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 Rust's grammar is defined over Unicode codepoints, each conventionally denoted
33 `U+XXXX`, for 4 or more hexadecimal digits `X`. _Most_ of Rust's grammar is
34 confined to the ASCII range of Unicode, and is described in this document by a
35 dialect of Extended Backus-Naur Form (EBNF), specifically a dialect of EBNF
36 supported by common automated LL(k) parsing tools such as `llgen`, rather than
37 the dialect given in ISO 14977. The dialect can be defined self-referentially
42 rule : nonterminal ':' productionrule ';' ;
43 productionrule : production [ '|' production ] * ;
45 term : element repeats ;
46 element : LITERAL | IDENTIFIER | '[' productionrule ']' ;
47 repeats : [ '*' | '+' ] NUMBER ? | NUMBER ? | '?' ;
52 - Whitespace in the grammar is ignored.
53 - Square brackets are used to group rules.
54 - `LITERAL` is a single printable ASCII character, or an escaped hexadecimal
55 ASCII code of the form `\xQQ`, in single quotes, denoting the corresponding
56 Unicode codepoint `U+00QQ`.
57 - `IDENTIFIER` is a nonempty string of ASCII letters and underscores.
58 - The `repeat` forms apply to the adjacent `element`, and are as follows:
59 - `?` means zero or one repetition
60 - `*` means zero or more repetitions
61 - `+` means one or more repetitions
62 - NUMBER trailing a repeat symbol gives a maximum repetition count
63 - NUMBER on its own gives an exact repetition count
65 This EBNF dialect should hopefully be familiar to many readers.
67 ## Unicode productions
69 A few productions in Rust's grammar permit Unicode codepoints outside the ASCII
70 range. We define these productions in terms of character properties specified
71 in the Unicode standard, rather than in terms of ASCII-range codepoints. The
72 section [Special Unicode Productions](#special-unicode-productions) lists these
75 ## String table productions
77 Some rules in the grammar — notably [unary
78 operators](#unary-operator-expressions), [binary
79 operators](#binary-operator-expressions), and [keywords](#keywords) — are
80 given in a simplified form: as a listing of a table of unquoted, printable
81 whitespace-separated strings. These cases form a subset of the rules regarding
82 the [token](#tokens) rule, and are assumed to be the result of a
83 lexical-analysis phase feeding the parser, driven by a DFA, operating over the
84 disjunction of all such string table entries.
86 When such a string enclosed in double-quotes (`"`) occurs inside the grammar,
87 it is an implicit reference to a single member of such a string table
88 production. See [tokens](#tokens) for more information.
94 Rust input is interpreted as a sequence of Unicode codepoints encoded in UTF-8.
95 Most Rust grammar rules are defined in terms of printable ASCII-range
96 codepoints, but a small number are defined in terms of Unicode properties or
97 explicit codepoint lists. [^inputformat]
99 [^inputformat]: Substitute definitions for the special Unicode productions are
100 provided to the grammar verifier, restricted to ASCII range, when verifying the
101 grammar in this document.
103 ## Special Unicode Productions
105 The following productions in the Rust grammar are defined in terms of Unicode
106 properties: `ident`, `non_null`, `non_star`, `non_eol`, `non_slash_or_star`,
107 `non_single_quote` and `non_double_quote`.
111 The `ident` production is any nonempty Unicode string of the following form:
113 - The first character has property `XID_start`
114 - The remaining characters have property `XID_continue`
116 that does _not_ occur in the set of [keywords](#keywords).
118 > **Note**: `XID_start` and `XID_continue` as character properties cover the
119 > character ranges used to form the more familiar C and Java language-family
122 ### Delimiter-restricted productions
124 Some productions are defined by exclusion of particular Unicode characters:
126 - `non_null` is any single Unicode character aside from `U+0000` (null)
127 - `non_eol` is `non_null` restricted to exclude `U+000A` (`'\n'`)
128 - `non_star` is `non_null` restricted to exclude `U+002A` (`*`)
129 - `non_slash_or_star` is `non_null` restricted to exclude `U+002F` (`/`) and `U+002A` (`*`)
130 - `non_single_quote` is `non_null` restricted to exclude `U+0027` (`'`)
131 - `non_double_quote` is `non_null` restricted to exclude `U+0022` (`"`)
136 comment : block_comment | line_comment ;
137 block_comment : "/*" block_comment_body * "*/" ;
138 block_comment_body : [block_comment | character] * ;
139 line_comment : "//" non_eol * ;
142 Comments in Rust code follow the general C++ style of line and block-comment
143 forms. Nested block comments are supported.
145 Line comments beginning with exactly _three_ slashes (`///`), and block
146 comments beginning with exactly one repeated asterisk in the block-open
147 sequence (`/**`), are interpreted as a special syntax for `doc`
148 [attributes](#attributes). That is, they are equivalent to writing
149 `#[doc="..."]` around the body of the comment (this includes the comment
150 characters themselves, ie `/// Foo` turns into `#[doc="/// Foo"]`).
152 Line comments beginning with `//!` and block comments beginning with `/*!` are
153 doc comments that apply to the parent of the comment, rather than the item
154 that follows. That is, they are equivalent to writing `#![doc="..."]` around
155 the body of the comment. `//!` comments are usually used to display
156 information on the crate index page.
158 Non-doc comments are interpreted as a form of whitespace.
163 whitespace_char : '\x20' | '\x09' | '\x0a' | '\x0d' ;
164 whitespace : [ whitespace_char | comment ] + ;
167 The `whitespace_char` production is any nonempty Unicode string consisting of
168 any of the following Unicode characters: `U+0020` (space, `' '`), `U+0009`
169 (tab, `'\t'`), `U+000A` (LF, `'\n'`), `U+000D` (CR, `'\r'`).
171 Rust is a "free-form" language, meaning that all forms of whitespace serve only
172 to separate _tokens_ in the grammar, and have no semantic significance.
174 A Rust program has identical meaning if each whitespace element is replaced
175 with any other legal whitespace element, such as a single space character.
180 simple_token : keyword | unop | binop ;
181 token : simple_token | ident | literal | symbol | whitespace token ;
184 Tokens are primitive productions in the grammar defined by regular
185 (non-recursive) languages. "Simple" tokens are given in [string table
186 production](#string-table-productions) form, and occur in the rest of the
187 grammar as double-quoted strings. Other tokens have exact rules given.
191 <p id="keyword-table-marker"></p>
194 |----------|----------|----------|----------|---------|
195 | abstract | alignof | as | become | box |
196 | break | const | continue | crate | do |
197 | else | enum | extern | false | final |
198 | fn | for | if | impl | in |
199 | let | loop | macro | match | mod |
200 | move | mut | offsetof | override | priv |
201 | proc | pub | pure | ref | return |
202 | Self | self | sizeof | static | struct |
203 | super | trait | true | type | typeof |
204 | unsafe | unsized | use | virtual | where |
205 | while | yield | | | |
208 Each of these keywords has special meaning in its grammar, and all of them are
209 excluded from the `ident` rule.
211 Note that some of these keywords are reserved, and do not currently do
216 A literal is an expression consisting of a single token, rather than a sequence
217 of tokens, that immediately and directly denotes the value it evaluates to,
218 rather than referring to it by name or some other evaluation rule. A literal is
219 a form of constant expression, so is evaluated (primarily) at compile time.
223 literal : [ string_lit | char_lit | byte_string_lit | byte_lit | num_lit ] lit_suffix ?;
226 The optional suffix is only used for certain numeric literals, but is
227 reserved for future extension, that is, the above gives the lexical
228 grammar, but a Rust parser will reject everything but the 12 special
229 cases mentioned in [Number literals](#number-literals) below.
233 ##### Characters and strings
235 | | Example | `#` sets | Characters | Escapes |
236 |----------------------------------------------|-----------------|------------|-------------|---------------------|
237 | [Character](#character-literals) | `'H'` | `N/A` | All Unicode | `\'` & [Byte](#byte-escapes) & [Unicode](#unicode-escapes) |
238 | [String](#string-literals) | `"hello"` | `N/A` | All Unicode | `\"` & [Byte](#byte-escapes) & [Unicode](#unicode-escapes) |
239 | [Raw](#raw-string-literals) | `r#"hello"#` | `0...` | All Unicode | `N/A` |
240 | [Byte](#byte-literals) | `b'H'` | `N/A` | All ASCII | `\'` & [Byte](#byte-escapes) |
241 | [Byte string](#byte-string-literals) | `b"hello"` | `N/A` | All ASCII | `\"` & [Byte](#byte-escapes) |
242 | [Raw byte string](#raw-byte-string-literals) | `br#"hello"#` | `0...` | All ASCII | `N/A` |
248 | `\x7F` | 8-bit character code (exactly 2 digits) |
250 | `\r` | Carriage return |
254 ##### Unicode escapes
257 | `\u{7FFF}` | 24-bit Unicode character code (up to 6 digits) |
261 | [Number literals](#number-literals)`*` | Example | Exponentiation | Suffixes |
262 |----------------------------------------|---------|----------------|----------|
263 | Decimal integer | `98_222` | `N/A` | Integer suffixes |
264 | Hex integer | `0xff` | `N/A` | Integer suffixes |
265 | Octal integer | `0o77` | `N/A` | Integer suffixes |
266 | Binary integer | `0b1111_0000` | `N/A` | Integer suffixes |
267 | Floating-point | `123.0E+77` | `Optional` | Floating-point suffixes |
269 `*` All number literals allow `_` as a visual separator: `1_234.0E+18f64`
272 | Integer | Floating-point |
273 |---------|----------------|
274 | `u8`, `i8`, `u16`, `i16`, `u32`, `i32`, `u64`, `i64`, `is` (`isize`), `us` (`usize`) | `f32`, `f64` |
276 #### Character and string literals
279 char_lit : '\x27' char_body '\x27' ;
280 string_lit : '"' string_body * '"' | 'r' raw_string ;
282 char_body : non_single_quote
283 | '\x5c' [ '\x27' | common_escape | unicode_escape ] ;
285 string_body : non_double_quote
286 | '\x5c' [ '\x22' | common_escape | unicode_escape ] ;
287 raw_string : '"' raw_string_body '"' | '#' raw_string '#' ;
289 common_escape : '\x5c'
290 | 'n' | 'r' | 't' | '0'
293 unicode_escape : 'u' '{' hex_digit+ 6 '}';
295 hex_digit : 'a' | 'b' | 'c' | 'd' | 'e' | 'f'
296 | 'A' | 'B' | 'C' | 'D' | 'E' | 'F'
298 oct_digit : '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' ;
299 dec_digit : '0' | nonzero_dec ;
300 nonzero_dec: '1' | '2' | '3' | '4'
301 | '5' | '6' | '7' | '8' | '9' ;
304 ##### Character literals
306 A _character literal_ is a single Unicode character enclosed within two
307 `U+0027` (single-quote) characters, with the exception of `U+0027` itself,
308 which must be _escaped_ by a preceding `U+005C` character (`\`).
310 ##### String literals
312 A _string literal_ is a sequence of any Unicode characters enclosed within two
313 `U+0022` (double-quote) characters, with the exception of `U+0022` itself,
314 which must be _escaped_ by a preceding `U+005C` character (`\`), or a _raw
317 A multi-line string literal may be defined by terminating each line with a
318 `U+005C` character (`\`) immediately before the newline. This causes the
319 `U+005C` character, the newline, and all whitespace at the beginning of the
320 next line to be ignored.
330 ##### Character escapes
332 Some additional _escapes_ are available in either character or non-raw string
333 literals. An escape starts with a `U+005C` (`\`) and continues with one of the
336 * An _8-bit codepoint escape_ escape starts with `U+0078` (`x`) and is
337 followed by exactly two _hex digits_. It denotes the Unicode codepoint
338 equal to the provided hex value.
339 * A _24-bit codepoint escape_ starts with `U+0075` (`u`) and is followed
340 by up to six _hex digits_ surrounded by braces `U+007B` (`{`) and `U+007D`
341 (`}`). It denotes the Unicode codepoint equal to the provided hex value.
342 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
343 (`r`), or `U+0074` (`t`), denoting the unicode values `U+000A` (LF),
344 `U+000D` (CR) or `U+0009` (HT) respectively.
345 * The _backslash escape_ is the character `U+005C` (`\`) which must be
346 escaped in order to denote *itself*.
348 ##### Raw string literals
350 Raw string literals do not process any escapes. They start with the character
351 `U+0072` (`r`), followed by zero or more of the character `U+0023` (`#`) and a
352 `U+0022` (double-quote) character. The _raw string body_ is not defined in the
353 EBNF grammar above: it can contain any sequence of Unicode characters and is
354 terminated only by another `U+0022` (double-quote) character, followed by the
355 same number of `U+0023` (`#`) characters that preceded the opening `U+0022`
356 (double-quote) character.
358 All Unicode characters contained in the raw string body represent themselves,
359 the characters `U+0022` (double-quote) (except when followed by at least as
360 many `U+0023` (`#`) characters as were used to start the raw string literal) or
361 `U+005C` (`\`) do not have any special meaning.
363 Examples for string literals:
366 "foo"; r"foo"; // foo
367 "\"foo\""; r#""foo""#; // "foo"
370 r##"foo #"# bar"##; // foo #"# bar
372 "\x52"; "R"; r"R"; // R
373 "\\x52"; r"\x52"; // \x52
376 #### Byte and byte string literals
379 byte_lit : "b\x27" byte_body '\x27' ;
380 byte_string_lit : "b\x22" string_body * '\x22' | "br" raw_byte_string ;
382 byte_body : ascii_non_single_quote
383 | '\x5c' [ '\x27' | common_escape ] ;
385 byte_string_body : ascii_non_double_quote
386 | '\x5c' [ '\x22' | common_escape ] ;
387 raw_byte_string : '"' raw_byte_string_body '"' | '#' raw_byte_string '#' ;
393 A _byte literal_ is a single ASCII character (in the `U+0000` to `U+007F`
394 range) enclosed within two `U+0027` (single-quote) characters, with the
395 exception of `U+0027` itself, which must be _escaped_ by a preceding U+005C
396 character (`\`), or a single _escape_. It is equivalent to a `u8` unsigned
397 8-bit integer _number literal_.
399 ##### Byte string literals
401 A non-raw _byte string literal_ is a sequence of ASCII characters and _escapes_,
402 preceded by the characters `U+0062` (`b`) and `U+0022` (double-quote), and
403 followed by the character `U+0022`. If the character `U+0022` is present within
404 the literal, it must be _escaped_ by a preceding `U+005C` (`\`) character.
405 Alternatively, a byte string literal can be a _raw byte string literal_, defined
406 below. A byte string literal is equivalent to a `&'static [u8]` borrowed array
407 of unsigned 8-bit integers.
409 Some additional _escapes_ are available in either byte or non-raw byte string
410 literals. An escape starts with a `U+005C` (`\`) and continues with one of the
413 * An _byte escape_ escape starts with `U+0078` (`x`) and is
414 followed by exactly two _hex digits_. It denotes the byte
415 equal to the provided hex value.
416 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
417 (`r`), or `U+0074` (`t`), denoting the bytes values `0x0A` (ASCII LF),
418 `0x0D` (ASCII CR) or `0x09` (ASCII HT) respectively.
419 * The _backslash escape_ is the character `U+005C` (`\`) which must be
420 escaped in order to denote its ASCII encoding `0x5C`.
422 ##### Raw byte string literals
424 Raw byte string literals do not process any escapes. They start with the
425 character `U+0062` (`b`), followed by `U+0072` (`r`), followed by zero or more
426 of the character `U+0023` (`#`), and a `U+0022` (double-quote) character. The
427 _raw string body_ is not defined in the EBNF grammar above: it can contain any
428 sequence of ASCII characters and is terminated only by another `U+0022`
429 (double-quote) character, followed by the same number of `U+0023` (`#`)
430 characters that preceded the opening `U+0022` (double-quote) character. A raw
431 byte string literal can not contain any non-ASCII byte.
433 All characters contained in the raw string body represent their ASCII encoding,
434 the characters `U+0022` (double-quote) (except when followed by at least as
435 many `U+0023` (`#`) characters as were used to start the raw string literal) or
436 `U+005C` (`\`) do not have any special meaning.
438 Examples for byte string literals:
441 b"foo"; br"foo"; // foo
442 b"\"foo\""; br#""foo""#; // "foo"
445 br##"foo #"# bar"##; // foo #"# bar
447 b"\x52"; b"R"; br"R"; // R
448 b"\\x52"; br"\x52"; // \x52
454 num_lit : nonzero_dec [ dec_digit | '_' ] * float_suffix ?
455 | '0' [ [ dec_digit | '_' ] * float_suffix ?
456 | 'b' [ '1' | '0' | '_' ] +
457 | 'o' [ oct_digit | '_' ] +
458 | 'x' [ hex_digit | '_' ] + ] ;
460 float_suffix : [ exponent | '.' dec_lit exponent ? ] ? ;
462 exponent : ['E' | 'e'] ['-' | '+' ] ? dec_lit ;
463 dec_lit : [ dec_digit | '_' ] + ;
466 A _number literal_ is either an _integer literal_ or a _floating-point
467 literal_. The grammar for recognizing the two kinds of literals is mixed.
469 ##### Integer literals
471 An _integer literal_ has one of four forms:
473 * A _decimal literal_ starts with a *decimal digit* and continues with any
474 mixture of *decimal digits* and _underscores_.
475 * A _hex literal_ starts with the character sequence `U+0030` `U+0078`
476 (`0x`) and continues as any mixture of hex digits and underscores.
477 * An _octal literal_ starts with the character sequence `U+0030` `U+006F`
478 (`0o`) and continues as any mixture of octal digits and underscores.
479 * A _binary literal_ starts with the character sequence `U+0030` `U+0062`
480 (`0b`) and continues as any mixture of binary digits and underscores.
482 Like any literal, an integer literal may be followed (immediately,
483 without any spaces) by an _integer suffix_, which forcibly sets the
484 type of the literal. The integer suffix must be the name of one of the
485 integral types: `u8`, `i8`, `u16`, `i16`, `u32`, `i32`, `u64`, `i64`,
488 The type of an _unsuffixed_ integer literal is determined by type inference.
489 If an integer type can be _uniquely_ determined from the surrounding program
490 context, the unsuffixed integer literal has that type. If the program context
491 underconstrains the type, it defaults to the signed 32-bit integer `i32`; if
492 the program context overconstrains the type, it is considered a static type
495 Examples of integer literals of various forms:
502 0o70_i16; // type i16
503 0b1111_1111_1001_0000_i32; // type i32
504 0usize; // type usize
507 ##### Floating-point literals
509 A _floating-point literal_ has one of two forms:
511 * A _decimal literal_ followed by a period character `U+002E` (`.`). This is
512 optionally followed by another decimal literal, with an optional _exponent_.
513 * A single _decimal literal_ followed by an _exponent_.
515 By default, a floating-point literal has a generic type, and, like integer
516 literals, the type must be uniquely determined from the context. There are two valid
517 _floating-point suffixes_, `f32` and `f64` (the 32-bit and 64-bit floating point
518 types), which explicitly determine the type of the literal.
520 Examples of floating-point literals of various forms:
523 123.0f64; // type f64
526 12E+99_f64; // type f64
527 let x: f64 = 2.; // type f64
530 This last example is different because it is not possible to use the suffix
531 syntax with a floating point literal ending in a period. `2.f64` would attempt
532 to call a method named `f64` on `2`.
534 The representation semantics of floating-point numbers are described in
535 ["Machine Types"](#machine-types).
537 #### Boolean literals
539 The two values of the boolean type are written `true` and `false`.
545 | '#' | '[' | ']' | '(' | ')' | '{' | '}'
549 Symbols are a general class of printable [token](#tokens) that play structural
550 roles in a variety of grammar productions. They are catalogued here for
551 completeness as the set of remaining miscellaneous printable tokens that do not
552 otherwise appear as [unary operators](#unary-operator-expressions), [binary
553 operators](#binary-operator-expressions), or [keywords](#keywords).
559 expr_path : [ "::" ] ident [ "::" expr_path_tail ] + ;
560 expr_path_tail : '<' type_expr [ ',' type_expr ] + '>'
563 type_path : ident [ type_path_tail ] + ;
564 type_path_tail : '<' type_expr [ ',' type_expr ] + '>'
568 A _path_ is a sequence of one or more path components _logically_ separated by
569 a namespace qualifier (`::`). If a path consists of only one component, it may
570 refer to either an [item](#items) or a [variable](#variables) in a local control
571 scope. If a path has multiple components, it refers to an item.
573 Every item has a _canonical path_ within its crate, but the path naming an item
574 is only meaningful within a given crate. There is no global namespace across
575 crates; an item's canonical path merely identifies it within the crate.
577 Two examples of simple paths consisting of only identifier components:
584 Path components are usually [identifiers](#identifiers), but the trailing
585 component of a path may be an angle-bracket-enclosed list of type arguments. In
586 [expression](#expressions) context, the type argument list is given after a
587 final (`::`) namespace qualifier in order to disambiguate it from a relational
588 expression involving the less-than symbol (`<`). In type expression context,
589 the final namespace qualifier is omitted.
591 Two examples of paths with type arguments:
594 # struct HashMap<K, V>(K,V);
596 # fn id<T>(t: T) -> T { t }
597 type T = HashMap<i32,String>; // Type arguments used in a type expression
598 let x = id::<i32>(10); // Type arguments used in a call expression
602 Paths can be denoted with various leading qualifiers to change the meaning of
605 * Paths starting with `::` are considered to be global paths where the
606 components of the path start being resolved from the crate root. Each
607 identifier in the path must resolve to an item.
615 ::a::foo(); // call a's foo function
621 * Paths starting with the keyword `super` begin resolution relative to the
622 parent module. Each further identifier must resolve to an item.
630 super::a::foo(); // call a's foo function
636 * Paths starting with the keyword `self` begin resolution relative to the
637 current module. Each further identifier must resolve to an item.
649 A number of minor features of Rust are not central enough to have their own
650 syntax, and yet are not implementable as functions. Instead, they are given
651 names, and invoked through a consistent syntax: `some_extension!(...)`.
653 Users of `rustc` can define new syntax extensions in two ways:
655 * [Compiler plugins][plugin] can include arbitrary
656 Rust code that manipulates syntax trees at compile time.
658 * [Macros](book/macros.html) define new syntax in a higher-level,
664 expr_macro_rules : "macro_rules" '!' ident '(' macro_rule * ')' ;
665 macro_rule : '(' matcher * ')' "=>" '(' transcriber * ')' ';' ;
666 matcher : '(' matcher * ')' | '[' matcher * ']'
667 | '{' matcher * '}' | '$' ident ':' ident
668 | '$' '(' matcher * ')' sep_token? [ '*' | '+' ]
669 | non_special_token ;
670 transcriber : '(' transcriber * ')' | '[' transcriber * ']'
671 | '{' transcriber * '}' | '$' ident
672 | '$' '(' transcriber * ')' sep_token? [ '*' | '+' ]
673 | non_special_token ;
676 `macro_rules` allows users to define syntax extension in a declarative way. We
677 call such extensions "macros by example" or simply "macros" — to be distinguished
678 from the "procedural macros" defined in [compiler plugins][plugin].
680 Currently, macros can expand to expressions, statements, items, or patterns.
682 (A `sep_token` is any token other than `*` and `+`. A `non_special_token` is
683 any token other than a delimiter or `$`.)
685 The macro expander looks up macro invocations by name, and tries each macro
686 rule in turn. It transcribes the first successful match. Matching and
687 transcription are closely related to each other, and we will describe them
692 The macro expander matches and transcribes every token that does not begin with
693 a `$` literally, including delimiters. For parsing reasons, delimiters must be
694 balanced, but they are otherwise not special.
696 In the matcher, `$` _name_ `:` _designator_ matches the nonterminal in the Rust
697 syntax named by _designator_. Valid designators are `item`, `block`, `stmt`,
698 `pat`, `expr`, `ty` (type), `ident`, `path`, `tt` (either side of the `=>`
699 in macro rules). In the transcriber, the designator is already known, and so
700 only the name of a matched nonterminal comes after the dollar sign.
702 In both the matcher and transcriber, the Kleene star-like operator indicates
703 repetition. The Kleene star operator consists of `$` and parens, optionally
704 followed by a separator token, followed by `*` or `+`. `*` means zero or more
705 repetitions, `+` means at least one repetition. The parens are not matched or
706 transcribed. On the matcher side, a name is bound to _all_ of the names it
707 matches, in a structure that mimics the structure of the repetition encountered
708 on a successful match. The job of the transcriber is to sort that structure
711 The rules for transcription of these repetitions are called "Macro By Example".
712 Essentially, one "layer" of repetition is discharged at a time, and all of them
713 must be discharged by the time a name is transcribed. Therefore, `( $( $i:ident
714 ),* ) => ( $i )` is an invalid macro, but `( $( $i:ident ),* ) => ( $( $i:ident
715 ),* )` is acceptable (if trivial).
717 When Macro By Example encounters a repetition, it examines all of the `$`
718 _name_ s that occur in its body. At the "current layer", they all must repeat
719 the same number of times, so ` ( $( $i:ident ),* ; $( $j:ident ),* ) => ( $(
720 ($i,$j) ),* )` is valid if given the argument `(a,b,c ; d,e,f)`, but not
721 `(a,b,c ; d,e)`. The repetition walks through the choices at that layer in
722 lockstep, so the former input transcribes to `( (a,d), (b,e), (c,f) )`.
724 Nested repetitions are allowed.
726 ### Parsing limitations
728 The parser used by the macro system is reasonably powerful, but the parsing of
729 Rust syntax is restricted in two ways:
731 1. The parser will always parse as much as possible. If it attempts to match
732 `$i:expr [ , ]` against `8 [ , ]`, it will attempt to parse `i` as an array
733 index operation and fail. Adding a separator can solve this problem.
734 2. The parser must have eliminated all ambiguity by the time it reaches a `$`
735 _name_ `:` _designator_. This requirement most often affects name-designator
736 pairs when they occur at the beginning of, or immediately after, a `$(...)*`;
737 requiring a distinctive token in front can solve the problem.
739 # Crates and source files
741 Rust is a *compiled* language. Its semantics obey a *phase distinction* between
742 compile-time and run-time. Those semantic rules that have a *static
743 interpretation* govern the success or failure of compilation. Those semantics
744 that have a *dynamic interpretation* govern the behavior of the program at
747 The compilation model centers on artifacts called _crates_. Each compilation
748 processes a single crate in source form, and if successful, produces a single
749 crate in binary form: either an executable or a library.[^cratesourcefile]
751 [^cratesourcefile]: A crate is somewhat analogous to an *assembly* in the
752 ECMA-335 CLI model, a *library* in the SML/NJ Compilation Manager, a *unit*
753 in the Owens and Flatt module system, or a *configuration* in Mesa.
755 A _crate_ is a unit of compilation and linking, as well as versioning,
756 distribution and runtime loading. A crate contains a _tree_ of nested
757 [module](#modules) scopes. The top level of this tree is a module that is
758 anonymous (from the point of view of paths within the module) and any item
759 within a crate has a canonical [module path](#paths) denoting its location
760 within the crate's module tree.
762 The Rust compiler is always invoked with a single source file as input, and
763 always produces a single output crate. The processing of that source file may
764 result in other source files being loaded as modules. Source files have the
767 A Rust source file describes a module, the name and location of which —
768 in the module tree of the current crate — are defined from outside the
769 source file: either by an explicit `mod_item` in a referencing source file, or
770 by the name of the crate itself.
772 Each source file contains a sequence of zero or more `item` definitions, and
773 may optionally begin with any number of `attributes` that apply to the
774 containing module. Attributes on the anonymous crate module define important
775 metadata that influences the behavior of the compiler.
779 #![crate_name = "projx"]
781 // Specify the output type
782 #![crate_type = "lib"]
785 #![warn(non_camel_case_types)]
788 A crate that contains a `main` function can be compiled to an executable. If a
789 `main` function is present, its return type must be [`unit`](#primitive-types)
790 and it must take no arguments.
792 # Items and attributes
794 Crates contain [items](#items), each of which may have some number of
795 [attributes](#attributes) attached to it.
800 item : extern_crate_decl | use_decl | mod_item | fn_item | type_item
801 | struct_item | enum_item | static_item | trait_item | impl_item
805 An _item_ is a component of a crate. Items are organized within a crate by a
806 nested set of [modules](#modules). Every crate has a single "outermost"
807 anonymous module; all further items within the crate have [paths](#paths)
808 within the module tree of the crate.
810 Items are entirely determined at compile-time, generally remain fixed during
811 execution, and may reside in read-only memory.
813 There are several kinds of item:
815 * [`extern crate` declarations](#extern-crate-declarations)
816 * [`use` declarations](#use-declarations)
817 * [modules](#modules)
818 * [functions](#functions)
819 * [type definitions](#type-definitions)
820 * [structures](#structures)
821 * [enumerations](#enumerations)
822 * [static items](#static-items)
824 * [implementations](#implementations)
826 Some items form an implicit scope for the declaration of sub-items. In other
827 words, within a function or module, declarations of items can (in many cases)
828 be mixed with the statements, control blocks, and similar artifacts that
829 otherwise compose the item body. The meaning of these scoped items is the same
830 as if the item was declared outside the scope — it is still a static item
831 — except that the item's *path name* within the module namespace is
832 qualified by the name of the enclosing item, or is private to the enclosing
833 item (in the case of functions). The grammar specifies the exact locations in
834 which sub-item declarations may appear.
838 All items except modules may be *parameterized* by type. Type parameters are
839 given as a comma-separated list of identifiers enclosed in angle brackets
840 (`<...>`), after the name of the item and before its definition. The type
841 parameters of an item are considered "part of the name", not part of the type
842 of the item. A referencing [path](#paths) must (in principle) provide type
843 arguments as a list of comma-separated types enclosed within angle brackets, in
844 order to refer to the type-parameterized item. In practice, the type-inference
845 system can usually infer such argument types from context. There are no
846 general type-parametric types, only type-parametric items. That is, Rust has
847 no notion of type abstraction: there are no first-class "forall" types.
852 mod_item : "mod" ident ( ';' | '{' mod '}' );
856 A module is a container for zero or more [items](#items).
858 A _module item_ is a module, surrounded in braces, named, and prefixed with the
859 keyword `mod`. A module item introduces a new, named module into the tree of
860 modules making up a crate. Modules can nest arbitrarily.
862 An example of a module:
866 type Complex = (f64, f64);
867 fn sin(f: f64) -> f64 {
871 fn cos(f: f64) -> f64 {
875 fn tan(f: f64) -> f64 {
882 Modules and types share the same namespace. Declaring a named type with
883 the same name as a module in scope is forbidden: that is, a type definition,
884 trait, struct, enumeration, or type parameter can't shadow the name of a module
885 in scope, or vice versa.
887 A module without a body is loaded from an external file, by default with the
888 same name as the module, plus the `.rs` extension. When a nested submodule is
889 loaded from an external file, it is loaded from a subdirectory path that
890 mirrors the module hierarchy.
893 // Load the `vec` module from `vec.rs`
897 // Load the `local_data` module from `thread/local_data.rs`
902 The directories and files used for loading external file modules can be
903 influenced with the `path` attribute.
906 #[path = "thread_files"]
908 // Load the `local_data` module from `thread_files/tls.rs`
914 ##### Extern crate declarations
917 extern_crate_decl : "extern" "crate" crate_name
918 crate_name: ident | ( string_lit "as" ident )
921 An _`extern crate` declaration_ specifies a dependency on an external crate.
922 The external crate is then bound into the declaring scope as the `ident`
923 provided in the `extern_crate_decl`.
925 The external crate is resolved to a specific `soname` at compile time, and a
926 runtime linkage requirement to that `soname` is passed to the linker for
927 loading at runtime. The `soname` is resolved at compile time by scanning the
928 compiler's library path and matching the optional `crateid` provided as a
929 string literal against the `crateid` attributes that were declared on the
930 external crate when it was compiled. If no `crateid` is provided, a default
931 `name` attribute is assumed, equal to the `ident` given in the
934 Three examples of `extern crate` declarations:
939 extern crate std; // equivalent to: extern crate std as std;
941 extern crate std as ruststd; // linking to 'std' under another name
944 ##### Use declarations
947 use_decl : "pub" ? "use" [ path "as" ident
950 path_glob : ident [ "::" [ path_glob
952 | '{' path_item [ ',' path_item ] * '}' ;
954 path_item : ident | "self" ;
957 A _use declaration_ creates one or more local name bindings synonymous with
958 some other [path](#paths). Usually a `use` declaration is used to shorten the
959 path required to refer to a module item. These declarations may appear at the
960 top of [modules](#modules) and [blocks](#blocks).
962 > **Note**: Unlike in many languages,
963 > `use` declarations in Rust do *not* declare linkage dependency with external crates.
964 > Rather, [`extern crate` declarations](#extern-crate-declarations) declare linkage dependencies.
966 Use declarations support a number of convenient shortcuts:
968 * Rebinding the target name as a new local name, using the syntax `use p::q::r as x;`
969 * Simultaneously binding a list of paths differing only in their final element,
970 using the glob-like brace syntax `use a::b::{c,d,e,f};`
971 * Binding all paths matching a given prefix, using the asterisk wildcard syntax
973 * Simultaneously binding a list of paths differing only in their final element
974 and their immediate parent module, using the `self` keyword, such as
975 `use a::b::{self, c, d};`
977 An example of `use` declarations:
980 use std::option::Option::{Some, None};
981 use std::collections::hash_map::{self, HashMap};
984 fn bar(map1: HashMap<String, usize>, map2: hash_map::HashMap<String, usize>){}
987 // Equivalent to 'foo(vec![std::option::Option::Some(1.0f64),
988 // std::option::Option::None]);'
989 foo(vec![Some(1.0f64), None]);
991 // Both `hash_map` and `HashMap` are in scope.
992 let map1 = HashMap::new();
993 let map2 = hash_map::HashMap::new();
998 Like items, `use` declarations are private to the containing module, by
999 default. Also like items, a `use` declaration can be public, if qualified by
1000 the `pub` keyword. Such a `use` declaration serves to _re-export_ a name. A
1001 public `use` declaration can therefore _redirect_ some public name to a
1002 different target definition: even a definition with a private canonical path,
1003 inside a different module. If a sequence of such redirections form a cycle or
1004 cannot be resolved unambiguously, they represent a compile-time error.
1006 An example of re-exporting:
1011 pub use quux::foo::{bar, baz};
1020 In this example, the module `quux` re-exports two public names defined in
1023 Also note that the paths contained in `use` items are relative to the crate
1024 root. So, in the previous example, the `use` refers to `quux::foo::{bar,
1025 baz}`, and not simply to `foo::{bar, baz}`. This also means that top-level
1026 module declarations should be at the crate root if direct usage of the declared
1027 modules within `use` items is desired. It is also possible to use `self` and
1028 `super` at the beginning of a `use` item to refer to the current and direct
1029 parent modules respectively. All rules regarding accessing declared modules in
1030 `use` declarations applies to both module declarations and `extern crate`
1033 An example of what will and will not work for `use` items:
1036 # #![allow(unused_imports)]
1037 use foo::baz::foobaz; // good: foo is at the root of the crate
1045 use foo::example::iter; // good: foo is at crate root
1046 // use example::iter; // bad: core is not at the crate root
1047 use self::baz::foobaz; // good: self refers to module 'foo'
1048 use foo::bar::foobar; // good: foo is at crate root
1055 use super::bar::foobar; // good: super refers to module 'foo'
1065 A _function item_ defines a sequence of [statements](#statements) and an
1066 optional final [expression](#expressions), along with a name and a set of
1067 parameters. Functions are declared with the keyword `fn`. Functions declare a
1068 set of *input* [*variables*](#variables) as parameters, through which the caller
1069 passes arguments into the function, and the *output* [*type*](#types)
1070 of the value the function will return to its caller on completion.
1072 A function may also be copied into a first-class *value*, in which case the
1073 value has the corresponding [*function type*](#function-types), and can be used
1074 otherwise exactly as a function item (with a minor additional cost of calling
1075 the function indirectly).
1077 Every control path in a function logically ends with a `return` expression or a
1078 diverging expression. If the outermost block of a function has a
1079 value-producing expression in its final-expression position, that expression is
1080 interpreted as an implicit `return` expression applied to the final-expression.
1082 An example of a function:
1085 fn add(x: i32, y: i32) -> i32 {
1090 As with `let` bindings, function arguments are irrefutable patterns, so any
1091 pattern that is valid in a let binding is also valid as an argument.
1094 fn first((value, _): (i32, i32)) -> i32 { value }
1098 #### Generic functions
1100 A _generic function_ allows one or more _parameterized types_ to appear in its
1101 signature. Each type parameter must be explicitly declared, in an
1102 angle-bracket-enclosed, comma-separated list following the function name.
1105 fn iter<T, F>(seq: &[T], f: F) where T: Copy, F: Fn(T) {
1106 for elt in seq { f(*elt); }
1108 fn map<T, U, F>(seq: &[T], f: F) -> Vec<U> where T: Copy, U: Copy, F: Fn(T) -> U {
1109 let mut acc = vec![];
1110 for elt in seq { acc.push(f(*elt)); }
1115 Inside the function signature and body, the name of the type parameter can be
1116 used as a type name. [Trait](#traits) bounds can be specified for type parameters
1117 to allow methods with that trait to be called on values of that type. This is
1118 specified using the `where` syntax, as in the above example.
1120 When a generic function is referenced, its type is instantiated based on the
1121 context of the reference. For example, calling the `iter` function defined
1122 above on `[1, 2]` will instantiate type parameter `T` with `i32`, and require
1123 the closure parameter to have type `Fn(i32)`.
1125 The type parameters can also be explicitly supplied in a trailing
1126 [path](#paths) component after the function name. This might be necessary if
1127 there is not sufficient context to determine the type parameters. For example,
1128 `mem::size_of::<u32>() == 4`.
1132 Unsafe operations are those that potentially violate the memory-safety
1133 guarantees of Rust's static semantics.
1135 The following language level features cannot be used in the safe subset of
1138 - Dereferencing a [raw pointer](#pointer-types).
1139 - Reading or writing a [mutable static variable](#mutable-statics).
1140 - Calling an unsafe function (including an intrinsic or foreign function).
1142 ##### Unsafe functions
1144 Unsafe functions are functions that are not safe in all contexts and/or for all
1145 possible inputs. Such a function must be prefixed with the keyword `unsafe` and
1146 can only be called from an `unsafe` block or another `unsafe` function.
1150 A block of code can be prefixed with the `unsafe` keyword, to permit calling
1151 `unsafe` functions or dereferencing raw pointers within a safe function.
1153 When a programmer has sufficient conviction that a sequence of potentially
1154 unsafe operations is actually safe, they can encapsulate that sequence (taken
1155 as a whole) within an `unsafe` block. The compiler will consider uses of such
1156 code safe, in the surrounding context.
1158 Unsafe blocks are used to wrap foreign libraries, make direct use of hardware
1159 or implement features not directly present in the language. For example, Rust
1160 provides the language features necessary to implement memory-safe concurrency
1161 in the language but the implementation of threads and message passing is in the
1164 Rust's type system is a conservative approximation of the dynamic safety
1165 requirements, so in some cases there is a performance cost to using safe code.
1166 For example, a doubly-linked list is not a tree structure and can only be
1167 represented with reference-counted pointers in safe code. By using `unsafe`
1168 blocks to represent the reverse links as raw pointers, it can be implemented
1171 ##### Behavior considered undefined
1173 The following is a list of behavior which is forbidden in all Rust code,
1174 including within `unsafe` blocks and `unsafe` functions. Type checking provides
1175 the guarantee that these issues are never caused by safe code.
1178 * Dereferencing a null/dangling raw pointer
1179 * Reads of [undef](http://llvm.org/docs/LangRef.html#undefined-values)
1180 (uninitialized) memory
1181 * Breaking the [pointer aliasing
1182 rules](http://llvm.org/docs/LangRef.html#pointer-aliasing-rules)
1183 with raw pointers (a subset of the rules used by C)
1184 * `&mut` and `&` follow LLVM’s scoped [noalias] model, except if the `&T`
1185 contains an `UnsafeCell<U>`. Unsafe code must not violate these aliasing
1187 * Mutating an immutable value/reference without `UnsafeCell<U>`
1188 * Invoking undefined behavior via compiler intrinsics:
1189 * Indexing outside of the bounds of an object with `std::ptr::offset`
1190 (`offset` intrinsic), with
1191 the exception of one byte past the end which is permitted.
1192 * Using `std::ptr::copy_nonoverlapping_memory` (`memcpy32`/`memcpy64`
1193 intrinsics) on overlapping buffers
1194 * Invalid values in primitive types, even in private fields/locals:
1195 * Dangling/null references or boxes
1196 * A value other than `false` (0) or `true` (1) in a `bool`
1197 * A discriminant in an `enum` not included in the type definition
1198 * A value in a `char` which is a surrogate or above `char::MAX`
1199 * Non-UTF-8 byte sequences in a `str`
1200 * Unwinding into Rust from foreign code or unwinding from Rust into foreign
1201 code. Rust's failure system is not compatible with exception handling in
1202 other languages. Unwinding must be caught and handled at FFI boundaries.
1204 [noalias]: http://llvm.org/docs/LangRef.html#noalias
1206 ##### Behaviour not considered unsafe
1208 This is a list of behaviour not considered *unsafe* in Rust terms, but that may
1212 * Reading data from private fields (`std::repr`)
1213 * Leaks due to reference count cycles, even in the global heap
1214 * Exiting without calling destructors
1216 * Accessing/modifying the file system
1217 * Unsigned integer overflow (well-defined as wrapping)
1218 * Signed integer overflow (well-defined as two’s complement representation
1221 #### Diverging functions
1223 A special kind of function can be declared with a `!` character where the
1224 output type would normally be. For example:
1227 fn my_err(s: &str) -> ! {
1233 We call such functions "diverging" because they never return a value to the
1234 caller. Every control path in a diverging function must end with a `panic!()` or
1235 a call to another diverging function on every control path. The `!` annotation
1236 does *not* denote a type.
1238 It might be necessary to declare a diverging function because as mentioned
1239 previously, the typechecker checks that every control path in a function ends
1240 with a [`return`](#return-expressions) or diverging expression. So, if `my_err`
1241 were declared without the `!` annotation, the following code would not
1245 # fn my_err(s: &str) -> ! { panic!() }
1247 fn f(i: i32) -> i32 {
1252 my_err("Bad number!");
1257 This will not compile without the `!` annotation on `my_err`, since the `else`
1258 branch of the conditional in `f` does not return an `i32`, as required by the
1259 signature of `f`. Adding the `!` annotation to `my_err` informs the
1260 typechecker that, should control ever enter `my_err`, no further type judgments
1261 about `f` need to hold, since control will never resume in any context that
1262 relies on those judgments. Thus the return type on `f` only needs to reflect
1263 the `if` branch of the conditional.
1265 #### Extern functions
1267 Extern functions are part of Rust's foreign function interface, providing the
1268 opposite functionality to [external blocks](#external-blocks). Whereas
1269 external blocks allow Rust code to call foreign code, extern functions with
1270 bodies defined in Rust code _can be called by foreign code_. They are defined
1271 in the same way as any other Rust function, except that they have the `extern`
1275 // Declares an extern fn, the ABI defaults to "C"
1276 extern fn new_i32() -> i32 { 0 }
1278 // Declares an extern fn with "stdcall" ABI
1279 extern "stdcall" fn new_i32_stdcall() -> i32 { 0 }
1282 Unlike normal functions, extern fns have an `extern "ABI" fn()`. This is the
1283 same type as the functions declared in an extern block.
1286 # extern fn new_i32() -> i32 { 0 }
1287 let fptr: extern "C" fn() -> i32 = new_i32;
1290 Extern functions may be called directly from Rust code as Rust uses large,
1291 contiguous stack segments like C.
1295 A _type alias_ defines a new name for an existing [type](#types). Type
1296 aliases are declared with the keyword `type`. Every value has a single,
1297 specific type, but may implement several different traits, or be compatible with
1298 several different type constraints.
1300 For example, the following defines the type `Point` as a synonym for the type
1301 `(u8, u8)`, the type of pairs of unsigned 8 bit integers.:
1304 type Point = (u8, u8);
1305 let p: Point = (41, 68);
1310 A _structure_ is a nominal [structure type](#structure-types) defined with the
1313 An example of a `struct` item and its use:
1316 struct Point {x: i32, y: i32}
1317 let p = Point {x: 10, y: 11};
1321 A _tuple structure_ is a nominal [tuple type](#tuple-types), also defined with
1322 the keyword `struct`. For example:
1325 struct Point(i32, i32);
1326 let p = Point(10, 11);
1327 let px: i32 = match p { Point(x, _) => x };
1330 A _unit-like struct_ is a structure without any fields, defined by leaving off
1331 the list of fields entirely. Such types will have a single value, just like
1332 the [unit value `()`](#unit-and-boolean-literals) of the unit type. For
1337 let c = [Cookie, Cookie, Cookie, Cookie];
1340 The precise memory layout of a structure is not specified. One can specify a
1341 particular layout using the [`repr` attribute](#ffi-attributes).
1345 An _enumeration_ is a simultaneous definition of a nominal [enumerated
1346 type](#enumerated-types) as well as a set of *constructors*, that can be used
1347 to create or pattern-match values of the corresponding enumerated type.
1349 Enumerations are declared with the keyword `enum`.
1351 An example of an `enum` item and its use:
1359 let mut a: Animal = Animal::Dog;
1363 Enumeration constructors can have either named or unnamed fields:
1368 Cat { name: String, weight: f64 }
1371 let mut a: Animal = Animal::Dog("Cocoa".to_string(), 37.2);
1372 a = Animal::Cat { name: "Spotty".to_string(), weight: 2.7 };
1375 In this example, `Cat` is a _struct-like enum variant_,
1376 whereas `Dog` is simply called an enum variant.
1378 Enums have a discriminant. You can assign them explicitly:
1386 If a discriminant isn't assigned, they start at zero, and add one for each
1389 You can cast an enum to get this value:
1392 # enum Foo { Bar = 123 }
1393 let x = Foo::Bar as u32; // x is now 123u32
1396 This only works as long as none of the variants have data attached. If
1397 it were `Bar(i32)`, this is disallowed.
1402 const_item : "const" ident ':' type '=' expr ';' ;
1405 A *constant item* is a named _constant value_ which is not associated with a
1406 specific memory location in the program. Constants are essentially inlined
1407 wherever they are used, meaning that they are copied directly into the relevant
1408 context when used. References to the same constant are not necessarily
1409 guaranteed to refer to the same memory address.
1411 Constant values must not have destructors, and otherwise permit most forms of
1412 data. Constants may refer to the address of other constants, in which case the
1413 address will have the `static` lifetime. The compiler is, however, still at
1414 liberty to translate the constant many times, so the address referred to may not
1417 Constants must be explicitly typed. The type may be `bool`, `char`, a number, or
1418 a type derived from those primitive types. The derived types are references with
1419 the `static` lifetime, fixed-size arrays, tuples, enum variants, and structs.
1422 const BIT1: u32 = 1 << 0;
1423 const BIT2: u32 = 1 << 1;
1425 const BITS: [u32; 2] = [BIT1, BIT2];
1426 const STRING: &'static str = "bitstring";
1428 struct BitsNStrings<'a> {
1433 const BITS_N_STRINGS: BitsNStrings<'static> = BitsNStrings {
1442 static_item : "static" ident ':' type '=' expr ';' ;
1445 A *static item* is similar to a *constant*, except that it represents a precise
1446 memory location in the program. A static is never "inlined" at the usage site,
1447 and all references to it refer to the same memory location. Static items have
1448 the `static` lifetime, which outlives all other lifetimes in a Rust program.
1449 Static items may be placed in read-only memory if they do not contain any
1450 interior mutability.
1452 Statics may contain interior mutability through the `UnsafeCell` language item.
1453 All access to a static is safe, but there are a number of restrictions on
1456 * Statics may not contain any destructors.
1457 * The types of static values must ascribe to `Sync` to allow threadsafe access.
1458 * Statics may not refer to other statics by value, only by reference.
1459 * Constants cannot refer to statics.
1461 Constants should in general be preferred over statics, unless large amounts of
1462 data are being stored, or single-address and mutability properties are required.
1464 #### Mutable statics
1466 If a static item is declared with the `mut` keyword, then it is allowed to
1467 be modified by the program. One of Rust's goals is to make concurrency bugs
1468 hard to run into, and this is obviously a very large source of race conditions
1469 or other bugs. For this reason, an `unsafe` block is required when either
1470 reading or writing a mutable static variable. Care should be taken to ensure
1471 that modifications to a mutable static are safe with respect to other threads
1472 running in the same process.
1474 Mutable statics are still very useful, however. They can be used with C
1475 libraries and can also be bound from C libraries (in an `extern` block).
1478 # fn atomic_add(_: &mut u32, _: u32) -> u32 { 2 }
1480 static mut LEVELS: u32 = 0;
1482 // This violates the idea of no shared state, and this doesn't internally
1483 // protect against races, so this function is `unsafe`
1484 unsafe fn bump_levels_unsafe1() -> u32 {
1490 // Assuming that we have an atomic_add function which returns the old value,
1491 // this function is "safe" but the meaning of the return value may not be what
1492 // callers expect, so it's still marked as `unsafe`
1493 unsafe fn bump_levels_unsafe2() -> u32 {
1494 return atomic_add(&mut LEVELS, 1);
1498 Mutable statics have the same restrictions as normal statics, except that the
1499 type of the value is not required to ascribe to `Sync`.
1503 A _trait_ describes a set of method types.
1505 Traits can include default implementations of methods, written in terms of some
1506 unknown [`self` type](#self-types); the `self` type may either be completely
1507 unspecified, or constrained by some other trait.
1509 Traits are implemented for specific types through separate
1510 [implementations](#implementations).
1513 # type Surface = i32;
1514 # type BoundingBox = i32;
1516 fn draw(&self, Surface);
1517 fn bounding_box(&self) -> BoundingBox;
1521 This defines a trait with two methods. All values that have
1522 [implementations](#implementations) of this trait in scope can have their
1523 `draw` and `bounding_box` methods called, using `value.bounding_box()`
1524 [syntax](#method-call-expressions).
1526 Type parameters can be specified for a trait to make it generic. These appear
1527 after the trait name, using the same syntax used in [generic
1528 functions](#generic-functions).
1532 fn len(&self) -> u32;
1533 fn elt_at(&self, n: u32) -> T;
1534 fn iter<F>(&self, F) where F: Fn(T);
1538 Generic functions may use traits as _bounds_ on their type parameters. This
1539 will have two effects: only types that have the trait may instantiate the
1540 parameter, and within the generic function, the methods of the trait can be
1541 called on values that have the parameter's type. For example:
1544 # type Surface = i32;
1545 # trait Shape { fn draw(&self, Surface); }
1546 fn draw_twice<T: Shape>(surface: Surface, sh: T) {
1552 Traits also define an [trait object](#trait-objects) with the same name as the
1553 trait. Values of this type are created by [casting](#type-cast-expressions)
1554 pointer values (pointing to a type for which an implementation of the given
1555 trait is in scope) to pointers to the trait name, used as a type.
1558 # trait Shape { fn dummy(&self) { } }
1559 # impl Shape for i32 { }
1560 # let mycircle = 0i32;
1561 let myshape: Box<Shape> = Box::new(mycircle) as Box<Shape>;
1564 The resulting value is a box containing the value that was cast, along with
1565 information that identifies the methods of the implementation that was used.
1566 Values with a trait type can have [methods called](#method-call-expressions) on
1567 them, for any method in the trait, and can be used to instantiate type
1568 parameters that are bounded by the trait.
1570 Trait methods may be static, which means that they lack a `self` argument.
1571 This means that they can only be called with function call syntax (`f(x)`) and
1572 not method call syntax (`obj.f()`). The way to refer to the name of a static
1573 method is to qualify it with the trait name, treating the trait name like a
1574 module. For example:
1578 fn from_i32(n: i32) -> Self;
1581 fn from_i32(n: i32) -> f64 { n as f64 }
1583 let x: f64 = Num::from_i32(42);
1586 Traits may inherit from other traits. For example, in
1589 trait Shape { fn area(&self) -> f64; }
1590 trait Circle : Shape { fn radius(&self) -> f64; }
1593 the syntax `Circle : Shape` means that types that implement `Circle` must also
1594 have an implementation for `Shape`. Multiple supertraits are separated by `+`,
1595 `trait Circle : Shape + PartialEq { }`. In an implementation of `Circle` for a
1596 given type `T`, methods can refer to `Shape` methods, since the typechecker
1597 checks that any type with an implementation of `Circle` also has an
1598 implementation of `Shape`.
1600 In type-parameterized functions, methods of the supertrait may be called on
1601 values of subtrait-bound type parameters. Referring to the previous example of
1602 `trait Circle : Shape`:
1605 # trait Shape { fn area(&self) -> f64; }
1606 # trait Circle : Shape { fn radius(&self) -> f64; }
1607 fn radius_times_area<T: Circle>(c: T) -> f64 {
1608 // `c` is both a Circle and a Shape
1609 c.radius() * c.area()
1613 Likewise, supertrait methods may also be called on trait objects.
1616 # trait Shape { fn area(&self) -> f64; }
1617 # trait Circle : Shape { fn radius(&self) -> f64; }
1618 # impl Shape for i32 { fn area(&self) -> f64 { 0.0 } }
1619 # impl Circle for i32 { fn radius(&self) -> f64 { 0.0 } }
1620 # let mycircle = 0i32;
1621 let mycircle = Box::new(mycircle) as Box<Circle>;
1622 let nonsense = mycircle.radius() * mycircle.area();
1627 An _implementation_ is an item that implements a [trait](#traits) for a
1630 Implementations are defined with the keyword `impl`.
1633 # #[derive(Copy, Clone)]
1634 # struct Point {x: f64, y: f64};
1635 # type Surface = i32;
1636 # struct BoundingBox {x: f64, y: f64, width: f64, height: f64};
1637 # trait Shape { fn draw(&self, Surface); fn bounding_box(&self) -> BoundingBox; }
1638 # fn do_draw_circle(s: Surface, c: Circle) { }
1644 impl Copy for Circle {}
1646 impl Clone for Circle {
1647 fn clone(&self) -> Circle { *self }
1650 impl Shape for Circle {
1651 fn draw(&self, s: Surface) { do_draw_circle(s, *self); }
1652 fn bounding_box(&self) -> BoundingBox {
1653 let r = self.radius;
1654 BoundingBox{x: self.center.x - r, y: self.center.y - r,
1655 width: 2.0 * r, height: 2.0 * r}
1660 It is possible to define an implementation without referring to a trait. The
1661 methods in such an implementation can only be used as direct calls on the
1662 values of the type that the implementation targets. In such an implementation,
1663 the trait type and `for` after `impl` are omitted. Such implementations are
1664 limited to nominal types (enums, structs), and the implementation must appear
1665 in the same module or a sub-module as the `self` type:
1668 struct Point {x: i32, y: i32}
1672 println!("Point is at ({}, {})", self.x, self.y);
1676 let my_point = Point {x: 10, y:11};
1680 When a trait _is_ specified in an `impl`, all methods declared as part of the
1681 trait must be implemented, with matching types and type parameter counts.
1683 An implementation can take type parameters, which can be different from the
1684 type parameters taken by the trait it implements. Implementation parameters
1685 are written after the `impl` keyword.
1688 # trait Seq<T> { fn dummy(&self, _: T) { } }
1689 impl<T> Seq<T> for Vec<T> {
1692 impl Seq<bool> for u32 {
1693 /* Treat the integer as a sequence of bits */
1700 extern_block_item : "extern" '{' extern_block '}' ;
1701 extern_block : [ foreign_fn ] * ;
1704 External blocks form the basis for Rust's foreign function interface.
1705 Declarations in an external block describe symbols in external, non-Rust
1708 Functions within external blocks are declared in the same way as other Rust
1709 functions, with the exception that they may not have a body and are instead
1710 terminated by a semicolon.
1712 Functions within external blocks may be called by Rust code, just like
1713 functions defined in Rust. The Rust compiler automatically translates between
1714 the Rust ABI and the foreign ABI.
1716 A number of [attributes](#attributes) control the behavior of external blocks.
1718 By default external blocks assume that the library they are calling uses the
1719 standard C "cdecl" ABI. Other ABIs may be specified using an `abi` string, as
1723 // Interface to the Windows API
1724 extern "stdcall" { }
1727 The `link` attribute allows the name of the library to be specified. When
1728 specified the compiler will attempt to link against the native library of the
1732 #[link(name = "crypto")]
1736 The type of a function declared in an extern block is `extern "abi" fn(A1, ...,
1737 An) -> R`, where `A1...An` are the declared types of its arguments and `R` is
1738 the declared return type.
1740 ## Visibility and Privacy
1742 These two terms are often used interchangeably, and what they are attempting to
1743 convey is the answer to the question "Can this item be used at this location?"
1745 Rust's name resolution operates on a global hierarchy of namespaces. Each level
1746 in the hierarchy can be thought of as some item. The items are one of those
1747 mentioned above, but also include external crates. Declaring or defining a new
1748 module can be thought of as inserting a new tree into the hierarchy at the
1749 location of the definition.
1751 To control whether interfaces can be used across modules, Rust checks each use
1752 of an item to see whether it should be allowed or not. This is where privacy
1753 warnings are generated, or otherwise "you used a private item of another module
1754 and weren't allowed to."
1756 By default, everything in Rust is *private*, with one exception. Enum variants
1757 in a `pub` enum are also public by default. You are allowed to alter this
1758 default visibility with the `priv` keyword. When an item is declared as `pub`,
1759 it can be thought of as being accessible to the outside world. For example:
1763 // Declare a private struct
1766 // Declare a public struct with a private field
1771 // Declare a public enum with two public variants
1773 PubliclyAccessibleState,
1774 PubliclyAccessibleState2,
1778 With the notion of an item being either public or private, Rust allows item
1779 accesses in two cases:
1781 1. If an item is public, then it can be used externally through any of its
1783 2. If an item is private, it may be accessed by the current module and its
1786 These two cases are surprisingly powerful for creating module hierarchies
1787 exposing public APIs while hiding internal implementation details. To help
1788 explain, here's a few use cases and what they would entail:
1790 * A library developer needs to expose functionality to crates which link
1791 against their library. As a consequence of the first case, this means that
1792 anything which is usable externally must be `pub` from the root down to the
1793 destination item. Any private item in the chain will disallow external
1796 * A crate needs a global available "helper module" to itself, but it doesn't
1797 want to expose the helper module as a public API. To accomplish this, the
1798 root of the crate's hierarchy would have a private module which then
1799 internally has a "public api". Because the entire crate is a descendant of
1800 the root, then the entire local crate can access this private module through
1803 * When writing unit tests for a module, it's often a common idiom to have an
1804 immediate child of the module to-be-tested named `mod test`. This module
1805 could access any items of the parent module through the second case, meaning
1806 that internal implementation details could also be seamlessly tested from the
1809 In the second case, it mentions that a private item "can be accessed" by the
1810 current module and its descendants, but the exact meaning of accessing an item
1811 depends on what the item is. Accessing a module, for example, would mean
1812 looking inside of it (to import more items). On the other hand, accessing a
1813 function would mean that it is invoked. Additionally, path expressions and
1814 import statements are considered to access an item in the sense that the
1815 import/expression is only valid if the destination is in the current visibility
1818 Here's an example of a program which exemplifies the three cases outlined
1822 // This module is private, meaning that no external crate can access this
1823 // module. Because it is private at the root of this current crate, however, any
1824 // module in the crate may access any publicly visible item in this module.
1825 mod crate_helper_module {
1827 // This function can be used by anything in the current crate
1828 pub fn crate_helper() {}
1830 // This function *cannot* be used by anything else in the crate. It is not
1831 // publicly visible outside of the `crate_helper_module`, so only this
1832 // current module and its descendants may access it.
1833 fn implementation_detail() {}
1836 // This function is "public to the root" meaning that it's available to external
1837 // crates linking against this one.
1838 pub fn public_api() {}
1840 // Similarly to 'public_api', this module is public so external crates may look
1843 use crate_helper_module;
1845 pub fn my_method() {
1846 // Any item in the local crate may invoke the helper module's public
1847 // interface through a combination of the two rules above.
1848 crate_helper_module::crate_helper();
1851 // This function is hidden to any module which is not a descendant of
1853 fn my_implementation() {}
1859 fn test_my_implementation() {
1860 // Because this module is a descendant of `submodule`, it's allowed
1861 // to access private items inside of `submodule` without a privacy
1863 super::my_implementation();
1871 For a rust program to pass the privacy checking pass, all paths must be valid
1872 accesses given the two rules above. This includes all use statements,
1873 expressions, types, etc.
1875 ### Re-exporting and Visibility
1877 Rust allows publicly re-exporting items through a `pub use` directive. Because
1878 this is a public directive, this allows the item to be used in the current
1879 module through the rules above. It essentially allows public access into the
1880 re-exported item. For example, this program is valid:
1883 pub use self::implementation::api;
1885 mod implementation {
1894 This means that any external crate referencing `implementation::api::f` would
1895 receive a privacy violation, while the path `api::f` would be allowed.
1897 When re-exporting a private item, it can be thought of as allowing the "privacy
1898 chain" being short-circuited through the reexport instead of passing through
1899 the namespace hierarchy as it normally would.
1904 attribute : '#' '!' ? '[' meta_item ']' ;
1905 meta_item : ident [ '=' literal
1906 | '(' meta_seq ')' ] ? ;
1907 meta_seq : meta_item [ ',' meta_seq ] ? ;
1910 Any item declaration may have an _attribute_ applied to it. Attributes in Rust
1911 are modeled on Attributes in ECMA-335, with the syntax coming from ECMA-334
1912 (C#). An attribute is a general, free-form metadatum that is interpreted
1913 according to name, convention, and language and compiler version. Attributes
1914 may appear as any of:
1916 * A single identifier, the attribute name
1917 * An identifier followed by the equals sign '=' and a literal, providing a
1919 * An identifier followed by a parenthesized list of sub-attribute arguments
1921 Attributes with a bang ("!") after the hash ("#") apply to the item that the
1922 attribute is declared within. Attributes that do not have a bang after the hash
1923 apply to the item that follows the attribute.
1925 An example of attributes:
1928 // General metadata applied to the enclosing module or crate.
1929 #![crate_type = "lib"]
1931 // A function marked as a unit test
1937 // A conditionally-compiled module
1938 #[cfg(target_os="linux")]
1943 // A lint attribute used to suppress a warning/error
1944 #[allow(non_camel_case_types)]
1948 > **Note:** At some point in the future, the compiler will distinguish between
1949 > language-reserved and user-available attributes. Until then, there is
1950 > effectively no difference between an attribute handled by a loadable syntax
1951 > extension and the compiler.
1953 ### Crate-only attributes
1955 - `crate_name` - specify the this crate's crate name.
1956 - `crate_type` - see [linkage](#linkage).
1957 - `feature` - see [compiler features](#compiler-features).
1958 - `no_builtins` - disable optimizing certain code patterns to invocations of
1959 library functions that are assumed to exist
1960 - `no_main` - disable emitting the `main` symbol. Useful when some other
1961 object being linked to defines `main`.
1962 - `no_start` - disable linking to the `native` crate, which specifies the
1963 "start" language item.
1964 - `no_std` - disable linking to the `std` crate.
1965 - `plugin` — load a list of named crates as compiler plugins, e.g.
1966 `#![plugin(foo, bar)]`. Optional arguments for each plugin,
1967 i.e. `#![plugin(foo(... args ...))]`, are provided to the plugin's
1968 registrar function. The `plugin` feature gate is required to use
1971 ### Module-only attributes
1973 - `no_implicit_prelude` - disable injecting `use std::prelude::*` in this
1975 - `path` - specifies the file to load the module from. `#[path="foo.rs"] mod
1976 bar;` is equivalent to `mod bar { /* contents of foo.rs */ }`. The path is
1977 taken relative to the directory that the current module is in.
1979 ### Function-only attributes
1981 - `main` - indicates that this function should be passed to the entry point,
1982 rather than the function in the crate root named `main`.
1983 - `plugin_registrar` - mark this function as the registration point for
1984 [compiler plugins][plugin], such as loadable syntax extensions.
1985 - `start` - indicates that this function should be used as the entry point,
1986 overriding the "start" language item. See the "start" [language
1987 item](#language-items) for more details.
1988 - `test` - indicates that this function is a test function, to only be compiled
1989 in case of `--test`.
1990 - `should_panic` - indicates that this test function should panic, inverting the success condition.
1991 - `cold` - The function is unlikely to be executed, so optimize it (and calls
1994 ### Static-only attributes
1996 - `thread_local` - on a `static mut`, this signals that the value of this
1997 static may change depending on the current thread. The exact consequences of
1998 this are implementation-defined.
2002 On an `extern` block, the following attributes are interpreted:
2004 - `link_args` - specify arguments to the linker, rather than just the library
2005 name and type. This is feature gated and the exact behavior is
2006 implementation-defined (due to variety of linker invocation syntax).
2007 - `link` - indicate that a native library should be linked to for the
2008 declarations in this block to be linked correctly. `link` supports an optional `kind`
2009 key with three possible values: `dylib`, `static`, and `framework`. See [external blocks](#external-blocks) for more about external blocks. Two
2010 examples: `#[link(name = "readline")]` and
2011 `#[link(name = "CoreFoundation", kind = "framework")]`.
2013 On declarations inside an `extern` block, the following attributes are
2016 - `link_name` - the name of the symbol that this function or static should be
2018 - `linkage` - on a static, this specifies the [linkage
2019 type](http://llvm.org/docs/LangRef.html#linkage-types).
2023 - `repr` - on C-like enums, this sets the underlying type used for
2024 representation. Takes one argument, which is the primitive
2025 type this enum should be represented for, or `C`, which specifies that it
2026 should be the default `enum` size of the C ABI for that platform. Note that
2027 enum representation in C is undefined, and this may be incorrect when the C
2028 code is compiled with certain flags.
2032 - `repr` - specifies the representation to use for this struct. Takes a list
2033 of options. The currently accepted ones are `C` and `packed`, which may be
2034 combined. `C` will use a C ABI compatible struct layout, and `packed` will
2035 remove any padding between fields (note that this is very fragile and may
2036 break platforms which require aligned access).
2038 ### Macro-related attributes
2040 - `macro_use` on a `mod` — macros defined in this module will be visible in the
2041 module's parent, after this module has been included.
2043 - `macro_use` on an `extern crate` — load macros from this crate. An optional
2044 list of names `#[macro_use(foo, bar)]` restricts the import to just those
2045 macros named. The `extern crate` must appear at the crate root, not inside
2046 `mod`, which ensures proper function of the [`$crate` macro
2047 variable](book/macros.html#the-variable-$crate).
2049 - `macro_reexport` on an `extern crate` — re-export the named macros.
2051 - `macro_export` - export a macro for cross-crate usage.
2053 - `no_link` on an `extern crate` — even if we load this crate for macros, don't
2054 link it into the output.
2056 See the [macros section of the
2057 book](book/macros.html#scoping-and-macro-import/export) for more information on
2061 ### Miscellaneous attributes
2063 - `export_name` - on statics and functions, this determines the name of the
2065 - `link_section` - on statics and functions, this specifies the section of the
2066 object file that this item's contents will be placed into.
2067 - `no_mangle` - on any item, do not apply the standard name mangling. Set the
2068 symbol for this item to its identifier.
2069 - `packed` - on structs or enums, eliminate any padding that would be used to
2071 - `simd` - on certain tuple structs, derive the arithmetic operators, which
2072 lower to the target's SIMD instructions, if any; the `simd` feature gate
2073 is necessary to use this attribute.
2074 - `static_assert` - on statics whose type is `bool`, terminates compilation
2075 with an error if it is not initialized to `true`.
2076 - `unsafe_destructor` - allow implementations of the "drop" language item
2077 where the type it is implemented for does not implement the "send" language
2078 item; the `unsafe_destructor` feature gate is needed to use this attribute
2079 - `unsafe_no_drop_flag` - on structs, remove the flag that prevents
2080 destructors from being run twice. Destructors might be run multiple times on
2081 the same object with this attribute.
2082 - `doc` - Doc comments such as `/// foo` are equivalent to `#[doc = "foo"]`.
2083 - `rustc_on_unimplemented` - Write a custom note to be shown along with the error
2084 when the trait is found to be unimplemented on a type.
2085 You may use format arguments like `{T}`, `{A}` to correspond to the
2086 types at the point of use corresponding to the type parameters of the
2087 trait of the same name. `{Self}` will be replaced with the type that is supposed
2088 to implement the trait but doesn't. To use this, the `on_unimplemented` feature gate
2091 ### Conditional compilation
2093 Sometimes one wants to have different compiler outputs from the same code,
2094 depending on build target, such as targeted operating system, or to enable
2097 There are two kinds of configuration options, one that is either defined or not
2098 (`#[cfg(foo)]`), and the other that contains a string that can be checked
2099 against (`#[cfg(bar = "baz")]` (currently only compiler-defined configuration
2100 options can have the latter form).
2103 // The function is only included in the build when compiling for OSX
2104 #[cfg(target_os = "macos")]
2109 // This function is only included when either foo or bar is defined
2110 #[cfg(any(foo, bar))]
2111 fn needs_foo_or_bar() {
2115 // This function is only included when compiling for a unixish OS with a 32-bit
2117 #[cfg(all(unix, target_pointer_width = "32"))]
2118 fn on_32bit_unix() {
2122 // This function is only included when foo is not defined
2124 fn needs_not_foo() {
2129 This illustrates some conditional compilation can be achieved using the
2130 `#[cfg(...)]` attribute. `any`, `all` and `not` can be used to assemble
2131 arbitrarily complex configurations through nesting.
2133 The following configurations must be defined by the implementation:
2135 * `target_arch = "..."`. Target CPU architecture, such as `"x86"`, `"x86_64"`
2136 `"mips"`, `"powerpc"`, `"arm"`, or `"aarch64"`.
2137 * `target_endian = "..."`. Endianness of the target CPU, either `"little"` or
2139 * `target_family = "..."`. Operating system family of the target, e. g.
2140 `"unix"` or `"windows"`. The value of this configuration option is defined
2141 as a configuration itself, like `unix` or `windows`.
2142 * `target_os = "..."`. Operating system of the target, examples include
2143 `"windows"`, `"macos"`, `"ios"`, `"linux"`, `"android"`, `"freebsd"`, `"dragonfly"`,
2144 `"bitrig"` or `"openbsd"`.
2145 * `target_pointer_width = "..."`. Target pointer width in bits. This is set
2146 to `"32"` for targets with 32-bit pointers, and likewise set to `"64"` for
2148 * `unix`. See `target_family`.
2149 * `windows`. See `target_family`.
2151 You can also set another attribute based on a `cfg` variable with `cfg_attr`:
2157 Will be the same as `#[b]` if `a` is set by `cfg`, and nothing otherwise.
2159 ### Lint check attributes
2161 A lint check names a potentially undesirable coding pattern, such as
2162 unreachable code or omitted documentation, for the static entity to which the
2165 For any lint check `C`:
2167 * `allow(C)` overrides the check for `C` so that violations will go
2169 * `deny(C)` signals an error after encountering a violation of `C`,
2170 * `forbid(C)` is the same as `deny(C)`, but also forbids changing the lint
2172 * `warn(C)` warns about violations of `C` but continues compilation.
2174 The lint checks supported by the compiler can be found via `rustc -W help`,
2175 along with their default settings. [Compiler
2176 plugins](book/plugins.html#lint-plugins) can provide additional lint checks.
2180 // Missing documentation is ignored here
2181 #[allow(missing_docs)]
2182 pub fn undocumented_one() -> i32 { 1 }
2184 // Missing documentation signals a warning here
2185 #[warn(missing_docs)]
2186 pub fn undocumented_too() -> i32 { 2 }
2188 // Missing documentation signals an error here
2189 #[deny(missing_docs)]
2190 pub fn undocumented_end() -> i32 { 3 }
2194 This example shows how one can use `allow` and `warn` to toggle a particular
2198 #[warn(missing_docs)]
2200 #[allow(missing_docs)]
2202 // Missing documentation is ignored here
2203 pub fn undocumented_one() -> i32 { 1 }
2205 // Missing documentation signals a warning here,
2206 // despite the allow above.
2207 #[warn(missing_docs)]
2208 pub fn undocumented_two() -> i32 { 2 }
2211 // Missing documentation signals a warning here
2212 pub fn undocumented_too() -> i32 { 3 }
2216 This example shows how one can use `forbid` to disallow uses of `allow` for
2220 #[forbid(missing_docs)]
2222 // Attempting to toggle warning signals an error here
2223 #[allow(missing_docs)]
2225 pub fn undocumented_too() -> i32 { 2 }
2231 Some primitive Rust operations are defined in Rust code, rather than being
2232 implemented directly in C or assembly language. The definitions of these
2233 operations have to be easy for the compiler to find. The `lang` attribute
2234 makes it possible to declare these operations. For example, the `str` module
2235 in the Rust standard library defines the string equality function:
2239 pub fn eq_slice(a: &str, b: &str) -> bool {
2244 The name `str_eq` has a special meaning to the Rust compiler, and the presence
2245 of this definition means that it will use this definition when generating calls
2246 to the string equality function.
2248 A complete list of the built-in language items will be added in the future.
2250 ### Inline attributes
2252 The inline attribute is used to suggest to the compiler to perform an inline
2253 expansion and place a copy of the function or static in the caller rather than
2254 generating code to call the function or access the static where it is defined.
2256 The compiler automatically inlines functions based on internal heuristics.
2257 Incorrectly inlining functions can actually making the program slower, so it
2258 should be used with care.
2260 Immutable statics are always considered inlineable unless marked with
2261 `#[inline(never)]`. It is undefined whether two different inlineable statics
2262 have the same memory address. In other words, the compiler is free to collapse
2263 duplicate inlineable statics together.
2265 `#[inline]` and `#[inline(always)]` always causes the function to be serialized
2266 into crate metadata to allow cross-crate inlining.
2268 There are three different types of inline attributes:
2270 * `#[inline]` hints the compiler to perform an inline expansion.
2271 * `#[inline(always)]` asks the compiler to always perform an inline expansion.
2272 * `#[inline(never)]` asks the compiler to never perform an inline expansion.
2276 The `derive` attribute allows certain traits to be automatically implemented
2277 for data structures. For example, the following will create an `impl` for the
2278 `PartialEq` and `Clone` traits for `Foo`, the type parameter `T` will be given
2279 the `PartialEq` or `Clone` constraints for the appropriate `impl`:
2282 #[derive(PartialEq, Clone)]
2289 The generated `impl` for `PartialEq` is equivalent to
2292 # struct Foo<T> { a: i32, b: T }
2293 impl<T: PartialEq> PartialEq for Foo<T> {
2294 fn eq(&self, other: &Foo<T>) -> bool {
2295 self.a == other.a && self.b == other.b
2298 fn ne(&self, other: &Foo<T>) -> bool {
2299 self.a != other.a || self.b != other.b
2304 ### Compiler Features
2306 Certain aspects of Rust may be implemented in the compiler, but they're not
2307 necessarily ready for every-day use. These features are often of "prototype
2308 quality" or "almost production ready", but may not be stable enough to be
2309 considered a full-fledged language feature.
2311 For this reason, Rust recognizes a special crate-level attribute of the form:
2314 #![feature(feature1, feature2, feature3)]
2317 This directive informs the compiler that the feature list: `feature1`,
2318 `feature2`, and `feature3` should all be enabled. This is only recognized at a
2319 crate-level, not at a module-level. Without this directive, all features are
2320 considered off, and using the features will result in a compiler error.
2322 The currently implemented features of the reference compiler are:
2324 * `advanced_slice_patterns` - See the [match expressions](#match-expressions)
2325 section for discussion; the exact semantics of
2326 slice patterns are subject to change, so some types
2329 * `slice_patterns` - OK, actually, slice patterns are just scary and
2330 completely unstable.
2332 * `asm` - The `asm!` macro provides a means for inline assembly. This is often
2333 useful, but the exact syntax for this feature along with its
2334 semantics are likely to change, so this macro usage must be opted
2337 * `associated_types` - Allows type aliases in traits. Experimental.
2339 * `box_patterns` - Allows `box` patterns, the exact semantics of which
2340 is subject to change.
2342 * `box_syntax` - Allows use of `box` expressions, the exact semantics of which
2343 is subject to change.
2345 * `concat_idents` - Allows use of the `concat_idents` macro, which is in many
2346 ways insufficient for concatenating identifiers, and may be
2347 removed entirely for something more wholesome.
2349 * `custom_attribute` - Allows the usage of attributes unknown to the compiler
2350 so that new attributes can be added in a backwards compatible
2353 * `custom_derive` - Allows the use of `#[derive(Foo,Bar)]` as sugar for
2354 `#[derive_Foo] #[derive_Bar]`, which can be user-defined syntax
2357 * `intrinsics` - Allows use of the "rust-intrinsics" ABI. Compiler intrinsics
2358 are inherently unstable and no promise about them is made.
2360 * `lang_items` - Allows use of the `#[lang]` attribute. Like `intrinsics`,
2361 lang items are inherently unstable and no promise about them
2364 * `link_args` - This attribute is used to specify custom flags to the linker,
2365 but usage is strongly discouraged. The compiler's usage of the
2366 system linker is not guaranteed to continue in the future, and
2367 if the system linker is not used then specifying custom flags
2368 doesn't have much meaning.
2370 * `link_llvm_intrinsics` – Allows linking to LLVM intrinsics via
2371 `#[link_name="llvm.*"]`.
2373 * `linkage` - Allows use of the `linkage` attribute, which is not portable.
2375 * `log_syntax` - Allows use of the `log_syntax` macro attribute, which is a
2376 nasty hack that will certainly be removed.
2378 * `main` - Allows use of the `#[main]` attribute, which changes the entry point
2379 into a Rust program. This capability is subject to change.
2381 * `macro_reexport` - Allows macros to be re-exported from one crate after being imported
2382 from another. This feature was originally designed with the sole
2383 use case of the Rust standard library in mind, and is subject to
2386 * `non_ascii_idents` - The compiler supports the use of non-ascii identifiers,
2387 but the implementation is a little rough around the
2388 edges, so this can be seen as an experimental feature
2389 for now until the specification of identifiers is fully
2392 * `no_std` - Allows the `#![no_std]` crate attribute, which disables the implicit
2393 `extern crate std`. This typically requires use of the unstable APIs
2394 behind the libstd "facade", such as libcore and libcollections. It
2395 may also cause problems when using syntax extensions, including
2398 * `on_unimplemented` - Allows the `#[rustc_on_unimplemented]` attribute, which allows
2399 trait definitions to add specialized notes to error messages
2400 when an implementation was expected but not found.
2402 * `optin_builtin_traits` - Allows the definition of default and negative trait
2403 implementations. Experimental.
2405 * `plugin` - Usage of [compiler plugins][plugin] for custom lints or syntax extensions.
2406 These depend on compiler internals and are subject to change.
2408 * `plugin_registrar` - Indicates that a crate provides [compiler plugins][plugin].
2410 * `quote` - Allows use of the `quote_*!` family of macros, which are
2411 implemented very poorly and will likely change significantly
2412 with a proper implementation.
2414 * `rustc_attrs` - Gates internal `#[rustc_*]` attributes which may be
2415 for internal use only or have meaning added to them in the future.
2417 * `rustc_diagnostic_macros`- A mysterious feature, used in the implementation
2418 of rustc, not meant for mortals.
2420 * `simd` - Allows use of the `#[simd]` attribute, which is overly simple and
2421 not the SIMD interface we want to expose in the long term.
2423 * `simd_ffi` - Allows use of SIMD vectors in signatures for foreign functions.
2424 The SIMD interface is subject to change.
2426 * `staged_api` - Allows usage of stability markers and `#![staged_api]` in a
2427 crate. Stability markers are also attributes: `#[stable]`,
2428 `#[unstable]`, and `#[deprecated]` are the three levels.
2430 * `static_assert` - The `#[static_assert]` functionality is experimental and
2431 unstable. The attribute can be attached to a `static` of
2432 type `bool` and the compiler will error if the `bool` is
2433 `false` at compile time. This version of this functionality
2434 is unintuitive and suboptimal.
2436 * `start` - Allows use of the `#[start]` attribute, which changes the entry point
2437 into a Rust program. This capability, especially the signature for the
2438 annotated function, is subject to change.
2440 * `struct_inherit` - Allows using struct inheritance, which is barely
2441 implemented and will probably be removed. Don't use this.
2443 * `struct_variant` - Structural enum variants (those with named fields). It is
2444 currently unknown whether this style of enum variant is as
2445 fully supported as the tuple-forms, and it's not certain
2446 that this style of variant should remain in the language.
2447 For now this style of variant is hidden behind a feature
2450 * `thread_local` - The usage of the `#[thread_local]` attribute is experimental
2451 and should be seen as unstable. This attribute is used to
2452 declare a `static` as being unique per-thread leveraging
2453 LLVM's implementation which works in concert with the kernel
2454 loader and dynamic linker. This is not necessarily available
2455 on all platforms, and usage of it is discouraged.
2457 * `trace_macros` - Allows use of the `trace_macros` macro, which is a nasty
2458 hack that will certainly be removed.
2460 * `unboxed_closures` - Rust's new closure design, which is currently a work in
2461 progress feature with many known bugs.
2463 * `unsafe_destructor` - Allows use of the `#[unsafe_destructor]` attribute,
2464 which is considered wildly unsafe and will be
2465 obsoleted by language improvements.
2467 * `unsafe_no_drop_flag` - Allows use of the `#[unsafe_no_drop_flag]` attribute,
2468 which removes hidden flag added to a type that
2469 implements the `Drop` trait. The design for the
2470 `Drop` flag is subject to change, and this feature
2471 may be removed in the future.
2473 * `unmarked_api` - Allows use of items within a `#![staged_api]` crate
2474 which have not been marked with a stability marker.
2475 Such items should not be allowed by the compiler to exist,
2476 so if you need this there probably is a compiler bug.
2478 * `visible_private_types` - Allows public APIs to expose otherwise private
2479 types, e.g. as the return type of a public function.
2480 This capability may be removed in the future.
2482 * `allow_internal_unstable` - Allows `macro_rules!` macros to be tagged with the
2483 `#[allow_internal_unstable]` attribute, designed
2484 to allow `std` macros to call
2485 `#[unstable]`/feature-gated functionality
2486 internally without imposing on callers
2487 (i.e. making them behave like function calls in
2488 terms of encapsulation).
2490 If a feature is promoted to a language feature, then all existing programs will
2491 start to receive compilation warnings about #[feature] directives which enabled
2492 the new feature (because the directive is no longer necessary). However, if a
2493 feature is decided to be removed from the language, errors will be issued (if
2494 there isn't a parser error first). The directive in this case is no longer
2495 necessary, and it's likely that existing code will break if the feature isn't
2498 If an unknown feature is found in a directive, it results in a compiler error.
2499 An unknown feature is one which has never been recognized by the compiler.
2501 # Statements and expressions
2503 Rust is _primarily_ an expression language. This means that most forms of
2504 value-producing or effect-causing evaluation are directed by the uniform syntax
2505 category of _expressions_. Each kind of expression can typically _nest_ within
2506 each other kind of expression, and rules for evaluation of expressions involve
2507 specifying both the value produced by the expression and the order in which its
2508 sub-expressions are themselves evaluated.
2510 In contrast, statements in Rust serve _mostly_ to contain and explicitly
2511 sequence expression evaluation.
2515 A _statement_ is a component of a block, which is in turn a component of an
2516 outer [expression](#expressions) or [function](#functions).
2518 Rust has two kinds of statement: [declaration
2519 statements](#declaration-statements) and [expression
2520 statements](#expression-statements).
2522 ### Declaration statements
2524 A _declaration statement_ is one that introduces one or more *names* into the
2525 enclosing statement block. The declared names may denote new variables or new
2528 #### Item declarations
2530 An _item declaration statement_ has a syntactic form identical to an
2531 [item](#items) declaration within a module. Declaring an item — a
2532 function, enumeration, structure, type, static, trait, implementation or module
2533 — locally within a statement block is simply a way of restricting its
2534 scope to a narrow region containing all of its uses; it is otherwise identical
2535 in meaning to declaring the item outside the statement block.
2537 > **Note**: there is no implicit capture of the function's dynamic environment when
2538 > declaring a function-local item.
2540 #### Variable declarations
2543 let_decl : "let" pat [':' type ] ? [ init ] ? ';' ;
2544 init : [ '=' ] expr ;
2547 A _variable declaration_ introduces a new set of variable, given by a pattern. The
2548 pattern may be followed by a type annotation, and/or an initializer expression.
2549 When no type annotation is given, the compiler will infer the type, or signal
2550 an error if insufficient type information is available for definite inference.
2551 Any variables introduced by a variable declaration are visible from the point of
2552 declaration until the end of the enclosing block scope.
2554 ### Expression statements
2556 An _expression statement_ is one that evaluates an [expression](#expressions)
2557 and ignores its result. The type of an expression statement `e;` is always
2558 `()`, regardless of the type of `e`. As a rule, an expression statement's
2559 purpose is to trigger the effects of evaluating its expression.
2563 An expression may have two roles: it always produces a *value*, and it may have
2564 *effects* (otherwise known as "side effects"). An expression *evaluates to* a
2565 value, and has effects during *evaluation*. Many expressions contain
2566 sub-expressions (operands). The meaning of each kind of expression dictates
2569 * Whether or not to evaluate the sub-expressions when evaluating the expression
2570 * The order in which to evaluate the sub-expressions
2571 * How to combine the sub-expressions' values to obtain the value of the expression
2573 In this way, the structure of expressions dictates the structure of execution.
2574 Blocks are just another kind of expression, so blocks, statements, expressions,
2575 and blocks again can recursively nest inside each other to an arbitrary depth.
2577 #### Lvalues, rvalues and temporaries
2579 Expressions are divided into two main categories: _lvalues_ and _rvalues_.
2580 Likewise within each expression, sub-expressions may occur in _lvalue context_
2581 or _rvalue context_. The evaluation of an expression depends both on its own
2582 category and the context it occurs within.
2584 An lvalue is an expression that represents a memory location. These expressions
2585 are [paths](#path-expressions) (which refer to local variables, function and
2586 method arguments, or static variables), dereferences (`*expr`), [indexing
2587 expressions](#index-expressions) (`expr[expr]`), and [field
2588 references](#field-expressions) (`expr.f`). All other expressions are rvalues.
2590 The left operand of an [assignment](#assignment-expressions) or
2591 [compound-assignment](#compound-assignment-expressions) expression is an lvalue
2592 context, as is the single operand of a unary
2593 [borrow](#unary-operator-expressions). All other expression contexts are
2596 When an lvalue is evaluated in an _lvalue context_, it denotes a memory
2597 location; when evaluated in an _rvalue context_, it denotes the value held _in_
2598 that memory location.
2600 When an rvalue is used in an lvalue context, a temporary un-named lvalue is
2601 created and used instead. A temporary's lifetime equals the largest lifetime
2602 of any reference that points to it.
2604 #### Moved and copied types
2606 When a [local variable](#variables) is used as an
2607 [rvalue](#lvalues,-rvalues-and-temporaries) the variable will either be moved
2608 or copied, depending on its type. All values whose type implements `Copy` are
2609 copied, all others are moved.
2611 ### Literal expressions
2613 A _literal expression_ consists of one of the [literal](#literals) forms
2614 described earlier. It directly describes a number, character, string, boolean
2615 value, or the unit value.
2619 "hello"; // string type
2620 '5'; // character type
2624 ### Path expressions
2626 A [path](#paths) used as an expression context denotes either a local variable
2627 or an item. Path expressions are [lvalues](#lvalues,-rvalues-and-temporaries).
2629 ### Tuple expressions
2631 Tuples are written by enclosing zero or more comma-separated expressions in
2632 parentheses. They are used to create [tuple-typed](#tuple-types) values.
2640 ### Unit expressions
2642 The expression `()` denotes the _unit value_, the only value of the type with
2645 ### Structure expressions
2648 struct_expr : expr_path '{' ident ':' expr
2649 [ ',' ident ':' expr ] *
2652 [ ',' expr ] * ')' |
2656 There are several forms of structure expressions. A _structure expression_
2657 consists of the [path](#paths) of a [structure item](#structures), followed by
2658 a brace-enclosed list of one or more comma-separated name-value pairs,
2659 providing the field values of a new instance of the structure. A field name
2660 can be any identifier, and is separated from its value expression by a colon.
2661 The location denoted by a structure field is mutable if and only if the
2662 enclosing structure is mutable.
2664 A _tuple structure expression_ consists of the [path](#paths) of a [structure
2665 item](#structures), followed by a parenthesized list of one or more
2666 comma-separated expressions (in other words, the path of a structure item
2667 followed by a tuple expression). The structure item must be a tuple structure
2670 A _unit-like structure expression_ consists only of the [path](#paths) of a
2671 [structure item](#structures).
2673 The following are examples of structure expressions:
2676 # struct Point { x: f64, y: f64 }
2677 # struct TuplePoint(f64, f64);
2678 # mod game { pub struct User<'a> { pub name: &'a str, pub age: u32, pub score: usize } }
2679 # struct Cookie; fn some_fn<T>(t: T) {}
2680 Point {x: 10.0, y: 20.0};
2681 TuplePoint(10.0, 20.0);
2682 let u = game::User {name: "Joe", age: 35, score: 100_000};
2683 some_fn::<Cookie>(Cookie);
2686 A structure expression forms a new value of the named structure type. Note
2687 that for a given *unit-like* structure type, this will always be the same
2690 A structure expression can terminate with the syntax `..` followed by an
2691 expression to denote a functional update. The expression following `..` (the
2692 base) must have the same structure type as the new structure type being formed.
2693 The entire expression denotes the result of constructing a new structure (with
2694 the same type as the base expression) with the given values for the fields that
2695 were explicitly specified and the values in the base expression for all other
2699 # struct Point3d { x: i32, y: i32, z: i32 }
2700 let base = Point3d {x: 1, y: 2, z: 3};
2701 Point3d {y: 0, z: 10, .. base};
2704 ### Block expressions
2707 block_expr : '{' [ stmt ';' | item ] *
2711 A _block expression_ is similar to a module in terms of the declarations that
2712 are possible. Each block conceptually introduces a new namespace scope. Use
2713 items can bring new names into scopes and declared items are in scope for only
2716 A block will execute each statement sequentially, and then execute the
2717 expression (if given). If the block ends in a statement, its value is `()`:
2720 let x: () = { println!("Hello."); };
2723 If it ends in an expression, its value and type are that of the expression:
2726 let x: i32 = { println!("Hello."); 5 };
2731 ### Method-call expressions
2734 method_call_expr : expr '.' ident paren_expr_list ;
2737 A _method call_ consists of an expression followed by a single dot, an
2738 identifier, and a parenthesized expression-list. Method calls are resolved to
2739 methods on specific traits, either statically dispatching to a method if the
2740 exact `self`-type of the left-hand-side is known, or dynamically dispatching if
2741 the left-hand-side expression is an indirect [trait object](#trait-objects).
2743 ### Field expressions
2746 field_expr : expr '.' ident ;
2749 A _field expression_ consists of an expression followed by a single dot and an
2750 identifier, when not immediately followed by a parenthesized expression-list
2751 (the latter is a [method call expression](#method-call-expressions)). A field
2752 expression denotes a field of a [structure](#structure-types).
2757 (Struct {a: 10, b: 20}).a;
2760 A field access is an [lvalue](#lvalues,-rvalues-and-temporaries) referring to
2761 the value of that field. When the type providing the field inherits mutability,
2762 it can be [assigned](#assignment-expressions) to.
2764 Also, if the type of the expression to the left of the dot is a pointer, it is
2765 automatically dereferenced to make the field access possible.
2767 ### Array expressions
2770 array_expr : '[' "mut" ? array_elems? ']' ;
2772 array_elems : [expr [',' expr]*] | [expr ';' expr] ;
2775 An [array](#array,-and-slice-types) _expression_ is written by enclosing zero
2776 or more comma-separated expressions of uniform type in square brackets.
2778 In the `[expr ';' expr]` form, the expression after the `';'` must be a
2779 constant expression that can be evaluated at compile time, such as a
2780 [literal](#literals) or a [static item](#static-items).
2784 ["a", "b", "c", "d"];
2785 [0; 128]; // array with 128 zeros
2786 [0u8, 0u8, 0u8, 0u8];
2789 ### Index expressions
2792 idx_expr : expr '[' expr ']' ;
2795 [Array](#array,-and-slice-types)-typed expressions can be indexed by
2796 writing a square-bracket-enclosed expression (the index) after them. When the
2797 array is mutable, the resulting [lvalue](#lvalues,-rvalues-and-temporaries) can
2800 Indices are zero-based, and may be of any integral type. Vector access is
2801 bounds-checked at run-time. When the check fails, it will put the thread in a
2806 (["a", "b"])[10]; // panics
2809 ### Unary operator expressions
2811 Rust defines three unary operators. They are all written as prefix operators,
2812 before the expression they apply to.
2815 : Negation. May only be applied to numeric types.
2817 : Dereference. When applied to a [pointer](#pointer-types) it denotes the
2818 pointed-to location. For pointers to mutable locations, the resulting
2819 [lvalue](#lvalues,-rvalues-and-temporaries) can be assigned to.
2820 On non-pointer types, it calls the `deref` method of the `std::ops::Deref`
2821 trait, or the `deref_mut` method of the `std::ops::DerefMut` trait (if
2822 implemented by the type and required for an outer expression that will or
2823 could mutate the dereference), and produces the result of dereferencing the
2824 `&` or `&mut` borrowed pointer returned from the overload method.
2827 : Logical negation. On the boolean type, this flips between `true` and
2828 `false`. On integer types, this inverts the individual bits in the
2829 two's complement representation of the value.
2831 ### Binary operator expressions
2834 binop_expr : expr binop expr ;
2837 Binary operators expressions are given in terms of [operator
2838 precedence](#operator-precedence).
2840 #### Arithmetic operators
2842 Binary arithmetic expressions are syntactic sugar for calls to built-in traits,
2843 defined in the `std::ops` module of the `std` library. This means that
2844 arithmetic operators can be overridden for user-defined types. The default
2845 meaning of the operators on standard types is given here.
2848 : Addition and array/string concatenation.
2849 Calls the `add` method on the `std::ops::Add` trait.
2852 Calls the `sub` method on the `std::ops::Sub` trait.
2855 Calls the `mul` method on the `std::ops::Mul` trait.
2858 Calls the `div` method on the `std::ops::Div` trait.
2861 Calls the `rem` method on the `std::ops::Rem` trait.
2863 #### Bitwise operators
2865 Like the [arithmetic operators](#arithmetic-operators), bitwise operators are
2866 syntactic sugar for calls to methods of built-in traits. This means that
2867 bitwise operators can be overridden for user-defined types. The default
2868 meaning of the operators on standard types is given here.
2872 Calls the `bitand` method of the `std::ops::BitAnd` trait.
2875 Calls the `bitor` method of the `std::ops::BitOr` trait.
2878 Calls the `bitxor` method of the `std::ops::BitXor` trait.
2881 Calls the `shl` method of the `std::ops::Shl` trait.
2884 Calls the `shr` method of the `std::ops::Shr` trait.
2886 #### Lazy boolean operators
2888 The operators `||` and `&&` may be applied to operands of boolean type. The
2889 `||` operator denotes logical 'or', and the `&&` operator denotes logical
2890 'and'. They differ from `|` and `&` in that the right-hand operand is only
2891 evaluated when the left-hand operand does not already determine the result of
2892 the expression. That is, `||` only evaluates its right-hand operand when the
2893 left-hand operand evaluates to `false`, and `&&` only when it evaluates to
2896 #### Comparison operators
2898 Comparison operators are, like the [arithmetic
2899 operators](#arithmetic-operators), and [bitwise operators](#bitwise-operators),
2900 syntactic sugar for calls to built-in traits. This means that comparison
2901 operators can be overridden for user-defined types. The default meaning of the
2902 operators on standard types is given here.
2906 Calls the `eq` method on the `std::cmp::PartialEq` trait.
2909 Calls the `ne` method on the `std::cmp::PartialEq` trait.
2912 Calls the `lt` method on the `std::cmp::PartialOrd` trait.
2915 Calls the `gt` method on the `std::cmp::PartialOrd` trait.
2917 : Less than or equal.
2918 Calls the `le` method on the `std::cmp::PartialOrd` trait.
2920 : Greater than or equal.
2921 Calls the `ge` method on the `std::cmp::PartialOrd` trait.
2923 #### Type cast expressions
2925 A type cast expression is denoted with the binary operator `as`.
2927 Executing an `as` expression casts the value on the left-hand side to the type
2928 on the right-hand side.
2930 An example of an `as` expression:
2933 # fn sum(v: &[f64]) -> f64 { 0.0 }
2934 # fn len(v: &[f64]) -> i32 { 0 }
2936 fn avg(v: &[f64]) -> f64 {
2937 let sum: f64 = sum(v);
2938 let sz: f64 = len(v) as f64;
2943 #### Assignment expressions
2945 An _assignment expression_ consists of an
2946 [lvalue](#lvalues,-rvalues-and-temporaries) expression followed by an equals
2947 sign (`=`) and an [rvalue](#lvalues,-rvalues-and-temporaries) expression.
2949 Evaluating an assignment expression [either copies or
2950 moves](#moved-and-copied-types) its right-hand operand to its left-hand
2960 #### Compound assignment expressions
2962 The `+`, `-`, `*`, `/`, `%`, `&`, `|`, `^`, `<<`, and `>>` operators may be
2963 composed with the `=` operator. The expression `lval OP= val` is equivalent to
2964 `lval = lval OP val`. For example, `x = x + 1` may be written as `x += 1`.
2966 Any such expression always has the [`unit`](#primitive-types) type.
2968 #### Operator precedence
2970 The precedence of Rust binary operators is ordered as follows, going from
2973 ```{.text .precedence}
2987 Operators at the same precedence level are evaluated left-to-right. [Unary
2988 operators](#unary-operator-expressions) have the same precedence level and are
2989 stronger than any of the binary operators.
2991 ### Grouped expressions
2993 An expression enclosed in parentheses evaluates to the result of the enclosed
2994 expression. Parentheses can be used to explicitly specify evaluation order
2995 within an expression.
2998 paren_expr : '(' expr ')' ;
3001 An example of a parenthesized expression:
3004 let x: i32 = (2 + 3) * 4;
3008 ### Call expressions
3011 expr_list : [ expr [ ',' expr ]* ] ? ;
3012 paren_expr_list : '(' expr_list ')' ;
3013 call_expr : expr paren_expr_list ;
3016 A _call expression_ invokes a function, providing zero or more input variables
3017 and an optional location to move the function's output into. If the function
3018 eventually returns, then the expression completes.
3020 Some examples of call expressions:
3023 # fn add(x: i32, y: i32) -> i32 { 0 }
3025 let x: i32 = add(1i32, 2i32);
3026 let pi: Result<f32, _> = "3.14".parse();
3029 ### Lambda expressions
3032 ident_list : [ ident [ ',' ident ]* ] ? ;
3033 lambda_expr : '|' ident_list '|' expr ;
3036 A _lambda expression_ (sometimes called an "anonymous function expression")
3037 defines a function and denotes it as a value, in a single expression. A lambda
3038 expression is a pipe-symbol-delimited (`|`) list of identifiers followed by an
3041 A lambda expression denotes a function that maps a list of parameters
3042 (`ident_list`) onto the expression that follows the `ident_list`. The
3043 identifiers in the `ident_list` are the parameters to the function. These
3044 parameters' types need not be specified, as the compiler infers them from
3047 Lambda expressions are most useful when passing functions as arguments to other
3048 functions, as an abbreviation for defining and capturing a separate function.
3050 Significantly, lambda expressions _capture their environment_, which regular
3051 [function definitions](#functions) do not. The exact type of capture depends
3052 on the [function type](#function-types) inferred for the lambda expression. In
3053 the simplest and least-expensive form (analogous to a ```|| { }``` expression),
3054 the lambda expression captures its environment by reference, effectively
3055 borrowing pointers to all outer variables mentioned inside the function.
3056 Alternately, the compiler may infer that a lambda expression should copy or
3057 move values (depending on their type) from the environment into the lambda
3058 expression's captured environment.
3060 In this example, we define a function `ten_times` that takes a higher-order
3061 function argument, and call it with a lambda expression as an argument:
3064 fn ten_times<F>(f: F) where F: Fn(i32) {
3072 ten_times(|j| println!("hello, {}", j));
3077 A `loop` expression denotes an infinite loop.
3080 loop_expr : [ lifetime ':' ] "loop" '{' block '}';
3083 A `loop` expression may optionally have a _label_. The label is written as
3084 a lifetime preceding the loop expression, as in `'foo: loop{ }`. If a
3085 label is present, then labeled `break` and `continue` expressions nested
3086 within this loop may exit out of this loop or return control to its head.
3087 See [Break expressions](#break-expressions) and [Continue
3088 expressions](#continue-expressions).
3090 ### Break expressions
3093 break_expr : "break" [ lifetime ];
3096 A `break` expression has an optional _label_. If the label is absent, then
3097 executing a `break` expression immediately terminates the innermost loop
3098 enclosing it. It is only permitted in the body of a loop. If the label is
3099 present, then `break 'foo` terminates the loop with label `'foo`, which need not
3100 be the innermost label enclosing the `break` expression, but must enclose it.
3102 ### Continue expressions
3105 continue_expr : "continue" [ lifetime ];
3108 A `continue` expression has an optional _label_. If the label is absent, then
3109 executing a `continue` expression immediately terminates the current iteration
3110 of the innermost loop enclosing it, returning control to the loop *head*. In
3111 the case of a `while` loop, the head is the conditional expression controlling
3112 the loop. In the case of a `for` loop, the head is the call-expression
3113 controlling the loop. If the label is present, then `continue 'foo` returns
3114 control to the head of the loop with label `'foo`, which need not be the
3115 innermost label enclosing the `break` expression, but must enclose it.
3117 A `continue` expression is only permitted in the body of a loop.
3122 while_expr : [ lifetime ':' ] "while" no_struct_literal_expr '{' block '}' ;
3125 A `while` loop begins by evaluating the boolean loop conditional expression.
3126 If the loop conditional expression evaluates to `true`, the loop body block
3127 executes and control returns to the loop conditional expression. If the loop
3128 conditional expression evaluates to `false`, the `while` expression completes.
3141 Like `loop` expressions, `while` loops can be controlled with `break` or
3142 `continue`, and may optionally have a _label_. See [infinite
3143 loops](#infinite-loops), [break expressions](#break-expressions), and
3144 [continue expressions](#continue-expressions) for more information.
3149 for_expr : [ lifetime ':' ] "for" pat "in" no_struct_literal_expr '{' block '}' ;
3152 A `for` expression is a syntactic construct for looping over elements provided
3153 by an implementation of `std::iter::Iterator`.
3155 An example of a for loop over the contents of an array:
3159 # fn bar(f: Foo) { }
3164 let v: &[Foo] = &[a, b, c];
3171 An example of a for loop over a series of integers:
3174 # fn bar(b:usize) { }
3180 Like `loop` expressions, `for` loops can be controlled with `break` or
3181 `continue`, and may optionally have a _label_. See [infinite
3182 loops](#infinite-loops), [break expressions](#break-expressions), and
3183 [continue expressions](#continue-expressions) for more information.
3188 if_expr : "if" no_struct_literal_expr '{' block '}'
3191 else_tail : "else" [ if_expr | if_let_expr
3195 An `if` expression is a conditional branch in program control. The form of an
3196 `if` expression is a condition expression, followed by a consequent block, any
3197 number of `else if` conditions and blocks, and an optional trailing `else`
3198 block. The condition expressions must have type `bool`. If a condition
3199 expression evaluates to `true`, the consequent block is executed and any
3200 subsequent `else if` or `else` block is skipped. If a condition expression
3201 evaluates to `false`, the consequent block is skipped and any subsequent `else
3202 if` condition is evaluated. If all `if` and `else if` conditions evaluate to
3203 `false` then any `else` block is executed.
3205 ### Match expressions
3208 match_expr : "match" no_struct_literal_expr '{' match_arm * '}' ;
3210 match_arm : attribute * match_pat "=>" [ expr "," | '{' block '}' ] ;
3212 match_pat : pat [ '|' pat ] * [ "if" expr ] ? ;
3215 A `match` expression branches on a *pattern*. The exact form of matching that
3216 occurs depends on the pattern. Patterns consist of some combination of
3217 literals, destructured arrays or enum constructors, structures and tuples,
3218 variable binding specifications, wildcards (`..`), and placeholders (`_`). A
3219 `match` expression has a *head expression*, which is the value to compare to
3220 the patterns. The type of the patterns must equal the type of the head
3223 In a pattern whose head expression has an `enum` type, a placeholder (`_`)
3224 stands for a *single* data field, whereas a wildcard `..` stands for *all* the
3225 fields of a particular variant.
3227 A `match` behaves differently depending on whether or not the head expression
3228 is an [lvalue or an rvalue](#lvalues,-rvalues-and-temporaries). If the head
3229 expression is an rvalue, it is first evaluated into a temporary location, and
3230 the resulting value is sequentially compared to the patterns in the arms until
3231 a match is found. The first arm with a matching pattern is chosen as the branch
3232 target of the `match`, any variables bound by the pattern are assigned to local
3233 variables in the arm's block, and control enters the block.
3235 When the head expression is an lvalue, the match does not allocate a temporary
3236 location (however, a by-value binding may copy or move from the lvalue). When
3237 possible, it is preferable to match on lvalues, as the lifetime of these
3238 matches inherits the lifetime of the lvalue, rather than being restricted to
3239 the inside of the match.
3241 An example of a `match` expression:
3247 1 => println!("one"),
3248 2 => println!("two"),
3249 3 => println!("three"),
3250 4 => println!("four"),
3251 5 => println!("five"),
3252 _ => println!("something else"),
3256 Patterns that bind variables default to binding to a copy or move of the
3257 matched value (depending on the matched value's type). This can be changed to
3258 bind to a reference by using the `ref` keyword, or to a mutable reference using
3261 Subpatterns can also be bound to variables by the use of the syntax `variable @
3262 subpattern`. For example:
3268 e @ 1 ... 5 => println!("got a range element {}", e),
3269 _ => println!("anything"),
3273 Patterns can also dereference pointers by using the `&`, `&mut` and `box`
3274 symbols, as appropriate. For example, these two matches on `x: &i32` are
3279 let y = match *x { 0 => "zero", _ => "some" };
3280 let z = match x { &0 => "zero", _ => "some" };
3285 A pattern that's just an identifier, like `Nil` in the previous example, could
3286 either refer to an enum variant that's in scope, or bind a new variable. The
3287 compiler resolves this ambiguity by forbidding variable bindings that occur in
3288 `match` patterns from shadowing names of variants that are in scope. For
3289 example, wherever `List` is in scope, a `match` pattern would not be able to
3290 bind `Nil` as a new name. The compiler interprets a variable pattern `x` as a
3291 binding _only_ if there is no variant named `x` in scope. A convention you can
3292 use to avoid conflicts is simply to name variants with upper-case letters, and
3293 local variables with lower-case letters.
3295 Multiple match patterns may be joined with the `|` operator. A range of values
3296 may be specified with `...`. For example:
3301 let message = match x {
3302 0 | 1 => "not many",
3308 Range patterns only work on scalar types (like integers and characters; not
3309 like arrays and structs, which have sub-components). A range pattern may not
3310 be a sub-range of another range pattern inside the same `match`.
3312 Finally, match patterns can accept *pattern guards* to further refine the
3313 criteria for matching a case. Pattern guards appear after the pattern and
3314 consist of a bool-typed expression following the `if` keyword. A pattern guard
3315 may refer to the variables bound within the pattern they follow.
3318 # let maybe_digit = Some(0);
3319 # fn process_digit(i: i32) { }
3320 # fn process_other(i: i32) { }
3322 let message = match maybe_digit {
3323 Some(x) if x < 10 => process_digit(x),
3324 Some(x) => process_other(x),
3329 ### If let expressions
3332 if_let_expr : "if" "let" pat '=' expr '{' block '}'
3334 else_tail : "else" [ if_expr | if_let_expr | '{' block '}' ] ;
3337 An `if let` expression is semantically identical to an `if` expression but in place
3338 of a condition expression it expects a refutable let statement. If the value of the
3339 expression on the right hand side of the let statement matches the pattern, the corresponding
3340 block will execute, otherwise flow proceeds to the first `else` block that follows.
3345 while_let_expr : "while" "let" pat '=' expr '{' block '}' ;
3348 A `while let` loop is semantically identical to a `while` loop but in place of a
3349 condition expression it expects a refutable let statement. If the value of the
3350 expression on the right hand side of the let statement matches the pattern, the
3351 loop body block executes and control returns to the pattern matching statement.
3352 Otherwise, the while expression completes.
3354 ### Return expressions
3357 return_expr : "return" expr ? ;
3360 Return expressions are denoted with the keyword `return`. Evaluating a `return`
3361 expression moves its argument into the designated output location for the
3362 current function call, destroys the current function activation frame, and
3363 transfers control to the caller frame.
3365 An example of a `return` expression:
3368 fn max(a: i32, b: i32) -> i32 {
3380 Every variable, item and value in a Rust program has a type. The _type_ of a
3381 *value* defines the interpretation of the memory holding it.
3383 Built-in types and type-constructors are tightly integrated into the language,
3384 in nontrivial ways that are not possible to emulate in user-defined types.
3385 User-defined types have limited capabilities.
3389 The primitive types are the following:
3391 * The "unit" type `()`, having the single "unit" value `()` (occasionally called
3393 * The boolean type `bool` with values `true` and `false`.
3394 * The machine types.
3395 * The machine-dependent integer and floating-point types.
3397 [^unittype]: The "unit" value `()` is *not* a sentinel "null pointer" value for
3398 reference variables; the "unit" type is the implicit return type from functions
3399 otherwise lacking a return type, and can be used in other contexts (such as
3400 message-sending or type-parametric code) as a zero-size type.]
3404 The machine types are the following:
3406 * The unsigned word types `u8`, `u16`, `u32` and `u64`, with values drawn from
3407 the integer intervals [0, 2^8 - 1], [0, 2^16 - 1], [0, 2^32 - 1] and
3408 [0, 2^64 - 1] respectively.
3410 * The signed two's complement word types `i8`, `i16`, `i32` and `i64`, with
3411 values drawn from the integer intervals [-(2^(7)), 2^7 - 1],
3412 [-(2^(15)), 2^15 - 1], [-(2^(31)), 2^31 - 1], [-(2^(63)), 2^63 - 1]
3415 * The IEEE 754-2008 `binary32` and `binary64` floating-point types: `f32` and
3416 `f64`, respectively.
3418 #### Machine-dependent integer types
3420 The `usize` type is an unsigned integer type with the same number of bits as the
3421 platform's pointer type. It can represent every memory address in the process.
3423 The `isize` type is a signed integer type with the same number of bits as the
3424 platform's pointer type. The theoretical upper bound on object and array size
3425 is the maximum `isize` value. This ensures that `isize` can be used to calculate
3426 differences between pointers into an object or array and can address every byte
3427 within an object along with one byte past the end.
3431 The types `char` and `str` hold textual data.
3433 A value of type `char` is a [Unicode scalar value](
3434 http://www.unicode.org/glossary/#unicode_scalar_value) (i.e. a code point that
3435 is not a surrogate), represented as a 32-bit unsigned word in the 0x0000 to
3436 0xD7FF or 0xE000 to 0x10FFFF range. A `[char]` array is effectively an UCS-4 /
3439 A value of type `str` is a Unicode string, represented as an array of 8-bit
3440 unsigned bytes holding a sequence of UTF-8 codepoints. Since `str` is of
3441 unknown size, it is not a _first-class_ type, but can only be instantiated
3442 through a pointer type, such as `&str` or `String`.
3446 A tuple *type* is a heterogeneous product of other types, called the *elements*
3447 of the tuple. It has no nominal name and is instead structurally typed.
3449 Tuple types and values are denoted by listing the types or values of their
3450 elements, respectively, in a parenthesized, comma-separated list.
3452 Because tuple elements don't have a name, they can only be accessed by
3453 pattern-matching or by using `N` directly as a field to access the
3456 An example of a tuple type and its use:
3459 type Pair<'a> = (i32, &'a str);
3460 let p: Pair<'static> = (10, "hello");
3462 assert!(b != "world");
3466 ### Array, and Slice types
3468 Rust has two different types for a list of items:
3470 * `[T; N]`, an 'array'.
3471 * `&[T]`, a 'slice'.
3473 An array has a fixed size, and can be allocated on either the stack or the
3476 A slice is a 'view' into an array. It doesn't own the data it points
3479 An example of each kind:
3482 let vec: Vec<i32> = vec![1, 2, 3];
3483 let arr: [i32; 3] = [1, 2, 3];
3484 let s: &[i32] = &vec[..];
3487 As you can see, the `vec!` macro allows you to create a `Vec<T>` easily. The
3488 `vec!` macro is also part of the standard library, rather than the language.
3490 All in-bounds elements of arrays, and slices are always initialized, and access
3491 to an array or slice is always bounds-checked.
3495 A `struct` *type* is a heterogeneous product of other types, called the
3496 *fields* of the type.[^structtype]
3498 [^structtype]: `struct` types are analogous `struct` types in C,
3499 the *record* types of the ML family,
3500 or the *structure* types of the Lisp family.
3502 New instances of a `struct` can be constructed with a [struct
3503 expression](#structure-expressions).
3505 The memory layout of a `struct` is undefined by default to allow for compiler
3506 optimizations like field reordering, but it can be fixed with the
3507 `#[repr(...)]` attribute. In either case, fields may be given in any order in
3508 a corresponding struct *expression*; the resulting `struct` value will always
3509 have the same memory layout.
3511 The fields of a `struct` may be qualified by [visibility
3512 modifiers](#re-exporting-and-visibility), to allow access to data in a
3513 structure outside a module.
3515 A _tuple struct_ type is just like a structure type, except that the fields are
3518 A _unit-like struct_ type is like a structure type, except that it has no
3519 fields. The one value constructed by the associated [structure
3520 expression](#structure-expressions) is the only value that inhabits such a
3523 ### Enumerated types
3525 An *enumerated type* is a nominal, heterogeneous disjoint union type, denoted
3526 by the name of an [`enum` item](#enumerations). [^enumtype]
3528 [^enumtype]: The `enum` type is analogous to a `data` constructor declaration in
3529 ML, or a *pick ADT* in Limbo.
3531 An [`enum` item](#enumerations) declares both the type and a number of *variant
3532 constructors*, each of which is independently named and takes an optional tuple
3535 New instances of an `enum` can be constructed by calling one of the variant
3536 constructors, in a [call expression](#call-expressions).
3538 Any `enum` value consumes as much memory as the largest variant constructor for
3539 its corresponding `enum` type.
3541 Enum types cannot be denoted *structurally* as types, but must be denoted by
3542 named reference to an [`enum` item](#enumerations).
3546 Nominal types — [enumerations](#enumerated-types) and
3547 [structures](#structure-types) — may be recursive. That is, each `enum`
3548 constructor or `struct` field may refer, directly or indirectly, to the
3549 enclosing `enum` or `struct` type itself. Such recursion has restrictions:
3551 * Recursive types must include a nominal type in the recursion
3552 (not mere [type definitions](#type-definitions),
3553 or other structural types such as [arrays](#array,-and-slice-types) or [tuples](#tuple-types)).
3554 * A recursive `enum` item must have at least one non-recursive constructor
3555 (in order to give the recursion a basis case).
3556 * The size of a recursive type must be finite;
3557 in other words the recursive fields of the type must be [pointer types](#pointer-types).
3558 * Recursive type definitions can cross module boundaries, but not module *visibility* boundaries,
3559 or crate boundaries (in order to simplify the module system and type checker).
3561 An example of a *recursive* type and its use:
3566 Cons(T, Box<List<T>>)
3569 let a: List<i32> = List::Cons(7, Box::new(List::Cons(13, Box::new(List::Nil))));
3574 All pointers in Rust are explicit first-class values. They can be copied,
3575 stored into data structures, and returned from functions. There are two
3576 varieties of pointer in Rust:
3579 : These point to memory _owned by some other value_.
3580 A reference type is written `&type` for some lifetime-variable `f`,
3581 or just `&'a type` when you need an explicit lifetime.
3582 Copying a reference is a "shallow" operation:
3583 it involves only copying the pointer itself.
3584 Releasing a reference typically has no effect on the value it points to,
3585 with the exception of temporary values, which are released when the last
3586 reference to them is released.
3588 * Raw pointers (`*`)
3589 : Raw pointers are pointers without safety or liveness guarantees.
3590 Raw pointers are written as `*const T` or `*mut T`,
3591 for example `*const int` means a raw pointer to an integer.
3592 Copying or dropping a raw pointer has no effect on the lifecycle of any
3593 other value. Dereferencing a raw pointer or converting it to any other
3594 pointer type is an [`unsafe` operation](#unsafe-functions).
3595 Raw pointers are generally discouraged in Rust code;
3596 they exist to support interoperability with foreign code,
3597 and writing performance-critical or low-level functions.
3599 The standard library contains additional 'smart pointer' types beyond references
3604 The function type constructor `fn` forms new function types. A function type
3605 consists of a possibly-empty set of function-type modifiers (such as `unsafe`
3606 or `extern`), a sequence of input types and an output type.
3608 An example of a `fn` type:
3611 fn add(x: i32, y: i32) -> i32 {
3615 let mut x = add(5,7);
3617 type Binop = fn(i32, i32) -> i32;
3618 let bo: Binop = add;
3624 ```{.ebnf .notation}
3625 closure_type := [ 'unsafe' ] [ '<' lifetime-list '>' ] '|' arg-list '|'
3626 [ ':' bound-list ] [ '->' type ]
3627 lifetime-list := lifetime | lifetime ',' lifetime-list
3628 arg-list := ident ':' type | ident ':' type ',' arg-list
3629 bound-list := bound | bound '+' bound-list
3630 bound := path | lifetime
3633 The type of a closure mapping an input of type `A` to an output of type `B` is
3634 `|A| -> B`. A closure with no arguments or return values has type `||`.
3636 An example of creating and calling a closure:
3639 let captured_var = 10;
3641 let closure_no_args = || println!("captured_var={}", captured_var);
3643 let closure_args = |arg: i32| -> i32 {
3644 println!("captured_var={}, arg={}", captured_var, arg);
3645 arg // Note lack of semicolon after 'arg'
3648 fn call_closure<F: Fn(), G: Fn(i32) -> i32>(c1: F, c2: G) {
3653 call_closure(closure_no_args, closure_args);
3659 Every trait item (see [traits](#traits)) defines a type with the same name as
3660 the trait. This type is called the _trait object_ of the trait. Trait objects
3661 permit "late binding" of methods, dispatched using _virtual method tables_
3662 ("vtables"). Whereas most calls to trait methods are "early bound" (statically
3663 resolved) to specific implementations at compile time, a call to a method on an
3664 trait objects is only resolved to a vtable entry at compile time. The actual
3665 implementation for each vtable entry can vary on an object-by-object basis.
3667 Given a pointer-typed expression `E` of type `&T` or `Box<T>`, where `T`
3668 implements trait `R`, casting `E` to the corresponding pointer type `&R` or
3669 `Box<R>` results in a value of the _trait object_ `R`. This result is
3670 represented as a pair of pointers: the vtable pointer for the `T`
3671 implementation of `R`, and the pointer value of `E`.
3673 An example of a trait object:
3677 fn stringify(&self) -> String;
3680 impl Printable for i32 {
3681 fn stringify(&self) -> String { self.to_string() }
3684 fn print(a: Box<Printable>) {
3685 println!("{}", a.stringify());
3689 print(Box::new(10) as Box<Printable>);
3693 In this example, the trait `Printable` occurs as a trait object in both the
3694 type signature of `print`, and the cast expression in `main`.
3698 Within the body of an item that has type parameter declarations, the names of
3699 its type parameters are types:
3702 fn map<A: Clone, B: Clone>(f: |A| -> B, xs: &[A]) -> Vec<B> {
3706 let first: B = f(xs[0].clone());
3707 let mut rest: Vec<B> = map(f, xs.slice(1, xs.len()));
3708 rest.insert(0, first);
3713 Here, `first` has type `B`, referring to `map`'s `B` type parameter; and `rest`
3714 has type `Vec<B>`, a vector type with element type `B`.
3718 The special type `self` has a meaning within methods inside an impl item. It
3719 refers to the type of the implicit `self` argument. For example, in:
3723 fn make_string(&self) -> String;
3726 impl Printable for String {
3727 fn make_string(&self) -> String {
3733 `self` refers to the value of type `String` that is the receiver for a call to
3734 the method `make_string`.
3738 Several traits define special evaluation behavior.
3742 The `Copy` trait changes the semantics of a type implementing it. Values whose
3743 type implements `Copy` are copied rather than moved upon assignment.
3745 ## The `Sized` trait
3747 The `Sized` trait indicates that the size of this type is known at compile-time.
3751 The `Drop` trait provides a destructor, to be run whenever a value of this type
3756 A Rust program's memory consists of a static set of *items* and a *heap*.
3757 Immutable portions of the heap may be safely shared between threads, mutable
3758 portions may not be safely shared, but several mechanisms for effectively-safe
3759 sharing of mutable values, built on unsafe code but enforcing a safe locking
3760 discipline, exist in the standard library.
3762 Allocations in the stack consist of *variables*, and allocations in the heap
3765 ### Memory allocation and lifetime
3767 The _items_ of a program are those functions, modules and types that have their
3768 value calculated at compile-time and stored uniquely in the memory image of the
3769 rust process. Items are neither dynamically allocated nor freed.
3771 The _heap_ is a general term that describes boxes. The lifetime of an
3772 allocation in the heap depends on the lifetime of the box values pointing to
3773 it. Since box values may themselves be passed in and out of frames, or stored
3774 in the heap, heap allocations may outlive the frame they are allocated within.
3776 ### Memory ownership
3778 When a stack frame is exited, its local allocations are all released, and its
3779 references to boxes are dropped.
3783 A _variable_ is a component of a stack frame, either a named function parameter,
3784 an anonymous [temporary](#lvalues,-rvalues-and-temporaries), or a named local
3787 A _local variable_ (or *stack-local* allocation) holds a value directly,
3788 allocated within the stack's memory. The value is a part of the stack frame.
3790 Local variables are immutable unless declared otherwise like: `let mut x = ...`.
3792 Function parameters are immutable unless declared with `mut`. The `mut` keyword
3793 applies only to the following parameter (so `|mut x, y|` and `fn f(mut x:
3794 Box<i32>, y: Box<i32>)` declare one mutable variable `x` and one immutable
3797 Methods that take either `self` or `Box<Self>` can optionally place them in a
3798 mutable variable by prefixing them with `mut` (similar to regular arguments):
3802 fn change(mut self) -> Self;
3803 fn modify(mut self: Box<Self>) -> Box<Self>;
3807 Local variables are not initialized when allocated; the entire frame worth of
3808 local variables are allocated at once, on frame-entry, in an uninitialized
3809 state. Subsequent statements within a function may or may not initialize the
3810 local variables. Local variables can be used only after they have been
3811 initialized; this is enforced by the compiler.
3815 The Rust compiler supports various methods to link crates together both
3816 statically and dynamically. This section will explore the various methods to
3817 link Rust crates together, and more information about native libraries can be
3818 found in the [ffi section of the book][ffi].
3820 In one session of compilation, the compiler can generate multiple artifacts
3821 through the usage of either command line flags or the `crate_type` attribute.
3822 If one or more command line flag is specified, all `crate_type` attributes will
3823 be ignored in favor of only building the artifacts specified by command line.
3825 * `--crate-type=bin`, `#[crate_type = "bin"]` - A runnable executable will be
3826 produced. This requires that there is a `main` function in the crate which
3827 will be run when the program begins executing. This will link in all Rust and
3828 native dependencies, producing a distributable binary.
3830 * `--crate-type=lib`, `#[crate_type = "lib"]` - A Rust library will be produced.
3831 This is an ambiguous concept as to what exactly is produced because a library
3832 can manifest itself in several forms. The purpose of this generic `lib` option
3833 is to generate the "compiler recommended" style of library. The output library
3834 will always be usable by rustc, but the actual type of library may change from
3835 time-to-time. The remaining output types are all different flavors of
3836 libraries, and the `lib` type can be seen as an alias for one of them (but the
3837 actual one is compiler-defined).
3839 * `--crate-type=dylib`, `#[crate_type = "dylib"]` - A dynamic Rust library will
3840 be produced. This is different from the `lib` output type in that this forces
3841 dynamic library generation. The resulting dynamic library can be used as a
3842 dependency for other libraries and/or executables. This output type will
3843 create `*.so` files on linux, `*.dylib` files on osx, and `*.dll` files on
3846 * `--crate-type=staticlib`, `#[crate_type = "staticlib"]` - A static system
3847 library will be produced. This is different from other library outputs in that
3848 the Rust compiler will never attempt to link to `staticlib` outputs. The
3849 purpose of this output type is to create a static library containing all of
3850 the local crate's code along with all upstream dependencies. The static
3851 library is actually a `*.a` archive on linux and osx and a `*.lib` file on
3852 windows. This format is recommended for use in situations such as linking
3853 Rust code into an existing non-Rust application because it will not have
3854 dynamic dependencies on other Rust code.
3856 * `--crate-type=rlib`, `#[crate_type = "rlib"]` - A "Rust library" file will be
3857 produced. This is used as an intermediate artifact and can be thought of as a
3858 "static Rust library". These `rlib` files, unlike `staticlib` files, are
3859 interpreted by the Rust compiler in future linkage. This essentially means
3860 that `rustc` will look for metadata in `rlib` files like it looks for metadata
3861 in dynamic libraries. This form of output is used to produce statically linked
3862 executables as well as `staticlib` outputs.
3864 Note that these outputs are stackable in the sense that if multiple are
3865 specified, then the compiler will produce each form of output at once without
3866 having to recompile. However, this only applies for outputs specified by the
3867 same method. If only `crate_type` attributes are specified, then they will all
3868 be built, but if one or more `--crate-type` command line flag is specified,
3869 then only those outputs will be built.
3871 With all these different kinds of outputs, if crate A depends on crate B, then
3872 the compiler could find B in various different forms throughout the system. The
3873 only forms looked for by the compiler, however, are the `rlib` format and the
3874 dynamic library format. With these two options for a dependent library, the
3875 compiler must at some point make a choice between these two formats. With this
3876 in mind, the compiler follows these rules when determining what format of
3877 dependencies will be used:
3879 1. If a static library is being produced, all upstream dependencies are
3880 required to be available in `rlib` formats. This requirement stems from the
3881 reason that a dynamic library cannot be converted into a static format.
3883 Note that it is impossible to link in native dynamic dependencies to a static
3884 library, and in this case warnings will be printed about all unlinked native
3885 dynamic dependencies.
3887 2. If an `rlib` file is being produced, then there are no restrictions on what
3888 format the upstream dependencies are available in. It is simply required that
3889 all upstream dependencies be available for reading metadata from.
3891 The reason for this is that `rlib` files do not contain any of their upstream
3892 dependencies. It wouldn't be very efficient for all `rlib` files to contain a
3893 copy of `libstd.rlib`!
3895 3. If an executable is being produced and the `-C prefer-dynamic` flag is not
3896 specified, then dependencies are first attempted to be found in the `rlib`
3897 format. If some dependencies are not available in an rlib format, then
3898 dynamic linking is attempted (see below).
3900 4. If a dynamic library or an executable that is being dynamically linked is
3901 being produced, then the compiler will attempt to reconcile the available
3902 dependencies in either the rlib or dylib format to create a final product.
3904 A major goal of the compiler is to ensure that a library never appears more
3905 than once in any artifact. For example, if dynamic libraries B and C were
3906 each statically linked to library A, then a crate could not link to B and C
3907 together because there would be two copies of A. The compiler allows mixing
3908 the rlib and dylib formats, but this restriction must be satisfied.
3910 The compiler currently implements no method of hinting what format a library
3911 should be linked with. When dynamically linking, the compiler will attempt to
3912 maximize dynamic dependencies while still allowing some dependencies to be
3913 linked in via an rlib.
3915 For most situations, having all libraries available as a dylib is recommended
3916 if dynamically linking. For other situations, the compiler will emit a
3917 warning if it is unable to determine which formats to link each library with.
3919 In general, `--crate-type=bin` or `--crate-type=lib` should be sufficient for
3920 all compilation needs, and the other options are just available if more
3921 fine-grained control is desired over the output format of a Rust crate.
3923 # Appendix: Rationales and design tradeoffs
3927 # Appendix: Influences
3929 Rust is not a particularly original language, with design elements coming from
3930 a wide range of sources. Some of these are listed below (including elements
3931 that have since been removed):
3933 * SML, OCaml: algebraic datatypes, pattern matching, type inference,
3934 semicolon statement separation
3935 * C++: references, RAII, smart pointers, move semantics, monomorphisation,
3937 * ML Kit, Cyclone: region based memory management
3938 * Haskell (GHC): typeclasses, type families
3939 * Newsqueak, Alef, Limbo: channels, concurrency
3940 * Erlang: message passing, task failure, ~~linked task failure~~,
3941 ~~lightweight concurrency~~
3942 * Swift: optional bindings
3943 * Scheme: hygienic macros
3945 * Ruby: ~~block syntax~~
3946 * NIL, Hermes: ~~typestate~~
3947 * [Unicode Annex #31](http://www.unicode.org/reports/tr31/): identifier and
3950 [ffi]: book/ffi.html
3951 [plugin]: book/plugins.html