1 % The Rust Reference Manual
5 This document is the reference manual for the Rust programming language. It
6 provides three kinds of material:
8 - Chapters that formally define the language grammar and, for each
9 construct, informally describe its semantics and give examples of its
11 - Chapters that informally describe the memory model, concurrency model,
12 runtime services, linkage model and debugging facilities.
13 - Appendix chapters providing rationale and references to languages that
14 influenced the design.
16 This document does not serve as a tutorial introduction to the
17 language. Background familiarity with the language is assumed. A separate
18 [tutorial] document is available to help acquire such background familiarity.
20 This document also does not serve as a reference to the [standard]
21 library included in the language distribution. Those libraries are
22 documented separately by extracting documentation attributes from their
25 [tutorial]: tutorial.html
26 [standard]: std/index.html
30 Rust is a work in progress. The language continues to evolve as the design
31 shifts and is fleshed out in working code. Certain parts work, certain parts
32 do not, certain parts will be removed or changed.
34 This manual is a snapshot written in the present tense. All features described
35 exist in working code unless otherwise noted, but some are quite primitive or
36 remain to be further modified by planned work. Some may be temporary. It is a
37 *draft*, and we ask that you not take anything you read here as final.
39 If you have suggestions to make, please try to focus them on *reductions* to
40 the language: possible features that can be combined or omitted. We aim to
41 keep the size and complexity of the language under control.
43 > **Note:** The grammar for Rust given in this document is rough and
44 > very incomplete; only a modest number of sections have accompanying grammar
45 > rules. Formalizing the grammar accepted by the Rust parser is ongoing work,
46 > but future versions of this document will contain a complete
47 > grammar. Moreover, we hope that this grammar will be extracted and verified
48 > as LL(1) by an automated grammar-analysis tool, and further tested against the
49 > Rust sources. Preliminary versions of this automation exist, but are not yet
54 Rust's grammar is defined over Unicode codepoints, each conventionally
55 denoted `U+XXXX`, for 4 or more hexadecimal digits `X`. _Most_ of Rust's
56 grammar is confined to the ASCII range of Unicode, and is described in this
57 document by a dialect of Extended Backus-Naur Form (EBNF), specifically a
58 dialect of EBNF supported by common automated LL(k) parsing tools such as
59 `llgen`, rather than the dialect given in ISO 14977. The dialect can be
60 defined self-referentially as follows:
62 ~~~~ {.ebnf .notation}
64 rule : nonterminal ':' productionrule ';' ;
65 productionrule : production [ '|' production ] * ;
67 term : element repeats ;
68 element : LITERAL | IDENTIFIER | '[' productionrule ']' ;
69 repeats : [ '*' | '+' ] NUMBER ? | NUMBER ? | '?' ;
74 - Whitespace in the grammar is ignored.
75 - Square brackets are used to group rules.
76 - `LITERAL` is a single printable ASCII character, or an escaped hexadecimal
77 ASCII code of the form `\xQQ`, in single quotes, denoting the corresponding
78 Unicode codepoint `U+00QQ`.
79 - `IDENTIFIER` is a nonempty string of ASCII letters and underscores.
80 - The `repeat` forms apply to the adjacent `element`, and are as follows:
81 - `?` means zero or one repetition
82 - `*` means zero or more repetitions
83 - `+` means one or more repetitions
84 - NUMBER trailing a repeat symbol gives a maximum repetition count
85 - NUMBER on its own gives an exact repetition count
87 This EBNF dialect should hopefully be familiar to many readers.
89 ## Unicode productions
91 A few productions in Rust's grammar permit Unicode codepoints outside the ASCII range.
92 We define these productions in terms of character properties specified in the Unicode standard,
93 rather than in terms of ASCII-range codepoints.
94 The section [Special Unicode Productions](#special-unicode-productions) lists these productions.
96 ## String table productions
98 Some rules in the grammar — notably [unary
99 operators](#unary-operator-expressions), [binary
100 operators](#binary-operator-expressions), and [keywords](#keywords) —
101 are given in a simplified form: as a listing of a table of unquoted,
102 printable whitespace-separated strings. These cases form a subset of
103 the rules regarding the [token](#tokens) rule, and are assumed to be
104 the result of a lexical-analysis phase feeding the parser, driven by a
105 DFA, operating over the disjunction of all such string table entries.
107 When such a string enclosed in double-quotes (`"`) occurs inside the
108 grammar, it is an implicit reference to a single member of such a string table
109 production. See [tokens](#tokens) for more information.
115 Rust input is interpreted as a sequence of Unicode codepoints encoded in UTF-8.
116 Most Rust grammar rules are defined in terms of printable ASCII-range codepoints,
117 but a small number are defined in terms of Unicode properties or explicit
118 codepoint lists. [^inputformat]
120 [^inputformat]: Substitute definitions for the special Unicode productions are
121 provided to the grammar verifier, restricted to ASCII range, when verifying
122 the grammar in this document.
124 ## Special Unicode Productions
126 The following productions in the Rust grammar are defined in terms of Unicode properties:
127 `ident`, `non_null`, `non_star`, `non_eol`, `non_slash_or_star`, `non_single_quote` and `non_double_quote`.
131 The `ident` production is any nonempty Unicode string of the following form:
133 - The first character has property `XID_start`
134 - The remaining characters have property `XID_continue`
136 that does _not_ occur in the set of [keywords](#keywords).
138 Note: `XID_start` and `XID_continue` as character properties cover the
139 character ranges used to form the more familiar C and Java language-family
142 ### Delimiter-restricted productions
144 Some productions are defined by exclusion of particular Unicode characters:
146 - `non_null` is any single Unicode character aside from `U+0000` (null)
147 - `non_eol` is `non_null` restricted to exclude `U+000A` (`'\n'`)
148 - `non_star` is `non_null` restricted to exclude `U+002A` (`*`)
149 - `non_slash_or_star` is `non_null` restricted to exclude `U+002F` (`/`) and `U+002A` (`*`)
150 - `non_single_quote` is `non_null` restricted to exclude `U+0027` (`'`)
151 - `non_double_quote` is `non_null` restricted to exclude `U+0022` (`"`)
156 comment : block_comment | line_comment ;
157 block_comment : "/*" block_comment_body * '*' + '/' ;
158 block_comment_body : [block_comment | character] * ;
159 line_comment : "//" non_eol * ;
162 Comments in Rust code follow the general C++ style of line and block-comment forms.
163 Nested block comments are supported.
165 Line comments beginning with exactly _three_ slashes (`///`), and block
166 comments beginning with exactly one repeated asterisk in the block-open
167 sequence (`/**`), are interpreted as a special syntax for `doc`
168 [attributes](#attributes). That is, they are equivalent to writing
169 `#[doc="..."]` around the body of the comment (this includes the comment
170 characters themselves, ie `/// Foo` turns into `#[doc="/// Foo"]`).
172 Non-doc comments are interpreted as a form of whitespace.
177 whitespace_char : '\x20' | '\x09' | '\x0a' | '\x0d' ;
178 whitespace : [ whitespace_char | comment ] + ;
181 The `whitespace_char` production is any nonempty Unicode string consisting of any
182 of the following Unicode characters: `U+0020` (space, `' '`), `U+0009` (tab,
183 `'\t'`), `U+000A` (LF, `'\n'`), `U+000D` (CR, `'\r'`).
185 Rust is a "free-form" language, meaning that all forms of whitespace serve
186 only to separate _tokens_ in the grammar, and have no semantic significance.
188 A Rust program has identical meaning if each whitespace element is replaced
189 with any other legal whitespace element, such as a single space character.
194 simple_token : keyword | unop | binop ;
195 token : simple_token | ident | literal | symbol | whitespace token ;
198 Tokens are primitive productions in the grammar defined by regular
199 (non-recursive) languages. "Simple" tokens are given in [string table
200 production](#string-table-productions) form, and occur in the rest of the
201 grammar as double-quoted strings. Other tokens have exact rules given.
205 The keywords are the following strings:
207 ~~~~ {.text .keyword}
218 self static struct super
224 Each of these keywords has special meaning in its grammar,
225 and all of them are excluded from the `ident` rule.
229 A literal is an expression consisting of a single token, rather than a
230 sequence of tokens, that immediately and directly denotes the value it
231 evaluates to, rather than referring to it by name or some other evaluation
232 rule. A literal is a form of constant expression, so is evaluated (primarily)
236 literal : string_lit | char_lit | byte_string_lit | byte_lit | num_lit ;
239 #### Character and string literals
242 char_lit : '\x27' char_body '\x27' ;
243 string_lit : '"' string_body * '"' | 'r' raw_string ;
245 char_body : non_single_quote
246 | '\x5c' [ '\x27' | common_escape | unicode_escape ] ;
248 string_body : non_double_quote
249 | '\x5c' [ '\x22' | common_escape | unicode_escape ] ;
250 raw_string : '"' raw_string_body '"' | '#' raw_string '#' ;
252 common_escape : '\x5c'
253 | 'n' | 'r' | 't' | '0'
255 unicode_escape : 'u' hex_digit 4
258 hex_digit : 'a' | 'b' | 'c' | 'd' | 'e' | 'f'
259 | 'A' | 'B' | 'C' | 'D' | 'E' | 'F'
261 oct_digit : '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' ;
262 dec_digit : '0' | nonzero_dec ;
263 nonzero_dec: '1' | '2' | '3' | '4'
264 | '5' | '6' | '7' | '8' | '9' ;
267 A _character literal_ is a single Unicode character enclosed within two
268 `U+0027` (single-quote) characters, with the exception of `U+0027` itself,
269 which must be _escaped_ by a preceding U+005C character (`\`).
271 A _string literal_ is a sequence of any Unicode characters enclosed within
272 two `U+0022` (double-quote) characters, with the exception of `U+0022`
273 itself, which must be _escaped_ by a preceding `U+005C` character (`\`),
274 or a _raw string literal_.
276 Some additional _escapes_ are available in either character or non-raw string
277 literals. An escape starts with a `U+005C` (`\`) and continues with one of
280 * An _8-bit codepoint escape_ escape starts with `U+0078` (`x`) and is
281 followed by exactly two _hex digits_. It denotes the Unicode codepoint
282 equal to the provided hex value.
283 * A _16-bit codepoint escape_ starts with `U+0075` (`u`) and is followed
284 by exactly four _hex digits_. It denotes the Unicode codepoint equal to
285 the provided hex value.
286 * A _32-bit codepoint escape_ starts with `U+0055` (`U`) and is followed
287 by exactly eight _hex digits_. It denotes the Unicode codepoint equal to
288 the provided hex value.
289 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
290 (`r`), or `U+0074` (`t`), denoting the unicode values `U+000A` (LF),
291 `U+000D` (CR) or `U+0009` (HT) respectively.
292 * The _backslash escape_ is the character `U+005C` (`\`) which must be
293 escaped in order to denote *itself*.
295 Raw string literals do not process any escapes. They start with the character
296 `U+0072` (`r`), followed by zero or more of the character `U+0023` (`#`) and a
297 `U+0022` (double-quote) character. The _raw string body_ is not defined in the
298 EBNF grammar above: it can contain any sequence of Unicode characters and is
299 terminated only by another `U+0022` (double-quote) character, followed by the
300 same number of `U+0023` (`#`) characters that preceded the opening `U+0022`
301 (double-quote) character.
303 All Unicode characters contained in the raw string body represent themselves,
304 the characters `U+0022` (double-quote) (except when followed by at least as
305 many `U+0023` (`#`) characters as were used to start the raw string literal) or
306 `U+005C` (`\`) do not have any special meaning.
308 Examples for string literals:
311 "foo"; r"foo"; // foo
312 "\"foo\""; r#""foo""#; // "foo"
315 r##"foo #"# bar"##; // foo #"# bar
317 "\x52"; "R"; r"R"; // R
318 "\\x52"; r"\x52"; // \x52
321 #### Byte and byte string literals
324 byte_lit : 'b' '\x27' byte_body '\x27' ;
325 byte_string_lit : 'b' '"' string_body * '"' | 'b' 'r' raw_byte_string ;
327 byte_body : ascii_non_single_quote
328 | '\x5c' [ '\x27' | common_escape ] ;
330 byte_string_body : ascii_non_double_quote
331 | '\x5c' [ '\x22' | common_escape ] ;
332 raw_byte_string : '"' raw_byte_string_body '"' | '#' raw_byte_string '#' ;
336 A _byte literal_ is a single ASCII character (in the `U+0000` to `U+007F` range)
337 enclosed within two `U+0027` (single-quote) characters,
338 with the exception of `U+0027` itself,
339 which must be _escaped_ by a preceding U+005C character (`\`),
340 or a single _escape_.
341 It is equivalent to a `u8` unsigned 8-bit integer _number literal_.
343 A _byte string literal_ is a sequence of ASCII characters and _escapes_
344 enclosed within two `U+0022` (double-quote) characters,
345 with the exception of `U+0022` itself,
346 which must be _escaped_ by a preceding `U+005C` character (`\`),
347 or a _raw byte string literal_.
348 It is equivalent to a `&'static [u8]` borrowed vector of unsigned 8-bit integers.
350 Some additional _escapes_ are available in either byte or non-raw byte string
351 literals. An escape starts with a `U+005C` (`\`) and continues with one of
354 * An _byte escape_ escape starts with `U+0078` (`x`) and is
355 followed by exactly two _hex digits_. It denotes the byte
356 equal to the provided hex value.
357 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
358 (`r`), or `U+0074` (`t`), denoting the bytes values `0x0A` (ASCII LF),
359 `0x0D` (ASCII CR) or `0x09` (ASCII HT) respectively.
360 * The _backslash escape_ is the character `U+005C` (`\`) which must be
361 escaped in order to denote its ASCII encoding `0x5C`.
363 Raw byte string literals do not process any escapes.
364 They start with the character `U+0072` (`r`),
365 followed by `U+0062` (`b`),
366 followed by zero or more of the character `U+0023` (`#`),
367 and a `U+0022` (double-quote) character.
368 The _raw string body_ is not defined in the EBNF grammar above:
369 it can contain any sequence of ASCII characters and is
370 terminated only by another `U+0022` (double-quote) character, followed by the
371 same number of `U+0023` (`#`) characters that preceded the opening `U+0022`
372 (double-quote) character.
373 A raw byte string literal can not contain any non-ASCII byte.
375 All characters contained in the raw string body represent their ASCII encoding,
376 the characters `U+0022` (double-quote) (except when followed by at least as
377 many `U+0023` (`#`) characters as were used to start the raw string literal) or
378 `U+005C` (`\`) do not have any special meaning.
380 Examples for byte string literals:
383 b"foo"; br"foo"; // foo
384 b"\"foo\""; br#""foo""#; // "foo"
387 br##"foo #"# bar"##; // foo #"# bar
389 b"\x52"; b"R"; br"R"; // R
390 b"\\x52"; br"\x52"; // \x52
396 num_lit : nonzero_dec [ dec_digit | '_' ] * num_suffix ?
397 | '0' [ [ dec_digit | '_' ] * num_suffix ?
398 | 'b' [ '1' | '0' | '_' ] + int_suffix ?
399 | 'o' [ oct_digit | '_' ] + int_suffix ?
400 | 'x' [ hex_digit | '_' ] + int_suffix ? ] ;
402 num_suffix : int_suffix | float_suffix ;
404 int_suffix : 'u' int_suffix_size ?
405 | 'i' int_suffix_size ? ;
406 int_suffix_size : [ '8' | '1' '6' | '3' '2' | '6' '4' ] ;
408 float_suffix : [ exponent | '.' dec_lit exponent ? ] ? float_suffix_ty ? ;
409 float_suffix_ty : 'f' [ '3' '2' | '6' '4' ] ;
410 exponent : ['E' | 'e'] ['-' | '+' ] ? dec_lit ;
411 dec_lit : [ dec_digit | '_' ] + ;
414 A _number literal_ is either an _integer literal_ or a _floating-point
415 literal_. The grammar for recognizing the two kinds of literals is mixed,
416 as they are differentiated by suffixes.
418 ##### Integer literals
420 An _integer literal_ has one of four forms:
422 * A _decimal literal_ starts with a *decimal digit* and continues with any
423 mixture of *decimal digits* and _underscores_.
424 * A _hex literal_ starts with the character sequence `U+0030` `U+0078`
425 (`0x`) and continues as any mixture hex digits and underscores.
426 * An _octal literal_ starts with the character sequence `U+0030` `U+006F`
427 (`0o`) and continues as any mixture octal digits and underscores.
428 * A _binary literal_ starts with the character sequence `U+0030` `U+0062`
429 (`0b`) and continues as any mixture binary digits and underscores.
431 An integer literal may be followed (immediately, without any spaces) by an
432 _integer suffix_, which changes the type of the literal. There are two kinds
433 of integer literal suffix:
435 * The `i` and `u` suffixes give the literal type `int` or `uint`,
437 * Each of the signed and unsigned machine types `u8`, `i8`,
438 `u16`, `i16`, `u32`, `i32`, `u64` and `i64`
439 give the literal the corresponding machine type.
441 The type of an _unsuffixed_ integer literal is determined by type inference.
442 If an integer type can be _uniquely_ determined from the surrounding program
443 context, the unsuffixed integer literal has that type. If the program context
444 underconstrains the type, it is considered a static type error;
445 if the program context overconstrains the type,
446 it is also considered a static type error.
448 Examples of integer literals of various forms:
455 0o70_i16; // type i16
456 0b1111_1111_1001_0000_i32; // type i32
459 ##### Floating-point literals
461 A _floating-point literal_ has one of two forms:
463 * Two _decimal literals_ separated by a period
464 character `U+002E` (`.`), with an optional _exponent_ trailing after the
465 second decimal literal.
466 * A single _decimal literal_ followed by an _exponent_.
468 By default, a floating-point literal has a generic type,
469 and, like integer literals, the type must be uniquely determined
471 A floating-point literal may be followed (immediately, without any
472 spaces) by a _floating-point suffix_, which changes the type of the literal.
473 There are two floating-point suffixes: `f32`, and `f64` (the 32-bit and 64-bit
474 floating point types).
476 Examples of floating-point literals of various forms:
479 123.0f64; // type f64
482 12E+99_f64; // type f64
485 ##### Unit and boolean literals
487 The _unit value_, the only value of the type that has the same name, is written as `()`.
488 The two values of the boolean type are written `true` and `false`.
494 | '#' | '[' | ']' | '(' | ')' | '{' | '}'
498 Symbols are a general class of printable [token](#tokens) that play structural
499 roles in a variety of grammar productions. They are catalogued here for
500 completeness as the set of remaining miscellaneous printable tokens that do not
501 otherwise appear as [unary operators](#unary-operator-expressions), [binary
502 operators](#binary-operator-expressions), or [keywords](#keywords).
508 expr_path : [ "::" ] ident [ "::" expr_path_tail ] + ;
509 expr_path_tail : '<' type_expr [ ',' type_expr ] + '>'
512 type_path : ident [ type_path_tail ] + ;
513 type_path_tail : '<' type_expr [ ',' type_expr ] + '>'
517 A _path_ is a sequence of one or more path components _logically_ separated by
518 a namespace qualifier (`::`). If a path consists of only one component, it may
519 refer to either an [item](#items) or a [slot](#memory-slots) in a local
520 control scope. If a path has multiple components, it refers to an item.
522 Every item has a _canonical path_ within its crate, but the path naming an
523 item is only meaningful within a given crate. There is no global namespace
524 across crates; an item's canonical path merely identifies it within the crate.
526 Two examples of simple paths consisting of only identifier components:
533 Path components are usually [identifiers](#identifiers), but the trailing
534 component of a path may be an angle-bracket-enclosed list of type
535 arguments. In [expression](#expressions) context, the type argument list is
536 given after a final (`::`) namespace qualifier in order to disambiguate it
537 from a relational expression involving the less-than symbol (`<`). In type
538 expression context, the final namespace qualifier is omitted.
540 Two examples of paths with type arguments:
543 # struct HashMap<K, V>;
545 # fn id<T>(t: T) -> T { t }
546 type T = HashMap<int,String>; // Type arguments used in a type expression
547 let x = id::<int>(10); // Type arguments used in a call expression
551 Paths can be denoted with various leading qualifiers to change the meaning of
554 * Paths starting with `::` are considered to be global paths where the
555 components of the path start being resolved from the crate root. Each
556 identifier in the path must resolve to an item.
564 ::a::foo(); // call a's foo function
570 * Paths starting with the keyword `super` begin resolution relative to the
571 parent module. Each further identifier must resolve to an item
579 super::a::foo(); // call a's foo function
585 * Paths starting with the keyword `self` begin resolution relative to the
586 current module. Each further identifier must resolve to an item.
598 A number of minor features of Rust are not central enough to have their own
599 syntax, and yet are not implementable as functions. Instead, they are given
600 names, and invoked through a consistent syntax: `name!(...)`. Examples
603 * `format!` : format data into a string
604 * `env!` : look up an environment variable's value at compile time
605 * `file!`: return the path to the file being compiled
606 * `stringify!` : pretty-print the Rust expression given as an argument
607 * `include!` : include the Rust expression in the given file
608 * `include_str!` : include the contents of the given file as a string
609 * `include_bin!` : include the contents of the given file as a binary blob
610 * `error!`, `warn!`, `info!`, `debug!` : provide diagnostic information.
612 All of the above extensions are expressions with values.
617 expr_macro_rules : "macro_rules" '!' ident '(' macro_rule * ')' ;
618 macro_rule : '(' matcher * ')' "=>" '(' transcriber * ')' ';' ;
619 matcher : '(' matcher * ')' | '[' matcher * ']'
620 | '{' matcher * '}' | '$' ident ':' ident
621 | '$' '(' matcher * ')' sep_token? [ '*' | '+' ]
622 | non_special_token ;
623 transcriber : '(' transcriber * ')' | '[' transcriber * ']'
624 | '{' transcriber * '}' | '$' ident
625 | '$' '(' transcriber * ')' sep_token? [ '*' | '+' ]
626 | non_special_token ;
629 User-defined syntax extensions are called "macros",
630 and the `macro_rules` syntax extension defines them.
631 Currently, user-defined macros can expand to expressions, statements, or items.
633 (A `sep_token` is any token other than `*` and `+`.
634 A `non_special_token` is any token other than a delimiter or `$`.)
636 The macro expander looks up macro invocations by name,
637 and tries each macro rule in turn.
638 It transcribes the first successful match.
639 Matching and transcription are closely related to each other,
640 and we will describe them together.
644 The macro expander matches and transcribes every token that does not begin with a `$` literally, including delimiters.
645 For parsing reasons, delimiters must be balanced, but they are otherwise not special.
647 In the matcher, `$` _name_ `:` _designator_ matches the nonterminal in the
648 Rust syntax named by _designator_. Valid designators are `item`, `block`,
649 `stmt`, `pat`, `expr`, `ty` (type), `ident`, `path`, `matchers` (lhs of the `=>` in macro rules),
650 `tt` (rhs of the `=>` in macro rules). In the transcriber, the designator is already known, and so only
651 the name of a matched nonterminal comes after the dollar sign.
653 In both the matcher and transcriber, the Kleene star-like operator indicates repetition.
654 The Kleene star operator consists of `$` and parens, optionally followed by a separator token, followed by `*` or `+`.
655 `*` means zero or more repetitions, `+` means at least one repetition.
656 The parens are not matched or transcribed.
657 On the matcher side, a name is bound to _all_ of the names it
658 matches, in a structure that mimics the structure of the repetition
659 encountered on a successful match. The job of the transcriber is to sort that
662 The rules for transcription of these repetitions are called "Macro By Example".
663 Essentially, one "layer" of repetition is discharged at a time, and all of
664 them must be discharged by the time a name is transcribed. Therefore,
665 `( $( $i:ident ),* ) => ( $i )` is an invalid macro, but
666 `( $( $i:ident ),* ) => ( $( $i:ident ),* )` is acceptable (if trivial).
668 When Macro By Example encounters a repetition, it examines all of the `$`
669 _name_ s that occur in its body. At the "current layer", they all must repeat
670 the same number of times, so
671 ` ( $( $i:ident ),* ; $( $j:ident ),* ) => ( $( ($i,$j) ),* )` is valid if
672 given the argument `(a,b,c ; d,e,f)`, but not `(a,b,c ; d,e)`. The repetition
673 walks through the choices at that layer in lockstep, so the former input
674 transcribes to `( (a,d), (b,e), (c,f) )`.
676 Nested repetitions are allowed.
678 ### Parsing limitations
680 The parser used by the macro system is reasonably powerful, but the parsing of
681 Rust syntax is restricted in two ways:
683 1. The parser will always parse as much as possible. If it attempts to match
684 `$i:expr [ , ]` against `8 [ , ]`, it will attempt to parse `i` as an array
685 index operation and fail. Adding a separator can solve this problem.
686 2. The parser must have eliminated all ambiguity by the time it reaches a `$` _name_ `:` _designator_.
687 This requirement most often affects name-designator pairs when they occur at the beginning of, or immediately after, a `$(...)*`; requiring a distinctive token in front can solve the problem.
689 ## Syntax extensions useful for the macro author
691 * `log_syntax!` : print out the arguments at compile time
692 * `trace_macros!` : supply `true` or `false` to enable or disable macro expansion logging
693 * `stringify!` : turn the identifier argument into a string literal
694 * `concat!` : concatenates a comma-separated list of literals
695 * `concat_idents!` : create a new identifier by concatenating the arguments
697 # Crates and source files
699 Rust is a *compiled* language.
700 Its semantics obey a *phase distinction* between compile-time and run-time.
701 Those semantic rules that have a *static interpretation* govern the success or failure of compilation.
702 We refer to these rules as "static semantics".
703 Semantic rules called "dynamic semantics" govern the behavior of programs at run-time.
704 A program that fails to compile due to violation of a compile-time rule has no defined dynamic semantics; the compiler should halt with an error report, and produce no executable artifact.
706 The compilation model centres on artifacts called _crates_.
707 Each compilation processes a single crate in source form, and if successful,
708 produces a single crate in binary form: either an executable or a
709 library.[^cratesourcefile]
711 [^cratesourcefile]: A crate is somewhat analogous to an *assembly* in the
712 ECMA-335 CLI model, a *library* in the SML/NJ Compilation Manager, a *unit*
713 in the Owens and Flatt module system, or a *configuration* in Mesa.
715 A _crate_ is a unit of compilation and linking, as well as versioning, distribution and runtime loading.
716 A crate contains a _tree_ of nested [module](#modules) scopes.
717 The top level of this tree is a module that is anonymous (from the point of view of paths within the module) and any item within a crate has a canonical [module path](#paths) denoting its location within the crate's module tree.
719 The Rust compiler is always invoked with a single source file as input, and always produces a single output crate.
720 The processing of that source file may result in other source files being loaded as modules.
721 Source files have the extension `.rs`.
723 A Rust source file describes a module, the name and
724 location of which — in the module tree of the current crate — are defined
725 from outside the source file: either by an explicit `mod_item` in
726 a referencing source file, or by the name of the crate itself.
728 Each source file contains a sequence of zero or more `item` definitions,
729 and may optionally begin with any number of `attributes` that apply to the containing module.
730 Attributes on the anonymous crate module define important metadata that influences
731 the behavior of the compiler.
734 # #![allow(unused_attribute)]
736 #![crate_id = "projx#2.5"]
738 // Additional metadata attributes
739 #![desc = "Project X"]
741 #![comment = "This is a comment on Project X."]
743 // Specify the output type
744 #![crate_type = "lib"]
747 #![warn(non_camel_case_types)]
750 A crate that contains a `main` function can be compiled to an executable.
751 If a `main` function is present, its return type must be [`unit`](#primitive-types) and it must take no arguments.
753 # Items and attributes
755 Crates contain [items](#items),
756 each of which may have some number of [attributes](#attributes) attached to it.
761 item : mod_item | fn_item | type_item | struct_item | enum_item
762 | static_item | trait_item | impl_item | extern_block ;
765 An _item_ is a component of a crate; some module items can be defined in crate
766 files, but most are defined in source files. Items are organized within a
767 crate by a nested set of [modules](#modules). Every crate has a single
768 "outermost" anonymous module; all further items within the crate have
769 [paths](#paths) within the module tree of the crate.
771 Items are entirely determined at compile-time, generally remain fixed during
772 execution, and may reside in read-only memory.
774 There are several kinds of item:
776 * [modules](#modules)
777 * [functions](#functions)
778 * [type definitions](#type-definitions)
779 * [structures](#structures)
780 * [enumerations](#enumerations)
781 * [static items](#static-items)
783 * [implementations](#implementations)
785 Some items form an implicit scope for the declaration of sub-items. In other
786 words, within a function or module, declarations of items can (in many cases)
787 be mixed with the statements, control blocks, and similar artifacts that
788 otherwise compose the item body. The meaning of these scoped items is the same
789 as if the item was declared outside the scope — it is still a static item —
790 except that the item's *path name* within the module namespace is qualified by
791 the name of the enclosing item, or is private to the enclosing item (in the
793 The grammar specifies the exact locations in which sub-item declarations may appear.
797 All items except modules may be *parameterized* by type. Type parameters are
798 given as a comma-separated list of identifiers enclosed in angle brackets
799 (`<...>`), after the name of the item and before its definition.
800 The type parameters of an item are considered "part of the name", not part of the type of the item.
801 A referencing [path](#paths) must (in principle) provide type arguments as a list of comma-separated types enclosed within angle brackets, in order to refer to the type-parameterized item.
802 In practice, the type-inference system can usually infer such argument types from context.
803 There are no general type-parametric types, only type-parametric items.
804 That is, Rust has no notion of type abstraction: there are no first-class "forall" types.
809 mod_item : "mod" ident ( ';' | '{' mod '}' );
810 mod : [ view_item | item ] * ;
813 A module is a container for zero or more [view items](#view-items) and zero or
814 more [items](#items). The view items manage the visibility of the items
815 defined within the module, as well as the visibility of names from outside the
816 module when referenced from inside the module.
818 A _module item_ is a module, surrounded in braces, named, and prefixed with
819 the keyword `mod`. A module item introduces a new, named module into the tree
820 of modules making up a crate. Modules can nest arbitrarily.
822 An example of a module:
826 type Complex = (f64, f64);
827 fn sin(f: f64) -> f64 {
831 fn cos(f: f64) -> f64 {
835 fn tan(f: f64) -> f64 {
842 Modules and types share the same namespace.
843 Declaring a named type that has the same name as a module in scope is forbidden:
844 that is, a type definition, trait, struct, enumeration, or type parameter
845 can't shadow the name of a module in scope, or vice versa.
847 A module without a body is loaded from an external file, by default with the same
848 name as the module, plus the `.rs` extension.
849 When a nested submodule is loaded from an external file,
850 it is loaded from a subdirectory path that mirrors the module hierarchy.
853 // Load the `vec` module from `vec.rs`
857 // Load the `local_data` module from `task/local_data.rs`
862 The directories and files used for loading external file modules can be influenced
863 with the `path` attribute.
866 #[path = "task_files"]
868 // Load the `local_data` module from `task_files/tls.rs`
877 view_item : extern_crate_decl | use_decl ;
880 A view item manages the namespace of a module.
881 View items do not define new items, but rather, simply change other items' visibility.
882 There are several kinds of view item:
884 * [`extern crate` declarations](#extern-crate-declarations)
885 * [`use` declarations](#use-declarations)
887 ##### Extern crate declarations
890 extern_crate_decl : "extern" "crate" ident [ '(' link_attrs ')' ] ? [ '=' string_lit ] ? ;
891 link_attrs : link_attr [ ',' link_attrs ] + ;
892 link_attr : ident '=' literal ;
895 An _`extern crate` declaration_ specifies a dependency on an external crate.
896 The external crate is then bound into the declaring scope as the `ident` provided
897 in the `extern_crate_decl`.
899 The external crate is resolved to a specific `soname` at compile time, and a
900 runtime linkage requirement to that `soname` is passed to the linker for
901 loading at runtime. The `soname` is resolved at compile time by scanning the
902 compiler's library path and matching the optional `crateid` provided as a string literal
903 against the `crateid` attributes that were declared on the external crate when
904 it was compiled. If no `crateid` is provided, a default `name` attribute is
905 assumed, equal to the `ident` given in the `extern_crate_decl`.
907 Four examples of `extern crate` declarations:
912 extern crate std; // equivalent to: extern crate std = "std";
914 extern crate ruststd = "std"; // linking to 'std' under another name
916 extern crate foo = "some/where/rust-foo#foo:1.0"; // a full crate ID for external tools
919 ##### Use declarations
922 use_decl : "pub" ? "use" [ ident '=' path
925 path_glob : ident [ "::" [ path_glob
927 | '{' ident [ ',' ident ] * '}' ;
930 A _use declaration_ creates one or more local name bindings synonymous
931 with some other [path](#paths).
932 Usually a `use` declaration is used to shorten the path required to refer to a
933 module item. These declarations may appear at the top of [modules](#modules) and
936 *Note*: Unlike in many languages,
937 `use` declarations in Rust do *not* declare linkage dependency with external crates.
938 Rather, [`extern crate` declarations](#extern-crate-declarations) declare linkage dependencies.
940 Use declarations support a number of convenient shortcuts:
942 * Rebinding the target name as a new local name, using the syntax `use x = p::q::r;`.
943 * Simultaneously binding a list of paths differing only in their final element,
944 using the glob-like brace syntax `use a::b::{c,d,e,f};`
945 * Binding all paths matching a given prefix, using the asterisk wildcard syntax `use a::b::*;`
947 An example of `use` declarations:
950 use std::iter::range_step;
951 use std::option::{Some, None};
956 // Equivalent to 'std::iter::range_step(0u, 10u, 2u);'
957 range_step(0u, 10u, 2u);
959 // Equivalent to 'foo(vec![std::option::Some(1.0f64),
960 // std::option::None]);'
961 foo(vec![Some(1.0f64), None]);
965 Like items, `use` declarations are private to the containing module, by default.
966 Also like items, a `use` declaration can be public, if qualified by the `pub` keyword.
967 Such a `use` declaration serves to _re-export_ a name.
968 A public `use` declaration can therefore _redirect_ some public name to a different target definition:
969 even a definition with a private canonical path, inside a different module.
970 If a sequence of such redirections form a cycle or cannot be resolved unambiguously,
971 they represent a compile-time error.
973 An example of re-exporting:
978 pub use quux::foo::{bar, baz};
987 In this example, the module `quux` re-exports two public names defined in `foo`.
989 Also note that the paths contained in `use` items are relative to the crate root.
990 So, in the previous example, the `use` refers to `quux::foo::{bar, baz}`, and not simply to `foo::{bar, baz}`.
991 This also means that top-level module declarations should be at the crate root if direct usage
992 of the declared modules within `use` items is desired. It is also possible to use `self` and `super`
993 at the beginning of a `use` item to refer to the current and direct parent modules respectively.
994 All rules regarding accessing declared modules in `use` declarations applies to both module declarations
995 and `extern crate` declarations.
997 An example of what will and will not work for `use` items:
1000 # #![allow(unused_imports)]
1001 use foo::native::start; // good: foo is at the root of the crate
1002 use foo::baz::foobaz; // good: foo is at the root of the crate
1005 extern crate native;
1007 use foo::native::start; // good: foo is at crate root
1008 // use native::start; // bad: native is not at the crate root
1009 use self::baz::foobaz; // good: self refers to module 'foo'
1010 use foo::bar::foobar; // good: foo is at crate root
1017 use super::bar::foobar; // good: super refers to module 'foo'
1027 A _function item_ defines a sequence of [statements](#statements) and an optional final [expression](#expressions), along with a name and a set of parameters.
1028 Functions are declared with the keyword `fn`.
1029 Functions declare a set of *input* [*slots*](#memory-slots) as parameters, through which the caller passes arguments into the function, and an *output* [*slot*](#memory-slots) through which the function passes results back to the caller.
1031 A function may also be copied into a first class *value*, in which case the
1032 value has the corresponding [*function type*](#function-types), and can be
1033 used otherwise exactly as a function item (with a minor additional cost of
1034 calling the function indirectly).
1036 Every control path in a function logically ends with a `return` expression or a
1037 diverging expression. If the outermost block of a function has a
1038 value-producing expression in its final-expression position, that expression
1039 is interpreted as an implicit `return` expression applied to the
1042 An example of a function:
1045 fn add(x: int, y: int) -> int {
1050 As with `let` bindings, function arguments are irrefutable patterns,
1051 so any pattern that is valid in a let binding is also valid as an argument.
1054 fn first((value, _): (int, int)) -> int { value }
1058 #### Generic functions
1060 A _generic function_ allows one or more _parameterized types_ to
1061 appear in its signature. Each type parameter must be explicitly
1062 declared, in an angle-bracket-enclosed, comma-separated list following
1066 fn iter<T>(seq: &[T], f: |T|) {
1067 for elt in seq.iter() { f(elt); }
1069 fn map<T, U>(seq: &[T], f: |T| -> U) -> Vec<U> {
1070 let mut acc = vec![];
1071 for elt in seq.iter() { acc.push(f(elt)); }
1076 Inside the function signature and body, the name of the type parameter
1077 can be used as a type name.
1079 When a generic function is referenced, its type is instantiated based
1080 on the context of the reference. For example, calling the `iter`
1081 function defined above on `[1, 2]` will instantiate type parameter `T`
1082 with `int`, and require the closure parameter to have type
1085 The type parameters can also be explicitly supplied in a trailing
1086 [path](#paths) component after the function name. This might be necessary
1087 if there is not sufficient context to determine the type parameters. For
1088 example, `mem::size_of::<u32>() == 4`.
1090 Since a parameter type is opaque to the generic function, the set of
1091 operations that can be performed on it is limited. Values of parameter
1092 type can only be moved, not copied.
1095 fn id<T>(x: T) -> T { x }
1098 Similarly, [trait](#traits) bounds can be specified for type
1099 parameters to allow methods with that trait to be called on values
1105 Unsafe operations are those that potentially violate the memory-safety guarantees of Rust's static semantics.
1107 The following language level features cannot be used in the safe subset of Rust:
1109 - Dereferencing a [raw pointer](#pointer-types).
1110 - Reading or writing a [mutable static variable](#mutable-statics).
1111 - Calling an unsafe function (including an intrinsic or foreign function).
1113 ##### Unsafe functions
1115 Unsafe functions are functions that are not safe in all contexts and/or for all possible inputs.
1116 Such a function must be prefixed with the keyword `unsafe`.
1120 A block of code can also be prefixed with the `unsafe` keyword, to permit calling `unsafe` functions
1121 or dereferencing raw pointers within a safe function.
1123 When a programmer has sufficient conviction that a sequence of potentially unsafe operations is
1124 actually safe, they can encapsulate that sequence (taken as a whole) within an `unsafe` block. The
1125 compiler will consider uses of such code safe, in the surrounding context.
1127 Unsafe blocks are used to wrap foreign libraries, make direct use of hardware or implement features
1128 not directly present in the language. For example, Rust provides the language features necessary to
1129 implement memory-safe concurrency in the language but the implementation of tasks and message
1130 passing is in the standard library.
1132 Rust's type system is a conservative approximation of the dynamic safety requirements, so in some
1133 cases there is a performance cost to using safe code. For example, a doubly-linked list is not a
1134 tree structure and can only be represented with managed or reference-counted pointers in safe code.
1135 By using `unsafe` blocks to represent the reverse links as raw pointers, it can be implemented with
1136 only owned pointers.
1138 ##### Behavior considered unsafe
1140 This is a list of behavior which is forbidden in all Rust code. Type checking provides the guarantee
1141 that these issues are never caused by safe code. An `unsafe` block or function is responsible for
1142 never invoking this behaviour or exposing an API making it possible for it to occur in safe code.
1145 * Dereferencing a null/dangling raw pointer
1146 * Mutating an immutable value/reference
1147 * Reads of [undef](http://llvm.org/docs/LangRef.html#undefined-values) (uninitialized) memory
1148 * Breaking the [pointer aliasing rules](http://llvm.org/docs/LangRef.html#pointer-aliasing-rules)
1149 with raw pointers (a subset of the rules used by C)
1150 * Invoking undefined behavior via compiler intrinsics:
1151 * Indexing outside of the bounds of an object with `std::ptr::offset` (`offset` intrinsic), with
1152 the exception of one byte past the end which is permitted.
1153 * Using `std::ptr::copy_nonoverlapping_memory` (`memcpy32`/`memcpy64` instrinsics) on
1155 * Invalid values in primitive types, even in private fields/locals:
1156 * Dangling/null pointers in non-raw pointers, or slices
1157 * A value other than `false` (0) or `true` (1) in a `bool`
1158 * A discriminant in an `enum` not included in the type definition
1159 * A value in a `char` which is a surrogate or above `char::MAX`
1160 * non-UTF-8 byte sequences in a `str`
1162 ##### Behaviour not considered unsafe
1164 This is a list of behaviour not considered *unsafe* in Rust terms, but that may be undesired.
1167 * Reading data from private fields (`std::repr`, `format!("{:?}", x)`)
1168 * Leaks due to reference count cycles, even in the global heap
1169 * Exiting without calling destructors
1171 * Accessing/modifying the file system
1172 * Unsigned integer overflow (well-defined as wrapping)
1173 * Signed integer overflow (well-defined as two's complement representation wrapping)
1175 #### Diverging functions
1177 A special kind of function can be declared with a `!` character where the
1178 output slot type would normally be. For example:
1181 fn my_err(s: &str) -> ! {
1187 We call such functions "diverging" because they never return a value to the
1188 caller. Every control path in a diverging function must end with a
1189 `fail!()` or a call to another diverging function on every
1190 control path. The `!` annotation does *not* denote a type. Rather, the result
1191 type of a diverging function is a special type called $\bot$ ("bottom") that
1192 unifies with any type. Rust has no syntax for $\bot$.
1194 It might be necessary to declare a diverging function because as mentioned
1195 previously, the typechecker checks that every control path in a function ends
1196 with a [`return`](#return-expressions) or diverging expression. So, if `my_err`
1197 were declared without the `!` annotation, the following code would not
1201 # fn my_err(s: &str) -> ! { fail!() }
1203 fn f(i: int) -> int {
1208 my_err("Bad number!");
1213 This will not compile without the `!` annotation on `my_err`,
1214 since the `else` branch of the conditional in `f` does not return an `int`,
1215 as required by the signature of `f`.
1216 Adding the `!` annotation to `my_err` informs the typechecker that,
1217 should control ever enter `my_err`, no further type judgments about `f` need to hold,
1218 since control will never resume in any context that relies on those judgments.
1219 Thus the return type on `f` only needs to reflect the `if` branch of the conditional.
1222 #### Extern functions
1224 Extern functions are part of Rust's foreign function interface,
1225 providing the opposite functionality to [external blocks](#external-blocks).
1226 Whereas external blocks allow Rust code to call foreign code,
1227 extern functions with bodies defined in Rust code _can be called by foreign
1228 code_. They are defined in the same way as any other Rust function,
1229 except that they have the `extern` modifier.
1232 // Declares an extern fn, the ABI defaults to "C"
1233 extern fn new_int() -> int { 0 }
1235 // Declares an extern fn with "stdcall" ABI
1236 extern "stdcall" fn new_int_stdcall() -> int { 0 }
1239 Unlike normal functions, extern fns have an `extern "ABI" fn()`.
1240 This is the same type as the functions declared in an extern
1244 # extern fn new_int() -> int { 0 }
1245 let fptr: extern "C" fn() -> int = new_int;
1248 Extern functions may be called directly from Rust code as Rust uses large,
1249 contiguous stack segments like C.
1251 ### Type definitions
1253 A _type definition_ defines a new name for an existing [type](#types). Type
1254 definitions are declared with the keyword `type`. Every value has a single,
1255 specific type; the type-specified aspects of a value include:
1257 * Whether the value is composed of sub-values or is indivisible.
1258 * Whether the value represents textual or numerical information.
1259 * Whether the value represents integral or floating-point information.
1260 * The sequence of memory operations required to access the value.
1261 * The [kind](#type-kinds) of the type.
1263 For example, the type `(u8, u8)` defines the set of immutable values that are composite pairs,
1264 each containing two unsigned 8-bit integers accessed by pattern-matching and laid out in memory with the `x` component preceding the `y` component.
1268 A _structure_ is a nominal [structure type](#structure-types) defined with the keyword `struct`.
1270 An example of a `struct` item and its use:
1273 struct Point {x: int, y: int}
1274 let p = Point {x: 10, y: 11};
1278 A _tuple structure_ is a nominal [tuple type](#tuple-types), also defined with the keyword `struct`.
1282 struct Point(int, int);
1283 let p = Point(10, 11);
1284 let px: int = match p { Point(x, _) => x };
1287 A _unit-like struct_ is a structure without any fields, defined by leaving off the list of fields entirely.
1288 Such types will have a single value, just like the [unit value `()`](#unit-and-boolean-literals) of the unit type.
1293 let c = [Cookie, Cookie, Cookie, Cookie];
1296 By using the `struct_inherit` feature gate, structures may use single inheritance. A Structure may only
1297 inherit from a single other structure, called the _super-struct_. The inheriting structure (sub-struct)
1298 acts as if all fields in the super-struct were present in the sub-struct. Fields declared in a sub-struct
1299 must not have the same name as any field in any (transitive) super-struct. All fields (both declared
1300 and inherited) must be specified in any initializers. Inheritance between structures does not give
1301 subtyping or coercion. The super-struct and sub-struct must be defined in the same crate. The super-struct
1302 must be declared using the `virtual` keyword.
1306 virtual struct Sup { x: int }
1307 struct Sub : Sup { y: int }
1308 let s = Sub {x: 10, y: 11};
1314 An _enumeration_ is a simultaneous definition of a nominal [enumerated type](#enumerated-types) as well as a set of *constructors*,
1315 that can be used to create or pattern-match values of the corresponding enumerated type.
1317 Enumerations are declared with the keyword `enum`.
1319 An example of an `enum` item and its use:
1327 let mut a: Animal = Dog;
1331 Enumeration constructors can have either named or unnamed fields:
1334 # #![feature(struct_variant)]
1338 Cat { name: String, weight: f64 }
1341 let mut a: Animal = Dog("Cocoa".to_string(), 37.2);
1342 a = Cat { name: "Spotty".to_string(), weight: 2.7 };
1346 In this example, `Cat` is a _struct-like enum variant_,
1347 whereas `Dog` is simply called an enum variant.
1352 static_item : "static" ident ':' type '=' expr ';' ;
1355 A *static item* is a named _constant value_ stored in the global data section of a crate.
1356 Immutable static items are stored in the read-only data section.
1357 The constant value bound to a static item is, like all constant values, evaluated at compile time.
1358 Static items have the `static` lifetime, which outlives all other lifetimes in a Rust program.
1359 Only values stored in the global data section (such as string constants
1360 and static items) can have the `static` lifetime;
1361 dynamically constructed values cannot safely be assigned the `static` lifetime.
1362 Static items are declared with the `static` keyword.
1363 A static item must have a _constant expression_ giving its definition.
1365 Static items must be explicitly typed.
1366 The type may be ```bool```, ```char```, a number, or a type derived from those primitive types.
1367 The derived types are references with the `static` lifetime,
1368 fixed-size arrays, tuples, and structs.
1371 static BIT1: uint = 1 << 0;
1372 static BIT2: uint = 1 << 1;
1374 static BITS: [uint, ..2] = [BIT1, BIT2];
1375 static STRING: &'static str = "bitstring";
1377 struct BitsNStrings<'a> {
1378 mybits: [uint, ..2],
1382 static bits_n_strings: BitsNStrings<'static> = BitsNStrings {
1388 #### Mutable statics
1390 If a static item is declared with the ```mut``` keyword, then it is allowed to
1391 be modified by the program. One of Rust's goals is to make concurrency bugs hard
1392 to run into, and this is obviously a very large source of race conditions or
1393 other bugs. For this reason, an ```unsafe``` block is required when either
1394 reading or writing a mutable static variable. Care should be taken to ensure
1395 that modifications to a mutable static are safe with respect to other tasks
1396 running in the same process.
1398 Mutable statics are still very useful, however. They can be used with C
1399 libraries and can also be bound from C libraries (in an ```extern``` block).
1402 # fn atomic_add(_: &mut uint, _: uint) -> uint { 2 }
1404 static mut LEVELS: uint = 0;
1406 // This violates the idea of no shared state, and this doesn't internally
1407 // protect against races, so this function is `unsafe`
1408 unsafe fn bump_levels_unsafe1() -> uint {
1414 // Assuming that we have an atomic_add function which returns the old value,
1415 // this function is "safe" but the meaning of the return value may not be what
1416 // callers expect, so it's still marked as `unsafe`
1417 unsafe fn bump_levels_unsafe2() -> uint {
1418 return atomic_add(&mut LEVELS, 1);
1424 A _trait_ describes a set of method types.
1426 Traits can include default implementations of methods,
1427 written in terms of some unknown [`self` type](#self-types);
1428 the `self` type may either be completely unspecified,
1429 or constrained by some other trait.
1431 Traits are implemented for specific types through separate [implementations](#implementations).
1434 # type Surface = int;
1435 # type BoundingBox = int;
1437 fn draw(&self, Surface);
1438 fn bounding_box(&self) -> BoundingBox;
1442 This defines a trait with two methods.
1443 All values that have [implementations](#implementations) of this trait in scope can have their `draw` and `bounding_box` methods called,
1444 using `value.bounding_box()` [syntax](#method-call-expressions).
1446 Type parameters can be specified for a trait to make it generic.
1447 These appear after the trait name, using the same syntax used in [generic functions](#generic-functions).
1451 fn len(&self) -> uint;
1452 fn elt_at(&self, n: uint) -> T;
1453 fn iter(&self, |T|);
1457 Generic functions may use traits as _bounds_ on their type parameters.
1458 This will have two effects: only types that have the trait may instantiate the parameter,
1459 and within the generic function,
1460 the methods of the trait can be called on values that have the parameter's type.
1464 # type Surface = int;
1465 # trait Shape { fn draw(&self, Surface); }
1466 fn draw_twice<T: Shape>(surface: Surface, sh: T) {
1472 Traits also define an [object type](#object-types) with the same name as the trait.
1473 Values of this type are created by [casting](#type-cast-expressions) pointer values
1474 (pointing to a type for which an implementation of the given trait is in scope)
1475 to pointers to the trait name, used as a type.
1479 # impl Shape for int { }
1480 # let mycircle = 0i;
1481 let myshape: Box<Shape> = box mycircle as Box<Shape>;
1484 The resulting value is a box containing the value that was cast,
1485 along with information that identifies the methods of the implementation that was used.
1486 Values with a trait type can have [methods called](#method-call-expressions) on them,
1487 for any method in the trait,
1488 and can be used to instantiate type parameters that are bounded by the trait.
1490 Trait methods may be static,
1491 which means that they lack a `self` argument.
1492 This means that they can only be called with function call syntax (`f(x)`)
1493 and not method call syntax (`obj.f()`).
1494 The way to refer to the name of a static method is to qualify it with the trait name,
1495 treating the trait name like a module.
1500 fn from_int(n: int) -> Self;
1503 fn from_int(n: int) -> f64 { n as f64 }
1505 let x: f64 = Num::from_int(42);
1508 Traits may inherit from other traits. For example, in
1511 trait Shape { fn area() -> f64; }
1512 trait Circle : Shape { fn radius() -> f64; }
1515 the syntax `Circle : Shape` means that types that implement `Circle` must also have an implementation for `Shape`.
1516 Multiple supertraits are separated by `+`, `trait Circle : Shape + PartialEq { }`.
1517 In an implementation of `Circle` for a given type `T`, methods can refer to `Shape` methods,
1518 since the typechecker checks that any type with an implementation of `Circle` also has an implementation of `Shape`.
1520 In type-parameterized functions,
1521 methods of the supertrait may be called on values of subtrait-bound type parameters.
1522 Referring to the previous example of `trait Circle : Shape`:
1525 # trait Shape { fn area(&self) -> f64; }
1526 # trait Circle : Shape { fn radius(&self) -> f64; }
1527 fn radius_times_area<T: Circle>(c: T) -> f64 {
1528 // `c` is both a Circle and a Shape
1529 c.radius() * c.area()
1533 Likewise, supertrait methods may also be called on trait objects.
1536 # trait Shape { fn area(&self) -> f64; }
1537 # trait Circle : Shape { fn radius(&self) -> f64; }
1538 # impl Shape for int { fn area(&self) -> f64 { 0.0 } }
1539 # impl Circle for int { fn radius(&self) -> f64 { 0.0 } }
1541 let mycircle: Circle = ~mycircle as ~Circle;
1542 let nonsense = mycircle.radius() * mycircle.area();
1547 An _implementation_ is an item that implements a [trait](#traits) for a specific type.
1549 Implementations are defined with the keyword `impl`.
1552 # struct Point {x: f64, y: f64};
1553 # type Surface = int;
1554 # struct BoundingBox {x: f64, y: f64, width: f64, height: f64};
1555 # trait Shape { fn draw(&self, Surface); fn bounding_box(&self) -> BoundingBox; }
1556 # fn do_draw_circle(s: Surface, c: Circle) { }
1562 impl Shape for Circle {
1563 fn draw(&self, s: Surface) { do_draw_circle(s, *self); }
1564 fn bounding_box(&self) -> BoundingBox {
1565 let r = self.radius;
1566 BoundingBox{x: self.center.x - r, y: self.center.y - r,
1567 width: 2.0 * r, height: 2.0 * r}
1572 It is possible to define an implementation without referring to a trait.
1573 The methods in such an implementation can only be used
1574 as direct calls on the values of the type that the implementation targets.
1575 In such an implementation, the trait type and `for` after `impl` are omitted.
1576 Such implementations are limited to nominal types (enums, structs),
1577 and the implementation must appear in the same module or a sub-module as the `self` type.
1579 When a trait _is_ specified in an `impl`,
1580 all methods declared as part of the trait must be implemented,
1581 with matching types and type parameter counts.
1583 An implementation can take type parameters,
1584 which can be different from the type parameters taken by the trait it implements.
1585 Implementation parameters are written after the `impl` keyword.
1589 impl<T> Seq<T> for Vec<T> {
1592 impl Seq<bool> for u32 {
1593 /* Treat the integer as a sequence of bits */
1600 extern_block_item : "extern" '{' extern_block '}' ;
1601 extern_block : [ foreign_fn ] * ;
1604 External blocks form the basis for Rust's foreign function interface.
1605 Declarations in an external block describe symbols
1606 in external, non-Rust libraries.
1608 Functions within external blocks
1609 are declared in the same way as other Rust functions,
1610 with the exception that they may not have a body
1611 and are instead terminated by a semicolon.
1615 use libc::{c_char, FILE};
1618 fn fopen(filename: *const c_char, mode: *const c_char) -> *mut FILE;
1623 Functions within external blocks may be called by Rust code,
1624 just like functions defined in Rust.
1625 The Rust compiler automatically translates
1626 between the Rust ABI and the foreign ABI.
1628 A number of [attributes](#attributes) control the behavior of external
1631 By default external blocks assume that the library they are calling
1632 uses the standard C "cdecl" ABI. Other ABIs may be specified using
1633 an `abi` string, as shown here:
1636 // Interface to the Windows API
1637 extern "stdcall" { }
1640 The `link` attribute allows the name of the library to be specified. When
1641 specified the compiler will attempt to link against the native library of the
1645 #[link(name = "crypto")]
1649 The type of a function declared in an extern block is `extern "abi" fn(A1,
1650 ..., An) -> R`, where `A1...An` are the declared types of its arguments and
1651 `R` is the declared return type.
1653 ## Visibility and Privacy
1655 These two terms are often used interchangeably, and what they are attempting to
1656 convey is the answer to the question "Can this item be used at this location?"
1658 Rust's name resolution operates on a global hierarchy of namespaces. Each level
1659 in the hierarchy can be thought of as some item. The items are one of those
1660 mentioned above, but also include external crates. Declaring or defining a new
1661 module can be thought of as inserting a new tree into the hierarchy at the
1662 location of the definition.
1664 To control whether interfaces can be used across modules, Rust checks each use
1665 of an item to see whether it should be allowed or not. This is where privacy
1666 warnings are generated, or otherwise "you used a private item of another module
1667 and weren't allowed to."
1669 By default, everything in rust is *private*, with one exception. Enum variants
1670 in a `pub` enum are also public by default. You are allowed to alter this
1671 default visibility with the `priv` keyword. When an item is declared as `pub`,
1672 it can be thought of as being accessible to the outside world. For example:
1676 // Declare a private struct
1679 // Declare a public struct with a private field
1684 // Declare a public enum with two public variants
1686 PubliclyAccessibleState,
1687 PubliclyAccessibleState2,
1691 With the notion of an item being either public or private, Rust allows item
1692 accesses in two cases:
1694 1. If an item is public, then it can be used externally through any of its
1696 2. If an item is private, it may be accessed by the current module and its
1699 These two cases are surprisingly powerful for creating module hierarchies
1700 exposing public APIs while hiding internal implementation details. To help
1701 explain, here's a few use cases and what they would entail.
1703 * A library developer needs to expose functionality to crates which link against
1704 their library. As a consequence of the first case, this means that anything
1705 which is usable externally must be `pub` from the root down to the destination
1706 item. Any private item in the chain will disallow external accesses.
1708 * A crate needs a global available "helper module" to itself, but it doesn't
1709 want to expose the helper module as a public API. To accomplish this, the root
1710 of the crate's hierarchy would have a private module which then internally has
1711 a "public api". Because the entire crate is a descendant of the root, then the
1712 entire local crate can access this private module through the second case.
1714 * When writing unit tests for a module, it's often a common idiom to have an
1715 immediate child of the module to-be-tested named `mod test`. This module could
1716 access any items of the parent module through the second case, meaning that
1717 internal implementation details could also be seamlessly tested from the child
1720 In the second case, it mentions that a private item "can be accessed" by the
1721 current module and its descendants, but the exact meaning of accessing an item
1722 depends on what the item is. Accessing a module, for example, would mean looking
1723 inside of it (to import more items). On the other hand, accessing a function
1724 would mean that it is invoked. Additionally, path expressions and import
1725 statements are considered to access an item in the sense that the
1726 import/expression is only valid if the destination is in the current visibility
1729 Here's an example of a program which exemplifies the three cases outlined above.
1732 // This module is private, meaning that no external crate can access this
1733 // module. Because it is private at the root of this current crate, however, any
1734 // module in the crate may access any publicly visible item in this module.
1735 mod crate_helper_module {
1737 // This function can be used by anything in the current crate
1738 pub fn crate_helper() {}
1740 // This function *cannot* be used by anything else in the crate. It is not
1741 // publicly visible outside of the `crate_helper_module`, so only this
1742 // current module and its descendants may access it.
1743 fn implementation_detail() {}
1746 // This function is "public to the root" meaning that it's available to external
1747 // crates linking against this one.
1748 pub fn public_api() {}
1750 // Similarly to 'public_api', this module is public so external crates may look
1753 use crate_helper_module;
1755 pub fn my_method() {
1756 // Any item in the local crate may invoke the helper module's public
1757 // interface through a combination of the two rules above.
1758 crate_helper_module::crate_helper();
1761 // This function is hidden to any module which is not a descendant of
1763 fn my_implementation() {}
1769 fn test_my_implementation() {
1770 // Because this module is a descendant of `submodule`, it's allowed
1771 // to access private items inside of `submodule` without a privacy
1773 super::my_implementation();
1781 For a rust program to pass the privacy checking pass, all paths must be valid
1782 accesses given the two rules above. This includes all use statements,
1783 expressions, types, etc.
1785 ### Re-exporting and Visibility
1787 Rust allows publicly re-exporting items through a `pub use` directive. Because
1788 this is a public directive, this allows the item to be used in the current
1789 module through the rules above. It essentially allows public access into the
1790 re-exported item. For example, this program is valid:
1793 pub use api = self::implementation;
1795 mod implementation {
1802 This means that any external crate referencing `implementation::f` would receive
1803 a privacy violation, while the path `api::f` would be allowed.
1805 When re-exporting a private item, it can be thought of as allowing the "privacy
1806 chain" being short-circuited through the reexport instead of passing through the
1807 namespace hierarchy as it normally would.
1809 ### Glob imports and Visibility
1811 Currently glob imports are considered an "experimental" language feature. For
1812 sanity purpose along with helping the implementation, glob imports will only
1813 import public items from their destination, not private items.
1815 > **Note:** This is subject to change, glob exports may be removed entirely or
1816 > they could possibly import private items for a privacy error to later be
1817 > issued if the item is used.
1822 attribute : '#' '!' ? '[' meta_item ']' ;
1823 meta_item : ident [ '=' literal
1824 | '(' meta_seq ')' ] ? ;
1825 meta_seq : meta_item [ ',' meta_seq ] ? ;
1828 Any item declaration may have an _attribute_ applied to it. Attributes in Rust
1829 are modeled on Attributes in ECMA-335, with the syntax coming from ECMA-334
1830 (C#). An attribute is a general, free-form metadatum that is interpreted
1831 according to name, convention, and language and compiler version. Attributes
1832 may appear as any of:
1834 * A single identifier, the attribute name
1835 * An identifier followed by the equals sign '=' and a literal, providing a
1837 * An identifier followed by a parenthesized list of sub-attribute arguments
1839 Attributes with a bang ("!") after the hash ("#") apply to the item that the
1840 attribute is declared within. Attributes that do not have a bang after the
1841 hash apply to the item that follows the attribute.
1843 An example of attributes:
1846 // General metadata applied to the enclosing module or crate.
1849 // A function marked as a unit test
1855 // A conditionally-compiled module
1856 #[cfg(target_os="linux")]
1861 // A lint attribute used to suppress a warning/error
1862 #[allow(non_camel_case_types)]
1866 > **Note:** At some point in the future, the compiler will distinguish between
1867 > language-reserved and user-available attributes. Until then, there is
1868 > effectively no difference between an attribute handled by a loadable syntax
1869 > extension and the compiler.
1871 ### Crate-only attributes
1873 - `crate_id` - specify the this crate's crate ID.
1874 - `crate_type` - see [linkage](#linkage).
1875 - `feature` - see [compiler features](#compiler-features).
1876 - `no_builtins` - disable optimizing certain code patterns to invocations of
1877 library functions that are assumed to exist
1878 - `no_main` - disable emitting the `main` symbol. Useful when some other
1879 object being linked to defines `main`.
1880 - `no_start` - disable linking to the `native` crate, which specifies the
1881 "start" language item.
1882 - `no_std` - disable linking to the `std` crate.
1884 ### Module-only attributes
1886 - `macro_escape` - macros defined in this module will be visible in the
1887 module's parent, after this module has been included.
1888 - `no_implicit_prelude` - disable injecting `use std::prelude::*` in this
1890 - `path` - specifies the file to load the module from. `#[path="foo.rs"] mod
1891 bar;` is equivalent to `mod bar { /* contents of foo.rs */ }`. The path is
1892 taken relative to the directory that the current module is in.
1894 ### Function-only attributes
1896 - `main` - indicates that this function should be passed to the entry point,
1897 rather than the function in the crate root named `main`.
1898 - `plugin_registrar` - mark this function as the registration point for
1899 compiler plugins, such as loadable syntax extensions.
1900 - `start` - indicates that this function should be used as the entry point,
1901 overriding the "start" language item. See the "start" [language
1902 item](#language-items) for more details.
1904 ### Static-only attributes
1906 - `thread_local` - on a `static mut`, this signals that the value of this
1907 static may change depending on the current thread. The exact consequences of
1908 this are implementation-defined.
1912 On an `extern` block, the following attributes are interpreted:
1914 - `link_args` - specify arguments to the linker, rather than just the library
1915 name and type. This is feature gated and the exact behavior is
1916 implementation-defined (due to variety of linker invocation syntax).
1917 - `link` - indicate that a native library should be linked to for the
1918 declarations in this block to be linked correctly. See [external
1919 blocks](#external-blocks)
1921 On declarations inside an `extern` block, the following attributes are
1924 - `link_name` - the name of the symbol that this function or static should be
1926 - `linkage` - on a static, this specifies the [linkage
1927 type](http://llvm.org/docs/LangRef.html#linkage-types).
1929 ### Miscellaneous attributes
1931 - `link_section` - on statics and functions, this specifies the section of the
1932 object file that this item's contents will be placed into.
1933 - `macro_export` - export a macro for cross-crate usage.
1934 - `no_mangle` - on any item, do not apply the standard name mangling. Set the
1935 symbol for this item to its identifier.
1936 - `packed` - on structs or enums, eliminate any padding that would be used to
1938 - `phase` - on `extern crate` statements, allows specifying which "phase" of
1939 compilation the crate should be loaded for. Currently, there are two
1940 choices: `link` and `plugin`. `link` is the default. `plugin` will load the
1941 crate at compile-time and use any syntax extensions or lints that the crate
1942 defines. They can both be specified, `#[phase(link, plugin)]` to use a crate
1943 both at runtime and compiletime.
1944 - `repr` - on C-like enums, this sets the underlying type used for
1945 representation. Useful for FFI. Takes one argument, which is the primitive
1946 type this enum should be represented for, or `C`, which specifies that it
1947 should be the default `enum` size of the C ABI for that platform. Note that
1948 enum representation in C is undefined, and this may be incorrect when the C
1949 code is compiled with certain flags.
1950 - `simd` - on certain tuple structs, derive the arithmetic operators, which
1951 lower to the target's SIMD instructions, if any; the `simd` feature gate
1952 is necessary to use this attribute.
1953 - `static_assert` - on statics whose type is `bool`, terminates compilation
1954 with an error if it is not initialized to `true`.
1955 - `unsafe_destructor` - allow implementations of the "drop" language item
1956 where the type it is implemented for does not implement the "send" language
1957 item; the `unsafe_destructor` feature gate is needed to use this attribute
1958 - `unsafe_no_drop_flag` - on structs, remove the flag that prevents
1959 destructors from being run twice. Destructors might be run multiple times on
1960 the same object with this attribute.
1962 ### Conditional compilation
1964 Sometimes one wants to have different compiler outputs from the same code,
1965 depending on build target, such as targeted operating system, or to enable
1968 There are two kinds of configuration options, one that is either defined or not
1969 (`#[cfg(foo)]`), and the other that contains a string that can be checked
1970 against (`#[cfg(bar = "baz")]` (currently only compiler-defined configuration
1971 options can have the latter form).
1974 // The function is only included in the build when compiling for OSX
1975 #[cfg(target_os = "macos")]
1980 // This function is only included when either foo or bar is defined
1983 fn needs_foo_or_bar() {
1987 // This function is only included when compiling for a unixish OS with a 32-bit
1989 #[cfg(unix, target_word_size = "32")]
1990 fn on_32bit_unix() {
1995 This illustrates some conditional compilation can be achieved using the
1996 `#[cfg(...)]` attribute. Note that `#[cfg(foo, bar)]` is a condition that needs
1997 both `foo` and `bar` to be defined while `#[cfg(foo)] #[cfg(bar)]` only needs
1998 one of `foo` and `bar` to be defined (this resembles in the disjunctive normal
1999 form). Additionally, one can reverse a condition by enclosing it in a
2000 `not(...)`, like e. g. `#[cfg(not(target_os = "win32"))]`.
2002 The following configurations must be defined by the implementation:
2004 * `target_arch = "..."`. Target CPU architecture, such as `"x86"`, `"x86_64"`
2005 `"mips"`, or `"arm"`.
2006 * `target_endian = "..."`. Endianness of the target CPU, either `"little"` or
2008 * `target_family = "..."`. Operating system family of the target, e. g.
2009 `"unix"` or `"windows"`. The value of this configuration option is defined as
2010 a configuration itself, like `unix` or `windows`.
2011 * `target_os = "..."`. Operating system of the target, examples include
2012 `"win32"`, `"macos"`, `"linux"`, `"android"`, `"freebsd"` or `"dragonfly"`.
2013 * `target_word_size = "..."`. Target word size in bits. This is set to `"32"`
2014 for targets with 32-bit pointers, and likewise set to `"64"` for 64-bit
2016 * `unix`. See `target_family`.
2017 * `windows`. See `target_family`.
2019 ### Lint check attributes
2021 A lint check names a potentially undesirable coding pattern, such as
2022 unreachable code or omitted documentation, for the static entity to
2023 which the attribute applies.
2025 For any lint check `C`:
2027 * `allow(C)` overrides the check for `C` so that violations will go
2029 * `deny(C)` signals an error after encountering a violation of `C`,
2030 * `forbid(C)` is the same as `deny(C)`, but also forbids changing the lint
2032 * `warn(C)` warns about violations of `C` but continues compilation.
2034 The lint checks supported by the compiler can be found via `rustc -W help`,
2035 along with their default settings.
2039 // Missing documentation is ignored here
2040 #[allow(missing_doc)]
2041 pub fn undocumented_one() -> int { 1 }
2043 // Missing documentation signals a warning here
2044 #[warn(missing_doc)]
2045 pub fn undocumented_too() -> int { 2 }
2047 // Missing documentation signals an error here
2048 #[deny(missing_doc)]
2049 pub fn undocumented_end() -> int { 3 }
2053 This example shows how one can use `allow` and `warn` to toggle
2054 a particular check on and off.
2057 #[warn(missing_doc)]
2059 #[allow(missing_doc)]
2061 // Missing documentation is ignored here
2062 pub fn undocumented_one() -> int { 1 }
2064 // Missing documentation signals a warning here,
2065 // despite the allow above.
2066 #[warn(missing_doc)]
2067 pub fn undocumented_two() -> int { 2 }
2070 // Missing documentation signals a warning here
2071 pub fn undocumented_too() -> int { 3 }
2075 This example shows how one can use `forbid` to disallow uses
2076 of `allow` for that lint check.
2079 #[forbid(missing_doc)]
2081 // Attempting to toggle warning signals an error here
2082 #[allow(missing_doc)]
2084 pub fn undocumented_too() -> int { 2 }
2090 Some primitive Rust operations are defined in Rust code, rather than being
2091 implemented directly in C or assembly language. The definitions of these
2092 operations have to be easy for the compiler to find. The `lang` attribute
2093 makes it possible to declare these operations. For example, the `str` module
2094 in the Rust standard library defines the string equality function:
2098 pub fn eq_slice(a: &str, b: &str) -> bool {
2103 The name `str_eq` has a special meaning to the Rust compiler,
2104 and the presence of this definition means that it will use this definition
2105 when generating calls to the string equality function.
2107 A complete list of the built-in language items follows:
2109 #### Built-in Traits
2112 : Types that do not move ownership when used by-value.
2116 : Able to be sent across task boundaries.
2118 : Has a size known at compile time.
2120 : Able to be safely shared between tasks when aliased.
2124 These language items are traits:
2127 : Elements can be added (for example, integers and floats).
2129 : Elements can be subtracted.
2131 : Elements can be multiplied.
2133 : Elements have a division operation.
2135 : Elements have a remainder operation.
2137 : Elements can be negated arithmetically.
2139 : Elements can be negated logically.
2141 : Elements have an exclusive-or operation.
2143 : Elements have a bitwise `and` operation.
2145 : Elements have a bitwise `or` operation.
2147 : Elements have a left shift operation.
2149 : Elements have a right shift operation.
2151 : Elements can be indexed.
2153 : ___Needs filling in___
2155 : Elements can be compared for equality.
2157 : Elements have a partial ordering.
2159 : `*` can be applied, yielding a reference to another type
2161 : `*` can be applied, yielding a mutable reference to another type
2163 These are functions:
2166 : ___Needs filling in___
2168 : ___Needs filling in___
2170 : ___Needs filling in___
2172 : Compare two strings (`&str`) for equality.
2174 : Return a new unique string
2175 containing a copy of the contents of a unique string.
2180 : The type returned by the `type_id` intrinsic.
2182 : A type whose contents can be mutated through an immutable reference
2186 These types help drive the compiler's analysis
2189 : ___Needs filling in___
2191 : This type does not implement "copy", even if eligible
2193 : This type does not implement "send", even if eligible
2195 : This type does not implement "sync", even if eligible
2197 : This type implements "managed"
2199 : ___Needs filling in___
2201 : Free memory that was allocated on the exchange heap.
2203 : Allocate memory on the exchange heap.
2204 * `closure_exchange_malloc`
2205 : ___Needs filling in___
2207 : Abort the program with an error.
2208 * `fail_bounds_check`
2209 : Abort the program with a bounds check error.
2211 : Free memory that was allocated on the managed heap.
2213 : ___Needs filling in___
2215 : ___Needs filling in___
2217 : ___Needs filling in___
2219 : ___Needs filling in___
2220 * `contravariant_lifetime`
2221 : The lifetime parameter should be considered contravariant
2222 * `covariant_lifetime`
2223 : The lifetime parameter should be considered covariant
2224 * `invariant_lifetime`
2225 : The lifetime parameter should be considered invariant
2227 : Allocate memory on the managed heap.
2229 : ___Needs filling in___
2231 : ___Needs filling in___
2233 : ___Needs filling in___
2235 : ___Needs filling in___
2236 * `contravariant_type`
2237 : The type parameter should be considered contravariant
2239 : The type parameter should be considered covariant
2241 : The type parameter should be considered invariant
2243 : ___Needs filling in___
2245 : ___Needs filling in___
2247 > **Note:** This list is likely to become out of date. We should auto-generate it
2248 > from `librustc/middle/lang_items.rs`.
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
2261 unless marked with `#[inline(never)]`.
2263 whether two different inlineable statics
2264 have the same memory address.
2266 the compiler is free
2267 to collapse duplicate inlineable statics together.
2269 `#[inline]` and `#[inline(always)]` always causes the function to be serialized
2270 into crate metadata to allow cross-crate inlining.
2272 There are three different types of inline attributes:
2274 * `#[inline]` hints the compiler to perform an inline expansion.
2275 * `#[inline(always)]` asks the compiler to always perform an inline expansion.
2276 * `#[inline(never)]` asks the compiler to never perform an inline expansion.
2280 The `deriving` attribute allows certain traits to be automatically
2281 implemented for data structures. For example, the following will
2282 create an `impl` for the `PartialEq` and `Clone` traits for `Foo`, the type
2283 parameter `T` will be given the `PartialEq` or `Clone` constraints for the
2287 #[deriving(PartialEq, Clone)]
2294 The generated `impl` for `PartialEq` is equivalent to
2297 # struct Foo<T> { a: int, b: T }
2298 impl<T: PartialEq> PartialEq for Foo<T> {
2299 fn eq(&self, other: &Foo<T>) -> bool {
2300 self.a == other.a && self.b == other.b
2303 fn ne(&self, other: &Foo<T>) -> bool {
2304 self.a != other.a || self.b != other.b
2309 Supported traits for `deriving` are:
2311 * Comparison traits: `PartialEq`, `Eq`, `PartialOrd`, `Ord`.
2312 * Serialization: `Encodable`, `Decodable`. These require `serialize`.
2313 * `Clone`, to create `T` from `&T` via a copy.
2314 * `Default`, to create an empty instance of a data type.
2315 * `FromPrimitive`, to create an instance from a numeric primitive.
2316 * `Hash`, to iterate over the bytes in a data type.
2317 * `Rand`, to create a random instance of a data type.
2318 * `Show`, to format a value using the `{}` formatter.
2319 * `Zero`, to create a zero instance of a numeric data type.
2323 One can indicate the stability of an API using the following attributes:
2325 * `deprecated`: This item should no longer be used, e.g. it has been
2326 replaced. No guarantee of backwards-compatibility.
2327 * `experimental`: This item was only recently introduced or is
2328 otherwise in a state of flux. It may change significantly, or even
2329 be removed. No guarantee of backwards-compatibility.
2330 * `unstable`: This item is still under development, but requires more
2331 testing to be considered stable. No guarantee of backwards-compatibility.
2332 * `stable`: This item is considered stable, and will not change
2333 significantly. Guarantee of backwards-compatibility.
2334 * `frozen`: This item is very stable, and is unlikely to
2335 change. Guarantee of backwards-compatibility.
2336 * `locked`: This item will never change unless a serious bug is
2337 found. Guarantee of backwards-compatibility.
2339 These levels are directly inspired by
2340 [Node.js' "stability index"](http://nodejs.org/api/documentation.html).
2342 Stability levels are inherited, so an items's stability attribute is the
2343 default stability for everything nested underneath it.
2345 There are lints for disallowing items marked with certain levels: `deprecated`,
2346 `experimental` and `unstable`. For now, only `deprecated` warns by default, but
2347 this will change once the standard library has been stabilized.
2348 Stability levels are meant to be promises at the crate
2349 level, so these lints only apply when referencing
2350 items from an _external_ crate, not to items defined within the
2351 current crate. Items with no stability level are considered
2352 to be unstable for the purposes of the lint. One can give an optional
2353 string that will be displayed when the lint flags the use of an item.
2355 For example, if we define one crate called `stability_levels`:
2358 #[deprecated="replaced by `best`"]
2360 // delete everything
2364 // delete fewer things
2373 then the lints will work as follows for a client crate:
2377 extern crate stability_levels;
2378 use stability_levels::{bad, better, best};
2381 bad(); // "warning: use of deprecated item: replaced by `best`"
2383 better(); // "warning: use of unmarked item"
2385 best(); // no warning
2389 > **Note:** Currently these are only checked when applied to
2390 > individual functions, structs, methods and enum variants, *not* to
2391 > entire modules, traits, impls or enums themselves.
2393 ### Compiler Features
2395 Certain aspects of Rust may be implemented in the compiler, but they're not
2396 necessarily ready for every-day use. These features are often of "prototype
2397 quality" or "almost production ready", but may not be stable enough to be
2398 considered a full-fledged language feature.
2400 For this reason, Rust recognizes a special crate-level attribute of the form:
2403 #![feature(feature1, feature2, feature3)]
2406 This directive informs the compiler that the feature list: `feature1`,
2407 `feature2`, and `feature3` should all be enabled. This is only recognized at a
2408 crate-level, not at a module-level. Without this directive, all features are
2409 considered off, and using the features will result in a compiler error.
2411 The currently implemented features of the reference compiler are:
2413 * `asm` - The `asm!` macro provides a means for inline assembly. This is often
2414 useful, but the exact syntax for this feature along with its semantics
2415 are likely to change, so this macro usage must be opted into.
2417 * `concat_idents` - Allows use of the `concat_idents` macro, which is in many
2418 ways insufficient for concatenating identifiers, and may
2419 be removed entirely for something more wholsome.
2421 * `default_type_params` - Allows use of default type parameters. The future of
2422 this feature is uncertain.
2424 * `globs` - Importing everything in a module through `*`. This is currently a
2425 large source of bugs in name resolution for Rust, and it's not clear
2426 whether this will continue as a feature or not. For these reasons,
2427 the glob import statement has been hidden behind this feature flag.
2429 * `intrinsics` - Allows use of the "rust-intrinsics" ABI. Compiler intrinsics
2430 are inherently unstable and no promise about them is made.
2432 * `lang_items` - Allows use of the `#[lang]` attribute. Like `intrinsics`,
2433 lang items are inherently unstable and no promise about
2436 * `link_args` - This attribute is used to specify custom flags to the linker,
2437 but usage is strongly discouraged. The compiler's usage of the
2438 system linker is not guaranteed to continue in the future, and
2439 if the system linker is not used then specifying custom flags
2440 doesn't have much meaning.
2442 * `linkage` - Allows use of the `linkage` attribute, which is not portable.
2444 * `log_syntax` - Allows use of the `log_syntax` macro attribute, which is a
2445 nasty hack that will certainly be removed.
2447 * `macro_rules` - The definition of new macros. This does not encompass
2448 macro-invocation, that is always enabled by default, this only
2449 covers the definition of new macros. There are currently
2450 various problems with invoking macros, how they interact with
2451 their environment, and possibly how they are used outside of
2452 location in which they are defined. Macro definitions are
2453 likely to change slightly in the future, so they are currently
2454 hidden behind this feature.
2456 * `managed_boxes` - Usage of `@` is gated due to many
2457 planned changes to this feature. In the past, this has meant
2458 "a GC pointer", but the current implementation uses
2459 reference counting and will likely change drastically over
2460 time. Additionally, the `@` syntax will no longer be used to
2463 * `non_ascii_idents` - The compiler supports the use of non-ascii identifiers,
2464 but the implementation is a little rough around the
2465 edges, so this can be seen as an experimental feature for
2466 now until the specification of identifiers is fully
2469 * `once_fns` - Onceness guarantees a closure is only executed once. Defining a
2470 closure as `once` is unlikely to be supported going forward. So
2471 they are hidden behind this feature until they are to be removed.
2473 * `overloaded_calls` - Allow implementing the `Fn*` family of traits on user
2474 types, allowing overloading the call operator (`()`).
2475 This feature may still undergo changes before being
2478 * `phase` - Usage of the `#[phase]` attribute allows loading compiler plugins
2479 for custom lints or syntax extensions. The implementation is considered
2480 unwholesome and in need of overhaul, and it is not clear what they
2481 will look like moving forward.
2483 * `plugin_registrar` - Indicates that a crate has compiler plugins that it
2484 wants to load. As with `phase`, the implementation is
2485 in need of a overhaul, and it is not clear that plugins
2486 defined using this will continue to work.
2488 * `quote` - Allows use of the `quote_*!` family of macros, which are
2489 implemented very poorly and will likely change significantly
2490 with a proper implementation.
2492 * `rustc_diagnostic_macros`- A mysterious feature, used in the implementation
2493 of rustc, not meant for mortals.
2495 * `simd` - Allows use of the `#[simd]` attribute, which is overly simple and
2496 not the SIMD interface we want to expose in the long term.
2498 * `struct_inherit` - Allows using struct inheritance, which is barely
2499 implemented and will probably be removed. Don't use this.
2501 * `struct_variant` - Structural enum variants (those with named fields). It is
2502 currently unknown whether this style of enum variant is as
2503 fully supported as the tuple-forms, and it's not certain
2504 that this style of variant should remain in the language.
2505 For now this style of variant is hidden behind a feature
2508 * `thread_local` - The usage of the `#[thread_local]` attribute is experimental
2509 and should be seen as unstable. This attribute is used to
2510 declare a `static` as being unique per-thread leveraging
2511 LLVM's implementation which works in concert with the kernel
2512 loader and dynamic linker. This is not necessarily available
2513 on all platforms, and usage of it is discouraged (rust
2514 focuses more on task-local data instead of thread-local
2517 * `trace_macros` - Allows use of the `trace_macros` macro, which is a nasty
2518 hack that will certainly be removed.
2520 * `unboxed_closure_sugar` - Allows using `|Foo| -> Bar` as a trait bound
2521 meaning one of the `Fn` traits. Still
2524 * `unboxed_closures` - A work in progress feature with many known bugs.
2526 * `unsafe_destructor` - Allows use of the `#[unsafe_destructor]` attribute,
2527 which is considered wildly unsafe and will be
2528 obsoleted by language improvements.
2530 If a feature is promoted to a language feature, then all existing programs will
2531 start to receive compilation warnings about #[feature] directives which enabled
2532 the new feature (because the directive is no longer necessary). However, if
2533 a feature is decided to be removed from the language, errors will be issued (if
2534 there isn't a parser error first). The directive in this case is no longer
2535 necessary, and it's likely that existing code will break if the feature isn't
2538 If a unknown feature is found in a directive, it results in a compiler error. An
2539 unknown feature is one which has never been recognized by the compiler.
2541 # Statements and expressions
2543 Rust is _primarily_ an expression language. This means that most forms of
2544 value-producing or effect-causing evaluation are directed by the uniform
2545 syntax category of _expressions_. Each kind of expression can typically _nest_
2546 within each other kind of expression, and rules for evaluation of expressions
2547 involve specifying both the value produced by the expression and the order in
2548 which its sub-expressions are themselves evaluated.
2550 In contrast, statements in Rust serve _mostly_ to contain and explicitly
2551 sequence expression evaluation.
2555 A _statement_ is a component of a block, which is in turn a component of an
2556 outer [expression](#expressions) or [function](#functions).
2558 Rust has two kinds of statement:
2559 [declaration statements](#declaration-statements) and
2560 [expression statements](#expression-statements).
2562 ### Declaration statements
2564 A _declaration statement_ is one that introduces one or more *names* into the enclosing statement block.
2565 The declared names may denote new slots or new items.
2567 #### Item declarations
2569 An _item declaration statement_ has a syntactic form identical to an
2570 [item](#items) declaration within a module. Declaring an item — a function,
2571 enumeration, structure, type, static, trait, implementation or module — locally
2572 within a statement block is simply a way of restricting its scope to a narrow
2573 region containing all of its uses; it is otherwise identical in meaning to
2574 declaring the item outside the statement block.
2576 Note: there is no implicit capture of the function's dynamic environment when
2577 declaring a function-local item.
2579 #### Slot declarations
2582 let_decl : "let" pat [':' type ] ? [ init ] ? ';' ;
2583 init : [ '=' ] expr ;
2586 A _slot declaration_ introduces a new set of slots, given by a pattern.
2587 The pattern may be followed by a type annotation, and/or an initializer expression.
2588 When no type annotation is given, the compiler will infer the type,
2589 or signal an error if insufficient type information is available for definite inference.
2590 Any slots introduced by a slot declaration are visible from the point of declaration until the end of the enclosing block scope.
2592 ### Expression statements
2594 An _expression statement_ is one that evaluates an [expression](#expressions)
2595 and ignores its result.
2596 The type of an expression statement `e;` is always `()`, regardless of the type of `e`.
2597 As a rule, an expression statement's purpose is to trigger the effects of evaluating its expression.
2601 An expression may have two roles: it always produces a *value*, and it may have *effects*
2602 (otherwise known as "side effects").
2603 An expression *evaluates to* a value, and has effects during *evaluation*.
2604 Many expressions contain sub-expressions (operands).
2605 The meaning of each kind of expression dictates several things:
2606 * Whether or not to evaluate the sub-expressions when evaluating the expression
2607 * The order in which to evaluate the sub-expressions
2608 * How to combine the sub-expressions' values to obtain the value of the expression.
2610 In this way, the structure of expressions dictates the structure of execution.
2611 Blocks are just another kind of expression,
2612 so blocks, statements, expressions, and blocks again can recursively nest inside each other
2613 to an arbitrary depth.
2615 #### Lvalues, rvalues and temporaries
2617 Expressions are divided into two main categories: _lvalues_ and _rvalues_.
2618 Likewise within each expression, sub-expressions may occur in _lvalue context_ or _rvalue context_.
2619 The evaluation of an expression depends both on its own category and the context it occurs within.
2621 An lvalue is an expression that represents a memory location. These
2622 expressions are [paths](#path-expressions) (which refer to local
2623 variables, function and method arguments, or static variables),
2624 dereferences (`*expr`), [indexing expressions](#index-expressions)
2625 (`expr[expr]`), and [field references](#field-expressions) (`expr.f`).
2626 All other expressions are rvalues.
2628 The left operand of an [assignment](#assignment-expressions) or
2629 [compound-assignment](#compound-assignment-expressions) expression is an lvalue context,
2630 as is the single operand of a unary [borrow](#unary-operator-expressions).
2631 All other expression contexts are rvalue contexts.
2633 When an lvalue is evaluated in an _lvalue context_, it denotes a memory location;
2634 when evaluated in an _rvalue context_, it denotes the value held _in_ that memory location.
2636 When an rvalue is used in lvalue context, a temporary un-named lvalue is created and used instead.
2637 A temporary's lifetime equals the largest lifetime of any reference that points to it.
2639 #### Moved and copied types
2641 When a [local variable](#memory-slots) is used
2642 as an [rvalue](#lvalues,-rvalues-and-temporaries)
2643 the variable will either be moved or copied, depending on its type.
2644 For types that contain [owning pointers](#pointer-types)
2645 or values that implement the special trait `Drop`,
2646 the variable is moved.
2647 All other types are copied.
2649 ### Literal expressions
2651 A _literal expression_ consists of one of the [literal](#literals)
2652 forms described earlier. It directly describes a number, character,
2653 string, boolean value, or the unit value.
2657 "hello"; // string type
2658 '5'; // character type
2662 ### Path expressions
2664 A [path](#paths) used as an expression context denotes either a local variable or an item.
2665 Path expressions are [lvalues](#lvalues,-rvalues-and-temporaries).
2667 ### Tuple expressions
2669 Tuples are written by enclosing one or more comma-separated
2670 expressions in parentheses. They are used to create [tuple-typed](#tuple-types)
2679 ### Structure expressions
2682 struct_expr : expr_path '{' ident ':' expr
2683 [ ',' ident ':' expr ] *
2686 [ ',' expr ] * ')' |
2690 There are several forms of structure expressions.
2691 A _structure expression_ consists of the [path](#paths) of a [structure item](#structures),
2692 followed by a brace-enclosed list of one or more comma-separated name-value pairs,
2693 providing the field values of a new instance of the structure.
2694 A field name can be any identifier, and is separated from its value expression by a colon.
2695 The location denoted by a structure field is mutable if and only if the enclosing structure is mutable.
2697 A _tuple structure expression_ consists of the [path](#paths) of a [structure item](#structures),
2698 followed by a parenthesized list of one or more comma-separated expressions
2699 (in other words, the path of a structure item followed by a tuple expression).
2700 The structure item must be a tuple structure item.
2702 A _unit-like structure expression_ consists only of the [path](#paths) of a [structure item](#structures).
2704 The following are examples of structure expressions:
2707 # struct Point { x: f64, y: f64 }
2708 # struct TuplePoint(f64, f64);
2709 # mod game { pub struct User<'a> { pub name: &'a str, pub age: uint, pub score: uint } }
2710 # struct Cookie; fn some_fn<T>(t: T) {}
2711 Point {x: 10.0, y: 20.0};
2712 TuplePoint(10.0, 20.0);
2713 let u = game::User {name: "Joe", age: 35, score: 100_000};
2714 some_fn::<Cookie>(Cookie);
2717 A structure expression forms a new value of the named structure type.
2718 Note that for a given *unit-like* structure type, this will always be the same value.
2720 A structure expression can terminate with the syntax `..` followed by an expression to denote a functional update.
2721 The expression following `..` (the base) must have the same structure type as the new structure type being formed.
2722 The entire expression denotes the result of constructing a new structure
2723 (with the same type as the base expression)
2724 with the given values for the fields that were explicitly specified
2725 and the values in the base expression for all other fields.
2728 # struct Point3d { x: int, y: int, z: int }
2729 let base = Point3d {x: 1, y: 2, z: 3};
2730 Point3d {y: 0, z: 10, .. base};
2733 ### Block expressions
2736 block_expr : '{' [ view_item ] *
2737 [ stmt ';' | item ] *
2741 A _block expression_ is similar to a module in terms of the declarations that
2742 are possible. Each block conceptually introduces a new namespace scope. View
2743 items can bring new names into scopes and declared items are in scope for only
2746 A block will execute each statement sequentially, and then execute the
2747 expression (if given). If the final expression is omitted, the type and return
2748 value of the block are `()`, but if it is provided, the type and return value
2749 of the block are that of the expression itself.
2751 ### Method-call expressions
2754 method_call_expr : expr '.' ident paren_expr_list ;
2757 A _method call_ consists of an expression followed by a single dot, an identifier, and a parenthesized expression-list.
2758 Method calls are resolved to methods on specific traits,
2759 either statically dispatching to a method if the exact `self`-type of the left-hand-side is known,
2760 or dynamically dispatching if the left-hand-side expression is an indirect [object type](#object-types).
2762 ### Field expressions
2765 field_expr : expr '.' ident ;
2768 A _field expression_ consists of an expression followed by a single dot and an identifier,
2769 when not immediately followed by a parenthesized expression-list (the latter is a [method call expression](#method-call-expressions)).
2770 A field expression denotes a field of a [structure](#structure-types).
2772 ~~~~ {.ignore .field}
2775 (Struct {a: 10, b: 20}).a;
2778 A field access is an [lvalue](#lvalues,-rvalues-and-temporaries) referring to the value of that field.
2779 When the type providing the field inherits mutabilty, it can be [assigned](#assignment-expressions) to.
2781 Also, if the type of the expression to the left of the dot is a pointer,
2782 it is automatically dereferenced to make the field access possible.
2784 ### Vector expressions
2787 vec_expr : '[' "mut" ? vec_elems? ']' ;
2789 vec_elems : [expr [',' expr]*] | [expr ',' ".." expr] ;
2792 A [_vector_](#vector-types) _expression_ is written by enclosing zero or
2793 more comma-separated expressions of uniform type in square brackets.
2795 In the `[expr ',' ".." expr]` form, the expression after the `".."`
2796 must be a constant expression that can be evaluated at compile time, such
2797 as a [literal](#literals) or a [static item](#static-items).
2801 ["a", "b", "c", "d"];
2802 [0i, ..128]; // vector with 128 zeros
2803 [0u8, 0u8, 0u8, 0u8];
2806 ### Index expressions
2809 idx_expr : expr '[' expr ']' ;
2812 [Vector](#vector-types)-typed expressions can be indexed by writing a
2813 square-bracket-enclosed expression (the index) after them. When the
2814 vector is mutable, the resulting [lvalue](#lvalues,-rvalues-and-temporaries) can be assigned to.
2816 Indices are zero-based, and may be of any integral type. Vector access
2817 is bounds-checked at run-time. When the check fails, it will put the
2818 task in a _failing state_.
2822 # task::spawn(proc() {
2825 (["a", "b"])[10]; // fails
2830 ### Unary operator expressions
2832 Rust defines six symbolic unary operators.
2833 They are all written as prefix operators,
2834 before the expression they apply to.
2837 : Negation. May only be applied to numeric types.
2839 : Dereference. When applied to a [pointer](#pointer-types) it denotes the pointed-to location.
2840 For pointers to mutable locations, the resulting [lvalue](#lvalues,-rvalues-and-temporaries) can be assigned to.
2841 On non-pointer types, it calls the `deref` method of the `std::ops::Deref` trait, or the
2842 `deref_mut` method of the `std::ops::DerefMut` trait (if implemented by the type and required
2843 for an outer expression that will or could mutate the dereference), and produces the
2844 result of dereferencing the `&` or `&mut` borrowed pointer returned from the overload method.
2847 : Logical negation. On the boolean type, this flips between `true` and
2848 `false`. On integer types, this inverts the individual bits in the
2849 two's complement representation of the value.
2851 : [Boxing](#pointer-types) operators. Allocate a box to hold the value they are applied to,
2852 and store the value in it. `box` creates an owned box.
2854 : Borrow operator. Returns a reference, pointing to its operand.
2855 The operand of a borrow is statically proven to outlive the resulting pointer.
2856 If the borrow-checker cannot prove this, it is a compilation error.
2858 ### Binary operator expressions
2861 binop_expr : expr binop expr ;
2864 Binary operators expressions are given in terms of
2865 [operator precedence](#operator-precedence).
2867 #### Arithmetic operators
2869 Binary arithmetic expressions are syntactic sugar for calls to built-in traits,
2870 defined in the `std::ops` module of the `std` library.
2871 This means that arithmetic operators can be overridden for user-defined types.
2872 The default meaning of the operators on standard types is given here.
2875 : Addition and vector/string concatenation.
2876 Calls the `add` method on the `std::ops::Add` trait.
2879 Calls the `sub` method on the `std::ops::Sub` trait.
2882 Calls the `mul` method on the `std::ops::Mul` trait.
2885 Calls the `div` method on the `std::ops::Div` trait.
2888 Calls the `rem` method on the `std::ops::Rem` trait.
2890 #### Bitwise operators
2892 Like the [arithmetic operators](#arithmetic-operators), bitwise operators
2893 are syntactic sugar for calls to methods of built-in traits.
2894 This means that bitwise operators can be overridden for user-defined types.
2895 The default meaning of the operators on standard types is given here.
2899 Calls the `bitand` method of the `std::ops::BitAnd` trait.
2902 Calls the `bitor` method of the `std::ops::BitOr` trait.
2905 Calls the `bitxor` method of the `std::ops::BitXor` trait.
2907 : Logical left shift.
2908 Calls the `shl` method of the `std::ops::Shl` trait.
2910 : Logical right shift.
2911 Calls the `shr` method of the `std::ops::Shr` trait.
2913 #### Lazy boolean operators
2915 The operators `||` and `&&` may be applied to operands of boolean type.
2916 The `||` operator denotes logical 'or', and the `&&` operator denotes logical 'and'.
2917 They differ from `|` and `&` in that the right-hand operand is only evaluated
2918 when the left-hand operand does not already determine the result of the expression.
2919 That is, `||` only evaluates its right-hand operand
2920 when the left-hand operand evaluates to `false`, and `&&` only when it evaluates to `true`.
2922 #### Comparison operators
2924 Comparison operators are, like the [arithmetic operators](#arithmetic-operators),
2925 and [bitwise operators](#bitwise-operators),
2926 syntactic sugar for calls to built-in traits.
2927 This means that comparison operators can be overridden for user-defined types.
2928 The default meaning of the operators on standard types is given here.
2932 Calls the `eq` method on the `std::cmp::PartialEq` trait.
2935 Calls the `ne` method on the `std::cmp::PartialEq` trait.
2938 Calls the `lt` method on the `std::cmp::PartialOrd` trait.
2941 Calls the `gt` method on the `std::cmp::PartialOrd` trait.
2943 : Less than or equal.
2944 Calls the `le` method on the `std::cmp::PartialOrd` trait.
2946 : Greater than or equal.
2947 Calls the `ge` method on the `std::cmp::PartialOrd` trait.
2949 #### Type cast expressions
2951 A type cast expression is denoted with the binary operator `as`.
2953 Executing an `as` expression casts the value on the left-hand side to the type
2954 on the right-hand side.
2956 A numeric value can be cast to any numeric type.
2957 A raw pointer value can be cast to or from any integral type or raw pointer type.
2958 Any other cast is unsupported and will fail to compile.
2960 An example of an `as` expression:
2963 # fn sum(v: &[f64]) -> f64 { 0.0 }
2964 # fn len(v: &[f64]) -> int { 0 }
2966 fn avg(v: &[f64]) -> f64 {
2967 let sum: f64 = sum(v);
2968 let sz: f64 = len(v) as f64;
2973 #### Assignment expressions
2975 An _assignment expression_ consists of an [lvalue](#lvalues,-rvalues-and-temporaries) expression followed by an
2976 equals sign (`=`) and an [rvalue](#lvalues,-rvalues-and-temporaries) expression.
2978 Evaluating an assignment expression [either copies or moves](#moved-and-copied-types) its right-hand operand to its left-hand operand.
2987 #### Compound assignment expressions
2989 The `+`, `-`, `*`, `/`, `%`, `&`, `|`, `^`, `<<`, and `>>`
2990 operators may be composed with the `=` operator. The expression `lval
2991 OP= val` is equivalent to `lval = lval OP val`. For example, `x = x +
2992 1` may be written as `x += 1`.
2994 Any such expression always has the [`unit`](#primitive-types) type.
2996 #### Operator precedence
2998 The precedence of Rust binary operators is ordered as follows, going
2999 from strong to weak:
3001 ~~~~ {.text .precedence}
3016 Operators at the same precedence level are evaluated left-to-right. [Unary operators](#unary-operator-expressions)
3017 have the same precedence level and it is stronger than any of the binary operators'.
3019 ### Grouped expressions
3021 An expression enclosed in parentheses evaluates to the result of the enclosed
3022 expression. Parentheses can be used to explicitly specify evaluation order
3023 within an expression.
3026 paren_expr : '(' expr ')' ;
3029 An example of a parenthesized expression:
3032 let x: int = (2 + 3) * 4;
3036 ### Call expressions
3039 expr_list : [ expr [ ',' expr ]* ] ? ;
3040 paren_expr_list : '(' expr_list ')' ;
3041 call_expr : expr paren_expr_list ;
3044 A _call expression_ invokes a function, providing zero or more input slots and
3045 an optional reference slot to serve as the function's output, bound to the
3046 `lval` on the right hand side of the call. If the function eventually returns,
3047 then the expression completes.
3049 Some examples of call expressions:
3052 # use std::from_str::FromStr;
3053 # fn add(x: int, y: int) -> int { 0 }
3055 let x: int = add(1, 2);
3056 let pi: Option<f32> = FromStr::from_str("3.14");
3059 ### Lambda expressions
3062 ident_list : [ ident [ ',' ident ]* ] ? ;
3063 lambda_expr : '|' ident_list '|' expr ;
3066 A _lambda expression_ (sometimes called an "anonymous function expression") defines a function and denotes it as a value,
3067 in a single expression.
3068 A lambda expression is a pipe-symbol-delimited (`|`) list of identifiers followed by an expression.
3070 A lambda expression denotes a function that maps a list of parameters (`ident_list`)
3071 onto the expression that follows the `ident_list`.
3072 The identifiers in the `ident_list` are the parameters to the function.
3073 These parameters' types need not be specified, as the compiler infers them from context.
3075 Lambda expressions are most useful when passing functions as arguments to other functions,
3076 as an abbreviation for defining and capturing a separate function.
3078 Significantly, lambda expressions _capture their environment_,
3079 which regular [function definitions](#functions) do not.
3080 The exact type of capture depends on the [function type](#function-types) inferred for the lambda expression.
3081 In the simplest and least-expensive form (analogous to a ```|| { }``` expression),
3082 the lambda expression captures its environment by reference,
3083 effectively borrowing pointers to all outer variables mentioned inside the function.
3084 Alternately, the compiler may infer that a lambda expression should copy or move values (depending on their type.)
3085 from the environment into the lambda expression's captured environment.
3087 In this example, we define a function `ten_times` that takes a higher-order function argument,
3088 and call it with a lambda expression as an argument.
3091 fn ten_times(f: |int|) {
3099 ten_times(|j| println!("hello, {}", j));
3105 while_expr : "while" no_struct_literal_expr '{' block '}' ;
3108 A `while` loop begins by evaluating the boolean loop conditional expression.
3109 If the loop conditional expression evaluates to `true`, the loop body block
3110 executes and control returns to the loop conditional expression. If the loop
3111 conditional expression evaluates to `false`, the `while` expression completes.
3126 A `loop` expression denotes an infinite loop.
3129 loop_expr : [ lifetime ':' ] "loop" '{' block '}';
3132 A `loop` expression may optionally have a _label_.
3133 If a label is present,
3134 then labeled `break` and `continue` expressions nested within this loop may exit out of this loop or return control to its head.
3135 See [Break expressions](#break-expressions) and [Continue expressions](#continue-expressions).
3137 ### Break expressions
3140 break_expr : "break" [ lifetime ];
3143 A `break` expression has an optional _label_.
3144 If the label is absent, then executing a `break` expression immediately terminates the innermost loop enclosing it.
3145 It is only permitted in the body of a loop.
3146 If the label is present, then `break foo` terminates the loop with label `foo`,
3147 which need not be the innermost label enclosing the `break` expression,
3148 but must enclose it.
3150 ### Continue expressions
3153 continue_expr : "continue" [ lifetime ];
3156 A `continue` expression has an optional _label_.
3157 If the label is absent,
3158 then executing a `continue` expression immediately terminates the current iteration of the innermost loop enclosing it,
3159 returning control to the loop *head*.
3160 In the case of a `while` loop,
3161 the head is the conditional expression controlling the loop.
3162 In the case of a `for` loop, the head is the call-expression controlling the loop.
3163 If the label is present, then `continue foo` returns control to the head of the loop with label `foo`,
3164 which need not be the innermost label enclosing the `break` expression,
3165 but must enclose it.
3167 A `continue` expression is only permitted in the body of a loop.
3172 for_expr : "for" pat "in" no_struct_literal_expr '{' block '}' ;
3175 A `for` expression is a syntactic construct for looping over elements
3176 provided by an implementation of `std::iter::Iterator`.
3178 An example of a for loop over the contents of a vector:
3182 # fn bar(f: Foo) { }
3187 let v: &[Foo] = &[a, b, c];
3194 An example of a for loop over a series of integers:
3197 # fn bar(b:uint) { }
3198 for i in range(0u, 256) {
3206 if_expr : "if" no_struct_literal_expr '{' block '}'
3209 else_tail : "else" [ if_expr
3213 An `if` expression is a conditional branch in program control. The form of
3214 an `if` expression is a condition expression, followed by a consequent
3215 block, any number of `else if` conditions and blocks, and an optional
3216 trailing `else` block. The condition expressions must have type
3217 `bool`. If a condition expression evaluates to `true`, the
3218 consequent block is executed and any subsequent `else if` or `else`
3219 block is skipped. If a condition expression evaluates to `false`, the
3220 consequent block is skipped and any subsequent `else if` condition is
3221 evaluated. If all `if` and `else if` conditions evaluate to `false`
3222 then any `else` block is executed.
3224 ### Match expressions
3227 match_expr : "match" no_struct_literal_expr '{' match_arm * '}' ;
3229 match_arm : attribute * match_pat "=>" [ expr "," | '{' block '}' ] ;
3231 match_pat : pat [ '|' pat ] * [ "if" expr ] ? ;
3234 A `match` expression branches on a *pattern*. The exact form of matching that
3235 occurs depends on the pattern. Patterns consist of some combination of
3236 literals, destructured vectors or enum constructors, structures and
3237 tuples, variable binding specifications, wildcards (`..`), and placeholders
3238 (`_`). A `match` expression has a *head expression*, which is the value to
3239 compare to the patterns. The type of the patterns must equal the type of the
3242 In a pattern whose head expression has an `enum` type, a placeholder (`_`)
3243 stands for a *single* data field, whereas a wildcard `..` stands for *all* the
3244 fields of a particular variant. For example:
3247 enum List<X> { Nil, Cons(X, Box<List<X>>) }
3249 let x: List<int> = Cons(10, box Cons(11, box Nil));
3252 Cons(_, box Nil) => fail!("singleton list"),
3254 Nil => fail!("empty list")
3258 The first pattern matches lists constructed by applying `Cons` to any head
3259 value, and a tail value of `box Nil`. The second pattern matches _any_ list
3260 constructed with `Cons`, ignoring the values of its arguments. The difference
3261 between `_` and `..` is that the pattern `C(_)` is only type-correct if `C` has
3262 exactly one argument, while the pattern `C(..)` is type-correct for any enum
3263 variant `C`, regardless of how many arguments `C` has.
3265 Used inside a vector pattern, `..` stands for any number of elements. This
3266 wildcard can be used at most once for a given vector, which implies that it
3267 cannot be used to specifically match elements that are at an unknown distance
3268 from both ends of a vector, like `[.., 42, ..]`. If followed by a variable name,
3269 it will bind the corresponding slice to the variable. Example:
3272 fn is_symmetric(list: &[uint]) -> bool {
3275 [x, ..inside, y] if x == y => is_symmetric(inside),
3281 let sym = &[0, 1, 4, 2, 4, 1, 0];
3282 let not_sym = &[0, 1, 7, 2, 4, 1, 0];
3283 assert!(is_symmetric(sym));
3284 assert!(!is_symmetric(not_sym));
3288 A `match` behaves differently depending on whether or not the head expression
3289 is an [lvalue or an rvalue](#lvalues,-rvalues-and-temporaries).
3290 If the head expression is an rvalue, it is
3291 first evaluated into a temporary location, and the resulting value
3292 is sequentially compared to the patterns in the arms until a match
3293 is found. The first arm with a matching pattern is chosen as the branch target
3294 of the `match`, any variables bound by the pattern are assigned to local
3295 variables in the arm's block, and control enters the block.
3297 When the head expression is an lvalue, the match does not allocate a
3298 temporary location (however, a by-value binding may copy or move from
3299 the lvalue). When possible, it is preferable to match on lvalues, as the
3300 lifetime of these matches inherits the lifetime of the lvalue, rather
3301 than being restricted to the inside of the match.
3303 An example of a `match` expression:
3306 # fn process_pair(a: int, b: int) { }
3307 # fn process_ten() { }
3309 enum List<X> { Nil, Cons(X, Box<List<X>>) }
3311 let x: List<int> = Cons(10, box Cons(11, box Nil));
3314 Cons(a, box Cons(b, _)) => {
3329 Patterns that bind variables
3330 default to binding to a copy or move of the matched value
3331 (depending on the matched value's type).
3332 This can be changed to bind to a reference by
3333 using the `ref` keyword,
3334 or to a mutable reference using `ref mut`.
3336 Subpatterns can also be bound to variables by the use of the syntax
3337 `variable @ subpattern`.
3341 enum List { Nil, Cons(uint, Box<List>) }
3343 fn is_sorted(list: &List) -> bool {
3345 Nil | Cons(_, box Nil) => true,
3346 Cons(x, ref r @ box Cons(_, _)) => {
3348 box Cons(y, _) => (x <= y) && is_sorted(&**r),
3356 let a = Cons(6, box Cons(7, box Cons(42, box Nil)));
3357 assert!(is_sorted(&a));
3362 Patterns can also dereference pointers by using the `&`,
3363 `box` or `@` symbols, as appropriate. For example, these two matches
3364 on `x: &int` are equivalent:
3368 let y = match *x { 0 => "zero", _ => "some" };
3369 let z = match x { &0 => "zero", _ => "some" };
3374 A pattern that's just an identifier, like `Nil` in the previous example,
3375 could either refer to an enum variant that's in scope, or bind a new variable.
3376 The compiler resolves this ambiguity by forbidding variable bindings that occur
3377 in `match` patterns from shadowing names of variants that are in scope.
3378 For example, wherever `List` is in scope,
3379 a `match` pattern would not be able to bind `Nil` as a new name.
3380 The compiler interprets a variable pattern `x` as a binding _only_ if there is
3381 no variant named `x` in scope.
3382 A convention you can use to avoid conflicts is simply to name variants with
3383 upper-case letters, and local variables with lower-case letters.
3385 Multiple match patterns may be joined with the `|` operator.
3386 A range of values may be specified with `..`.
3392 let message = match x {
3393 0 | 1 => "not many",
3399 Range patterns only work on scalar types
3400 (like integers and characters; not like vectors and structs, which have sub-components).
3401 A range pattern may not be a sub-range of another range pattern inside the same `match`.
3403 Finally, match patterns can accept *pattern guards* to further refine the
3404 criteria for matching a case. Pattern guards appear after the pattern and
3405 consist of a bool-typed expression following the `if` keyword. A pattern
3406 guard may refer to the variables bound within the pattern they follow.
3409 # let maybe_digit = Some(0);
3410 # fn process_digit(i: int) { }
3411 # fn process_other(i: int) { }
3413 let message = match maybe_digit {
3414 Some(x) if x < 10 => process_digit(x),
3415 Some(x) => process_other(x),
3420 ### Return expressions
3423 return_expr : "return" expr ? ;
3426 Return expressions are denoted with the keyword `return`. Evaluating a `return`
3427 expression moves its argument into the output slot of the current
3428 function, destroys the current function activation frame, and transfers
3429 control to the caller frame.
3431 An example of a `return` expression:
3434 fn max(a: int, b: int) -> int {
3446 Every slot, item and value in a Rust program has a type. The _type_ of a *value*
3447 defines the interpretation of the memory holding it.
3449 Built-in types and type-constructors are tightly integrated into the language,
3450 in nontrivial ways that are not possible to emulate in user-defined
3451 types. User-defined types have limited capabilities.
3455 The primitive types are the following:
3457 * The "unit" type `()`, having the single "unit" value `()` (occasionally called
3459 * The boolean type `bool` with values `true` and `false`.
3460 * The machine types.
3461 * The machine-dependent integer and floating-point types.
3463 [^unittype]: The "unit" value `()` is *not* a sentinel "null pointer" value for
3464 reference slots; the "unit" type is the implicit return type from functions
3465 otherwise lacking a return type, and can be used in other contexts (such as
3466 message-sending or type-parametric code) as a zero-size type.]
3470 The machine types are the following:
3472 * The unsigned word types `u8`, `u16`, `u32` and `u64`, with values drawn from
3473 the integer intervals [0, 2^8 - 1], [0, 2^16 - 1], [0, 2^32 - 1] and
3474 [0, 2^64 - 1] respectively.
3476 * The signed two's complement word types `i8`, `i16`, `i32` and `i64`, with
3477 values drawn from the integer intervals [-(2^(7)), 2^7 - 1],
3478 [-(2^(15)), 2^15 - 1], [-(2^(31)), 2^31 - 1], [-(2^(63)), 2^63 - 1]
3481 * The IEEE 754-2008 `binary32` and `binary64` floating-point types: `f32` and
3482 `f64`, respectively.
3484 #### Machine-dependent integer types
3486 The Rust type `uint` [^rustuint] is an
3487 unsigned integer type with target-machine-dependent size. Its size, in
3488 bits, is equal to the number of bits required to hold any memory address on
3491 The Rust type `int` [^rustint] is a
3492 two's complement signed integer type with target-machine-dependent size. Its
3493 size, in bits, is equal to the size of the rust type `uint` on the same target
3496 [^rustuint]: A Rust `uint` is analogous to a C99 `uintptr_t`.
3497 [^rustint]: A Rust `int` is analogous to a C99 `intptr_t`.
3501 The types `char` and `str` hold textual data.
3503 A value of type `char` is a [Unicode scalar value](
3504 http://www.unicode.org/glossary/#unicode_scalar_value)
3505 (ie. a code point that is not a surrogate),
3506 represented as a 32-bit unsigned word in the 0x0000 to 0xD7FF
3507 or 0xE000 to 0x10FFFF range.
3508 A `[char]` vector is effectively an UCS-4 / UTF-32 string.
3510 A value of type `str` is a Unicode string,
3511 represented as a vector of 8-bit unsigned bytes holding a sequence of UTF-8 codepoints.
3512 Since `str` is of unknown size, it is not a _first class_ type,
3513 but can only be instantiated through a pointer type,
3514 such as `&str` or `String`.
3518 A tuple *type* is a heterogeneous product of other types, called the *elements*
3519 of the tuple. It has no nominal name and is instead structurally typed.
3521 Tuple types and values are denoted by listing the types or values of their
3522 elements, respectively, in a parenthesized, comma-separated
3525 Because tuple elements don't have a name, they can only be accessed by pattern-matching.
3527 The members of a tuple are laid out in memory contiguously, in
3528 order specified by the tuple type.
3530 An example of a tuple type and its use:
3533 type Pair<'a> = (int, &'a str);
3534 let p: Pair<'static> = (10, "hello");
3536 assert!(b != "world");
3541 The vector type constructor represents a homogeneous array of values of a given type.
3542 A vector has a fixed size.
3543 (Operations like `vec.push` operate solely on owned vectors.)
3544 A vector type can be annotated with a _definite_ size, such as `[int, ..10]`.
3545 Such a definite-sized vector type is a first-class type, since its size is known statically.
3546 A vector without such a size is said to be of _indefinite_ size,
3547 and is therefore not a _first-class_ type.
3548 An indefinite-size vector can only be instantiated through a pointer type,
3549 such as `&[T]` or `Vec<T>`.
3550 The kind of a vector type depends on the kind of its element type,
3551 as with other simple structural types.
3553 Expressions producing vectors of definite size cannot be evaluated in a
3554 context expecting a vector of indefinite size; one must copy the
3555 definite-sized vector contents into a distinct vector of indefinite size.
3557 An example of a vector type and its use:
3560 let v: &[int] = &[7, 5, 3];
3565 All in-bounds elements of a vector are always initialized,
3566 and access to a vector is always bounds-checked.
3570 A `struct` *type* is a heterogeneous product of other types, called the *fields*
3571 of the type.[^structtype]
3573 [^structtype]: `struct` types are analogous `struct` types in C,
3574 the *record* types of the ML family,
3575 or the *structure* types of the Lisp family.
3577 New instances of a `struct` can be constructed with a [struct expression](#structure-expressions).
3579 The memory layout of a `struct` is undefined by default to allow for compiler optimziations like
3580 field reordering, but it can be fixed with the `#[repr(...)]` attribute.
3581 In either case, fields may be given in any order in a corresponding struct *expression*;
3582 the resulting `struct` value will always have the same memory layout.
3584 The fields of a `struct` may be qualified by [visibility modifiers](#re-exporting-and-visibility),
3585 to allow access to data in a structure outside a module.
3587 A _tuple struct_ type is just like a structure type, except that the fields are anonymous.
3589 A _unit-like struct_ type is like a structure type, except that it has no fields.
3590 The one value constructed by the associated [structure expression](#structure-expressions)
3591 is the only value that inhabits such a type.
3593 ### Enumerated types
3595 An *enumerated type* is a nominal, heterogeneous disjoint union type,
3596 denoted by the name of an [`enum` item](#enumerations). [^enumtype]
3598 [^enumtype]: The `enum` type is analogous to a `data` constructor declaration in
3599 ML, or a *pick ADT* in Limbo.
3601 An [`enum` item](#enumerations) declares both the type and a number of *variant constructors*,
3602 each of which is independently named and takes an optional tuple of arguments.
3604 New instances of an `enum` can be constructed by calling one of the variant constructors,
3605 in a [call expression](#call-expressions).
3607 Any `enum` value consumes as much memory as the largest variant constructor for its corresponding `enum` type.
3609 Enum types cannot be denoted *structurally* as types,
3610 but must be denoted by named reference to an [`enum` item](#enumerations).
3614 Nominal types — [enumerations](#enumerated-types) and [structures](#structure-types) — may be recursive.
3615 That is, each `enum` constructor or `struct` field may refer, directly or indirectly, to the enclosing `enum` or `struct` type itself.
3616 Such recursion has restrictions:
3618 * Recursive types must include a nominal type in the recursion
3619 (not mere [type definitions](#type-definitions),
3620 or other structural types such as [vectors](#vector-types) or [tuples](#tuple-types)).
3621 * A recursive `enum` item must have at least one non-recursive constructor
3622 (in order to give the recursion a basis case).
3623 * The size of a recursive type must be finite;
3624 in other words the recursive fields of the type must be [pointer types](#pointer-types).
3625 * Recursive type definitions can cross module boundaries, but not module *visibility* boundaries,
3626 or crate boundaries (in order to simplify the module system and type checker).
3628 An example of a *recursive* type and its use:
3633 Cons(T, Box<List<T>>)
3636 let a: List<int> = Cons(7, box Cons(13, box Nil));
3641 All pointers in Rust are explicit first-class values.
3642 They can be copied, stored into data structures, and returned from functions.
3643 There are four varieties of pointer in Rust:
3645 * Owning pointers (`Box`)
3646 : These point to owned heap allocations (or "boxes") in the shared, inter-task heap.
3647 Each owned box has a single owning pointer; pointer and pointee retain a 1:1 relationship at all times.
3648 Owning pointers are written `Box<content>`,
3649 for example `Box<int>` means an owning pointer to an owned box containing an integer.
3650 Copying an owned box is a "deep" operation:
3651 it involves allocating a new owned box and copying the contents of the old box into the new box.
3652 Releasing an owning pointer immediately releases its corresponding owned box.
3655 : These point to memory _owned by some other value_.
3656 References arise by (automatic) conversion from owning pointers, managed pointers,
3657 or by applying the borrowing operator `&` to some other value,
3658 including [lvalues, rvalues or temporaries](#lvalues,-rvalues-and-temporaries).
3659 A borrow expression is written `&content`.
3661 A reference type is written `&'f type` for some lifetime-variable `f`,
3662 or just `&type` when the lifetime can be elided;
3663 for example `&int` means a reference to an integer.
3664 Copying a reference is a "shallow" operation:
3665 it involves only copying the pointer itself.
3666 Releasing a reference typically has no effect on the value it points to,
3667 with the exception of temporary values,
3668 which are released when the last reference to them is released.
3670 * Raw pointers (`*`)
3671 : Raw pointers are pointers without safety or liveness guarantees.
3672 Raw pointers are written as `*const T` or `*mut T`,
3673 for example `*const int` means a raw pointer to an integer.
3674 Copying or dropping a raw pointer has no effect on the lifecycle of any
3675 other value. Dereferencing a raw pointer or converting it to any other
3676 pointer type is an [`unsafe` operation](#unsafe-functions).
3677 Raw pointers are generally discouraged in Rust code;
3678 they exist to support interoperability with foreign code,
3679 and writing performance-critical or low-level functions.
3683 The function type constructor `fn` forms new function types.
3684 A function type consists of a possibly-empty set of function-type modifiers
3685 (such as `unsafe` or `extern`), a sequence of input types and an output type.
3687 An example of a `fn` type:
3690 fn add(x: int, y: int) -> int {
3694 let mut x = add(5,7);
3696 type Binop<'a> = |int,int|: 'a -> int;
3697 let bo: Binop = add;
3703 ~~~~ {.ebnf .notation}
3704 closure_type := [ 'unsafe' ] [ '<' lifetime-list '>' ] '|' arg-list '|'
3705 [ ':' bound-list ] [ '->' type ]
3706 procedure_type := 'proc' [ '<' lifetime-list '>' ] '(' arg-list ')'
3707 [ ':' bound-list ] [ '->' type ]
3708 lifetime-list := lifetime | lifetime ',' lifetime-list
3709 arg-list := ident ':' type | ident ':' type ',' arg-list
3710 bound-list := bound | bound '+' bound-list
3711 bound := path | lifetime
3714 The type of a closure mapping an input of type `A` to an output of type `B` is
3715 `|A| -> B`. A closure with no arguments or return values has type `||`.
3716 Similarly, a procedure mapping `A` to `B` is `proc(A) -> B` and a no-argument
3717 and no-return value closure has type `proc()`.
3719 An example of creating and calling a closure:
3722 let captured_var = 10i;
3724 let closure_no_args = || println!("captured_var={}", captured_var);
3726 let closure_args = |arg: int| -> int {
3727 println!("captured_var={}, arg={}", captured_var, arg);
3728 arg // Note lack of semicolon after 'arg'
3731 fn call_closure(c1: ||, c2: |int| -> int) {
3736 call_closure(closure_no_args, closure_args);
3740 Unlike closures, procedures may only be invoked once, but own their
3741 environment, and are allowed to move out of their environment. Procedures are
3742 allocated on the heap (unlike closures). An example of creating and calling a
3746 let string = "Hello".to_string();
3748 // Creates a new procedure, passing it to the `spawn` function.
3750 println!("{} world!", string);
3753 // the variable `string` has been moved into the previous procedure, so it is
3754 // no longer usable.
3757 // Create an invoke a procedure. Note that the procedure is *moved* when
3758 // invoked, so it cannot be invoked again.
3759 let f = proc(n: int) { n + 22 };
3760 println!("answer: {}", f(20));
3766 Every trait item (see [traits](#traits)) defines a type with the same name as the trait.
3767 This type is called the _object type_ of the trait.
3768 Object types permit "late binding" of methods, dispatched using _virtual method tables_ ("vtables").
3769 Whereas most calls to trait methods are "early bound" (statically resolved) to specific implementations at compile time,
3770 a call to a method on an object type is only resolved to a vtable entry at compile time.
3771 The actual implementation for each vtable entry can vary on an object-by-object basis.
3773 Given a pointer-typed expression `E` of type `&T` or `Box<T>`, where `T` implements trait `R`,
3774 casting `E` to the corresponding pointer type `&R` or `Box<R>` results in a value of the _object type_ `R`.
3775 This result is represented as a pair of pointers:
3776 the vtable pointer for the `T` implementation of `R`, and the pointer value of `E`.
3778 An example of an object type:
3782 fn stringify(&self) -> String;
3785 impl Printable for int {
3786 fn stringify(&self) -> String { self.to_string() }
3789 fn print(a: Box<Printable>) {
3790 println!("{}", a.stringify());
3794 print(box 10i as Box<Printable>);
3798 In this example, the trait `Printable` occurs as an object type in both the type signature of `print`,
3799 and the cast expression in `main`.
3803 Within the body of an item that has type parameter declarations, the names of its type parameters are types:
3806 fn map<A: Clone, B: Clone>(f: |A| -> B, xs: &[A]) -> Vec<B> {
3810 let first: B = f(xs[0].clone());
3811 let rest: Vec<B> = map(f, xs.slice(1, xs.len()));
3812 return vec![first].append(rest.as_slice());
3816 Here, `first` has type `B`, referring to `map`'s `B` type parameter;
3817 and `rest` has type `Vec<B>`, a vector type with element type `B`.
3821 The special type `self` has a meaning within methods inside an
3822 impl item. It refers to the type of the implicit `self` argument. For
3827 fn make_string(&self) -> String;
3830 impl Printable for String {
3831 fn make_string(&self) -> String {
3837 `self` refers to the value of type `String` that is the receiver for a
3838 call to the method `make_string`.
3842 Types in Rust are categorized into kinds, based on various properties of the components of the type.
3846 : Types of this kind can be safely sent between tasks.
3847 This kind includes scalars, owning pointers, owned closures, and
3848 structural types containing only other owned types.
3849 All `Send` types are `'static`.
3851 : Types of this kind consist of "Plain Old Data"
3852 which can be copied by simply moving bits.
3853 All values of this kind can be implicitly copied.
3854 This kind includes scalars and immutable references,
3855 as well as structural types containing other `Copy` types.
3857 : Types of this kind do not contain any references (except for
3858 references with the `static` lifetime, which are allowed).
3859 This can be a useful guarantee for code
3860 that breaks borrowing assumptions
3861 using [`unsafe` operations](#unsafe-functions).
3863 : This is not strictly a kind,
3864 but its presence interacts with kinds:
3865 the `Drop` trait provides a single method `drop`
3866 that takes no parameters,
3867 and is run when values of the type are dropped.
3868 Such a method is called a "destructor",
3869 and are always executed in "top-down" order:
3870 a value is completely destroyed
3871 before any of the values it owns run their destructors.
3872 Only `Send` types can implement `Drop`.
3875 : Types with destructors, closure environments,
3876 and various other _non-first-class_ types,
3877 are not copyable at all.
3878 Such types can usually only be accessed through pointers,
3879 or in some cases, moved between mutable locations.
3881 Kinds can be supplied as _bounds_ on type parameters, like traits,
3882 in which case the parameter is constrained to types satisfying that kind.
3884 By default, type parameters do not carry any assumed kind-bounds at all.
3885 When instantiating a type parameter,
3886 the kind bounds on the parameter are checked
3887 to be the same or narrower than the kind
3888 of the type that it is instantiated with.
3890 Sending operations are not part of the Rust language,
3891 but are implemented in the library.
3892 Generic functions that send values
3893 bound the kind of these values to sendable.
3895 # Memory and concurrency models
3897 Rust has a memory model centered around concurrently-executing _tasks_. Thus
3898 its memory model and its concurrency model are best discussed simultaneously,
3899 as parts of each only make sense when considered from the perspective of the
3902 When reading about the memory model, keep in mind that it is partitioned in
3903 order to support tasks; and when reading about tasks, keep in mind that their
3904 isolation and communication mechanisms are only possible due to the ownership
3905 and lifetime semantics of the memory model.
3909 A Rust program's memory consists of a static set of *items*, a set of
3910 [tasks](#tasks) each with its own *stack*, and a *heap*. Immutable portions of
3911 the heap may be shared between tasks, mutable portions may not.
3913 Allocations in the stack consist of *slots*, and allocations in the heap
3916 ### Memory allocation and lifetime
3918 The _items_ of a program are those functions, modules and types
3919 that have their value calculated at compile-time and stored uniquely in the
3920 memory image of the rust process. Items are neither dynamically allocated nor
3923 A task's _stack_ consists of activation frames automatically allocated on
3924 entry to each function as the task executes. A stack allocation is reclaimed
3925 when control leaves the frame containing it.
3927 The _heap_ is a general term that describes two separate sets of boxes:
3928 managed boxes — which may be subject to garbage collection — and owned
3929 boxes. The lifetime of an allocation in the heap depends on the lifetime of
3930 the box values pointing to it. Since box values may themselves be passed in
3931 and out of frames, or stored in the heap, heap allocations may outlive the
3932 frame they are allocated within.
3934 ### Memory ownership
3936 A task owns all memory it can *safely* reach through local variables,
3937 as well as managed, owned boxes and references.
3939 When a task sends a value that has the `Send` trait to another task,
3940 it loses ownership of the value sent and can no longer refer to it.
3941 This is statically guaranteed by the combined use of "move semantics",
3942 and the compiler-checked _meaning_ of the `Send` trait:
3943 it is only instantiated for (transitively) sendable kinds of data constructor and pointers,
3944 never including managed boxes or references.
3946 When a stack frame is exited, its local allocations are all released, and its
3947 references to boxes (both managed and owned) are dropped.
3949 A managed box may (in the case of a recursive, mutable managed type) be cyclic;
3950 in this case the release of memory inside the managed structure may be deferred
3951 until task-local garbage collection can reclaim it. Code can ensure no such
3952 delayed deallocation occurs by restricting itself to owned boxes and similar
3953 unmanaged kinds of data.
3955 When a task finishes, its stack is necessarily empty and it therefore has no
3956 references to any boxes; the remainder of its heap is immediately freed.
3960 A task's stack contains slots.
3962 A _slot_ is a component of a stack frame, either a function parameter,
3963 a [temporary](#lvalues,-rvalues-and-temporaries), or a local variable.
3965 A _local variable_ (or *stack-local* allocation) holds a value directly,
3966 allocated within the stack's memory. The value is a part of the stack frame.
3968 Local variables are immutable unless declared otherwise like: `let mut x = ...`.
3970 Function parameters are immutable unless declared with `mut`. The
3971 `mut` keyword applies only to the following parameter (so `|mut x, y|`
3972 and `fn f(mut x: Box<int>, y: Box<int>)` declare one mutable variable `x` and
3973 one immutable variable `y`).
3975 Methods that take either `self` or `Box<Self>` can optionally place them in a
3976 mutable slot by prefixing them with `mut` (similar to regular arguments):
3980 fn change(mut self) -> Self;
3981 fn modify(mut self: Box<Self>) -> Box<Self>;
3985 Local variables are not initialized when allocated; the entire frame worth of
3986 local variables are allocated at once, on frame-entry, in an uninitialized
3987 state. Subsequent statements within a function may or may not initialize the
3988 local variables. Local variables can be used only after they have been
3989 initialized; this is enforced by the compiler.
3993 An _owned box_ is a reference to a heap allocation holding another value, which is constructed
3994 by the prefix operator `box`. When the standard library is in use, the type of an owned box is
3995 `std::owned::Box<T>`.
3997 An example of an owned box type and value:
4000 let x: Box<int> = box 10;
4003 Owned box values exist in 1:1 correspondence with their heap allocation,
4004 copying an owned box value makes a shallow copy of the pointer.
4005 Rust will consider a shallow copy of an owned box to move ownership of the value. After a value has been moved, the source location cannot be used unless it is reinitialized.
4008 let x: Box<int> = box 10;
4010 // attempting to use `x` will result in an error here
4017 An executing Rust program consists of a tree of tasks.
4018 A Rust _task_ consists of an entry function, a stack,
4019 a set of outgoing communication channels and incoming communication ports,
4020 and ownership of some portion of the heap of a single operating-system process.
4021 (We expect that many programs will not use channels and ports directly,
4022 but will instead use higher-level abstractions provided in standard libraries,
4025 Multiple Rust tasks may coexist in a single operating-system process.
4026 The runtime scheduler maps tasks to a certain number of operating-system threads.
4027 By default, the scheduler chooses the number of threads based on
4028 the number of concurrent physical CPUs detected at startup.
4029 It's also possible to override this choice at runtime.
4030 When the number of tasks exceeds the number of threads — which is likely —
4031 the scheduler multiplexes the tasks onto threads.[^mnscheduler]
4033 [^mnscheduler]: This is an M:N scheduler, which is known to give suboptimal
4034 results for CPU-bound concurrency problems. In such cases, running with the
4035 same number of threads and tasks can yield better results. Rust has M:N
4036 scheduling in order to support very large numbers of tasks in contexts where
4037 threads are too resource-intensive to use in large number. The cost of
4038 threads varies substantially per operating system, and is sometimes quite
4039 low, so this flexibility is not always worth exploiting.
4041 ### Communication between tasks
4043 Rust tasks are isolated and generally unable to interfere with one another's memory directly,
4044 except through [`unsafe` code](#unsafe-functions).
4045 All contact between tasks is mediated by safe forms of ownership transfer,
4046 and data races on memory are prohibited by the type system.
4048 Inter-task communication and co-ordination facilities are provided in the standard library.
4051 - synchronous and asynchronous communication channels with various communication topologies
4052 - read-only and read-write shared variables with various safe mutual exclusion patterns
4053 - simple locks and semaphores
4055 When such facilities carry values, the values are restricted to the [`Send` type-kind](#type-kinds).
4056 Restricting communication interfaces to this kind ensures that no references or managed pointers move between tasks.
4057 Thus access to an entire data structure can be mediated through its owning "root" value;
4058 no further locking or copying is required to avoid data races within the substructure of such a value.
4062 The _lifecycle_ of a task consists of a finite set of states and events
4063 that cause transitions between the states. The lifecycle states of a task are:
4070 A task begins its lifecycle — once it has been spawned — in the *running*
4071 state. In this state it executes the statements of its entry function, and any
4072 functions called by the entry function.
4074 A task may transition from the *running* state to the *blocked*
4075 state any time it makes a blocking communication call. When the
4076 call can be completed — when a message arrives at a sender, or a
4077 buffer opens to receive a message — then the blocked task will
4078 unblock and transition back to *running*.
4080 A task may transition to the *failing* state at any time, due being
4081 killed by some external event or internally, from the evaluation of a
4082 `fail!()` macro. Once *failing*, a task unwinds its stack and
4083 transitions to the *dead* state. Unwinding the stack of a task is done by
4084 the task itself, on its own control stack. If a value with a destructor is
4085 freed during unwinding, the code for the destructor is run, also on the task's
4086 control stack. Running the destructor code causes a temporary transition to a
4087 *running* state, and allows the destructor code to cause any subsequent
4088 state transitions. The original task of unwinding and failing thereby may
4089 suspend temporarily, and may involve (recursive) unwinding of the stack of a
4090 failed destructor. Nonetheless, the outermost unwinding activity will continue
4091 until the stack is unwound and the task transitions to the *dead*
4092 state. There is no way to "recover" from task failure. Once a task has
4093 temporarily suspended its unwinding in the *failing* state, failure
4094 occurring from within this destructor results in *hard* failure.
4095 A hard failure currently results in the process aborting.
4097 A task in the *dead* state cannot transition to other states; it exists
4098 only to have its termination status inspected by other tasks, and/or to await
4099 reclamation when the last reference to it drops.
4103 The currently scheduled task is given a finite *time slice* in which to
4104 execute, after which it is *descheduled* at a loop-edge or similar
4105 preemption point, and another task within is scheduled, pseudo-randomly.
4107 An executing task can yield control at any time, by making a library call to
4108 `std::task::yield`, which deschedules it immediately. Entering any other
4109 non-executing state (blocked, dead) similarly deschedules the task.
4111 # Runtime services, linkage and debugging
4113 The Rust _runtime_ is a relatively compact collection of C++ and Rust code
4114 that provides fundamental services and datatypes to all Rust tasks at
4115 run-time. It is smaller and simpler than many modern language runtimes. It is
4116 tightly integrated into the language's execution model of memory, tasks,
4117 communication and logging.
4119 > **Note:** The runtime library will merge with the `std` library in future versions of Rust.
4121 ### Memory allocation
4123 The runtime memory-management system is based on a _service-provider interface_,
4124 through which the runtime requests blocks of memory from its environment
4125 and releases them back to its environment when they are no longer needed.
4126 The default implementation of the service-provider interface
4127 consists of the C runtime functions `malloc` and `free`.
4129 The runtime memory-management system, in turn, supplies Rust tasks with
4130 facilities for allocating releasing stacks, as well as allocating and freeing
4135 The runtime provides C and Rust code to assist with various built-in types,
4136 such as vectors, strings, and the low level communication system (ports,
4139 Support for other built-in types such as simple types, tuples and
4140 enums is open-coded by the Rust compiler.
4142 ### Task scheduling and communication
4144 The runtime provides code to manage inter-task communication. This includes
4145 the system of task-lifecycle state transitions depending on the contents of
4146 queues, as well as code to copy values between queues and their recipients and
4147 to serialize values for transmission over operating-system inter-process
4148 communication facilities.
4152 The Rust compiler supports various methods to link crates together both
4153 statically and dynamically. This section will explore the various methods to
4154 link Rust crates together, and more information about native libraries can be
4155 found in the [ffi tutorial][ffi].
4157 In one session of compilation, the compiler can generate multiple artifacts
4158 through the usage of either command line flags or the `crate_type` attribute.
4159 If one or more command line flag is specified, all `crate_type` attributes will
4160 be ignored in favor of only building the artifacts specified by command line.
4162 * `--crate-type=bin`, `#[crate_type = "bin"]` - A runnable executable will be
4163 produced. This requires that there is a `main` function in the crate which
4164 will be run when the program begins executing. This will link in all Rust and
4165 native dependencies, producing a distributable binary.
4167 * `--crate-type=lib`, `#[crate_type = "lib"]` - A Rust library will be produced.
4168 This is an ambiguous concept as to what exactly is produced because a library
4169 can manifest itself in several forms. The purpose of this generic `lib` option
4170 is to generate the "compiler recommended" style of library. The output library
4171 will always be usable by rustc, but the actual type of library may change from
4172 time-to-time. The remaining output types are all different flavors of
4173 libraries, and the `lib` type can be seen as an alias for one of them (but the
4174 actual one is compiler-defined).
4176 * `--crate-type=dylib`, `#[crate_type = "dylib"]` - A dynamic Rust library will
4177 be produced. This is different from the `lib` output type in that this forces
4178 dynamic library generation. The resulting dynamic library can be used as a
4179 dependency for other libraries and/or executables. This output type will
4180 create `*.so` files on linux, `*.dylib` files on osx, and `*.dll` files on
4183 * `--crate-type=staticlib`, `#[crate_type = "staticlib"]` - A static system
4184 library will be produced. This is different from other library outputs in that
4185 the Rust compiler will never attempt to link to `staticlib` outputs. The
4186 purpose of this output type is to create a static library containing all of
4187 the local crate's code along with all upstream dependencies. The static
4188 library is actually a `*.a` archive on linux and osx and a `*.lib` file on
4189 windows. This format is recommended for use in situtations such as linking
4190 Rust code into an existing non-Rust application because it will not have
4191 dynamic dependencies on other Rust code.
4193 * `--crate-type=rlib`, `#[crate_type = "rlib"]` - A "Rust library" file will be
4194 produced. This is used as an intermediate artifact and can be thought of as a
4195 "static Rust library". These `rlib` files, unlike `staticlib` files, are
4196 interpreted by the Rust compiler in future linkage. This essentially means
4197 that `rustc` will look for metadata in `rlib` files like it looks for metadata
4198 in dynamic libraries. This form of output is used to produce statically linked
4199 executables as well as `staticlib` outputs.
4201 Note that these outputs are stackable in the sense that if multiple are
4202 specified, then the compiler will produce each form of output at once without
4203 having to recompile. However, this only applies for outputs specified by the same
4204 method. If only `crate_type` attributes are specified, then they will all be
4205 built, but if one or more `--crate-type` command line flag is specified,
4206 then only those outputs will be built.
4208 With all these different kinds of outputs, if crate A depends on crate B, then
4209 the compiler could find B in various different forms throughout the system. The
4210 only forms looked for by the compiler, however, are the `rlib` format and the
4211 dynamic library format. With these two options for a dependent library, the
4212 compiler must at some point make a choice between these two formats. With this
4213 in mind, the compiler follows these rules when determining what format of
4214 dependencies will be used:
4216 1. If a static library is being produced, all upstream dependencies are
4217 required to be available in `rlib` formats. This requirement stems from the
4218 reason that a dynamic library cannot be converted into a static format.
4220 Note that it is impossible to link in native dynamic dependencies to a static
4221 library, and in this case warnings will be printed about all unlinked native
4222 dynamic dependencies.
4224 2. If an `rlib` file is being produced, then there are no restrictions on what
4225 format the upstream dependencies are available in. It is simply required that
4226 all upstream dependencies be available for reading metadata from.
4228 The reason for this is that `rlib` files do not contain any of their upstream
4229 dependencies. It wouldn't be very efficient for all `rlib` files to contain a
4230 copy of `libstd.rlib`!
4232 3. If an executable is being produced and the `-C prefer-dynamic` flag is not
4233 specified, then dependencies are first attempted to be found in the `rlib`
4234 format. If some dependencies are not available in an rlib format, then
4235 dynamic linking is attempted (see below).
4237 4. If a dynamic library or an executable that is being dynamically linked is
4238 being produced, then the compiler will attempt to reconcile the available
4239 dependencies in either the rlib or dylib format to create a final product.
4241 A major goal of the compiler is to ensure that a library never appears more
4242 than once in any artifact. For example, if dynamic libraries B and C were
4243 each statically linked to library A, then a crate could not link to B and C
4244 together because there would be two copies of A. The compiler allows mixing
4245 the rlib and dylib formats, but this restriction must be satisfied.
4247 The compiler currently implements no method of hinting what format a library
4248 should be linked with. When dynamically linking, the compiler will attempt to
4249 maximize dynamic dependencies while still allowing some dependencies to be
4250 linked in via an rlib.
4252 For most situations, having all libraries available as a dylib is recommended
4253 if dynamically linking. For other situations, the compiler will emit a
4254 warning if it is unable to determine which formats to link each library with.
4256 In general, `--crate-type=bin` or `--crate-type=lib` should be sufficient for
4257 all compilation needs, and the other options are just available if more
4258 fine-grained control is desired over the output format of a Rust crate.
4262 The runtime contains a system for directing [logging
4263 expressions](#logging-expressions) to a logging console and/or internal logging
4264 buffers. Logging can be enabled per module.
4266 Logging output is enabled by setting the `RUST_LOG` environment
4267 variable. `RUST_LOG` accepts a logging specification made up of a
4268 comma-separated list of paths, with optional log levels. For each
4269 module containing log expressions, if `RUST_LOG` contains the path to
4270 that module or a parent of that module, then logs of the appropriate
4271 level will be output to the console.
4273 The path to a module consists of the crate name, any parent modules,
4274 then the module itself, all separated by double colons (`::`). The
4275 optional log level can be appended to the module path with an equals
4276 sign (`=`) followed by the log level, from 1 to 4, inclusive. Level 1
4277 is the error level, 2 is warning, 3 info, and 4 debug. You can also
4278 use the symbolic constants `error`, `warn`, `info`, and `debug`. Any
4279 logs less than or equal to the specified level will be output. If not
4280 specified then log level 4 is assumed. Debug messages can be omitted
4281 by passing `--cfg ndebug` to `rustc`.
4283 As an example, to see all the logs generated by the compiler, you would set
4284 `RUST_LOG` to `rustc`, which is the crate name (as specified in its `crate_id`
4285 [attribute](#attributes)). To narrow down the logs to just crate resolution,
4286 you would set it to `rustc::metadata::creader`. To see just error logging
4289 Note that when compiling source files that don't specify a
4290 crate name the crate is given a default name that matches the source file,
4291 with the extension removed. In that case, to turn on logging for a program
4292 compiled from, e.g. `helloworld.rs`, `RUST_LOG` should be set to `helloworld`.
4294 #### Logging Expressions
4296 Rust provides several macros to log information. Here's a simple Rust program
4297 that demonstrates all four of them:
4301 #[phase(plugin, link)] extern crate log;
4304 error!("This is an error log")
4305 warn!("This is a warn log")
4306 info!("this is an info log")
4307 debug!("This is a debug log")
4311 These four log levels correspond to levels 1-4, as controlled by `RUST_LOG`:
4314 $ RUST_LOG=rust=3 ./rust
4315 This is an error log
4320 # Appendix: Rationales and design tradeoffs
4324 # Appendix: Influences and further references
4328 > The essential problem that must be solved in making a fault-tolerant
4329 > software system is therefore that of fault-isolation. Different programmers
4330 > will write different modules, some modules will be correct, others will have
4331 > errors. We do not want the errors in one module to adversely affect the
4332 > behaviour of a module which does not have any errors.
4334 > — Joe Armstrong
4336 > In our approach, all data is private to some process, and processes can
4337 > only communicate through communications channels. *Security*, as used
4338 > in this paper, is the property which guarantees that processes in a system
4339 > cannot affect each other except by explicit communication.
4341 > When security is absent, nothing which can be proven about a single module
4342 > in isolation can be guaranteed to hold when that module is embedded in a
4345 > — Robert Strom and Shaula Yemini
4347 > Concurrent and applicative programming complement each other. The
4348 > ability to send messages on channels provides I/O without side effects,
4349 > while the avoidance of shared data helps keep concurrent processes from
4354 Rust is not a particularly original language. It may however appear unusual
4355 by contemporary standards, as its design elements are drawn from a number of
4356 "historical" languages that have, with a few exceptions, fallen out of
4357 favour. Five prominent lineages contribute the most, though their influences
4358 have come and gone during the course of Rust's development:
4360 * The NIL (1981) and Hermes (1990) family. These languages were developed by
4361 Robert Strom, Shaula Yemini, David Bacon and others in their group at IBM
4362 Watson Research Center (Yorktown Heights, NY, USA).
4364 * The Erlang (1987) language, developed by Joe Armstrong, Robert Virding, Claes
4365 Wikström, Mike Williams and others in their group at the Ericsson Computer
4366 Science Laboratory (Älvsjö, Stockholm, Sweden) .
4368 * The Sather (1990) language, developed by Stephen Omohundro, Chu-Cheow Lim,
4369 Heinz Schmidt and others in their group at The International Computer
4370 Science Institute of the University of California, Berkeley (Berkeley, CA,
4373 * The Newsqueak (1988), Alef (1995), and Limbo (1996) family. These
4374 languages were developed by Rob Pike, Phil Winterbottom, Sean Dorward and
4375 others in their group at Bell Labs Computing Sciences Research Center
4376 (Murray Hill, NJ, USA).
4378 * The Napier (1985) and Napier88 (1988) family. These languages were
4379 developed by Malcolm Atkinson, Ron Morrison and others in their group at
4380 the University of St. Andrews (St. Andrews, Fife, UK).
4382 Additional specific influences can be seen from the following languages:
4384 * The structural algebraic types and compilation manager of SML.
4385 * The attribute and assembly systems of C#.
4386 * The references and deterministic destructor system of C++.
4387 * The memory region systems of the ML Kit and Cyclone.
4388 * The typeclass system of Haskell.
4389 * The lexical identifier rule of Python.
4390 * The block syntax of Ruby.
4392 [ffi]: guide-ffi.html