address in the process.
3359 The `isize` type is a signed integer type with the same number of bits as the
3360 platform's pointer type. The theoretical upper bound on object and array size
3361 is the maximum `isize` value. This ensures that `isize` can be used to calculate
3362 differences between pointers into an object or array and can address every byte
3363 within an object along with one byte past the end.
3367 The types `char` and `str` hold textual data.
3369 A value of type `char` is a [Unicode scalar value](
3370 http://www.unicode.org/glossary/#unicode_scalar_value) (i.e. a code point that
3371 is not a surrogate), represented as a 32-bit unsigned word in the 0x0000 to
3372 0xD7FF or 0xE000 to 0x10FFFF range. A `[char]` array is effectively an UCS-4 /
3375 A value of type `str` is a Unicode string, represented as an array of 8-bit
3376 unsigned bytes holding a sequence of UTF-8 code points. Since `str` is of
3377 unknown size, it is not a _first-class_ type, but can only be instantiated
3378 through a pointer type, such as `&str`.
3382 A tuple *type* is a heterogeneous product of other types, called the *elements*
3383 of the tuple. It has no nominal name and is instead structurally typed.
3385 Tuple types and values are denoted by listing the types or values of their
3386 elements, respectively, in a parenthesized, comma-separated list.
3388 Because tuple elements don't have a name, they can only be accessed by
3389 pattern-matching or by using `N` directly as a field to access the
3392 An example of a tuple type and its use:
3395 type Pair<'a> = (i32, &'a str);
3396 let p: Pair<'static> = (10, "ten");
3400 assert_eq!(b, "ten");
3401 assert_eq!(p.0, 10);
3402 assert_eq!(p.1, "ten");
3405 For historical reasons and convenience, the tuple type with no elements (`()`)
3406 is often called ‘unit’ or ‘the unit type’.
3408 ### Array, and Slice types
3410 Rust has two different types for a list of items:
3412 * `[T; N]`, an 'array'
3415 An array has a fixed size, and can be allocated on either the stack or the
3418 A slice is a 'view' into an array. It doesn't own the data it points
3424 // A stack-allocated array
3425 let array: [i32; 3] = [1, 2, 3];
3427 // A heap-allocated array
3428 let vector: Vec<i32> = vec![1, 2, 3];
3430 // A slice into an array
3431 let slice: &[i32] = &vector[..];
3434 As you can see, the `vec!` macro allows you to create a `Vec<T>` easily. The
3435 `vec!` macro is also part of the standard library, rather than the language.
3437 All in-bounds elements of arrays and slices are always initialized, and access
3438 to an array or slice is always bounds-checked.
3442 A `struct` *type* is a heterogeneous product of other types, called the
3443 *fields* of the type.[^structtype]
3445 [^structtype]: `struct` types are analogous to `struct` types in C,
3446 the *record* types of the ML family,
3447 or the *struct* types of the Lisp family.
3449 New instances of a `struct` can be constructed with a [struct
3450 expression](#struct-expressions).
3452 The memory layout of a `struct` is undefined by default to allow for compiler
3453 optimizations like field reordering, but it can be fixed with the
3454 `#[repr(...)]` attribute. In either case, fields may be given in any order in
3455 a corresponding struct *expression*; the resulting `struct` value will always
3456 have the same memory layout.
3458 The fields of a `struct` may be qualified by [visibility
3459 modifiers](#visibility-and-privacy), to allow access to data in a
3460 struct outside a module.
3462 A _tuple struct_ type is just like a struct type, except that the fields are
3465 A _unit-like struct_ type is like a struct type, except that it has no
3466 fields. The one value constructed by the associated [struct
3467 expression](#struct-expressions) is the only value that inhabits such a
3470 ### Enumerated types
3472 An *enumerated type* is a nominal, heterogeneous disjoint union type, denoted
3473 by the name of an [`enum` item](#enumerations). [^enumtype]
3475 [^enumtype]: The `enum` type is analogous to a `data` constructor declaration in
3476 ML, or a *pick ADT* in Limbo.
3478 An [`enum` item](#enumerations) declares both the type and a number of *variant
3479 constructors*, each of which is independently named and takes an optional tuple
3482 New instances of an `enum` can be constructed by calling one of the variant
3483 constructors, in a [call expression](#call-expressions).
3485 Any `enum` value consumes as much memory as the largest variant constructor for
3486 its corresponding `enum` type.
3488 Enum types cannot be denoted *structurally* as types, but must be denoted by
3489 named reference to an [`enum` item](#enumerations).
3493 Nominal types — [enumerations](#enumerated-types) and
3494 [structs](#struct-types) — may be recursive. That is, each `enum`
3495 constructor or `struct` field may refer, directly or indirectly, to the
3496 enclosing `enum` or `struct` type itself. Such recursion has restrictions:
3498 * Recursive types must include a nominal type in the recursion
3499 (not mere [type definitions](grammar.html#type-definitions),
3500 or other structural types such as [arrays](#array-and-slice-types) or [tuples](#tuple-types)).
3501 * A recursive `enum` item must have at least one non-recursive constructor
3502 (in order to give the recursion a basis case).
3503 * The size of a recursive type must be finite;
3504 in other words the recursive fields of the type must be [pointer types](#pointer-types).
3505 * Recursive type definitions can cross module boundaries, but not module *visibility* boundaries,
3506 or crate boundaries (in order to simplify the module system and type checker).
3508 An example of a *recursive* type and its use:
3513 Cons(T, Box<List<T>>)
3516 let a: List<i32> = List::Cons(7, Box::new(List::Cons(13, Box::new(List::Nil))));
3521 All pointers in Rust are explicit first-class values. They can be copied,
3522 stored into data structs, and returned from functions. There are two
3523 varieties of pointer in Rust:
3526 : These point to memory _owned by some other value_.
3527 A reference type is written `&type`,
3528 or `&'a type` when you need to specify an explicit lifetime.
3529 Copying a reference is a "shallow" operation:
3530 it involves only copying the pointer itself.
3531 Releasing a reference has no effect on the value it points to,
3532 but a reference of a temporary value will keep it alive during the scope
3533 of the reference itself.
3535 * Raw pointers (`*`)
3536 : Raw pointers are pointers without safety or liveness guarantees.
3537 Raw pointers are written as `*const T` or `*mut T`,
3538 for example `*const i32` means a raw pointer to a 32-bit integer.
3539 Copying or dropping a raw pointer has no effect on the lifecycle of any
3540 other value. Dereferencing a raw pointer or converting it to any other
3541 pointer type is an [`unsafe` operation](#unsafe-functions).
3542 Raw pointers are generally discouraged in Rust code;
3543 they exist to support interoperability with foreign code,
3544 and writing performance-critical or low-level functions.
3546 The standard library contains additional 'smart pointer' types beyond references
3551 The function type constructor `fn` forms new function types. A function type
3552 consists of a possibly-empty set of function-type modifiers (such as `unsafe`
3553 or `extern`), a sequence of input types and an output type.
3555 An example of a `fn` type:
3558 fn add(x: i32, y: i32) -> i32 {
3562 let mut x = add(5,7);
3564 type Binop = fn(i32, i32) -> i32;
3565 let bo: Binop = add;
3569 #### Function types for specific items
3571 Internal to the compiler, there are also function types that are specific to a particular
3572 function item. In the following snippet, for example, the internal types of the functions
3573 `foo` and `bar` are different, despite the fact that they have the same signature:
3580 The types of `foo` and `bar` can both be implicitly coerced to the fn
3581 pointer type `fn()`. There is currently no syntax for unique fn types,
3582 though the compiler will emit a type like `fn() {foo}` in error
3583 messages to indicate "the unique fn type for the function `foo`".
3587 A [lambda expression](#lambda-expressions) produces a closure value with
3588 a unique, anonymous type that cannot be written out.
3590 Depending on the requirements of the closure, its type implements one or
3591 more of the closure traits:
3594 : The closure can be called once. A closure called as `FnOnce`
3595 can move out values from its environment.
3598 : The closure can be called multiple times as mutable. A closure called as
3599 `FnMut` can mutate values from its environment. `FnMut` inherits from
3600 `FnOnce` (i.e. anything implementing `FnMut` also implements `FnOnce`).
3603 : The closure can be called multiple times through a shared reference.
3604 A closure called as `Fn` can neither move out from nor mutate values
3605 from its environment. `Fn` inherits from `FnMut`, which itself
3606 inherits from `FnOnce`.
3611 In Rust, a type like `&SomeTrait` or `Box<SomeTrait>` is called a _trait object_.
3612 Each instance of a trait object includes:
3614 - a pointer to an instance of a type `T` that implements `SomeTrait`
3615 - a _virtual method table_, often just called a _vtable_, which contains, for
3616 each method of `SomeTrait` that `T` implements, a pointer to `T`'s
3617 implementation (i.e. a function pointer).
3619 The purpose of trait objects is to permit "late binding" of methods. Calling a
3620 method on a trait object results in virtual dispatch at runtime: that is, a
3621 function pointer is loaded from the trait object vtable and invoked indirectly.
3622 The actual implementation for each vtable entry can vary on an object-by-object
3625 Note that for a trait object to be instantiated, the trait must be
3626 _object-safe_. Object safety rules are defined in [RFC 255].
3628 [RFC 255]: https://github.com/rust-lang/rfcs/blob/master/text/0255-object-safety.md
3630 Given a pointer-typed expression `E` of type `&T` or `Box<T>`, where `T`
3631 implements trait `R`, casting `E` to the corresponding pointer type `&R` or
3632 `Box<R>` results in a value of the _trait object_ `R`. This result is
3633 represented as a pair of pointers: the vtable pointer for the `T`
3634 implementation of `R`, and the pointer value of `E`.
3636 An example of a trait object:
3640 fn stringify(&self) -> String;
3643 impl Printable for i32 {
3644 fn stringify(&self) -> String { self.to_string() }
3647 fn print(a: Box<Printable>) {
3648 println!("{}", a.stringify());
3652 print(Box::new(10) as Box<Printable>);
3656 In this example, the trait `Printable` occurs as a trait object in both the
3657 type signature of `print`, and the cast expression in `main`.
3661 Within the body of an item that has type parameter declarations, the names of
3662 its type parameters are types:
3665 fn to_vec<A: Clone>(xs: &[A]) -> Vec<A> {
3669 let first: A = xs[0].clone();
3670 let mut rest: Vec<A> = to_vec(&xs[1..]);
3671 rest.insert(0, first);
3676 Here, `first` has type `A`, referring to `to_vec`'s `A` type parameter; and `rest`
3677 has type `Vec<A>`, a vector with element type `A`.
3681 The special type `Self` has a meaning within traits and impls. In a trait definition, it refers
3682 to an implicit type parameter representing the "implementing" type. In an impl,
3683 it is an alias for the implementing type. For example, in:
3687 fn make_string(&self) -> String;
3690 impl Printable for String {
3691 fn make_string(&self) -> String {
3697 The notation `&self` is a shorthand for `self: &Self`. In this case,
3698 in the impl, `Self` refers to the value of type `String` that is the
3699 receiver for a call to the method `make_string`.
3703 Subtyping is implicit and can occur at any stage in type checking or
3704 inference. Subtyping in Rust is very restricted and occurs only due to
3705 variance with respect to lifetimes and between types with higher ranked
3706 lifetimes. If we were to erase lifetimes from types, then the only subtyping
3707 would be due to type equality.
3709 Consider the following example: string literals always have `'static`
3710 lifetime. Nevertheless, we can assign `s` to `t`:
3714 let s: &'static str = "hi";
3718 Since `'static` "lives longer" than `'a`, `&'static str` is a subtype of
3723 Coercions are defined in [RFC401]. A coercion is implicit and has no syntax.
3725 [RFC401]: https://github.com/rust-lang/rfcs/blob/master/text/0401-coercions.md
3729 A coercion can only occur at certain coercion sites in a program; these are
3730 typically places where the desired type is explicit or can be derived by
3731 propagation from explicit types (without type inference). Possible coercion
3734 * `let` statements where an explicit type is given.
3736 For example, `42` is coerced to have type `i8` in the following:
3742 * `static` and `const` statements (similar to `let` statements).
3744 * Arguments for function calls
3746 The value being coerced is the actual parameter, and it is coerced to
3747 the type of the formal parameter.
3749 For example, `42` is coerced to have type `i8` in the following:
3759 * Instantiations of struct or variant fields
3761 For example, `42` is coerced to have type `i8` in the following:
3764 struct Foo { x: i8 }
3771 * Function results, either the final line of a block if it is not
3772 semicolon-terminated or any expression in a `return` statement
3774 For example, `42` is coerced to have type `i8` in the following:
3782 If the expression in one of these coercion sites is a coercion-propagating
3783 expression, then the relevant sub-expressions in that expression are also
3784 coercion sites. Propagation recurses from these new coercion sites.
3785 Propagating expressions and their relevant sub-expressions are:
3787 * Array literals, where the array has type `[U; n]`. Each sub-expression in
3788 the array literal is a coercion site for coercion to type `U`.
3790 * Array literals with repeating syntax, where the array has type `[U; n]`. The
3791 repeated sub-expression is a coercion site for coercion to type `U`.
3793 * Tuples, where a tuple is a coercion site to type `(U_0, U_1, ..., U_n)`.
3794 Each sub-expression is a coercion site to the respective type, e.g. the
3795 zeroth sub-expression is a coercion site to type `U_0`.
3797 * Parenthesized sub-expressions (`(e)`): if the expression has type `U`, then
3798 the sub-expression is a coercion site to `U`.
3800 * Blocks: if a block has type `U`, then the last expression in the block (if
3801 it is not semicolon-terminated) is a coercion site to `U`. This includes
3802 blocks which are part of control flow statements, such as `if`/`else`, if
3803 the block has a known type.
3807 Coercion is allowed between the following types:
3809 * `T` to `U` if `T` is a subtype of `U` (*reflexive case*)
3811 * `T_1` to `T_3` where `T_1` coerces to `T_2` and `T_2` coerces to `T_3`
3814 Note that this is not fully supported yet
3818 * `*mut T` to `*const T`
3820 * `&T` to `*const T`
3822 * `&mut T` to `*mut T`
3824 * `&T` to `&U` if `T` implements `Deref<Target = U>`. For example:
3827 use std::ops::Deref;
3829 struct CharContainer {
3833 impl Deref for CharContainer {
3836 fn deref<'a>(&'a self) -> &'a char {
3841 fn foo(arg: &char) {}
3844 let x = &mut CharContainer { value: 'y' };
3845 foo(x); //&mut CharContainer is coerced to &char.
3849 * `&mut T` to `&mut U` if `T` implements `DerefMut<Target = U>`.
3851 * TyCtor(`T`) to TyCtor(coerce_inner(`T`)), where TyCtor(`T`) is one of
3859 - coerce_inner(`[T, ..n]`) = `[T]`
3860 - coerce_inner(`T`) = `U` where `T` is a concrete type which implements the
3863 In the future, coerce_inner will be recursively extended to tuples and
3864 structs. In addition, coercions from sub-traits to super-traits will be
3865 added. See [RFC401] for more details.
3869 Several traits define special evaluation behavior.
3873 The `Copy` trait changes the semantics of a type implementing it. Values whose
3874 type implements `Copy` are copied rather than moved upon assignment.
3876 ## The `Sized` trait
3878 The `Sized` trait indicates that the size of this type is known at compile-time.
3882 The `Drop` trait provides a destructor, to be run whenever a value of this type
3885 ## The `Deref` trait
3887 The `Deref<Target = U>` trait allows a type to implicitly implement all the methods
3888 of the type `U`. When attempting to resolve a method call, the compiler will search
3889 the top-level type for the implementation of the called method. If no such method is
3890 found, `.deref()` is called and the compiler continues to search for the method
3891 implementation in the returned type `U`.
3895 A Rust program's memory consists of a static set of *items* and a *heap*.
3896 Immutable portions of the heap may be safely shared between threads, mutable
3897 portions may not be safely shared, but several mechanisms for effectively-safe
3898 sharing of mutable values, built on unsafe code but enforcing a safe locking
3899 discipline, exist in the standard library.
3901 Allocations in the stack consist of *variables*, and allocations in the heap
3904 ### Memory allocation and lifetime
3906 The _items_ of a program are those functions, modules and types that have their
3907 value calculated at compile-time and stored uniquely in the memory image of the
3908 rust process. Items are neither dynamically allocated nor freed.
3910 The _heap_ is a general term that describes boxes. The lifetime of an
3911 allocation in the heap depends on the lifetime of the box values pointing to
3912 it. Since box values may themselves be passed in and out of frames, or stored
3913 in the heap, heap allocations may outlive the frame they are allocated within.
3914 An allocation in the heap is guaranteed to reside at a single location in the
3915 heap for the whole lifetime of the allocation - it will never be relocated as
3916 a result of moving a box value.
3918 ### Memory ownership
3920 When a stack frame is exited, its local allocations are all released, and its
3921 references to boxes are dropped.
3925 A _variable_ is a component of a stack frame, either a named function parameter,
3926 an anonymous [temporary](#lvalues-rvalues-and-temporaries), or a named local
3929 A _local variable_ (or *stack-local* allocation) holds a value directly,
3930 allocated within the stack's memory. The value is a part of the stack frame.
3932 Local variables are immutable unless declared otherwise like: `let mut x = ...`.
3934 Function parameters are immutable unless declared with `mut`. The `mut` keyword
3935 applies only to the following parameter (so `|mut x, y|` and `fn f(mut x:
3936 Box<i32>, y: Box<i32>)` declare one mutable variable `x` and one immutable
3939 Methods that take either `self` or `Box<Self>` can optionally place them in a
3940 mutable variable by prefixing them with `mut` (similar to regular arguments):
3944 fn change(mut self) -> Self;
3945 fn modify(mut self: Box<Self>) -> Box<Self>;
3949 Local variables are not initialized when allocated; the entire frame worth of
3950 local variables are allocated at once, on frame-entry, in an uninitialized
3951 state. Subsequent statements within a function may or may not initialize the
3952 local variables. Local variables can be used only after they have been
3953 initialized; this is enforced by the compiler.
3957 The Rust compiler supports various methods to link crates together both
3958 statically and dynamically. This section will explore the various methods to
3959 link Rust crates together, and more information about native libraries can be
3960 found in the [FFI section of the book][ffi].
3962 In one session of compilation, the compiler can generate multiple artifacts
3963 through the usage of either command line flags or the `crate_type` attribute.
3964 If one or more command line flags are specified, all `crate_type` attributes will
3965 be ignored in favor of only building the artifacts specified by command line.
3967 * `--crate-type=bin`, `#[crate_type = "bin"]` - A runnable executable will be
3968 produced. This requires that there is a `main` function in the crate which
3969 will be run when the program begins executing. This will link in all Rust and
3970 native dependencies, producing a distributable binary.
3972 * `--crate-type=lib`, `#[crate_type = "lib"]` - A Rust library will be produced.
3973 This is an ambiguous concept as to what exactly is produced because a library
3974 can manifest itself in several forms. The purpose of this generic `lib` option
3975 is to generate the "compiler recommended" style of library. The output library
3976 will always be usable by rustc, but the actual type of library may change from
3977 time-to-time. The remaining output types are all different flavors of
3978 libraries, and the `lib` type can be seen as an alias for one of them (but the
3979 actual one is compiler-defined).
3981 * `--crate-type=dylib`, `#[crate_type = "dylib"]` - A dynamic Rust library will
3982 be produced. This is different from the `lib` output type in that this forces
3983 dynamic library generation. The resulting dynamic library can be used as a
3984 dependency for other libraries and/or executables. This output type will
3985 create `*.so` files on linux, `*.dylib` files on osx, and `*.dll` files on
3988 * `--crate-type=staticlib`, `#[crate_type = "staticlib"]` - A static system
3989 library will be produced. This is different from other library outputs in that
3990 the Rust compiler will never attempt to link to `staticlib` outputs. The
3991 purpose of this output type is to create a static library containing all of
3992 the local crate's code along with all upstream dependencies. The static
3993 library is actually a `*.a` archive on linux and osx and a `*.lib` file on
3994 windows. This format is recommended for use in situations such as linking
3995 Rust code into an existing non-Rust application because it will not have
3996 dynamic dependencies on other Rust code.
3998 * `--crate-type=rlib`, `#[crate_type = "rlib"]` - A "Rust library" file will be
3999 produced. This is used as an intermediate artifact and can be thought of as a
4000 "static Rust library". These `rlib` files, unlike `staticlib` files, are
4001 interpreted by the Rust compiler in future linkage. This essentially means
4002 that `rustc` will look for metadata in `rlib` files like it looks for metadata
4003 in dynamic libraries. This form of output is used to produce statically linked
4004 executables as well as `staticlib` outputs.
4006 Note that these outputs are stackable in the sense that if multiple are
4007 specified, then the compiler will produce each form of output at once without
4008 having to recompile. However, this only applies for outputs specified by the
4009 same method. If only `crate_type` attributes are specified, then they will all
4010 be built, but if one or more `--crate-type` command line flags are specified,
4011 then only those outputs will be built.
4013 With all these different kinds of outputs, if crate A depends on crate B, then
4014 the compiler could find B in various different forms throughout the system. The
4015 only forms looked for by the compiler, however, are the `rlib` format and the
4016 dynamic library format. With these two options for a dependent library, the
4017 compiler must at some point make a choice between these two formats. With this
4018 in mind, the compiler follows these rules when determining what format of
4019 dependencies will be used:
4021 1. If a static library is being produced, all upstream dependencies are
4022 required to be available in `rlib` formats. This requirement stems from the
4023 reason that a dynamic library cannot be converted into a static format.
4025 Note that it is impossible to link in native dynamic dependencies to a static
4026 library, and in this case warnings will be printed about all unlinked native
4027 dynamic dependencies.
4029 2. If an `rlib` file is being produced, then there are no restrictions on what
4030 format the upstream dependencies are available in. It is simply required that
4031 all upstream dependencies be available for reading metadata from.
4033 The reason for this is that `rlib` files do not contain any of their upstream
4034 dependencies. It wouldn't be very efficient for all `rlib` files to contain a
4035 copy of `libstd.rlib`!
4037 3. If an executable is being produced and the `-C prefer-dynamic` flag is not
4038 specified, then dependencies are first attempted to be found in the `rlib`
4039 format. If some dependencies are not available in an rlib format, then
4040 dynamic linking is attempted (see below).
4042 4. If a dynamic library or an executable that is being dynamically linked is
4043 being produced, then the compiler will attempt to reconcile the available
4044 dependencies in either the rlib or dylib format to create a final product.
4046 A major goal of the compiler is to ensure that a library never appears more
4047 than once in any artifact. For example, if dynamic libraries B and C were
4048 each statically linked to library A, then a crate could not link to B and C
4049 together because there would be two copies of A. The compiler allows mixing
4050 the rlib and dylib formats, but this restriction must be satisfied.
4052 The compiler currently implements no method of hinting what format a library
4053 should be linked with. When dynamically linking, the compiler will attempt to
4054 maximize dynamic dependencies while still allowing some dependencies to be
4055 linked in via an rlib.
4057 For most situations, having all libraries available as a dylib is recommended
4058 if dynamically linking. For other situations, the compiler will emit a
4059 warning if it is unable to determine which formats to link each library with.
4061 In general, `--crate-type=bin` or `--crate-type=lib` should be sufficient for
4062 all compilation needs, and the other options are just available if more
4063 fine-grained control is desired over the output format of a Rust crate.
4067 Unsafe operations are those that potentially violate the memory-safety
4068 guarantees of Rust's static semantics.
4070 The following language level features cannot be used in the safe subset of
4073 - Dereferencing a [raw pointer](#pointer-types).
4074 - Reading or writing a [mutable static variable](#mutable-statics).
4075 - Calling an unsafe function (including an intrinsic or foreign function).
4079 Unsafe functions are functions that are not safe in all contexts and/or for all
4080 possible inputs. Such a function must be prefixed with the keyword `unsafe` and
4081 can only be called from an `unsafe` block or another `unsafe` function.
4085 A block of code can be prefixed with the `unsafe` keyword, to permit calling
4086 `unsafe` functions or dereferencing raw pointers within a safe function.
4088 When a programmer has sufficient conviction that a sequence of potentially
4089 unsafe operations is actually safe, they can encapsulate that sequence (taken
4090 as a whole) within an `unsafe` block. The compiler will consider uses of such
4091 code safe, in the surrounding context.
4093 Unsafe blocks are used to wrap foreign libraries, make direct use of hardware
4094 or implement features not directly present in the language. For example, Rust
4095 provides the language features necessary to implement memory-safe concurrency
4096 in the language but the implementation of threads and message passing is in the
4099 Rust's type system is a conservative approximation of the dynamic safety
4100 requirements, so in some cases there is a performance cost to using safe code.
4101 For example, a doubly-linked list is not a tree structure and can only be
4102 represented with reference-counted pointers in safe code. By using `unsafe`
4103 blocks to represent the reverse links as raw pointers, it can be implemented
4106 ## Behavior considered undefined
4108 The following is a list of behavior which is forbidden in all Rust code,
4109 including within `unsafe` blocks and `unsafe` functions. Type checking provides
4110 the guarantee that these issues are never caused by safe code.
4113 * Dereferencing a null/dangling raw pointer
4114 * Reads of [undef](http://llvm.org/docs/LangRef.html#undefined-values)
4115 (uninitialized) memory
4116 * Breaking the [pointer aliasing
4117 rules](http://llvm.org/docs/LangRef.html#pointer-aliasing-rules)
4118 with raw pointers (a subset of the rules used by C)
4119 * `&mut T` and `&T` follow LLVM’s scoped [noalias] model, except if the `&T`
4120 contains an `UnsafeCell<U>`. Unsafe code must not violate these aliasing
4122 * Mutating non-mutable data (that is, data reached through a shared reference or
4123 data owned by a `let` binding), unless that data is contained within an `UnsafeCell<U>`.
4124 * Invoking undefined behavior via compiler intrinsics:
4125 * Indexing outside of the bounds of an object with `std::ptr::offset`
4126 (`offset` intrinsic), with
4127 the exception of one byte past the end which is permitted.
4128 * Using `std::ptr::copy_nonoverlapping_memory` (`memcpy32`/`memcpy64`
4129 intrinsics) on overlapping buffers
4130 * Invalid values in primitive types, even in private fields/locals:
4131 * Dangling/null references or boxes
4132 * A value other than `false` (0) or `true` (1) in a `bool`
4133 * A discriminant in an `enum` not included in the type definition
4134 * A value in a `char` which is a surrogate or above `char::MAX`
4135 * Non-UTF-8 byte sequences in a `str`
4136 * Unwinding into Rust from foreign code or unwinding from Rust into foreign
4137 code. Rust's failure system is not compatible with exception handling in
4138 other languages. Unwinding must be caught and handled at FFI boundaries.
4140 [noalias]: http://llvm.org/docs/LangRef.html#noalias
4142 ## Behavior not considered unsafe
4144 This is a list of behavior not considered *unsafe* in Rust terms, but that may
4148 * Leaks of memory and other resources
4149 * Exiting without calling destructors
4151 - Overflow is considered "unexpected" behavior and is always user-error,
4152 unless the `wrapping` primitives are used. In non-optimized builds, the compiler
4153 will insert debug checks that panic on overflow, but in optimized builds overflow
4154 instead results in wrapped values. See [RFC 560] for the rationale and more details.
4156 [RFC 560]: https://github.com/rust-lang/rfcs/blob/master/text/0560-integer-overflow.md
4158 # Appendix: Influences
4160 Rust is not a particularly original language, with design elements coming from
4161 a wide range of sources. Some of these are listed below (including elements
4162 that have since been removed):
4164 * SML, OCaml: algebraic data types, pattern matching, type inference,
4165 semicolon statement separation
4166 * C++: references, RAII, smart pointers, move semantics, monomorphization,
4168 * ML Kit, Cyclone: region based memory management
4169 * Haskell (GHC): typeclasses, type families
4170 * Newsqueak, Alef, Limbo: channels, concurrency
4171 * Erlang: message passing, thread failure, ~~linked thread failure~~,
4172 ~~lightweight concurrency~~
4173 * Swift: optional bindings
4174 * Scheme: hygienic macros
4176 * Ruby: ~~block syntax~~
4177 * NIL, Hermes: ~~typestate~~
4178 * [Unicode Annex #31](http://www.unicode.org/reports/tr31/): identifier and
4181 [ffi]: book/ffi.html
4182 [plugin]: book/compiler-plugins.html