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 | `\'` & [Byte](#byte-escapes) & [Unicode](#unicode-escapes) |
151 | [String](#string-literals) | `"hello"` | `N/A` | All Unicode | `\"` & [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 | `\'` & [Byte](#byte-escapes) |
154 | [Byte string](#byte-string-literals) | `b"hello"` | `N/A` | All ASCII | `\"` & [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 |
167 ##### Unicode escapes
170 | `\u{7FFF}` | 24-bit Unicode character code (up to 6 digits) |
174 | [Number literals](#number-literals)`*` | Example | Exponentiation | Suffixes |
175 |----------------------------------------|---------|----------------|----------|
176 | Decimal integer | `98_222` | `N/A` | Integer suffixes |
177 | Hex integer | `0xff` | `N/A` | Integer suffixes |
178 | Octal integer | `0o77` | `N/A` | Integer suffixes |
179 | Binary integer | `0b1111_0000` | `N/A` | Integer suffixes |
180 | Floating-point | `123.0E+77` | `Optional` | Floating-point suffixes |
182 `*` All number literals allow `_` as a visual separator: `1_234.0E+18f64`
185 | Integer | Floating-point |
186 |---------|----------------|
187 | `u8`, `i8`, `u16`, `i16`, `u32`, `i32`, `u64`, `i64`, `isize`, `usize` | `f32`, `f64` |
189 #### Character and string literals
191 ##### Character literals
193 A _character literal_ is a single Unicode character enclosed within two
194 `U+0027` (single-quote) characters, with the exception of `U+0027` itself,
195 which must be _escaped_ by a preceding `U+005C` character (`\`).
197 ##### String literals
199 A _string literal_ is a sequence of any Unicode characters enclosed within two
200 `U+0022` (double-quote) characters, with the exception of `U+0022` itself,
201 which must be _escaped_ by a preceding `U+005C` character (`\`).
203 Line-break characters are allowed in string literals. Normally they represent
204 themselves (i.e. no translation), but as a special exception, when a `U+005C`
205 character (`\`) occurs immediately before the newline, the `U+005C` character,
206 the newline, and all whitespace at the beginning of the next line are ignored.
207 Thus `a` and `b` are equal:
217 ##### Character escapes
219 Some additional _escapes_ are available in either character or non-raw string
220 literals. An escape starts with a `U+005C` (`\`) and continues with one of the
223 * An _8-bit code point escape_ starts with `U+0078` (`x`) and is
224 followed by exactly two _hex digits_. It denotes the Unicode code point
225 equal to the provided hex value.
226 * A _24-bit code point escape_ starts with `U+0075` (`u`) and is followed
227 by up to six _hex digits_ surrounded by braces `U+007B` (`{`) and `U+007D`
228 (`}`). It denotes the Unicode code point equal to the provided hex value.
229 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
230 (`r`), or `U+0074` (`t`), denoting the Unicode values `U+000A` (LF),
231 `U+000D` (CR) or `U+0009` (HT) respectively.
232 * The _backslash escape_ is the character `U+005C` (`\`) which must be
233 escaped in order to denote *itself*.
235 ##### Raw string literals
237 Raw string literals do not process any escapes. They start with the character
238 `U+0072` (`r`), followed by zero or more of the character `U+0023` (`#`) and a
239 `U+0022` (double-quote) character. The _raw string body_ can contain any sequence
240 of Unicode characters and is terminated only by another `U+0022` (double-quote)
241 character, followed by the same number of `U+0023` (`#`) characters that preceded
242 the opening `U+0022` (double-quote) character.
244 All Unicode characters contained in the raw string body represent themselves,
245 the characters `U+0022` (double-quote) (except when followed by at least as
246 many `U+0023` (`#`) characters as were used to start the raw string literal) or
247 `U+005C` (`\`) do not have any special meaning.
249 Examples for string literals:
252 "foo"; r"foo"; // foo
253 "\"foo\""; r#""foo""#; // "foo"
256 r##"foo #"# bar"##; // foo #"# bar
258 "\x52"; "R"; r"R"; // R
259 "\\x52"; r"\x52"; // \x52
262 #### Byte and byte string literals
266 A _byte literal_ is a single ASCII character (in the `U+0000` to `U+007F`
267 range) or a single _escape_ preceded by the characters `U+0062` (`b`) and
268 `U+0027` (single-quote), and followed by the character `U+0027`. If the character
269 `U+0027` is present within the literal, it must be _escaped_ by a preceding
270 `U+005C` (`\`) character. It is equivalent to a `u8` unsigned 8-bit integer
273 ##### Byte string literals
275 A non-raw _byte string literal_ is a sequence of ASCII characters and _escapes_,
276 preceded by the characters `U+0062` (`b`) and `U+0022` (double-quote), and
277 followed by the character `U+0022`. If the character `U+0022` is present within
278 the literal, it must be _escaped_ by a preceding `U+005C` (`\`) character.
279 Alternatively, a byte string literal can be a _raw byte string literal_, defined
280 below. A byte string literal of length `n` is equivalent to a `&'static [u8; n]` borrowed fixed-sized array
281 of unsigned 8-bit integers.
283 Some additional _escapes_ are available in either byte or non-raw byte string
284 literals. An escape starts with a `U+005C` (`\`) and continues with one of the
287 * A _byte escape_ escape starts with `U+0078` (`x`) and is
288 followed by exactly two _hex digits_. It denotes the byte
289 equal to the provided hex value.
290 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
291 (`r`), or `U+0074` (`t`), denoting the bytes values `0x0A` (ASCII LF),
292 `0x0D` (ASCII CR) or `0x09` (ASCII HT) respectively.
293 * The _backslash escape_ is the character `U+005C` (`\`) which must be
294 escaped in order to denote its ASCII encoding `0x5C`.
296 ##### Raw byte string literals
298 Raw byte string literals do not process any escapes. They start with the
299 character `U+0062` (`b`), followed by `U+0072` (`r`), followed by zero or more
300 of the character `U+0023` (`#`), and a `U+0022` (double-quote) character. The
301 _raw string body_ can contain any sequence of ASCII characters and is terminated
302 only by another `U+0022` (double-quote) character, followed by the same number of
303 `U+0023` (`#`) characters that preceded the opening `U+0022` (double-quote)
304 character. A raw byte string literal can not contain any non-ASCII byte.
306 All characters contained in the raw string body represent their ASCII encoding,
307 the characters `U+0022` (double-quote) (except when followed by at least as
308 many `U+0023` (`#`) characters as were used to start the raw string literal) or
309 `U+005C` (`\`) do not have any special meaning.
311 Examples for byte string literals:
314 b"foo"; br"foo"; // foo
315 b"\"foo\""; br#""foo""#; // "foo"
318 br##"foo #"# bar"##; // foo #"# bar
320 b"\x52"; b"R"; br"R"; // R
321 b"\\x52"; br"\x52"; // \x52
326 A _number literal_ is either an _integer literal_ or a _floating-point
327 literal_. The grammar for recognizing the two kinds of literals is mixed.
329 ##### Integer literals
331 An _integer literal_ has one of four forms:
333 * A _decimal literal_ starts with a *decimal digit* and continues with any
334 mixture of *decimal digits* and _underscores_.
335 * A _hex literal_ starts with the character sequence `U+0030` `U+0078`
336 (`0x`) and continues as any mixture of hex digits and underscores.
337 * An _octal literal_ starts with the character sequence `U+0030` `U+006F`
338 (`0o`) and continues as any mixture of octal digits and underscores.
339 * A _binary literal_ starts with the character sequence `U+0030` `U+0062`
340 (`0b`) and continues as any mixture of binary digits and underscores.
342 Like any literal, an integer literal may be followed (immediately,
343 without any spaces) by an _integer suffix_, which forcibly sets the
344 type of the literal. The integer suffix must be the name of one of the
345 integral types: `u8`, `i8`, `u16`, `i16`, `u32`, `i32`, `u64`, `i64`,
348 The type of an _unsuffixed_ integer literal is determined by type inference:
350 * If an integer type can be _uniquely_ determined from the surrounding
351 program context, the unsuffixed integer literal has that type.
353 * If the program context under-constrains the type, it defaults to the
354 signed 32-bit integer `i32`.
356 * If the program context over-constrains the type, it is considered a
359 Examples of integer literals of various forms:
366 0o70_i16; // type i16
367 0b1111_1111_1001_0000_i32; // type i32
368 0usize; // type usize
371 ##### Floating-point literals
373 A _floating-point literal_ has one of two forms:
375 * A _decimal literal_ followed by a period character `U+002E` (`.`). This is
376 optionally followed by another decimal literal, with an optional _exponent_.
377 * A single _decimal literal_ followed by an _exponent_.
379 Like integer literals, a floating-point literal may be followed by a
380 suffix, so long as the pre-suffix part does not end with `U+002E` (`.`).
381 The suffix forcibly sets the type of the literal. There are two valid
382 _floating-point suffixes_, `f32` and `f64` (the 32-bit and 64-bit floating point
383 types), which explicitly determine the type of the literal.
385 The type of an _unsuffixed_ floating-point literal is determined by
388 * If a floating-point type can be _uniquely_ determined from the
389 surrounding program context, the unsuffixed floating-point literal
392 * If the program context under-constrains the type, it defaults to `f64`.
394 * If the program context over-constrains the type, it is considered a
397 Examples of floating-point literals of various forms:
400 123.0f64; // type f64
403 12E+99_f64; // type f64
404 let x: f64 = 2.; // type f64
407 This last example is different because it is not possible to use the suffix
408 syntax with a floating point literal ending in a period. `2.f64` would attempt
409 to call a method named `f64` on `2`.
411 The representation semantics of floating-point numbers are described in
412 ["Machine Types"](#machine-types).
414 #### Boolean literals
416 The two values of the boolean type are written `true` and `false`.
420 Symbols are a general class of printable [tokens](#tokens) that play structural
421 roles in a variety of grammar productions. They are a
422 set of remaining miscellaneous printable tokens that do not
423 otherwise appear as [unary operators](#unary-operator-expressions), [binary
424 operators](#binary-operator-expressions), or [keywords][keywords].
425 They are catalogued in [the Symbols section][symbols] of the Grammar document.
427 [symbols]: grammar.html#symbols
432 A _path_ is a sequence of one or more path components _logically_ separated by
433 a namespace qualifier (`::`). If a path consists of only one component, it may
434 refer to either an [item](#items) or a [variable](#variables) in a local control
435 scope. If a path has multiple components, it refers to an item.
437 Every item has a _canonical path_ within its crate, but the path naming an item
438 is only meaningful within a given crate. There is no global namespace across
439 crates; an item's canonical path merely identifies it within the crate.
441 Two examples of simple paths consisting of only identifier components:
448 Path components are usually [identifiers](#identifiers), but they may
449 also include angle-bracket-enclosed lists of type arguments. In
450 [expression](#expressions) context, the type argument list is given
451 after a `::` namespace qualifier in order to disambiguate it from a
452 relational expression involving the less-than symbol (`<`). In type
453 expression context, the final namespace qualifier is omitted.
455 Two examples of paths with type arguments:
458 # struct HashMap<K, V>(K,V);
460 # fn id<T>(t: T) -> T { t }
461 type T = HashMap<i32,String>; // Type arguments used in a type expression
462 let x = id::<i32>(10); // Type arguments used in a call expression
466 Paths can be denoted with various leading qualifiers to change the meaning of
469 * Paths starting with `::` are considered to be global paths where the
470 components of the path start being resolved from the crate root. Each
471 identifier in the path must resolve to an item.
479 ::a::foo(); // call a's foo function
485 * Paths starting with the keyword `super` begin resolution relative to the
486 parent module. Each further identifier must resolve to an item.
494 super::a::foo(); // call a's foo function
500 * Paths starting with the keyword `self` begin resolution relative to the
501 current module. Each further identifier must resolve to an item.
513 A number of minor features of Rust are not central enough to have their own
514 syntax, and yet are not implementable as functions. Instead, they are given
515 names, and invoked through a consistent syntax: `some_extension!(...)`.
517 Users of `rustc` can define new syntax extensions in two ways:
519 * [Compiler plugins][plugin] can include arbitrary Rust code that
520 manipulates syntax trees at compile time. Note that the interface
521 for compiler plugins is considered highly unstable.
523 * [Macros](book/macros.html) define new syntax in a higher-level,
528 `macro_rules` allows users to define syntax extension in a declarative way. We
529 call such extensions "macros by example" or simply "macros" — to be distinguished
530 from the "procedural macros" defined in [compiler plugins][plugin].
532 Currently, macros can expand to expressions, statements, items, or patterns.
534 (A `sep_token` is any token other than `*` and `+`. A `non_special_token` is
535 any token other than a delimiter or `$`.)
537 The macro expander looks up macro invocations by name, and tries each macro
538 rule in turn. It transcribes the first successful match. Matching and
539 transcription are closely related to each other, and we will describe them
544 The macro expander matches and transcribes every token that does not begin with
545 a `$` literally, including delimiters. For parsing reasons, delimiters must be
546 balanced, but they are otherwise not special.
548 In the matcher, `$` _name_ `:` _designator_ matches the nonterminal in the Rust
549 syntax named by _designator_. Valid designators are:
551 * `item`: an [item](#items)
552 * `block`: a [block](#block-expressions)
553 * `stmt`: a [statement](#statements)
554 * `pat`: a [pattern](#match-expressions)
555 * `expr`: an [expression](#expressions)
556 * `ty`: a [type](#types)
557 * `ident`: an [identifier](#identifiers)
558 * `path`: a [path](#paths)
559 * `tt`: either side of the `=>` in macro rules
560 * `meta`: the contents of an [attribute](#attributes)
562 In the transcriber, the
563 designator is already known, and so only the name of a matched nonterminal comes
564 after the dollar sign.
566 In both the matcher and transcriber, the Kleene star-like operator indicates
567 repetition. The Kleene star operator consists of `$` and parentheses, optionally
568 followed by a separator token, followed by `*` or `+`. `*` means zero or more
569 repetitions, `+` means at least one repetition. The parentheses are not matched or
570 transcribed. On the matcher side, a name is bound to _all_ of the names it
571 matches, in a structure that mimics the structure of the repetition encountered
572 on a successful match. The job of the transcriber is to sort that structure
575 The rules for transcription of these repetitions are called "Macro By Example".
576 Essentially, one "layer" of repetition is discharged at a time, and all of them
577 must be discharged by the time a name is transcribed. Therefore, `( $( $i:ident
578 ),* ) => ( $i )` is an invalid macro, but `( $( $i:ident ),* ) => ( $( $i:ident
579 ),* )` is acceptable (if trivial).
581 When Macro By Example encounters a repetition, it examines all of the `$`
582 _name_ s that occur in its body. At the "current layer", they all must repeat
583 the same number of times, so ` ( $( $i:ident ),* ; $( $j:ident ),* ) => ( $(
584 ($i,$j) ),* )` is valid if given the argument `(a,b,c ; d,e,f)`, but not
585 `(a,b,c ; d,e)`. The repetition walks through the choices at that layer in
586 lockstep, so the former input transcribes to `(a,d), (b,e), (c,f)`.
588 Nested repetitions are allowed.
590 ### Parsing limitations
592 The parser used by the macro system is reasonably powerful, but the parsing of
593 Rust syntax is restricted in two ways:
595 1. Macro definitions are required to include suitable separators after parsing
596 expressions and other bits of the Rust grammar. This implies that
597 a macro definition like `$i:expr [ , ]` is not legal, because `[` could be part
598 of an expression. A macro definition like `$i:expr,` or `$i:expr;` would be legal,
599 however, because `,` and `;` are legal separators. See [RFC 550] for more information.
600 2. The parser must have eliminated all ambiguity by the time it reaches a `$`
601 _name_ `:` _designator_. This requirement most often affects name-designator
602 pairs when they occur at the beginning of, or immediately after, a `$(...)*`;
603 requiring a distinctive token in front can solve the problem.
605 [RFC 550]: https://github.com/rust-lang/rfcs/blob/master/text/0550-macro-future-proofing.md
607 # Crates and source files
609 Although Rust, like any other language, can be implemented by an interpreter as
610 well as a compiler, the only existing implementation is a compiler,
612 always been designed to be compiled. For these reasons, this section assumes a
615 Rust's semantics obey a *phase distinction* between compile-time and
616 run-time.[^phase-distinction] Semantic rules that have a *static
617 interpretation* govern the success or failure of compilation, while
619 that have a *dynamic interpretation* govern the behavior of the program at
622 [^phase-distinction]: This distinction would also exist in an interpreter.
623 Static checks like syntactic analysis, type checking, and lints should
624 happen before the program is executed regardless of when it is executed.
626 The compilation model centers on artifacts called _crates_. Each compilation
627 processes a single crate in source form, and if successful, produces a single
628 crate in binary form: either an executable or some sort of
629 library.[^cratesourcefile]
631 [^cratesourcefile]: A crate is somewhat analogous to an *assembly* in the
632 ECMA-335 CLI model, a *library* in the SML/NJ Compilation Manager, a *unit*
633 in the Owens and Flatt module system, or a *configuration* in Mesa.
635 A _crate_ is a unit of compilation and linking, as well as versioning,
636 distribution and runtime loading. A crate contains a _tree_ of nested
637 [module](#modules) scopes. The top level of this tree is a module that is
638 anonymous (from the point of view of paths within the module) and any item
639 within a crate has a canonical [module path](#paths) denoting its location
640 within the crate's module tree.
642 The Rust compiler is always invoked with a single source file as input, and
643 always produces a single output crate. The processing of that source file may
644 result in other source files being loaded as modules. Source files have the
647 A Rust source file describes a module, the name and location of which —
648 in the module tree of the current crate — are defined from outside the
649 source file: either by an explicit `mod_item` in a referencing source file, or
650 by the name of the crate itself. Every source file is a module, but not every
651 module needs its own source file: [module definitions](#modules) can be nested
654 Each source file contains a sequence of zero or more `item` definitions, and
655 may optionally begin with any number of [attributes](#items-and-attributes)
656 that apply to the containing module, most of which influence the behavior of
657 the compiler. The anonymous crate module can have additional attributes that
658 apply to the crate as a whole.
661 // Specify the crate name.
662 #![crate_name = "projx"]
664 // Specify the type of output artifact.
665 #![crate_type = "lib"]
667 // Turn on a warning.
668 // This can be done in any module, not just the anonymous crate module.
669 #![warn(non_camel_case_types)]
672 A crate that contains a `main` function can be compiled to an executable. If a
673 `main` function is present, its return type must be [`unit`](#tuple-types)
674 and it must take no arguments.
676 # Items and attributes
678 Crates contain [items](#items), each of which may have some number of
679 [attributes](#attributes) attached to it.
683 An _item_ is a component of a crate. Items are organized within a crate by a
684 nested set of [modules](#modules). Every crate has a single "outermost"
685 anonymous module; all further items within the crate have [paths](#paths)
686 within the module tree of the crate.
688 Items are entirely determined at compile-time, generally remain fixed during
689 execution, and may reside in read-only memory.
691 There are several kinds of item:
693 * [`extern crate` declarations](#extern-crate-declarations)
694 * [`use` declarations](#use-declarations)
695 * [modules](#modules)
696 * [functions](#functions)
697 * [type definitions](grammar.html#type-definitions)
698 * [structs](#structs)
699 * [enumerations](#enumerations)
700 * [constant items](#constant-items)
701 * [static items](#static-items)
703 * [implementations](#implementations)
705 Some items form an implicit scope for the declaration of sub-items. In other
706 words, within a function or module, declarations of items can (in many cases)
707 be mixed with the statements, control blocks, and similar artifacts that
708 otherwise compose the item body. The meaning of these scoped items is the same
709 as if the item was declared outside the scope — it is still a static item
710 — except that the item's *path name* within the module namespace is
711 qualified by the name of the enclosing item, or is private to the enclosing
712 item (in the case of functions). The grammar specifies the exact locations in
713 which sub-item declarations may appear.
717 All items except modules, constants and statics may be *parameterized* by type.
718 Type parameters are given as a comma-separated list of identifiers enclosed in
719 angle brackets (`<...>`), after the name of the item and before its definition.
720 The type parameters of an item are considered "part of the name", not part of
721 the type of the item. A referencing [path](#paths) must (in principle) provide
722 type arguments as a list of comma-separated types enclosed within angle
723 brackets, in order to refer to the type-parameterized item. In practice, the
724 type-inference system can usually infer such argument types from context. There
725 are no general type-parametric types, only type-parametric items. That is, Rust
726 has no notion of type abstraction: there are no higher-ranked (or "forall") types
727 abstracted over other types, though higher-ranked types do exist for lifetimes.
731 A module is a container for zero or more [items](#items).
733 A _module item_ is a module, surrounded in braces, named, and prefixed with the
734 keyword `mod`. A module item introduces a new, named module into the tree of
735 modules making up a crate. Modules can nest arbitrarily.
737 An example of a module:
741 type Complex = (f64, f64);
742 fn sin(f: f64) -> f64 {
746 fn cos(f: f64) -> f64 {
750 fn tan(f: f64) -> f64 {
757 Modules and types share the same namespace. Declaring a named type with
758 the same name as a module in scope is forbidden: that is, a type definition,
759 trait, struct, enumeration, or type parameter can't shadow the name of a module
760 in scope, or vice versa.
762 A module without a body is loaded from an external file, by default with the
763 same name as the module, plus the `.rs` extension. When a nested submodule is
764 loaded from an external file, it is loaded from a subdirectory path that
765 mirrors the module hierarchy.
768 // Load the `vec` module from `vec.rs`
772 // Load the `local_data` module from `thread/local_data.rs`
773 // or `thread/local_data/mod.rs`.
778 The directories and files used for loading external file modules can be
779 influenced with the `path` attribute.
782 #[path = "thread_files"]
784 // Load the `local_data` module from `thread_files/tls.rs`
790 #### Extern crate declarations
792 An _`extern crate` declaration_ specifies a dependency on an external crate.
793 The external crate is then bound into the declaring scope as the `ident`
794 provided in the `extern_crate_decl`.
796 The external crate is resolved to a specific `soname` at compile time, and a
797 runtime linkage requirement to that `soname` is passed to the linker for
798 loading at runtime. The `soname` is resolved at compile time by scanning the
799 compiler's library path and matching the optional `crateid` provided against
800 the `crateid` attributes that were declared on the external crate when it was
801 compiled. If no `crateid` is provided, a default `name` attribute is assumed,
802 equal to the `ident` given in the `extern_crate_decl`.
804 Three examples of `extern crate` declarations:
809 extern crate std; // equivalent to: extern crate std as std;
811 extern crate std as ruststd; // linking to 'std' under another name
814 #### Use declarations
816 A _use declaration_ creates one or more local name bindings synonymous with
817 some other [path](#paths). Usually a `use` declaration is used to shorten the
818 path required to refer to a module item. These declarations may appear at the
819 top of [modules](#modules) and [blocks](grammar.html#block-expressions).
821 > **Note**: Unlike in many languages,
822 > `use` declarations in Rust do *not* declare linkage dependency with external crates.
823 > Rather, [`extern crate` declarations](#extern-crate-declarations) declare linkage dependencies.
825 Use declarations support a number of convenient shortcuts:
827 * Rebinding the target name as a new local name, using the syntax `use p::q::r as x;`
828 * Simultaneously binding a list of paths differing only in their final element,
829 using the glob-like brace syntax `use a::b::{c,d,e,f};`
830 * Binding all paths matching a given prefix, using the asterisk wildcard syntax
832 * Simultaneously binding a list of paths differing only in their final element
833 and their immediate parent module, using the `self` keyword, such as
834 `use a::b::{self, c, d};`
836 An example of `use` declarations:
839 use std::option::Option::{Some, None};
840 use std::collections::hash_map::{self, HashMap};
843 fn bar(map1: HashMap<String, usize>, map2: hash_map::HashMap<String, usize>){}
846 // Equivalent to 'foo(vec![std::option::Option::Some(1.0f64),
847 // std::option::Option::None]);'
848 foo(vec![Some(1.0f64), None]);
850 // Both `hash_map` and `HashMap` are in scope.
851 let map1 = HashMap::new();
852 let map2 = hash_map::HashMap::new();
857 Like items, `use` declarations are private to the containing module, by
858 default. Also like items, a `use` declaration can be public, if qualified by
859 the `pub` keyword. Such a `use` declaration serves to _re-export_ a name. A
860 public `use` declaration can therefore _redirect_ some public name to a
861 different target definition: even a definition with a private canonical path,
862 inside a different module. If a sequence of such redirections form a cycle or
863 cannot be resolved unambiguously, they represent a compile-time error.
865 An example of re-exporting:
870 pub use quux::foo::{bar, baz};
879 In this example, the module `quux` re-exports two public names defined in
882 Also note that the paths contained in `use` items are relative to the crate
883 root. So, in the previous example, the `use` refers to `quux::foo::{bar,
884 baz}`, and not simply to `foo::{bar, baz}`. This also means that top-level
885 module declarations should be at the crate root if direct usage of the declared
886 modules within `use` items is desired. It is also possible to use `self` and
887 `super` at the beginning of a `use` item to refer to the current and direct
888 parent modules respectively. All rules regarding accessing declared modules in
889 `use` declarations apply to both module declarations and `extern crate`
892 An example of what will and will not work for `use` items:
895 # #![allow(unused_imports)]
896 use foo::baz::foobaz; // good: foo is at the root of the crate
904 use foo::example::iter; // good: foo is at crate root
905 // use example::iter; // bad: example is not at the crate root
906 use self::baz::foobaz; // good: self refers to module 'foo'
907 use foo::bar::foobar; // good: foo is at crate root
914 use super::bar::foobar; // good: super refers to module 'foo'
924 A _function item_ defines a sequence of [statements](#statements) and a
925 final [expression](#expressions), along with a name and a set of
926 parameters. Other than a name, all these are optional.
927 Functions are declared with the keyword `fn`. Functions may declare a
928 set of *input* [*variables*](#variables) as parameters, through which the caller
929 passes arguments into the function, and the *output* [*type*](#types)
930 of the value the function will return to its caller on completion.
932 A function may also be copied into a first-class *value*, in which case the
933 value has the corresponding [*function type*](#function-types), and can be used
934 otherwise exactly as a function item (with a minor additional cost of calling
935 the function indirectly).
937 Every control path in a function logically ends with a `return` expression or a
938 diverging expression. If the outermost block of a function has a
939 value-producing expression in its final-expression position, that expression is
940 interpreted as an implicit `return` expression applied to the final-expression.
942 An example of a function:
945 fn add(x: i32, y: i32) -> i32 {
950 As with `let` bindings, function arguments are irrefutable patterns, so any
951 pattern that is valid in a let binding is also valid as an argument.
954 fn first((value, _): (i32, i32)) -> i32 { value }
958 #### Generic functions
960 A _generic function_ allows one or more _parameterized types_ to appear in its
961 signature. Each type parameter must be explicitly declared, in an
962 angle-bracket-enclosed, comma-separated list following the function name.
965 // foo is generic over A and B
967 fn foo<A, B>(x: A, y: B) {
970 Inside the function signature and body, the name of the type parameter can be
971 used as a type name. [Trait](#traits) bounds can be specified for type parameters
972 to allow methods with that trait to be called on values of that type. This is
973 specified using the `where` syntax:
976 fn foo<T>(x: T) where T: Debug {
979 When a generic function is referenced, its type is instantiated based on the
980 context of the reference. For example, calling the `foo` function here:
985 fn foo<T>(x: &[T]) where T: Debug {
993 will instantiate type parameter `T` with `i32`.
995 The type parameters can also be explicitly supplied in a trailing
996 [path](#paths) component after the function name. This might be necessary if
997 there is not sufficient context to determine the type parameters. For example,
998 `mem::size_of::<u32>() == 4`.
1000 #### Diverging functions
1002 A special kind of function can be declared with a `!` character where the
1003 output type would normally be. For example:
1006 fn my_err(s: &str) -> ! {
1012 We call such functions "diverging" because they never return a value to the
1013 caller. Every control path in a diverging function must end with a `panic!()` or
1014 a call to another diverging function on every control path. The `!` annotation
1015 does *not* denote a type.
1017 It might be necessary to declare a diverging function because as mentioned
1018 previously, the typechecker checks that every control path in a function ends
1019 with a [`return`](#return-expressions) or diverging expression. So, if `my_err`
1020 were declared without the `!` annotation, the following code would not
1024 # fn my_err(s: &str) -> ! { panic!() }
1026 fn f(i: i32) -> i32 {
1031 my_err("Bad number!");
1036 This will not compile without the `!` annotation on `my_err`, since the `else`
1037 branch of the conditional in `f` does not return an `i32`, as required by the
1038 signature of `f`. Adding the `!` annotation to `my_err` informs the
1039 typechecker that, should control ever enter `my_err`, no further type judgments
1040 about `f` need to hold, since control will never resume in any context that
1041 relies on those judgments. Thus the return type on `f` only needs to reflect
1042 the `if` branch of the conditional.
1044 #### Extern functions
1046 Extern functions are part of Rust's foreign function interface, providing the
1047 opposite functionality to [external blocks](#external-blocks). Whereas
1048 external blocks allow Rust code to call foreign code, extern functions with
1049 bodies defined in Rust code _can be called by foreign code_. They are defined
1050 in the same way as any other Rust function, except that they have the `extern`
1054 // Declares an extern fn, the ABI defaults to "C"
1055 extern fn new_i32() -> i32 { 0 }
1057 // Declares an extern fn with "stdcall" ABI
1058 extern "stdcall" fn new_i32_stdcall() -> i32 { 0 }
1061 Unlike normal functions, extern fns have type `extern "ABI" fn()`. This is the
1062 same type as the functions declared in an extern block.
1065 # extern fn new_i32() -> i32 { 0 }
1066 let fptr: extern "C" fn() -> i32 = new_i32;
1069 Extern functions may be called directly from Rust code as Rust uses large,
1070 contiguous stack segments like C.
1074 A _type alias_ defines a new name for an existing [type](#types). Type
1075 aliases are declared with the keyword `type`. Every value has a single,
1076 specific type, but may implement several different traits, or be compatible with
1077 several different type constraints.
1079 For example, the following defines the type `Point` as a synonym for the type
1080 `(u8, u8)`, the type of pairs of unsigned 8 bit integers:
1083 type Point = (u8, u8);
1084 let p: Point = (41, 68);
1089 A _struct_ is a nominal [struct type](#struct-types) defined with the
1092 An example of a `struct` item and its use:
1095 struct Point {x: i32, y: i32}
1096 let p = Point {x: 10, y: 11};
1100 A _tuple struct_ is a nominal [tuple type](#tuple-types), also defined with
1101 the keyword `struct`. For example:
1104 struct Point(i32, i32);
1105 let p = Point(10, 11);
1106 let px: i32 = match p { Point(x, _) => x };
1109 A _unit-like struct_ is a struct without any fields, defined by leaving off
1110 the list of fields entirely. Such a struct implicitly defines a constant of
1111 its type with the same name. For example:
1114 # #![feature(braced_empty_structs)]
1116 let c = [Cookie, Cookie {}, Cookie, Cookie {}];
1122 # #![feature(braced_empty_structs)]
1124 const Cookie: Cookie = Cookie {};
1125 let c = [Cookie, Cookie {}, Cookie, Cookie {}];
1128 The precise memory layout of a struct is not specified. One can specify a
1129 particular layout using the [`repr` attribute](#ffi-attributes).
1133 An _enumeration_ is a simultaneous definition of a nominal [enumerated
1134 type](#enumerated-types) as well as a set of *constructors*, that can be used
1135 to create or pattern-match values of the corresponding enumerated type.
1137 Enumerations are declared with the keyword `enum`.
1139 An example of an `enum` item and its use:
1147 let mut a: Animal = Animal::Dog;
1151 Enumeration constructors can have either named or unnamed fields:
1156 Cat { name: String, weight: f64 }
1159 let mut a: Animal = Animal::Dog("Cocoa".to_string(), 37.2);
1160 a = Animal::Cat { name: "Spotty".to_string(), weight: 2.7 };
1163 In this example, `Cat` is a _struct-like enum variant_,
1164 whereas `Dog` is simply called an enum variant.
1166 Enums have a discriminant. You can assign them explicitly:
1174 If a discriminant isn't assigned, they start at zero, and add one for each
1177 You can cast an enum to get this value:
1180 # enum Foo { Bar = 123 }
1181 let x = Foo::Bar as u32; // x is now 123u32
1184 This only works as long as none of the variants have data attached. If
1185 it were `Bar(i32)`, this is disallowed.
1189 A *constant item* is a named _constant value_ which is not associated with a
1190 specific memory location in the program. Constants are essentially inlined
1191 wherever they are used, meaning that they are copied directly into the relevant
1192 context when used. References to the same constant are not necessarily
1193 guaranteed to refer to the same memory address.
1195 Constant values must not have destructors, and otherwise permit most forms of
1196 data. Constants may refer to the address of other constants, in which case the
1197 address will have the `static` lifetime. The compiler is, however, still at
1198 liberty to translate the constant many times, so the address referred to may not
1201 Constants must be explicitly typed. The type may be `bool`, `char`, a number, or
1202 a type derived from those primitive types. The derived types are references with
1203 the `static` lifetime, fixed-size arrays, tuples, enum variants, and structs.
1206 const BIT1: u32 = 1 << 0;
1207 const BIT2: u32 = 1 << 1;
1209 const BITS: [u32; 2] = [BIT1, BIT2];
1210 const STRING: &'static str = "bitstring";
1212 struct BitsNStrings<'a> {
1217 const BITS_N_STRINGS: BitsNStrings<'static> = BitsNStrings {
1225 A *static item* is similar to a *constant*, except that it represents a precise
1226 memory location in the program. A static is never "inlined" at the usage site,
1227 and all references to it refer to the same memory location. Static items have
1228 the `static` lifetime, which outlives all other lifetimes in a Rust program.
1229 Static items may be placed in read-only memory if they do not contain any
1230 interior mutability.
1232 Statics may contain interior mutability through the `UnsafeCell` language item.
1233 All access to a static is safe, but there are a number of restrictions on
1236 * Statics may not contain any destructors.
1237 * The types of static values must ascribe to `Sync` to allow thread-safe access.
1238 * Statics may not refer to other statics by value, only by reference.
1239 * Constants cannot refer to statics.
1241 Constants should in general be preferred over statics, unless large amounts of
1242 data are being stored, or single-address and mutability properties are required.
1244 #### Mutable statics
1246 If a static item is declared with the `mut` keyword, then it is allowed to
1247 be modified by the program. One of Rust's goals is to make concurrency bugs
1248 hard to run into, and this is obviously a very large source of race conditions
1249 or other bugs. For this reason, an `unsafe` block is required when either
1250 reading or writing a mutable static variable. Care should be taken to ensure
1251 that modifications to a mutable static are safe with respect to other threads
1252 running in the same process.
1254 Mutable statics are still very useful, however. They can be used with C
1255 libraries and can also be bound from C libraries (in an `extern` block).
1258 # fn atomic_add(_: &mut u32, _: u32) -> u32 { 2 }
1260 static mut LEVELS: u32 = 0;
1262 // This violates the idea of no shared state, and this doesn't internally
1263 // protect against races, so this function is `unsafe`
1264 unsafe fn bump_levels_unsafe1() -> u32 {
1270 // Assuming that we have an atomic_add function which returns the old value,
1271 // this function is "safe" but the meaning of the return value may not be what
1272 // callers expect, so it's still marked as `unsafe`
1273 unsafe fn bump_levels_unsafe2() -> u32 {
1274 return atomic_add(&mut LEVELS, 1);
1278 Mutable statics have the same restrictions as normal statics, except that the
1279 type of the value is not required to ascribe to `Sync`.
1283 A _trait_ describes an abstract interface that types can
1284 implement. This interface consists of associated items, which come in
1291 Associated functions whose first parameter is named `self` are called
1292 methods and may be invoked using `.` notation (e.g., `x.foo()`).
1294 All traits define an implicit type parameter `Self` that refers to
1295 "the type that is implementing this interface". Traits may also
1296 contain additional type parameters. These type parameters (including
1297 `Self`) may be constrained by other traits and so forth as usual.
1299 Trait bounds on `Self` are considered "supertraits". These are
1300 required to be acyclic. Supertraits are somewhat different from other
1301 constraints in that they affect what methods are available in the
1302 vtable when the trait is used as a [trait object](#trait-objects).
1304 Traits are implemented for specific types through separate
1305 [implementations](#implementations).
1307 Consider the following trait:
1310 # type Surface = i32;
1311 # type BoundingBox = i32;
1313 fn draw(&self, Surface);
1314 fn bounding_box(&self) -> BoundingBox;
1318 This defines a trait with two methods. All values that have
1319 [implementations](#implementations) of this trait in scope can have their
1320 `draw` and `bounding_box` methods called, using `value.bounding_box()`
1321 [syntax](#method-call-expressions).
1323 Traits can include default implementations of methods, as in:
1328 fn baz(&self) { println!("We called baz."); }
1332 Here the `baz` method has a default implementation, so types that implement
1333 `Foo` need only implement `bar`. It is also possible for implementing types
1334 to override a method that has a default implementation.
1336 Type parameters can be specified for a trait to make it generic. These appear
1337 after the trait name, using the same syntax used in [generic
1338 functions](#generic-functions).
1342 fn len(&self) -> u32;
1343 fn elt_at(&self, n: u32) -> T;
1344 fn iter<F>(&self, F) where F: Fn(T);
1348 It is also possible to define associated types for a trait. Consider the
1349 following example of a `Container` trait. Notice how the type is available
1350 for use in the method signatures:
1356 fn insert(&mut self, Self::E);
1360 In order for a type to implement this trait, it must not only provide
1361 implementations for every method, but it must specify the type `E`. Here's
1362 an implementation of `Container` for the standard library type `Vec`:
1367 # fn empty() -> Self;
1368 # fn insert(&mut self, Self::E);
1370 impl<T> Container for Vec<T> {
1372 fn empty() -> Vec<T> { Vec::new() }
1373 fn insert(&mut self, x: T) { self.push(x); }
1377 Generic functions may use traits as _bounds_ on their type parameters. This
1378 will have two effects:
1380 - Only types that have the trait may instantiate the parameter.
1381 - Within the generic function, the methods of the trait can be
1382 called on values that have the parameter's type.
1387 # type Surface = i32;
1388 # trait Shape { fn draw(&self, Surface); }
1389 fn draw_twice<T: Shape>(surface: Surface, sh: T) {
1395 Traits also define a [trait object](#trait-objects) with the same
1396 name as the trait. Values of this type are created by coercing from a
1397 pointer of some specific type to a pointer of trait type. For example,
1398 `&T` could be coerced to `&Shape` if `T: Shape` holds (and similarly
1399 for `Box<T>`). This coercion can either be implicit or
1400 [explicit](#type-cast-expressions). Here is an example of an explicit
1405 impl Shape for i32 { }
1406 let mycircle = 0i32;
1407 let myshape: Box<Shape> = Box::new(mycircle) as Box<Shape>;
1410 The resulting value is a box containing the value that was cast, along with
1411 information that identifies the methods of the implementation that was used.
1412 Values with a trait type can have [methods called](#method-call-expressions) on
1413 them, for any method in the trait, and can be used to instantiate type
1414 parameters that are bounded by the trait.
1416 Trait methods may be static, which means that they lack a `self` argument.
1417 This means that they can only be called with function call syntax (`f(x)`) and
1418 not method call syntax (`obj.f()`). The way to refer to the name of a static
1419 method is to qualify it with the trait name, treating the trait name like a
1420 module. For example:
1424 fn from_i32(n: i32) -> Self;
1427 fn from_i32(n: i32) -> f64 { n as f64 }
1429 let x: f64 = Num::from_i32(42);
1432 Traits may inherit from other traits. Consider the following example:
1435 trait Shape { fn area(&self) -> f64; }
1436 trait Circle : Shape { fn radius(&self) -> f64; }
1439 The syntax `Circle : Shape` means that types that implement `Circle` must also
1440 have an implementation for `Shape`. Multiple supertraits are separated by `+`,
1441 `trait Circle : Shape + PartialEq { }`. In an implementation of `Circle` for a
1442 given type `T`, methods can refer to `Shape` methods, since the typechecker
1443 checks that any type with an implementation of `Circle` also has an
1444 implementation of `Shape`:
1449 trait Shape { fn area(&self) -> f64; }
1450 trait Circle : Shape { fn radius(&self) -> f64; }
1451 impl Shape for Foo {
1452 fn area(&self) -> f64 {
1456 impl Circle for Foo {
1457 fn radius(&self) -> f64 {
1458 println!("calling area: {}", self.area());
1468 In type-parameterized functions, methods of the supertrait may be called on
1469 values of subtrait-bound type parameters. Referring to the previous example of
1470 `trait Circle : Shape`:
1473 # trait Shape { fn area(&self) -> f64; }
1474 # trait Circle : Shape { fn radius(&self) -> f64; }
1475 fn radius_times_area<T: Circle>(c: T) -> f64 {
1476 // `c` is both a Circle and a Shape
1477 c.radius() * c.area()
1481 Likewise, supertrait methods may also be called on trait objects.
1484 # trait Shape { fn area(&self) -> f64; }
1485 # trait Circle : Shape { fn radius(&self) -> f64; }
1486 # impl Shape for i32 { fn area(&self) -> f64 { 0.0 } }
1487 # impl Circle for i32 { fn radius(&self) -> f64 { 0.0 } }
1488 # let mycircle = 0i32;
1489 let mycircle = Box::new(mycircle) as Box<Circle>;
1490 let nonsense = mycircle.radius() * mycircle.area();
1495 An _implementation_ is an item that implements a [trait](#traits) for a
1498 Implementations are defined with the keyword `impl`.
1501 # #[derive(Copy, Clone)]
1502 # struct Point {x: f64, y: f64};
1503 # type Surface = i32;
1504 # struct BoundingBox {x: f64, y: f64, width: f64, height: f64};
1505 # trait Shape { fn draw(&self, Surface); fn bounding_box(&self) -> BoundingBox; }
1506 # fn do_draw_circle(s: Surface, c: Circle) { }
1512 impl Copy for Circle {}
1514 impl Clone for Circle {
1515 fn clone(&self) -> Circle { *self }
1518 impl Shape for Circle {
1519 fn draw(&self, s: Surface) { do_draw_circle(s, *self); }
1520 fn bounding_box(&self) -> BoundingBox {
1521 let r = self.radius;
1523 x: self.center.x - r,
1524 y: self.center.y - r,
1532 It is possible to define an implementation without referring to a trait. The
1533 methods in such an implementation can only be used as direct calls on the values
1534 of the type that the implementation targets. In such an implementation, the
1535 trait type and `for` after `impl` are omitted. Such implementations are limited
1536 to nominal types (enums, structs, trait objects), and the implementation must
1537 appear in the same crate as the `self` type:
1540 struct Point {x: i32, y: i32}
1544 println!("Point is at ({}, {})", self.x, self.y);
1548 let my_point = Point {x: 10, y:11};
1552 When a trait _is_ specified in an `impl`, all methods declared as part of the
1553 trait must be implemented, with matching types and type parameter counts.
1555 An implementation can take type parameters, which can be different from the
1556 type parameters taken by the trait it implements. Implementation parameters
1557 are written after the `impl` keyword.
1560 # trait Seq<T> { fn dummy(&self, _: T) { } }
1561 impl<T> Seq<T> for Vec<T> {
1564 impl Seq<bool> for u32 {
1565 /* Treat the integer as a sequence of bits */
1571 External blocks form the basis for Rust's foreign function interface.
1572 Declarations in an external block describe symbols in external, non-Rust
1575 Functions within external blocks are declared in the same way as other Rust
1576 functions, with the exception that they may not have a body and are instead
1577 terminated by a semicolon.
1579 Functions within external blocks may be called by Rust code, just like
1580 functions defined in Rust. The Rust compiler automatically translates between
1581 the Rust ABI and the foreign ABI.
1583 A number of [attributes](#attributes) control the behavior of external blocks.
1585 By default external blocks assume that the library they are calling uses the
1586 standard C "cdecl" ABI. Other ABIs may be specified using an `abi` string, as
1590 // Interface to the Windows API
1591 extern "stdcall" { }
1594 The `link` attribute allows the name of the library to be specified. When
1595 specified the compiler will attempt to link against the native library of the
1599 #[link(name = "crypto")]
1603 The type of a function declared in an extern block is `extern "abi" fn(A1, ...,
1604 An) -> R`, where `A1...An` are the declared types of its arguments and `R` is
1605 the declared return type.
1607 It is valid to add the `link` attribute on an empty extern block. You can use
1608 this to satisfy the linking requirements of extern blocks elsewhere in your code
1609 (including upstream crates) instead of adding the attribute to each extern block.
1611 ## Visibility and Privacy
1613 These two terms are often used interchangeably, and what they are attempting to
1614 convey is the answer to the question "Can this item be used at this location?"
1616 Rust's name resolution operates on a global hierarchy of namespaces. Each level
1617 in the hierarchy can be thought of as some item. The items are one of those
1618 mentioned above, but also include external crates. Declaring or defining a new
1619 module can be thought of as inserting a new tree into the hierarchy at the
1620 location of the definition.
1622 To control whether interfaces can be used across modules, Rust checks each use
1623 of an item to see whether it should be allowed or not. This is where privacy
1624 warnings are generated, or otherwise "you used a private item of another module
1625 and weren't allowed to."
1627 By default, everything in Rust is *private*, with one exception. Enum variants
1628 in a `pub` enum are also public by default. When an item is declared as `pub`,
1629 it can be thought of as being accessible to the outside world. For example:
1633 // Declare a private struct
1636 // Declare a public struct with a private field
1641 // Declare a public enum with two public variants
1643 PubliclyAccessibleState,
1644 PubliclyAccessibleState2,
1648 With the notion of an item being either public or private, Rust allows item
1649 accesses in two cases:
1651 1. If an item is public, then it can be used externally through any of its
1653 2. If an item is private, it may be accessed by the current module and its
1656 These two cases are surprisingly powerful for creating module hierarchies
1657 exposing public APIs while hiding internal implementation details. To help
1658 explain, here's a few use cases and what they would entail:
1660 * A library developer needs to expose functionality to crates which link
1661 against their library. As a consequence of the first case, this means that
1662 anything which is usable externally must be `pub` from the root down to the
1663 destination item. Any private item in the chain will disallow external
1666 * A crate needs a global available "helper module" to itself, but it doesn't
1667 want to expose the helper module as a public API. To accomplish this, the
1668 root of the crate's hierarchy would have a private module which then
1669 internally has a "public API". Because the entire crate is a descendant of
1670 the root, then the entire local crate can access this private module through
1673 * When writing unit tests for a module, it's often a common idiom to have an
1674 immediate child of the module to-be-tested named `mod test`. This module
1675 could access any items of the parent module through the second case, meaning
1676 that internal implementation details could also be seamlessly tested from the
1679 In the second case, it mentions that a private item "can be accessed" by the
1680 current module and its descendants, but the exact meaning of accessing an item
1681 depends on what the item is. Accessing a module, for example, would mean
1682 looking inside of it (to import more items). On the other hand, accessing a
1683 function would mean that it is invoked. Additionally, path expressions and
1684 import statements are considered to access an item in the sense that the
1685 import/expression is only valid if the destination is in the current visibility
1688 Here's an example of a program which exemplifies the three cases outlined
1692 // This module is private, meaning that no external crate can access this
1693 // module. Because it is private at the root of this current crate, however, any
1694 // module in the crate may access any publicly visible item in this module.
1695 mod crate_helper_module {
1697 // This function can be used by anything in the current crate
1698 pub fn crate_helper() {}
1700 // This function *cannot* be used by anything else in the crate. It is not
1701 // publicly visible outside of the `crate_helper_module`, so only this
1702 // current module and its descendants may access it.
1703 fn implementation_detail() {}
1706 // This function is "public to the root" meaning that it's available to external
1707 // crates linking against this one.
1708 pub fn public_api() {}
1710 // Similarly to 'public_api', this module is public so external crates may look
1713 use crate_helper_module;
1715 pub fn my_method() {
1716 // Any item in the local crate may invoke the helper module's public
1717 // interface through a combination of the two rules above.
1718 crate_helper_module::crate_helper();
1721 // This function is hidden to any module which is not a descendant of
1723 fn my_implementation() {}
1729 fn test_my_implementation() {
1730 // Because this module is a descendant of `submodule`, it's allowed
1731 // to access private items inside of `submodule` without a privacy
1733 super::my_implementation();
1741 For a rust program to pass the privacy checking pass, all paths must be valid
1742 accesses given the two rules above. This includes all use statements,
1743 expressions, types, etc.
1745 ### Re-exporting and Visibility
1747 Rust allows publicly re-exporting items through a `pub use` directive. Because
1748 this is a public directive, this allows the item to be used in the current
1749 module through the rules above. It essentially allows public access into the
1750 re-exported item. For example, this program is valid:
1753 pub use self::implementation::api;
1755 mod implementation {
1764 This means that any external crate referencing `implementation::api::f` would
1765 receive a privacy violation, while the path `api::f` would be allowed.
1767 When re-exporting a private item, it can be thought of as allowing the "privacy
1768 chain" being short-circuited through the reexport instead of passing through
1769 the namespace hierarchy as it normally would.
1773 Any item declaration may have an _attribute_ applied to it. Attributes in Rust
1774 are modeled on Attributes in ECMA-335, with the syntax coming from ECMA-334
1775 (C#). An attribute is a general, free-form metadatum that is interpreted
1776 according to name, convention, and language and compiler version. Attributes
1777 may appear as any of:
1779 * A single identifier, the attribute name
1780 * An identifier followed by the equals sign '=' and a literal, providing a
1782 * An identifier followed by a parenthesized list of sub-attribute arguments
1784 Attributes with a bang ("!") after the hash ("#") apply to the item that the
1785 attribute is declared within. Attributes that do not have a bang after the hash
1786 apply to the item that follows the attribute.
1788 An example of attributes:
1791 // General metadata applied to the enclosing module or crate.
1792 #![crate_type = "lib"]
1794 // A function marked as a unit test
1800 // A conditionally-compiled module
1801 #[cfg(target_os="linux")]
1806 // A lint attribute used to suppress a warning/error
1807 #[allow(non_camel_case_types)]
1811 > **Note:** At some point in the future, the compiler will distinguish between
1812 > language-reserved and user-available attributes. Until then, there is
1813 > effectively no difference between an attribute handled by a loadable syntax
1814 > extension and the compiler.
1816 ### Crate-only attributes
1818 - `crate_name` - specify the crate's crate name.
1819 - `crate_type` - see [linkage](#linkage).
1820 - `feature` - see [compiler features](#compiler-features).
1821 - `no_builtins` - disable optimizing certain code patterns to invocations of
1822 library functions that are assumed to exist
1823 - `no_main` - disable emitting the `main` symbol. Useful when some other
1824 object being linked to defines `main`.
1825 - `no_start` - disable linking to the `native` crate, which specifies the
1826 "start" language item.
1827 - `no_std` - disable linking to the `std` crate.
1828 - `plugin` - load a list of named crates as compiler plugins, e.g.
1829 `#![plugin(foo, bar)]`. Optional arguments for each plugin,
1830 i.e. `#![plugin(foo(... args ...))]`, are provided to the plugin's
1831 registrar function. The `plugin` feature gate is required to use
1833 - `recursion_limit` - Sets the maximum depth for potentially
1834 infinitely-recursive compile-time operations like
1835 auto-dereference or macro expansion. The default is
1836 `#![recursion_limit="64"]`.
1838 ### Module-only attributes
1840 - `no_implicit_prelude` - disable injecting `use std::prelude::*` in this
1842 - `path` - specifies the file to load the module from. `#[path="foo.rs"] mod
1843 bar;` is equivalent to `mod bar { /* contents of foo.rs */ }`. The path is
1844 taken relative to the directory that the current module is in.
1846 ### Function-only attributes
1848 - `main` - indicates that this function should be passed to the entry point,
1849 rather than the function in the crate root named `main`.
1850 - `plugin_registrar` - mark this function as the registration point for
1851 [compiler plugins][plugin], such as loadable syntax extensions.
1852 - `start` - indicates that this function should be used as the entry point,
1853 overriding the "start" language item. See the "start" [language
1854 item](#language-items) for more details.
1855 - `test` - indicates that this function is a test function, to only be compiled
1856 in case of `--test`.
1857 - `should_panic` - indicates that this test function should panic, inverting the success condition.
1858 - `cold` - The function is unlikely to be executed, so optimize it (and calls
1861 ### Static-only attributes
1863 - `thread_local` - on a `static mut`, this signals that the value of this
1864 static may change depending on the current thread. The exact consequences of
1865 this are implementation-defined.
1869 On an `extern` block, the following attributes are interpreted:
1871 - `link_args` - specify arguments to the linker, rather than just the library
1872 name and type. This is feature gated and the exact behavior is
1873 implementation-defined (due to variety of linker invocation syntax).
1874 - `link` - indicate that a native library should be linked to for the
1875 declarations in this block to be linked correctly. `link` supports an optional
1876 `kind` key with three possible values: `dylib`, `static`, and `framework`. See
1877 [external blocks](#external-blocks) for more about external blocks. Two
1878 examples: `#[link(name = "readline")]` and
1879 `#[link(name = "CoreFoundation", kind = "framework")]`.
1880 - `linked_from` - indicates what native library this block of FFI items is
1881 coming from. This attribute is of the form `#[linked_from = "foo"]` where
1882 `foo` is the name of a library in either `#[link]` or a `-l` flag. This
1883 attribute is currently required to export symbols from a Rust dynamic library
1884 on Windows, and it is feature gated behind the `linked_from` feature.
1886 On declarations inside an `extern` block, the following attributes are
1889 - `link_name` - the name of the symbol that this function or static should be
1891 - `linkage` - on a static, this specifies the [linkage
1892 type](http://llvm.org/docs/LangRef.html#linkage-types).
1896 - `repr` - on C-like enums, this sets the underlying type used for
1897 representation. Takes one argument, which is the primitive
1898 type this enum should be represented for, or `C`, which specifies that it
1899 should be the default `enum` size of the C ABI for that platform. Note that
1900 enum representation in C is undefined, and this may be incorrect when the C
1901 code is compiled with certain flags.
1905 - `repr` - specifies the representation to use for this struct. Takes a list
1906 of options. The currently accepted ones are `C` and `packed`, which may be
1907 combined. `C` will use a C ABI compatible struct layout, and `packed` will
1908 remove any padding between fields (note that this is very fragile and may
1909 break platforms which require aligned access).
1911 ### Macro-related attributes
1913 - `macro_use` on a `mod` — macros defined in this module will be visible in the
1914 module's parent, after this module has been included.
1916 - `macro_use` on an `extern crate` — load macros from this crate. An optional
1917 list of names `#[macro_use(foo, bar)]` restricts the import to just those
1918 macros named. The `extern crate` must appear at the crate root, not inside
1919 `mod`, which ensures proper function of the [`$crate` macro
1920 variable](book/macros.html#the-variable-crate).
1922 - `macro_reexport` on an `extern crate` — re-export the named macros.
1924 - `macro_export` - export a macro for cross-crate usage.
1926 - `no_link` on an `extern crate` — even if we load this crate for macros, don't
1927 link it into the output.
1929 See the [macros section of the
1930 book](book/macros.html#scoping-and-macro-importexport) for more information on
1934 ### Miscellaneous attributes
1936 - `export_name` - on statics and functions, this determines the name of the
1938 - `link_section` - on statics and functions, this specifies the section of the
1939 object file that this item's contents will be placed into.
1940 - `no_mangle` - on any item, do not apply the standard name mangling. Set the
1941 symbol for this item to its identifier.
1942 - `simd` - on certain tuple structs, derive the arithmetic operators, which
1943 lower to the target's SIMD instructions, if any; the `simd` feature gate
1944 is necessary to use this attribute.
1945 - `unsafe_destructor_blind_to_params` - on `Drop::drop` method, asserts that the
1946 destructor code (and all potential specializations of that code) will
1947 never attempt to read from nor write to any references with lifetimes
1948 that come in via generic parameters. This is a constraint we cannot
1949 currently express via the type system, and therefore we rely on the
1950 programmer to assert that it holds. Adding this to a Drop impl causes
1951 the associated destructor to be considered "uninteresting" by the
1952 Drop-Check rule, and thus it can help sidestep data ordering
1953 constraints that would otherwise be introduced by the Drop-Check
1954 rule. Such sidestepping of the constraints, if done incorrectly, can
1955 lead to undefined behavior (in the form of reading or writing to data
1956 outside of its dynamic extent), and thus this attribute has the word
1957 "unsafe" in its name. To use this, the
1958 `unsafe_destructor_blind_to_params` feature gate must be enabled.
1959 - `unsafe_no_drop_flag` - on structs, remove the flag that prevents
1960 destructors from being run twice. Destructors might be run multiple times on
1961 the same object with this attribute. To use this, the `unsafe_no_drop_flag` feature
1962 gate must be enabled.
1963 - `doc` - Doc comments such as `/// foo` are equivalent to `#[doc = "foo"]`.
1964 - `rustc_on_unimplemented` - Write a custom note to be shown along with the error
1965 when the trait is found to be unimplemented on a type.
1966 You may use format arguments like `{T}`, `{A}` to correspond to the
1967 types at the point of use corresponding to the type parameters of the
1968 trait of the same name. `{Self}` will be replaced with the type that is supposed
1969 to implement the trait but doesn't. To use this, the `on_unimplemented` feature gate
1972 ### Conditional compilation
1974 Sometimes one wants to have different compiler outputs from the same code,
1975 depending on build target, such as targeted operating system, or to enable
1978 There are two kinds of configuration options, one that is either defined or not
1979 (`#[cfg(foo)]`), and the other that contains a string that can be checked
1980 against (`#[cfg(bar = "baz")]`). Currently, only compiler-defined configuration
1981 options can have the latter form.
1984 // The function is only included in the build when compiling for OSX
1985 #[cfg(target_os = "macos")]
1990 // This function is only included when either foo or bar is defined
1991 #[cfg(any(foo, bar))]
1992 fn needs_foo_or_bar() {
1996 // This function is only included when compiling for a unixish OS with a 32-bit
1998 #[cfg(all(unix, target_pointer_width = "32"))]
1999 fn on_32bit_unix() {
2003 // This function is only included when foo is not defined
2005 fn needs_not_foo() {
2010 This illustrates some conditional compilation can be achieved using the
2011 `#[cfg(...)]` attribute. `any`, `all` and `not` can be used to assemble
2012 arbitrarily complex configurations through nesting.
2014 The following configurations must be defined by the implementation:
2016 * `debug_assertions` - Enabled by default when compiling without optimizations.
2017 This can be used to enable extra debugging code in development but not in
2018 production. For example, it controls the behavior of the standard library's
2019 `debug_assert!` macro.
2020 * `target_arch = "..."` - Target CPU architecture, such as `"x86"`, `"x86_64"`
2021 `"mips"`, `"powerpc"`, `"arm"`, or `"aarch64"`.
2022 * `target_endian = "..."` - Endianness of the target CPU, either `"little"` or
2024 * `target_env = ".."` - An option provided by the compiler by default
2025 describing the runtime environment of the target platform. Some examples of
2026 this are `musl` for builds targeting the MUSL libc implementation, `msvc` for
2027 Windows builds targeting MSVC, and `gnu` frequently the rest of the time. This
2028 option may also be blank on some platforms.
2029 * `target_family = "..."` - Operating system family of the target, e. g.
2030 `"unix"` or `"windows"`. The value of this configuration option is defined
2031 as a configuration itself, like `unix` or `windows`.
2032 * `target_os = "..."` - Operating system of the target, examples include
2033 `"windows"`, `"macos"`, `"ios"`, `"linux"`, `"android"`, `"freebsd"`, `"dragonfly"`,
2034 `"bitrig"` , `"openbsd"` or `"netbsd"`.
2035 * `target_pointer_width = "..."` - Target pointer width in bits. This is set
2036 to `"32"` for targets with 32-bit pointers, and likewise set to `"64"` for
2038 * `target_vendor = "..."` - Vendor of the target, for example `apple`, `pc`, or
2040 * `test` - Enabled when compiling the test harness (using the `--test` flag).
2041 * `unix` - See `target_family`.
2042 * `windows` - See `target_family`.
2044 You can also set another attribute based on a `cfg` variable with `cfg_attr`:
2050 Will be the same as `#[b]` if `a` is set by `cfg`, and nothing otherwise.
2052 ### Lint check attributes
2054 A lint check names a potentially undesirable coding pattern, such as
2055 unreachable code or omitted documentation, for the static entity to which the
2058 For any lint check `C`:
2060 * `allow(C)` overrides the check for `C` so that violations will go
2062 * `deny(C)` signals an error after encountering a violation of `C`,
2063 * `forbid(C)` is the same as `deny(C)`, but also forbids changing the lint
2065 * `warn(C)` warns about violations of `C` but continues compilation.
2067 The lint checks supported by the compiler can be found via `rustc -W help`,
2068 along with their default settings. [Compiler
2069 plugins](book/compiler-plugins.html#lint-plugins) can provide additional lint checks.
2073 // Missing documentation is ignored here
2074 #[allow(missing_docs)]
2075 pub fn undocumented_one() -> i32 { 1 }
2077 // Missing documentation signals a warning here
2078 #[warn(missing_docs)]
2079 pub fn undocumented_too() -> i32 { 2 }
2081 // Missing documentation signals an error here
2082 #[deny(missing_docs)]
2083 pub fn undocumented_end() -> i32 { 3 }
2087 This example shows how one can use `allow` and `warn` to toggle a particular
2091 #[warn(missing_docs)]
2093 #[allow(missing_docs)]
2095 // Missing documentation is ignored here
2096 pub fn undocumented_one() -> i32 { 1 }
2098 // Missing documentation signals a warning here,
2099 // despite the allow above.
2100 #[warn(missing_docs)]
2101 pub fn undocumented_two() -> i32 { 2 }
2104 // Missing documentation signals a warning here
2105 pub fn undocumented_too() -> i32 { 3 }
2109 This example shows how one can use `forbid` to disallow uses of `allow` for
2113 #[forbid(missing_docs)]
2115 // Attempting to toggle warning signals an error here
2116 #[allow(missing_docs)]
2118 pub fn undocumented_too() -> i32 { 2 }
2124 Some primitive Rust operations are defined in Rust code, rather than being
2125 implemented directly in C or assembly language. The definitions of these
2126 operations have to be easy for the compiler to find. The `lang` attribute
2127 makes it possible to declare these operations. For example, the `str` module
2128 in the Rust standard library defines the string equality function:
2132 pub fn eq_slice(a: &str, b: &str) -> bool {
2137 The name `str_eq` has a special meaning to the Rust compiler, and the presence
2138 of this definition means that it will use this definition when generating calls
2139 to the string equality function.
2141 The set of language items is currently considered unstable. A complete
2142 list of the built-in language items will be added in the future.
2144 ### Inline attributes
2146 The inline attribute suggests that the compiler should place a copy of
2147 the function or static in the caller, rather than generating code to
2148 call the function or access the static where it is defined.
2150 The compiler automatically inlines functions based on internal heuristics.
2151 Incorrectly inlining functions can actually make the program slower, so it
2152 should be used with care.
2154 `#[inline]` and `#[inline(always)]` always cause the function to be serialized
2155 into the crate metadata to allow cross-crate inlining.
2157 There are three different types of inline attributes:
2159 * `#[inline]` hints the compiler to perform an inline expansion.
2160 * `#[inline(always)]` asks the compiler to always perform an inline expansion.
2161 * `#[inline(never)]` asks the compiler to never perform an inline expansion.
2165 The `derive` attribute allows certain traits to be automatically implemented
2166 for data structures. For example, the following will create an `impl` for the
2167 `PartialEq` and `Clone` traits for `Foo`, the type parameter `T` will be given
2168 the `PartialEq` or `Clone` constraints for the appropriate `impl`:
2171 #[derive(PartialEq, Clone)]
2178 The generated `impl` for `PartialEq` is equivalent to
2181 # struct Foo<T> { a: i32, b: T }
2182 impl<T: PartialEq> PartialEq for Foo<T> {
2183 fn eq(&self, other: &Foo<T>) -> bool {
2184 self.a == other.a && self.b == other.b
2187 fn ne(&self, other: &Foo<T>) -> bool {
2188 self.a != other.a || self.b != other.b
2193 ### Compiler Features
2195 Certain aspects of Rust may be implemented in the compiler, but they're not
2196 necessarily ready for every-day use. These features are often of "prototype
2197 quality" or "almost production ready", but may not be stable enough to be
2198 considered a full-fledged language feature.
2200 For this reason, Rust recognizes a special crate-level attribute of the form:
2203 #![feature(feature1, feature2, feature3)]
2206 This directive informs the compiler that the feature list: `feature1`,
2207 `feature2`, and `feature3` should all be enabled. This is only recognized at a
2208 crate-level, not at a module-level. Without this directive, all features are
2209 considered off, and using the features will result in a compiler error.
2211 The currently implemented features of the reference compiler are:
2213 * `advanced_slice_patterns` - See the [match expressions](#match-expressions)
2214 section for discussion; the exact semantics of
2215 slice patterns are subject to change, so some types
2218 * `slice_patterns` - OK, actually, slice patterns are just scary and
2219 completely unstable.
2221 * `asm` - The `asm!` macro provides a means for inline assembly. This is often
2222 useful, but the exact syntax for this feature along with its
2223 semantics are likely to change, so this macro usage must be opted
2226 * `associated_consts` - Allows constants to be defined in `impl` and `trait`
2227 blocks, so that they can be associated with a type or
2228 trait in a similar manner to methods and associated
2231 * `box_patterns` - Allows `box` patterns, the exact semantics of which
2232 is subject to change.
2234 * `box_syntax` - Allows use of `box` expressions, the exact semantics of which
2235 is subject to change.
2237 * `cfg_target_vendor` - Allows conditional compilation using the `target_vendor`
2238 matcher which is subject to change.
2240 * `concat_idents` - Allows use of the `concat_idents` macro, which is in many
2241 ways insufficient for concatenating identifiers, and may be
2242 removed entirely for something more wholesome.
2244 * `custom_attribute` - Allows the usage of attributes unknown to the compiler
2245 so that new attributes can be added in a backwards compatible
2248 * `custom_derive` - Allows the use of `#[derive(Foo,Bar)]` as sugar for
2249 `#[derive_Foo] #[derive_Bar]`, which can be user-defined syntax
2252 * `intrinsics` - Allows use of the "rust-intrinsics" ABI. Compiler intrinsics
2253 are inherently unstable and no promise about them is made.
2255 * `lang_items` - Allows use of the `#[lang]` attribute. Like `intrinsics`,
2256 lang items are inherently unstable and no promise about them
2259 * `link_args` - This attribute is used to specify custom flags to the linker,
2260 but usage is strongly discouraged. The compiler's usage of the
2261 system linker is not guaranteed to continue in the future, and
2262 if the system linker is not used then specifying custom flags
2263 doesn't have much meaning.
2265 * `link_llvm_intrinsics` – Allows linking to LLVM intrinsics via
2266 `#[link_name="llvm.*"]`.
2268 * `linkage` - Allows use of the `linkage` attribute, which is not portable.
2270 * `log_syntax` - Allows use of the `log_syntax` macro attribute, which is a
2271 nasty hack that will certainly be removed.
2273 * `main` - Allows use of the `#[main]` attribute, which changes the entry point
2274 into a Rust program. This capability is subject to change.
2276 * `macro_reexport` - Allows macros to be re-exported from one crate after being imported
2277 from another. This feature was originally designed with the sole
2278 use case of the Rust standard library in mind, and is subject to
2281 * `non_ascii_idents` - The compiler supports the use of non-ascii identifiers,
2282 but the implementation is a little rough around the
2283 edges, so this can be seen as an experimental feature
2284 for now until the specification of identifiers is fully
2287 * `no_std` - Allows the `#![no_std]` crate attribute, which disables the implicit
2288 `extern crate std`. This typically requires use of the unstable APIs
2289 behind the libstd "facade", such as libcore and libcollections. It
2290 may also cause problems when using syntax extensions, including
2293 * `on_unimplemented` - Allows the `#[rustc_on_unimplemented]` attribute, which allows
2294 trait definitions to add specialized notes to error messages
2295 when an implementation was expected but not found.
2297 * `optin_builtin_traits` - Allows the definition of default and negative trait
2298 implementations. Experimental.
2300 * `plugin` - Usage of [compiler plugins][plugin] for custom lints or syntax extensions.
2301 These depend on compiler internals and are subject to change.
2303 * `plugin_registrar` - Indicates that a crate provides [compiler plugins][plugin].
2305 * `quote` - Allows use of the `quote_*!` family of macros, which are
2306 implemented very poorly and will likely change significantly
2307 with a proper implementation.
2309 * `rustc_attrs` - Gates internal `#[rustc_*]` attributes which may be
2310 for internal use only or have meaning added to them in the future.
2312 * `rustc_diagnostic_macros`- A mysterious feature, used in the implementation
2313 of rustc, not meant for mortals.
2315 * `simd` - Allows use of the `#[simd]` attribute, which is overly simple and
2316 not the SIMD interface we want to expose in the long term.
2318 * `simd_ffi` - Allows use of SIMD vectors in signatures for foreign functions.
2319 The SIMD interface is subject to change.
2321 * `staged_api` - Allows usage of stability markers and `#![staged_api]` in a
2322 crate. Stability markers are also attributes: `#[stable]`,
2323 `#[unstable]`, and `#[deprecated]` are the three levels.
2325 * `start` - Allows use of the `#[start]` attribute, which changes the entry point
2326 into a Rust program. This capability, especially the signature for the
2327 annotated function, is subject to change.
2329 * `thread_local` - The usage of the `#[thread_local]` attribute is experimental
2330 and should be seen as unstable. This attribute is used to
2331 declare a `static` as being unique per-thread leveraging
2332 LLVM's implementation which works in concert with the kernel
2333 loader and dynamic linker. This is not necessarily available
2334 on all platforms, and usage of it is discouraged.
2336 * `trace_macros` - Allows use of the `trace_macros` macro, which is a nasty
2337 hack that will certainly be removed.
2339 * `unboxed_closures` - Rust's new closure design, which is currently a work in
2340 progress feature with many known bugs.
2342 * `unsafe_no_drop_flag` - Allows use of the `#[unsafe_no_drop_flag]` attribute,
2343 which removes hidden flag added to a type that
2344 implements the `Drop` trait. The design for the
2345 `Drop` flag is subject to change, and this feature
2346 may be removed in the future.
2348 * `unmarked_api` - Allows use of items within a `#![staged_api]` crate
2349 which have not been marked with a stability marker.
2350 Such items should not be allowed by the compiler to exist,
2351 so if you need this there probably is a compiler bug.
2353 * `visible_private_types` - Allows public APIs to expose otherwise private
2354 types, e.g. as the return type of a public function.
2355 This capability may be removed in the future.
2357 * `allow_internal_unstable` - Allows `macro_rules!` macros to be tagged with the
2358 `#[allow_internal_unstable]` attribute, designed
2359 to allow `std` macros to call
2360 `#[unstable]`/feature-gated functionality
2361 internally without imposing on callers
2362 (i.e. making them behave like function calls in
2363 terms of encapsulation).
2364 * - `default_type_parameter_fallback` - Allows type parameter defaults to
2365 influence type inference.
2366 * - `braced_empty_structs` - Allows use of empty structs and enum variants with braces.
2368 If a feature is promoted to a language feature, then all existing programs will
2369 start to receive compilation warnings about `#![feature]` directives which enabled
2370 the new feature (because the directive is no longer necessary). However, if a
2371 feature is decided to be removed from the language, errors will be issued (if
2372 there isn't a parser error first). The directive in this case is no longer
2373 necessary, and it's likely that existing code will break if the feature isn't
2376 If an unknown feature is found in a directive, it results in a compiler error.
2377 An unknown feature is one which has never been recognized by the compiler.
2379 # Statements and expressions
2381 Rust is _primarily_ an expression language. This means that most forms of
2382 value-producing or effect-causing evaluation are directed by the uniform syntax
2383 category of _expressions_. Each kind of expression can typically _nest_ within
2384 each other kind of expression, and rules for evaluation of expressions involve
2385 specifying both the value produced by the expression and the order in which its
2386 sub-expressions are themselves evaluated.
2388 In contrast, statements in Rust serve _mostly_ to contain and explicitly
2389 sequence expression evaluation.
2393 A _statement_ is a component of a block, which is in turn a component of an
2394 outer [expression](#expressions) or [function](#functions).
2396 Rust has two kinds of statement: [declaration
2397 statements](#declaration-statements) and [expression
2398 statements](#expression-statements).
2400 ### Declaration statements
2402 A _declaration statement_ is one that introduces one or more *names* into the
2403 enclosing statement block. The declared names may denote new variables or new
2406 #### Item declarations
2408 An _item declaration statement_ has a syntactic form identical to an
2409 [item](#items) declaration within a module. Declaring an item — a
2410 function, enumeration, struct, type, static, trait, implementation or module
2411 — locally within a statement block is simply a way of restricting its
2412 scope to a narrow region containing all of its uses; it is otherwise identical
2413 in meaning to declaring the item outside the statement block.
2415 > **Note**: there is no implicit capture of the function's dynamic environment when
2416 > declaring a function-local item.
2418 #### Variable declarations
2420 A _variable declaration_ introduces a new set of variable, given by a pattern. The
2421 pattern may be followed by a type annotation, and/or an initializer expression.
2422 When no type annotation is given, the compiler will infer the type, or signal
2423 an error if insufficient type information is available for definite inference.
2424 Any variables introduced by a variable declaration are visible from the point of
2425 declaration until the end of the enclosing block scope.
2427 ### Expression statements
2429 An _expression statement_ is one that evaluates an [expression](#expressions)
2430 and ignores its result. The type of an expression statement `e;` is always
2431 `()`, regardless of the type of `e`. As a rule, an expression statement's
2432 purpose is to trigger the effects of evaluating its expression.
2436 An expression may have two roles: it always produces a *value*, and it may have
2437 *effects* (otherwise known as "side effects"). An expression *evaluates to* a
2438 value, and has effects during *evaluation*. Many expressions contain
2439 sub-expressions (operands). The meaning of each kind of expression dictates
2442 * Whether or not to evaluate the sub-expressions when evaluating the expression
2443 * The order in which to evaluate the sub-expressions
2444 * How to combine the sub-expressions' values to obtain the value of the expression
2446 In this way, the structure of expressions dictates the structure of execution.
2447 Blocks are just another kind of expression, so blocks, statements, expressions,
2448 and blocks again can recursively nest inside each other to an arbitrary depth.
2450 #### Lvalues, rvalues and temporaries
2452 Expressions are divided into two main categories: _lvalues_ and _rvalues_.
2453 Likewise within each expression, sub-expressions may occur in _lvalue context_
2454 or _rvalue context_. The evaluation of an expression depends both on its own
2455 category and the context it occurs within.
2457 An lvalue is an expression that represents a memory location. These expressions
2458 are [paths](#path-expressions) (which refer to local variables, function and
2459 method arguments, or static variables), dereferences (`*expr`), [indexing
2460 expressions](#index-expressions) (`expr[expr]`), and [field
2461 references](#field-expressions) (`expr.f`). All other expressions are rvalues.
2463 The left operand of an [assignment](#assignment-expressions) or
2464 [compound-assignment](#compound-assignment-expressions) expression is
2465 an lvalue context, as is the single operand of a unary
2466 [borrow](#unary-operator-expressions). The discriminant or subject of
2467 a [match expression](#match-expressions) may be an lvalue context, if
2468 ref bindings are made, but is otherwise an rvalue context. All other
2469 expression contexts are rvalue contexts.
2471 When an lvalue is evaluated in an _lvalue context_, it denotes a memory
2472 location; when evaluated in an _rvalue context_, it denotes the value held _in_
2473 that memory location.
2475 ##### Temporary lifetimes
2477 When an rvalue is used in an lvalue context, a temporary un-named
2478 lvalue is created and used instead. The lifetime of temporary values
2479 is typically the innermost enclosing statement; the tail expression of
2480 a block is considered part of the statement that encloses the block.
2482 When a temporary rvalue is being created that is assigned into a `let`
2483 declaration, however, the temporary is created with the lifetime of
2484 the enclosing block instead, as using the enclosing statement (the
2485 `let` declaration) would be a guaranteed error (since a pointer to the
2486 temporary would be stored into a variable, but the temporary would be
2487 freed before the variable could be used). The compiler uses simple
2488 syntactic rules to decide which values are being assigned into a `let`
2489 binding, and therefore deserve a longer temporary lifetime.
2491 Here are some examples:
2493 - `let x = foo(&temp())`. The expression `temp()` is an rvalue. As it
2494 is being borrowed, a temporary is created which will be freed after
2495 the innermost enclosing statement (the `let` declaration, in this case).
2496 - `let x = temp().foo()`. This is the same as the previous example,
2497 except that the value of `temp()` is being borrowed via autoref on a
2498 method-call. Here we are assuming that `foo()` is an `&self` method
2499 defined in some trait, say `Foo`. In other words, the expression
2500 `temp().foo()` is equivalent to `Foo::foo(&temp())`.
2501 - `let x = &temp()`. Here, the same temporary is being assigned into
2502 `x`, rather than being passed as a parameter, and hence the
2503 temporary's lifetime is considered to be the enclosing block.
2504 - `let x = SomeStruct { foo: &temp() }`. As in the previous case, the
2505 temporary is assigned into a struct which is then assigned into a
2506 binding, and hence it is given the lifetime of the enclosing block.
2507 - `let x = [ &temp() ]`. As in the previous case, the
2508 temporary is assigned into an array which is then assigned into a
2509 binding, and hence it is given the lifetime of the enclosing block.
2510 - `let ref x = temp()`. In this case, the temporary is created using a ref binding,
2511 but the result is the same: the lifetime is extended to the enclosing block.
2513 #### Moved and copied types
2515 When a [local variable](#variables) is used as an
2516 [rvalue](#lvalues-rvalues-and-temporaries), the variable will be copied
2517 if its type implements `Copy`. All others are moved.
2519 ### Literal expressions
2521 A _literal expression_ consists of one of the [literal](#literals) forms
2522 described earlier. It directly describes a number, character, string, boolean
2523 value, or the unit value.
2527 "hello"; // string type
2528 '5'; // character type
2532 ### Path expressions
2534 A [path](#paths) used as an expression context denotes either a local variable
2535 or an item. Path expressions are [lvalues](#lvalues-rvalues-and-temporaries).
2537 ### Tuple expressions
2539 Tuples are written by enclosing zero or more comma-separated expressions in
2540 parentheses. They are used to create [tuple-typed](#tuple-types) values.
2544 ("a", 4usize, true);
2547 You can disambiguate a single-element tuple from a value in parentheses with a
2551 (0,); // single-element tuple
2552 (0); // zero in parentheses
2555 ### Struct expressions
2557 There are several forms of struct expressions. A _struct expression_
2558 consists of the [path](#paths) of a [struct item](#structs), followed by
2559 a brace-enclosed list of one or more comma-separated name-value pairs,
2560 providing the field values of a new instance of the struct. A field name
2561 can be any identifier, and is separated from its value expression by a colon.
2562 The location denoted by a struct field is mutable if and only if the
2563 enclosing struct is mutable.
2565 A _tuple struct expression_ consists of the [path](#paths) of a [struct
2566 item](#structs), followed by a parenthesized list of one or more
2567 comma-separated expressions (in other words, the path of a struct item
2568 followed by a tuple expression). The struct item must be a tuple struct
2571 A _unit-like struct expression_ consists only of the [path](#paths) of a
2572 [struct item](#structs).
2574 The following are examples of struct expressions:
2577 # struct Point { x: f64, y: f64 }
2578 # struct TuplePoint(f64, f64);
2579 # mod game { pub struct User<'a> { pub name: &'a str, pub age: u32, pub score: usize } }
2580 # struct Cookie; fn some_fn<T>(t: T) {}
2581 Point {x: 10.0, y: 20.0};
2582 TuplePoint(10.0, 20.0);
2583 let u = game::User {name: "Joe", age: 35, score: 100_000};
2584 some_fn::<Cookie>(Cookie);
2587 A struct expression forms a new value of the named struct type. Note
2588 that for a given *unit-like* struct type, this will always be the same
2591 A struct expression can terminate with the syntax `..` followed by an
2592 expression to denote a functional update. The expression following `..` (the
2593 base) must have the same struct type as the new struct type being formed.
2594 The entire expression denotes the result of constructing a new struct (with
2595 the same type as the base expression) with the given values for the fields that
2596 were explicitly specified and the values in the base expression for all other
2600 # struct Point3d { x: i32, y: i32, z: i32 }
2601 let base = Point3d {x: 1, y: 2, z: 3};
2602 Point3d {y: 0, z: 10, .. base};
2605 ### Block expressions
2607 A _block expression_ is similar to a module in terms of the declarations that
2608 are possible. Each block conceptually introduces a new namespace scope. Use
2609 items can bring new names into scopes and declared items are in scope for only
2612 A block will execute each statement sequentially, and then execute the
2613 expression (if given). If the block ends in a statement, its value is `()`:
2616 let x: () = { println!("Hello."); };
2619 If it ends in an expression, its value and type are that of the expression:
2622 let x: i32 = { println!("Hello."); 5 };
2627 ### Method-call expressions
2629 A _method call_ consists of an expression followed by a single dot, an
2630 identifier, and a parenthesized expression-list. Method calls are resolved to
2631 methods on specific traits, either statically dispatching to a method if the
2632 exact `self`-type of the left-hand-side is known, or dynamically dispatching if
2633 the left-hand-side expression is an indirect [trait object](#trait-objects).
2635 ### Field expressions
2637 A _field expression_ consists of an expression followed by a single dot and an
2638 identifier, when not immediately followed by a parenthesized expression-list
2639 (the latter is a [method call expression](#method-call-expressions)). A field
2640 expression denotes a field of a [struct](#struct-types).
2645 (Struct {a: 10, b: 20}).a;
2648 A field access is an [lvalue](#lvalues-rvalues-and-temporaries) referring to
2649 the value of that field. When the type providing the field inherits mutability,
2650 it can be [assigned](#assignment-expressions) to.
2652 Also, if the type of the expression to the left of the dot is a
2653 pointer, it is automatically dereferenced as many times as necessary
2654 to make the field access possible. In cases of ambiguity, we prefer
2655 fewer autoderefs to more.
2657 ### Array expressions
2659 An [array](#array-and-slice-types) _expression_ is written by enclosing zero
2660 or more comma-separated expressions of uniform type in square brackets.
2662 In the `[expr ';' expr]` form, the expression after the `';'` must be a
2663 constant expression that can be evaluated at compile time, such as a
2664 [literal](#literals) or a [static item](#static-items).
2668 ["a", "b", "c", "d"];
2669 [0; 128]; // array with 128 zeros
2670 [0u8, 0u8, 0u8, 0u8];
2673 ### Index expressions
2675 [Array](#array-and-slice-types)-typed expressions can be indexed by
2676 writing a square-bracket-enclosed expression (the index) after them. When the
2677 array is mutable, the resulting [lvalue](#lvalues-rvalues-and-temporaries) can
2680 Indices are zero-based, and may be of any integral type. Vector access is
2681 bounds-checked at compile-time for constant arrays being accessed with a constant index value.
2682 Otherwise a check will be performed at run-time that will put the thread in a _panicked state_ if it fails.
2687 let x = (["a", "b"])[10]; // compiler error: const index-expr is out of bounds
2690 let y = (["a", "b"])[n]; // panics
2692 let arr = ["a", "b"];
2696 Also, if the type of the expression to the left of the brackets is a
2697 pointer, it is automatically dereferenced as many times as necessary
2698 to make the indexing possible. In cases of ambiguity, we prefer fewer
2701 ### Range expressions
2703 The `..` operator will construct an object of one of the `std::ops::Range` variants.
2706 1..2; // std::ops::Range
2707 3..; // std::ops::RangeFrom
2708 ..4; // std::ops::RangeTo
2709 ..; // std::ops::RangeFull
2712 The following expressions are equivalent.
2715 let x = std::ops::Range {start: 0, end: 10};
2721 ### Unary operator expressions
2723 Rust defines the following unary operators. They are all written as prefix operators,
2724 before the expression they apply to.
2727 : Negation. May only be applied to numeric types.
2729 : Dereference. When applied to a [pointer](#pointer-types) it denotes the
2730 pointed-to location. For pointers to mutable locations, the resulting
2731 [lvalue](#lvalues-rvalues-and-temporaries) can be assigned to.
2732 On non-pointer types, it calls the `deref` method of the `std::ops::Deref`
2733 trait, or the `deref_mut` method of the `std::ops::DerefMut` trait (if
2734 implemented by the type and required for an outer expression that will or
2735 could mutate the dereference), and produces the result of dereferencing the
2736 `&` or `&mut` borrowed pointer returned from the overload method.
2738 : Logical negation. On the boolean type, this flips between `true` and
2739 `false`. On integer types, this inverts the individual bits in the
2740 two's complement representation of the value.
2742 : Borrowing. When applied to an lvalue, these operators produce a
2743 reference (pointer) to the lvalue. The lvalue is also placed into
2744 a borrowed state for the duration of the reference. For a shared
2745 borrow (`&`), this implies that the lvalue may not be mutated, but
2746 it may be read or shared again. For a mutable borrow (`&mut`), the
2747 lvalue may not be accessed in any way until the borrow expires.
2748 If the `&` or `&mut` operators are applied to an rvalue, a
2749 temporary value is created; the lifetime of this temporary value
2750 is defined by [syntactic rules](#temporary-lifetimes).
2752 ### Binary operator expressions
2754 Binary operators expressions are given in terms of [operator
2755 precedence](#operator-precedence).
2757 #### Arithmetic operators
2759 Binary arithmetic expressions are syntactic sugar for calls to built-in traits,
2760 defined in the `std::ops` module of the `std` library. This means that
2761 arithmetic operators can be overridden for user-defined types. The default
2762 meaning of the operators on standard types is given here.
2765 : Addition and array/string concatenation.
2766 Calls the `add` method on the `std::ops::Add` trait.
2769 Calls the `sub` method on the `std::ops::Sub` trait.
2772 Calls the `mul` method on the `std::ops::Mul` trait.
2775 Calls the `div` method on the `std::ops::Div` trait.
2778 Calls the `rem` method on the `std::ops::Rem` trait.
2780 #### Bitwise operators
2782 Like the [arithmetic operators](#arithmetic-operators), bitwise operators are
2783 syntactic sugar for calls to methods of built-in traits. This means that
2784 bitwise operators can be overridden for user-defined types. The default
2785 meaning of the operators on standard types is given here. Bitwise `&`, `|` and
2786 `^` applied to boolean arguments are equivalent to logical `&&`, `||` and `!=`
2787 evaluated in non-lazy fashion.
2791 Calls the `bitand` method of the `std::ops::BitAnd` trait.
2793 : Bitwise inclusive OR.
2794 Calls the `bitor` method of the `std::ops::BitOr` trait.
2796 : Bitwise exclusive OR.
2797 Calls the `bitxor` method of the `std::ops::BitXor` trait.
2800 Calls the `shl` method of the `std::ops::Shl` trait.
2802 : Right shift (arithmetic).
2803 Calls the `shr` method of the `std::ops::Shr` trait.
2805 #### Lazy boolean operators
2807 The operators `||` and `&&` may be applied to operands of boolean type. The
2808 `||` operator denotes logical 'or', and the `&&` operator denotes logical
2809 'and'. They differ from `|` and `&` in that the right-hand operand is only
2810 evaluated when the left-hand operand does not already determine the result of
2811 the expression. That is, `||` only evaluates its right-hand operand when the
2812 left-hand operand evaluates to `false`, and `&&` only when it evaluates to
2815 #### Comparison operators
2817 Comparison operators are, like the [arithmetic
2818 operators](#arithmetic-operators), and [bitwise operators](#bitwise-operators),
2819 syntactic sugar for calls to built-in traits. This means that comparison
2820 operators can be overridden for user-defined types. The default meaning of the
2821 operators on standard types is given here.
2825 Calls the `eq` method on the `std::cmp::PartialEq` trait.
2828 Calls the `ne` method on the `std::cmp::PartialEq` trait.
2831 Calls the `lt` method on the `std::cmp::PartialOrd` trait.
2834 Calls the `gt` method on the `std::cmp::PartialOrd` trait.
2836 : Less than or equal.
2837 Calls the `le` method on the `std::cmp::PartialOrd` trait.
2839 : Greater than or equal.
2840 Calls the `ge` method on the `std::cmp::PartialOrd` trait.
2842 #### Type cast expressions
2844 A type cast expression is denoted with the binary operator `as`.
2846 Executing an `as` expression casts the value on the left-hand side to the type
2847 on the right-hand side.
2849 An example of an `as` expression:
2852 # fn sum(values: &[f64]) -> f64 { 0.0 }
2853 # fn len(values: &[f64]) -> i32 { 0 }
2855 fn average(values: &[f64]) -> f64 {
2856 let sum: f64 = sum(values);
2857 let size: f64 = len(values) as f64;
2862 Some of the conversions which can be done through the `as` operator
2863 can also be done implicitly at various points in the program, such as
2864 argument passing and assignment to a `let` binding with an explicit
2865 type. Implicit conversions are limited to "harmless" conversions that
2866 do not lose information and which have minimal or no risk of
2867 surprising side-effects on the dynamic execution semantics.
2869 #### Assignment expressions
2871 An _assignment expression_ consists of an
2872 [lvalue](#lvalues-rvalues-and-temporaries) expression followed by an equals
2873 sign (`=`) and an [rvalue](#lvalues-rvalues-and-temporaries) expression.
2875 Evaluating an assignment expression [either copies or
2876 moves](#moved-and-copied-types) its right-hand operand to its left-hand
2885 #### Compound assignment expressions
2887 The `+`, `-`, `*`, `/`, `%`, `&`, `|`, `^`, `<<`, and `>>` operators may be
2888 composed with the `=` operator. The expression `lval OP= val` is equivalent to
2889 `lval = lval OP val`. For example, `x = x + 1` may be written as `x += 1`.
2891 Any such expression always has the [`unit`](#tuple-types) type.
2893 #### Operator precedence
2895 The precedence of Rust binary operators is ordered as follows, going from
2898 ```{.text .precedence}
2912 Operators at the same precedence level are evaluated left-to-right. [Unary
2913 operators](#unary-operator-expressions) have the same precedence level and are
2914 stronger than any of the binary operators.
2916 ### Grouped expressions
2918 An expression enclosed in parentheses evaluates to the result of the enclosed
2919 expression. Parentheses can be used to explicitly specify evaluation order
2920 within an expression.
2922 An example of a parenthesized expression:
2925 let x: i32 = (2 + 3) * 4;
2929 ### Call expressions
2931 A _call expression_ invokes a function, providing zero or more input variables
2932 and an optional location to move the function's output into. If the function
2933 eventually returns, then the expression completes.
2935 Some examples of call expressions:
2938 # fn add(x: i32, y: i32) -> i32 { 0 }
2940 let x: i32 = add(1i32, 2i32);
2941 let pi: Result<f32, _> = "3.14".parse();
2944 ### Lambda expressions
2946 A _lambda expression_ (sometimes called an "anonymous function expression")
2947 defines a function and denotes it as a value, in a single expression. A lambda
2948 expression is a pipe-symbol-delimited (`|`) list of identifiers followed by an
2951 A lambda expression denotes a function that maps a list of parameters
2952 (`ident_list`) onto the expression that follows the `ident_list`. The
2953 identifiers in the `ident_list` are the parameters to the function. These
2954 parameters' types need not be specified, as the compiler infers them from
2957 Lambda expressions are most useful when passing functions as arguments to other
2958 functions, as an abbreviation for defining and capturing a separate function.
2960 Significantly, lambda expressions _capture their environment_, which regular
2961 [function definitions](#functions) do not. The exact type of capture depends
2962 on the [function type](#function-types) inferred for the lambda expression. In
2963 the simplest and least-expensive form (analogous to a ```|| { }``` expression),
2964 the lambda expression captures its environment by reference, effectively
2965 borrowing pointers to all outer variables mentioned inside the function.
2966 Alternately, the compiler may infer that a lambda expression should copy or
2967 move values (depending on their type) from the environment into the lambda
2968 expression's captured environment.
2970 In this example, we define a function `ten_times` that takes a higher-order
2971 function argument, and we then call it with a lambda expression as an argument:
2974 fn ten_times<F>(f: F) where F: Fn(i32) {
2975 for index in 0..10 {
2980 ten_times(|j| println!("hello, {}", j));
2985 A `loop` expression denotes an infinite loop.
2987 A `loop` expression may optionally have a _label_. The label is written as
2988 a lifetime preceding the loop expression, as in `'foo: loop{ }`. If a
2989 label is present, then labeled `break` and `continue` expressions nested
2990 within this loop may exit out of this loop or return control to its head.
2991 See [break expressions](#break-expressions) and [continue
2992 expressions](#continue-expressions).
2994 ### `break` expressions
2996 A `break` expression has an optional _label_. If the label is absent, then
2997 executing a `break` expression immediately terminates the innermost loop
2998 enclosing it. It is only permitted in the body of a loop. If the label is
2999 present, then `break 'foo` terminates the loop with label `'foo`, which need not
3000 be the innermost label enclosing the `break` expression, but must enclose it.
3002 ### `continue` expressions
3004 A `continue` expression has an optional _label_. If the label is absent, then
3005 executing a `continue` expression immediately terminates the current iteration
3006 of the innermost loop enclosing it, returning control to the loop *head*. In
3007 the case of a `while` loop, the head is the conditional expression controlling
3008 the loop. In the case of a `for` loop, the head is the call-expression
3009 controlling the loop. If the label is present, then `continue 'foo` returns
3010 control to the head of the loop with label `'foo`, which need not be the
3011 innermost label enclosing the `break` expression, but must enclose it.
3013 A `continue` expression is only permitted in the body of a loop.
3017 A `while` loop begins by evaluating the boolean loop conditional expression.
3018 If the loop conditional expression evaluates to `true`, the loop body block
3019 executes and control returns to the loop conditional expression. If the loop
3020 conditional expression evaluates to `false`, the `while` expression completes.
3033 Like `loop` expressions, `while` loops can be controlled with `break` or
3034 `continue`, and may optionally have a _label_. See [infinite
3035 loops](#infinite-loops), [break expressions](#break-expressions), and
3036 [continue expressions](#continue-expressions) for more information.
3038 ### `for` expressions
3040 A `for` expression is a syntactic construct for looping over elements provided
3041 by an implementation of `std::iter::IntoIterator`.
3043 An example of a `for` loop over the contents of an array:
3047 # fn bar(f: &Foo) { }
3052 let v: &[Foo] = &[a, b, c];
3059 An example of a for loop over a series of integers:
3062 # fn bar(b:usize) { }
3068 Like `loop` expressions, `for` loops can be controlled with `break` or
3069 `continue`, and may optionally have a _label_. See [infinite
3070 loops](#infinite-loops), [break expressions](#break-expressions), and
3071 [continue expressions](#continue-expressions) for more information.
3073 ### `if` expressions
3075 An `if` expression is a conditional branch in program control. The form of an
3076 `if` expression is a condition expression, followed by a consequent block, any
3077 number of `else if` conditions and blocks, and an optional trailing `else`
3078 block. The condition expressions must have type `bool`. If a condition
3079 expression evaluates to `true`, the consequent block is executed and any
3080 subsequent `else if` or `else` block is skipped. If a condition expression
3081 evaluates to `false`, the consequent block is skipped and any subsequent `else
3082 if` condition is evaluated. If all `if` and `else if` conditions evaluate to
3083 `false` then any `else` block is executed.
3085 ### `match` expressions
3087 A `match` expression branches on a *pattern*. The exact form of matching that
3088 occurs depends on the pattern. Patterns consist of some combination of
3089 literals, destructured arrays or enum constructors, structs and tuples,
3090 variable binding specifications, wildcards (`..`), and placeholders (`_`). A
3091 `match` expression has a *head expression*, which is the value to compare to
3092 the patterns. The type of the patterns must equal the type of the head
3095 In a pattern whose head expression has an `enum` type, a placeholder (`_`)
3096 stands for a *single* data field, whereas a wildcard `..` stands for *all* the
3097 fields of a particular variant.
3099 A `match` behaves differently depending on whether or not the head expression
3100 is an [lvalue or an rvalue](#lvalues-rvalues-and-temporaries). If the head
3101 expression is an rvalue, it is first evaluated into a temporary location, and
3102 the resulting value is sequentially compared to the patterns in the arms until
3103 a match is found. The first arm with a matching pattern is chosen as the branch
3104 target of the `match`, any variables bound by the pattern are assigned to local
3105 variables in the arm's block, and control enters the block.
3107 When the head expression is an lvalue, the match does not allocate a temporary
3108 location (however, a by-value binding may copy or move from the lvalue). When
3109 possible, it is preferable to match on lvalues, as the lifetime of these
3110 matches inherits the lifetime of the lvalue, rather than being restricted to
3111 the inside of the match.
3113 An example of a `match` expression:
3119 1 => println!("one"),
3120 2 => println!("two"),
3121 3 => println!("three"),
3122 4 => println!("four"),
3123 5 => println!("five"),
3124 _ => println!("something else"),
3128 Patterns that bind variables default to binding to a copy or move of the
3129 matched value (depending on the matched value's type). This can be changed to
3130 bind to a reference by using the `ref` keyword, or to a mutable reference using
3133 Subpatterns can also be bound to variables by the use of the syntax `variable @
3134 subpattern`. For example:
3140 e @ 1 ... 5 => println!("got a range element {}", e),
3141 _ => println!("anything"),
3145 Patterns can also dereference pointers by using the `&`, `&mut` and `box`
3146 symbols, as appropriate. For example, these two matches on `x: &i32` are
3151 let y = match *x { 0 => "zero", _ => "some" };
3152 let z = match x { &0 => "zero", _ => "some" };
3157 Multiple match patterns may be joined with the `|` operator. A range of values
3158 may be specified with `...`. For example:
3163 let message = match x {
3164 0 | 1 => "not many",
3170 Range patterns only work on scalar types (like integers and characters; not
3171 like arrays and structs, which have sub-components). A range pattern may not
3172 be a sub-range of another range pattern inside the same `match`.
3174 Finally, match patterns can accept *pattern guards* to further refine the
3175 criteria for matching a case. Pattern guards appear after the pattern and
3176 consist of a bool-typed expression following the `if` keyword. A pattern guard
3177 may refer to the variables bound within the pattern they follow.
3180 # let maybe_digit = Some(0);
3181 # fn process_digit(i: i32) { }
3182 # fn process_other(i: i32) { }
3184 let message = match maybe_digit {
3185 Some(x) if x < 10 => process_digit(x),
3186 Some(x) => process_other(x),
3191 ### `if let` expressions
3193 An `if let` expression is semantically identical to an `if` expression but in place
3194 of a condition expression it expects a refutable let statement. If the value of the
3195 expression on the right hand side of the let statement matches the pattern, the corresponding
3196 block will execute, otherwise flow proceeds to the first `else` block that follows.
3199 let dish = ("Ham", "Eggs");
3201 // this body will be skipped because the pattern is refuted
3202 if let ("Bacon", b) = dish {
3203 println!("Bacon is served with {}", b);
3206 // this body will execute
3207 if let ("Ham", b) = dish {
3208 println!("Ham is served with {}", b);
3212 ### `while let` loops
3214 A `while let` loop is semantically identical to a `while` loop but in place of a
3215 condition expression it expects a refutable let statement. If the value of the
3216 expression on the right hand side of the let statement matches the pattern, the
3217 loop body block executes and control returns to the pattern matching statement.
3218 Otherwise, the while expression completes.
3220 ### `return` expressions
3222 Return expressions are denoted with the keyword `return`. Evaluating a `return`
3223 expression moves its argument into the designated output location for the
3224 current function call, destroys the current function activation frame, and
3225 transfers control to the caller frame.
3227 An example of a `return` expression:
3230 fn max(a: i32, b: i32) -> i32 {
3242 Every variable, item and value in a Rust program has a type. The _type_ of a
3243 *value* defines the interpretation of the memory holding it.
3245 Built-in types and type-constructors are tightly integrated into the language,
3246 in nontrivial ways that are not possible to emulate in user-defined types.
3247 User-defined types have limited capabilities.
3251 The primitive types are the following:
3253 * The boolean type `bool` with values `true` and `false`.
3254 * The machine types (integer and floating-point).
3255 * The machine-dependent integer types.
3259 The machine types are the following:
3261 * The unsigned word types `u8`, `u16`, `u32` and `u64`, with values drawn from
3262 the integer intervals [0, 2^8 - 1], [0, 2^16 - 1], [0, 2^32 - 1] and
3263 [0, 2^64 - 1] respectively.
3265 * The signed two's complement word types `i8`, `i16`, `i32` and `i64`, with
3266 values drawn from the integer intervals [-(2^(7)), 2^7 - 1],
3267 [-(2^(15)), 2^15 - 1], [-(2^(31)), 2^31 - 1], [-(2^(63)), 2^63 - 1]
3270 * The IEEE 754-2008 `binary32` and `binary64` floating-point types: `f32` and
3271 `f64`, respectively.
3273 #### Machine-dependent integer types
3275 The `usize` type is an unsigned integer type with the same number of bits as the
3276 platform's pointer type. It can represent every memory address in the process.
3278 The `isize` type is a signed integer type with the same number of bits as the
3279 platform's pointer type. The theoretical upper bound on object and array size
3280 is the maximum `isize` value. This ensures that `isize` can be used to calculate
3281 differences between pointers into an object or array and can address every byte
3282 within an object along with one byte past the end.
3286 The types `char` and `str` hold textual data.
3288 A value of type `char` is a [Unicode scalar value](
3289 http://www.unicode.org/glossary/#unicode_scalar_value) (i.e. a code point that
3290 is not a surrogate), represented as a 32-bit unsigned word in the 0x0000 to
3291 0xD7FF or 0xE000 to 0x10FFFF range. A `[char]` array is effectively an UCS-4 /
3294 A value of type `str` is a Unicode string, represented as an array of 8-bit
3295 unsigned bytes holding a sequence of UTF-8 code points. Since `str` is of
3296 unknown size, it is not a _first-class_ type, but can only be instantiated
3297 through a pointer type, such as `&str`.
3301 A tuple *type* is a heterogeneous product of other types, called the *elements*
3302 of the tuple. It has no nominal name and is instead structurally typed.
3304 Tuple types and values are denoted by listing the types or values of their
3305 elements, respectively, in a parenthesized, comma-separated list.
3307 Because tuple elements don't have a name, they can only be accessed by
3308 pattern-matching or by using `N` directly as a field to access the
3311 An example of a tuple type and its use:
3314 type Pair<'a> = (i32, &'a str);
3315 let p: Pair<'static> = (10, "ten");
3319 assert_eq!(b, "ten");
3320 assert_eq!(p.0, 10);
3321 assert_eq!(p.1, "ten");
3324 For historical reasons and convenience, the tuple type with no elements (`()`)
3325 is often called ‘unit’ or ‘the unit type’.
3327 ### Array, and Slice types
3329 Rust has two different types for a list of items:
3331 * `[T; N]`, an 'array'
3334 An array has a fixed size, and can be allocated on either the stack or the
3337 A slice is a 'view' into an array. It doesn't own the data it points
3343 // A stack-allocated array
3344 let array: [i32; 3] = [1, 2, 3];
3346 // A heap-allocated array
3347 let vector: Vec<i32> = vec![1, 2, 3];
3349 // A slice into an array
3350 let slice: &[i32] = &vector[..];
3353 As you can see, the `vec!` macro allows you to create a `Vec<T>` easily. The
3354 `vec!` macro is also part of the standard library, rather than the language.
3356 All in-bounds elements of arrays and slices are always initialized, and access
3357 to an array or slice is always bounds-checked.
3361 A `struct` *type* is a heterogeneous product of other types, called the
3362 *fields* of the type.[^structtype]
3364 [^structtype]: `struct` types are analogous to `struct` types in C,
3365 the *record* types of the ML family,
3366 or the *struct* types of the Lisp family.
3368 New instances of a `struct` can be constructed with a [struct
3369 expression](#struct-expressions).
3371 The memory layout of a `struct` is undefined by default to allow for compiler
3372 optimizations like field reordering, but it can be fixed with the
3373 `#[repr(...)]` attribute. In either case, fields may be given in any order in
3374 a corresponding struct *expression*; the resulting `struct` value will always
3375 have the same memory layout.
3377 The fields of a `struct` may be qualified by [visibility
3378 modifiers](#visibility-and-privacy), to allow access to data in a
3379 struct outside a module.
3381 A _tuple struct_ type is just like a struct type, except that the fields are
3384 A _unit-like struct_ type is like a struct type, except that it has no
3385 fields. The one value constructed by the associated [struct
3386 expression](#struct-expressions) is the only value that inhabits such a
3389 ### Enumerated types
3391 An *enumerated type* is a nominal, heterogeneous disjoint union type, denoted
3392 by the name of an [`enum` item](#enumerations). [^enumtype]
3394 [^enumtype]: The `enum` type is analogous to a `data` constructor declaration in
3395 ML, or a *pick ADT* in Limbo.
3397 An [`enum` item](#enumerations) declares both the type and a number of *variant
3398 constructors*, each of which is independently named and takes an optional tuple
3401 New instances of an `enum` can be constructed by calling one of the variant
3402 constructors, in a [call expression](#call-expressions).
3404 Any `enum` value consumes as much memory as the largest variant constructor for
3405 its corresponding `enum` type.
3407 Enum types cannot be denoted *structurally* as types, but must be denoted by
3408 named reference to an [`enum` item](#enumerations).
3412 Nominal types — [enumerations](#enumerated-types) and
3413 [structs](#struct-types) — may be recursive. That is, each `enum`
3414 constructor or `struct` field may refer, directly or indirectly, to the
3415 enclosing `enum` or `struct` type itself. Such recursion has restrictions:
3417 * Recursive types must include a nominal type in the recursion
3418 (not mere [type definitions](grammar.html#type-definitions),
3419 or other structural types such as [arrays](#array-and-slice-types) or [tuples](#tuple-types)).
3420 * A recursive `enum` item must have at least one non-recursive constructor
3421 (in order to give the recursion a basis case).
3422 * The size of a recursive type must be finite;
3423 in other words the recursive fields of the type must be [pointer types](#pointer-types).
3424 * Recursive type definitions can cross module boundaries, but not module *visibility* boundaries,
3425 or crate boundaries (in order to simplify the module system and type checker).
3427 An example of a *recursive* type and its use:
3432 Cons(T, Box<List<T>>)
3435 let a: List<i32> = List::Cons(7, Box::new(List::Cons(13, Box::new(List::Nil))));
3440 All pointers in Rust are explicit first-class values. They can be copied,
3441 stored into data structs, and returned from functions. There are two
3442 varieties of pointer in Rust:
3445 : These point to memory _owned by some other value_.
3446 A reference type is written `&type`,
3447 or `&'a type` when you need to specify an explicit lifetime.
3448 Copying a reference is a "shallow" operation:
3449 it involves only copying the pointer itself.
3450 Releasing a reference has no effect on the value it points to,
3451 but a reference of a temporary value will keep it alive during the scope
3452 of the reference itself.
3454 * Raw pointers (`*`)
3455 : Raw pointers are pointers without safety or liveness guarantees.
3456 Raw pointers are written as `*const T` or `*mut T`,
3457 for example `*const i32` means a raw pointer to a 32-bit integer.
3458 Copying or dropping a raw pointer has no effect on the lifecycle of any
3459 other value. Dereferencing a raw pointer or converting it to any other
3460 pointer type is an [`unsafe` operation](#unsafe-functions).
3461 Raw pointers are generally discouraged in Rust code;
3462 they exist to support interoperability with foreign code,
3463 and writing performance-critical or low-level functions.
3465 The standard library contains additional 'smart pointer' types beyond references
3470 The function type constructor `fn` forms new function types. A function type
3471 consists of a possibly-empty set of function-type modifiers (such as `unsafe`
3472 or `extern`), a sequence of input types and an output type.
3474 An example of a `fn` type:
3477 fn add(x: i32, y: i32) -> i32 {
3481 let mut x = add(5,7);
3483 type Binop = fn(i32, i32) -> i32;
3484 let bo: Binop = add;
3488 #### Function types for specific items
3490 Internal to the compiler, there are also function types that are specific to a particular
3491 function item. In the following snippet, for example, the internal types of the functions
3492 `foo` and `bar` are different, despite the fact that they have the same signature:
3499 The types of `foo` and `bar` can both be implicitly coerced to the fn
3500 pointer type `fn()`. There is currently no syntax for unique fn types,
3501 though the compiler will emit a type like `fn() {foo}` in error
3502 messages to indicate "the unique fn type for the function `foo`".
3506 A [lambda expression](#lambda-expressions) produces a closure value with
3507 a unique, anonymous type that cannot be written out.
3509 Depending on the requirements of the closure, its type implements one or
3510 more of the closure traits:
3513 : The closure can be called once. A closure called as `FnOnce`
3514 can move out values from its environment.
3517 : The closure can be called multiple times as mutable. A closure called as
3518 `FnMut` can mutate values from its environment. `FnMut` inherits from
3519 `FnOnce` (i.e. anything implementing `FnMut` also implements `FnOnce`).
3522 : The closure can be called multiple times through a shared reference.
3523 A closure called as `Fn` can neither move out from nor mutate values
3524 from its environment. `Fn` inherits from `FnMut`, which itself
3525 inherits from `FnOnce`.
3530 In Rust, a type like `&SomeTrait` or `Box<SomeTrait>` is called a _trait object_.
3531 Each instance of a trait object includes:
3533 - a pointer to an instance of a type `T` that implements `SomeTrait`
3534 - a _virtual method table_, often just called a _vtable_, which contains, for
3535 each method of `SomeTrait` that `T` implements, a pointer to `T`'s
3536 implementation (i.e. a function pointer).
3538 The purpose of trait objects is to permit "late binding" of methods. A call to
3539 a method on a trait object is only resolved to a vtable entry at compile time.
3540 The actual implementation for each vtable entry can vary on an object-by-object
3543 Note that for a trait object to be instantiated, the trait must be
3544 _object-safe_. Object safety rules are defined in [RFC 255].
3546 [RFC 255]: https://github.com/rust-lang/rfcs/blob/master/text/0255-object-safety.md
3548 Given a pointer-typed expression `E` of type `&T` or `Box<T>`, where `T`
3549 implements trait `R`, casting `E` to the corresponding pointer type `&R` or
3550 `Box<R>` results in a value of the _trait object_ `R`. This result is
3551 represented as a pair of pointers: the vtable pointer for the `T`
3552 implementation of `R`, and the pointer value of `E`.
3554 An example of a trait object:
3558 fn stringify(&self) -> String;
3561 impl Printable for i32 {
3562 fn stringify(&self) -> String { self.to_string() }
3565 fn print(a: Box<Printable>) {
3566 println!("{}", a.stringify());
3570 print(Box::new(10) as Box<Printable>);
3574 In this example, the trait `Printable` occurs as a trait object in both the
3575 type signature of `print`, and the cast expression in `main`.
3579 Within the body of an item that has type parameter declarations, the names of
3580 its type parameters are types:
3583 fn to_vec<A: Clone>(xs: &[A]) -> Vec<A> {
3587 let first: A = xs[0].clone();
3588 let mut rest: Vec<A> = to_vec(&xs[1..]);
3589 rest.insert(0, first);
3594 Here, `first` has type `A`, referring to `to_vec`'s `A` type parameter; and `rest`
3595 has type `Vec<A>`, a vector with element type `A`.
3599 The special type `Self` has a meaning within traits and impls. In a trait definition, it refers
3600 to an implicit type parameter representing the "implementing" type. In an impl,
3601 it is an alias for the implementing type. For example, in:
3605 fn make_string(&self) -> String;
3608 impl Printable for String {
3609 fn make_string(&self) -> String {
3615 The notation `&self` is a shorthand for `self: &Self`. In this case,
3616 in the impl, `Self` refers to the value of type `String` that is the
3617 receiver for a call to the method `make_string`.
3621 Subtyping is implicit and can occur at any stage in type checking or
3622 inference. Subtyping in Rust is very restricted and occurs only due to
3623 variance with respect to lifetimes and between types with higher ranked
3624 lifetimes. If we were to erase lifetimes from types, then the only subtyping
3625 would be due to type equality.
3627 Consider the following example: string literals always have `'static`
3628 lifetime. Nevertheless, we can assign `s` to `t`:
3632 let s: &'static str = "hi";
3636 Since `'static` "lives longer" than `'a`, `&'static str` is a subtype of
3641 Coercions are defined in [RFC401]. A coercion is implicit and has no syntax.
3643 [RFC401]: https://github.com/rust-lang/rfcs/blob/master/text/0401-coercions.md
3647 A coercion can only occur at certain coercion sites in a program; these are
3648 typically places where the desired type is explicit or can be derived by
3649 propagation from explicit types (without type inference). Possible coercion
3652 * `let` statements where an explicit type is given.
3654 For example, `128` is coerced to have type `i8` in the following:
3660 * `static` and `const` statements (similar to `let` statements).
3662 * Arguments for function calls
3664 The value being coerced is the actual parameter, and it is coerced to
3665 the type of the formal parameter.
3667 For example, `128` is coerced to have type `i8` in the following:
3677 * Instantiations of struct or variant fields
3679 For example, `128` is coerced to have type `i8` in the following:
3682 struct Foo { x: i8 }
3689 * Function results, either the final line of a block if it is not
3690 semicolon-terminated or any expression in a `return` statement
3692 For example, `128` is coerced to have type `i8` in the following:
3700 If the expression in one of these coercion sites is a coercion-propagating
3701 expression, then the relevant sub-expressions in that expression are also
3702 coercion sites. Propagation recurses from these new coercion sites.
3703 Propagating expressions and their relevant sub-expressions are:
3705 * Array literals, where the array has type `[U; n]`. Each sub-expression in
3706 the array literal is a coercion site for coercion to type `U`.
3708 * Array literals with repeating syntax, where the array has type `[U; n]`. The
3709 repeated sub-expression is a coercion site for coercion to type `U`.
3711 * Tuples, where a tuple is a coercion site to type `(U_0, U_1, ..., U_n)`.
3712 Each sub-expression is a coercion site to the respective type, e.g. the
3713 zeroth sub-expression is a coercion site to type `U_0`.
3715 * Parenthesized sub-expressions (`(e)`): if the expression has type `U`, then
3716 the sub-expression is a coercion site to `U`.
3718 * Blocks: if a block has type `U`, then the last expression in the block (if
3719 it is not semicolon-terminated) is a coercion site to `U`. This includes
3720 blocks which are part of control flow statements, such as `if`/`else`, if
3721 the block has a known type.
3725 Coercion is allowed between the following types:
3727 * `T` to `U` if `T` is a subtype of `U` (*reflexive case*)
3729 * `T_1` to `T_3` where `T_1` coerces to `T_2` and `T_2` coerces to `T_3`
3732 Note that this is not fully supported yet
3736 * `*mut T` to `*const T`
3738 * `&T` to `*const T`
3740 * `&mut T` to `*mut T`
3742 * `&T` to `&U` if `T` implements `Deref<Target = U>`. For example:
3745 use std::ops::Deref;
3747 struct CharContainer {
3751 impl Deref for CharContainer {
3754 fn deref<'a>(&'a self) -> &'a char {
3759 fn foo(arg: &char) {}
3762 let x = &mut CharContainer { value: 'y' };
3763 foo(x); //&mut CharContainer is coerced to &char.
3767 * `&mut T` to `&mut U` if `T` implements `DerefMut<Target = U>`.
3769 * TyCtor(`T`) to TyCtor(coerce_inner(`T`)), where TyCtor(`T`) is one of
3777 - coerce_inner(`[T, ..n]`) = `[T]`
3778 - coerce_inner(`T`) = `U` where `T` is a concrete type which implements the
3781 In the future, coerce_inner will be recursively extended to tuples and
3782 structs. In addition, coercions from sub-traits to super-traits will be
3783 added. See [RFC401] for more details.
3787 Several traits define special evaluation behavior.
3791 The `Copy` trait changes the semantics of a type implementing it. Values whose
3792 type implements `Copy` are copied rather than moved upon assignment.
3794 ## The `Sized` trait
3796 The `Sized` trait indicates that the size of this type is known at compile-time.
3800 The `Drop` trait provides a destructor, to be run whenever a value of this type
3803 ## The `Deref` trait
3805 The `Deref<Target = U>` trait allows a type to implicitly implement all the methods
3806 of the type `U`. When attempting to resolve a method call, the compiler will search
3807 the top-level type for the implementation of the called method. If no such method is
3808 found, `.deref()` is called and the compiler continues to search for the method
3809 implementation in the returned type `U`.
3813 A Rust program's memory consists of a static set of *items* and a *heap*.
3814 Immutable portions of the heap may be safely shared between threads, mutable
3815 portions may not be safely shared, but several mechanisms for effectively-safe
3816 sharing of mutable values, built on unsafe code but enforcing a safe locking
3817 discipline, exist in the standard library.
3819 Allocations in the stack consist of *variables*, and allocations in the heap
3822 ### Memory allocation and lifetime
3824 The _items_ of a program are those functions, modules and types that have their
3825 value calculated at compile-time and stored uniquely in the memory image of the
3826 rust process. Items are neither dynamically allocated nor freed.
3828 The _heap_ is a general term that describes boxes. The lifetime of an
3829 allocation in the heap depends on the lifetime of the box values pointing to
3830 it. Since box values may themselves be passed in and out of frames, or stored
3831 in the heap, heap allocations may outlive the frame they are allocated within.
3833 ### Memory ownership
3835 When a stack frame is exited, its local allocations are all released, and its
3836 references to boxes are dropped.
3840 A _variable_ is a component of a stack frame, either a named function parameter,
3841 an anonymous [temporary](#lvalues-rvalues-and-temporaries), or a named local
3844 A _local variable_ (or *stack-local* allocation) holds a value directly,
3845 allocated within the stack's memory. The value is a part of the stack frame.
3847 Local variables are immutable unless declared otherwise like: `let mut x = ...`.
3849 Function parameters are immutable unless declared with `mut`. The `mut` keyword
3850 applies only to the following parameter (so `|mut x, y|` and `fn f(mut x:
3851 Box<i32>, y: Box<i32>)` declare one mutable variable `x` and one immutable
3854 Methods that take either `self` or `Box<Self>` can optionally place them in a
3855 mutable variable by prefixing them with `mut` (similar to regular arguments):
3859 fn change(mut self) -> Self;
3860 fn modify(mut self: Box<Self>) -> Box<Self>;
3864 Local variables are not initialized when allocated; the entire frame worth of
3865 local variables are allocated at once, on frame-entry, in an uninitialized
3866 state. Subsequent statements within a function may or may not initialize the
3867 local variables. Local variables can be used only after they have been
3868 initialized; this is enforced by the compiler.
3872 The Rust compiler supports various methods to link crates together both
3873 statically and dynamically. This section will explore the various methods to
3874 link Rust crates together, and more information about native libraries can be
3875 found in the [FFI section of the book][ffi].
3877 In one session of compilation, the compiler can generate multiple artifacts
3878 through the usage of either command line flags or the `crate_type` attribute.
3879 If one or more command line flags are specified, all `crate_type` attributes will
3880 be ignored in favor of only building the artifacts specified by command line.
3882 * `--crate-type=bin`, `#[crate_type = "bin"]` - A runnable executable will be
3883 produced. This requires that there is a `main` function in the crate which
3884 will be run when the program begins executing. This will link in all Rust and
3885 native dependencies, producing a distributable binary.
3887 * `--crate-type=lib`, `#[crate_type = "lib"]` - A Rust library will be produced.
3888 This is an ambiguous concept as to what exactly is produced because a library
3889 can manifest itself in several forms. The purpose of this generic `lib` option
3890 is to generate the "compiler recommended" style of library. The output library
3891 will always be usable by rustc, but the actual type of library may change from
3892 time-to-time. The remaining output types are all different flavors of
3893 libraries, and the `lib` type can be seen as an alias for one of them (but the
3894 actual one is compiler-defined).
3896 * `--crate-type=dylib`, `#[crate_type = "dylib"]` - A dynamic Rust library will
3897 be produced. This is different from the `lib` output type in that this forces
3898 dynamic library generation. The resulting dynamic library can be used as a
3899 dependency for other libraries and/or executables. This output type will
3900 create `*.so` files on linux, `*.dylib` files on osx, and `*.dll` files on
3903 * `--crate-type=staticlib`, `#[crate_type = "staticlib"]` - A static system
3904 library will be produced. This is different from other library outputs in that
3905 the Rust compiler will never attempt to link to `staticlib` outputs. The
3906 purpose of this output type is to create a static library containing all of
3907 the local crate's code along with all upstream dependencies. The static
3908 library is actually a `*.a` archive on linux and osx and a `*.lib` file on
3909 windows. This format is recommended for use in situations such as linking
3910 Rust code into an existing non-Rust application because it will not have
3911 dynamic dependencies on other Rust code.
3913 * `--crate-type=rlib`, `#[crate_type = "rlib"]` - A "Rust library" file will be
3914 produced. This is used as an intermediate artifact and can be thought of as a
3915 "static Rust library". These `rlib` files, unlike `staticlib` files, are
3916 interpreted by the Rust compiler in future linkage. This essentially means
3917 that `rustc` will look for metadata in `rlib` files like it looks for metadata
3918 in dynamic libraries. This form of output is used to produce statically linked
3919 executables as well as `staticlib` outputs.
3921 Note that these outputs are stackable in the sense that if multiple are
3922 specified, then the compiler will produce each form of output at once without
3923 having to recompile. However, this only applies for outputs specified by the
3924 same method. If only `crate_type` attributes are specified, then they will all
3925 be built, but if one or more `--crate-type` command line flags are specified,
3926 then only those outputs will be built.
3928 With all these different kinds of outputs, if crate A depends on crate B, then
3929 the compiler could find B in various different forms throughout the system. The
3930 only forms looked for by the compiler, however, are the `rlib` format and the
3931 dynamic library format. With these two options for a dependent library, the
3932 compiler must at some point make a choice between these two formats. With this
3933 in mind, the compiler follows these rules when determining what format of
3934 dependencies will be used:
3936 1. If a static library is being produced, all upstream dependencies are
3937 required to be available in `rlib` formats. This requirement stems from the
3938 reason that a dynamic library cannot be converted into a static format.
3940 Note that it is impossible to link in native dynamic dependencies to a static
3941 library, and in this case warnings will be printed about all unlinked native
3942 dynamic dependencies.
3944 2. If an `rlib` file is being produced, then there are no restrictions on what
3945 format the upstream dependencies are available in. It is simply required that
3946 all upstream dependencies be available for reading metadata from.
3948 The reason for this is that `rlib` files do not contain any of their upstream
3949 dependencies. It wouldn't be very efficient for all `rlib` files to contain a
3950 copy of `libstd.rlib`!
3952 3. If an executable is being produced and the `-C prefer-dynamic` flag is not
3953 specified, then dependencies are first attempted to be found in the `rlib`
3954 format. If some dependencies are not available in an rlib format, then
3955 dynamic linking is attempted (see below).
3957 4. If a dynamic library or an executable that is being dynamically linked is
3958 being produced, then the compiler will attempt to reconcile the available
3959 dependencies in either the rlib or dylib format to create a final product.
3961 A major goal of the compiler is to ensure that a library never appears more
3962 than once in any artifact. For example, if dynamic libraries B and C were
3963 each statically linked to library A, then a crate could not link to B and C
3964 together because there would be two copies of A. The compiler allows mixing
3965 the rlib and dylib formats, but this restriction must be satisfied.
3967 The compiler currently implements no method of hinting what format a library
3968 should be linked with. When dynamically linking, the compiler will attempt to
3969 maximize dynamic dependencies while still allowing some dependencies to be
3970 linked in via an rlib.
3972 For most situations, having all libraries available as a dylib is recommended
3973 if dynamically linking. For other situations, the compiler will emit a
3974 warning if it is unable to determine which formats to link each library with.
3976 In general, `--crate-type=bin` or `--crate-type=lib` should be sufficient for
3977 all compilation needs, and the other options are just available if more
3978 fine-grained control is desired over the output format of a Rust crate.
3982 Unsafe operations are those that potentially violate the memory-safety
3983 guarantees of Rust's static semantics.
3985 The following language level features cannot be used in the safe subset of
3988 - Dereferencing a [raw pointer](#pointer-types).
3989 - Reading or writing a [mutable static variable](#mutable-statics).
3990 - Calling an unsafe function (including an intrinsic or foreign function).
3994 Unsafe functions are functions that are not safe in all contexts and/or for all
3995 possible inputs. Such a function must be prefixed with the keyword `unsafe` and
3996 can only be called from an `unsafe` block or another `unsafe` function.
4000 A block of code can be prefixed with the `unsafe` keyword, to permit calling
4001 `unsafe` functions or dereferencing raw pointers within a safe function.
4003 When a programmer has sufficient conviction that a sequence of potentially
4004 unsafe operations is actually safe, they can encapsulate that sequence (taken
4005 as a whole) within an `unsafe` block. The compiler will consider uses of such
4006 code safe, in the surrounding context.
4008 Unsafe blocks are used to wrap foreign libraries, make direct use of hardware
4009 or implement features not directly present in the language. For example, Rust
4010 provides the language features necessary to implement memory-safe concurrency
4011 in the language but the implementation of threads and message passing is in the
4014 Rust's type system is a conservative approximation of the dynamic safety
4015 requirements, so in some cases there is a performance cost to using safe code.
4016 For example, a doubly-linked list is not a tree structure and can only be
4017 represented with reference-counted pointers in safe code. By using `unsafe`
4018 blocks to represent the reverse links as raw pointers, it can be implemented
4021 ## Behavior considered undefined
4023 The following is a list of behavior which is forbidden in all Rust code,
4024 including within `unsafe` blocks and `unsafe` functions. Type checking provides
4025 the guarantee that these issues are never caused by safe code.
4028 * Dereferencing a null/dangling raw pointer
4029 * Reads of [undef](http://llvm.org/docs/LangRef.html#undefined-values)
4030 (uninitialized) memory
4031 * Breaking the [pointer aliasing
4032 rules](http://llvm.org/docs/LangRef.html#pointer-aliasing-rules)
4033 with raw pointers (a subset of the rules used by C)
4034 * `&mut` and `&` follow LLVM’s scoped [noalias] model, except if the `&T`
4035 contains an `UnsafeCell<U>`. Unsafe code must not violate these aliasing
4037 * Mutating non-mutable data (that is, data reached through a shared reference or
4038 data owned by a `let` binding), unless that data is contained within an `UnsafeCell<U>`.
4039 * Invoking undefined behavior via compiler intrinsics:
4040 * Indexing outside of the bounds of an object with `std::ptr::offset`
4041 (`offset` intrinsic), with
4042 the exception of one byte past the end which is permitted.
4043 * Using `std::ptr::copy_nonoverlapping_memory` (`memcpy32`/`memcpy64`
4044 intrinsics) on overlapping buffers
4045 * Invalid values in primitive types, even in private fields/locals:
4046 * Dangling/null references or boxes
4047 * A value other than `false` (0) or `true` (1) in a `bool`
4048 * A discriminant in an `enum` not included in the type definition
4049 * A value in a `char` which is a surrogate or above `char::MAX`
4050 * Non-UTF-8 byte sequences in a `str`
4051 * Unwinding into Rust from foreign code or unwinding from Rust into foreign
4052 code. Rust's failure system is not compatible with exception handling in
4053 other languages. Unwinding must be caught and handled at FFI boundaries.
4055 [noalias]: http://llvm.org/docs/LangRef.html#noalias
4057 ## Behavior not considered unsafe
4059 This is a list of behavior not considered *unsafe* in Rust terms, but that may
4063 * Leaks of memory and other resources
4064 * Exiting without calling destructors
4066 - Overflow is considered "unexpected" behavior and is always user-error,
4067 unless the `wrapping` primitives are used. In non-optimized builds, the compiler
4068 will insert debug checks that panic on overflow, but in optimized builds overflow
4069 instead results in wrapped values. See [RFC 560] for the rationale and more details.
4071 [RFC 560]: https://github.com/rust-lang/rfcs/blob/master/text/0560-integer-overflow.md
4073 # Appendix: Influences
4075 Rust is not a particularly original language, with design elements coming from
4076 a wide range of sources. Some of these are listed below (including elements
4077 that have since been removed):
4079 * SML, OCaml: algebraic data types, pattern matching, type inference,
4080 semicolon statement separation
4081 * C++: references, RAII, smart pointers, move semantics, monomorphization,
4083 * ML Kit, Cyclone: region based memory management
4084 * Haskell (GHC): typeclasses, type families
4085 * Newsqueak, Alef, Limbo: channels, concurrency
4086 * Erlang: message passing, thread failure, ~~linked thread failure~~,
4087 ~~lightweight concurrency~~
4088 * Swift: optional bindings
4089 * Scheme: hygienic macros
4091 * Ruby: ~~block syntax~~
4092 * NIL, Hermes: ~~typestate~~
4093 * [Unicode Annex #31](http://www.unicode.org/reports/tr31/): identifier and
4096 [ffi]: book/ffi.html
4097 [plugin]: book/compiler-plugins.html