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 normalized to Unicode normalization form NFKC.
117 Most Rust grammar rules are defined in terms of printable ASCII-range codepoints,
118 but a small number are defined in terms of Unicode properties or explicit
119 codepoint lists. [^inputformat]
121 [^inputformat]: Substitute definitions for the special Unicode productions are
122 provided to the grammar verifier, restricted to ASCII range, when verifying
123 the grammar in this document.
125 ## Special Unicode Productions
127 The following productions in the Rust grammar are defined in terms of Unicode properties:
128 `ident`, `non_null`, `non_star`, `non_eol`, `non_slash_or_star`, `non_single_quote` and `non_double_quote`.
132 The `ident` production is any nonempty Unicode string of the following form:
134 - The first character has property `XID_start`
135 - The remaining characters have property `XID_continue`
137 that does _not_ occur in the set of [keywords](#keywords).
139 Note: `XID_start` and `XID_continue` as character properties cover the
140 character ranges used to form the more familiar C and Java language-family
143 ### Delimiter-restricted productions
145 Some productions are defined by exclusion of particular Unicode characters:
147 - `non_null` is any single Unicode character aside from `U+0000` (null)
148 - `non_eol` is `non_null` restricted to exclude `U+000A` (`'\n'`)
149 - `non_star` is `non_null` restricted to exclude `U+002A` (`*`)
150 - `non_slash_or_star` is `non_null` restricted to exclude `U+002F` (`/`) and `U+002A` (`*`)
151 - `non_single_quote` is `non_null` restricted to exclude `U+0027` (`'`)
152 - `non_double_quote` is `non_null` restricted to exclude `U+0022` (`"`)
157 comment : block_comment | line_comment ;
158 block_comment : "/*" block_comment_body * '*' + '/' ;
159 block_comment_body : [block_comment | character] * ;
160 line_comment : "//" non_eol * ;
163 Comments in Rust code follow the general C++ style of line and block-comment forms.
164 Nested block comments are supported.
166 Line comments beginning with exactly _three_ slashes (`///`), and block
167 comments beginning with exactly one repeated asterisk in the block-open
168 sequence (`/**`), are interpreted as a special syntax for `doc`
169 [attributes](#attributes). That is, they are equivalent to writing
170 `#[doc="..."]` around the body of the comment (this includes the comment
171 characters themselves, ie `/// Foo` turns into `#[doc="/// Foo"]`).
173 Non-doc comments are interpreted as a form of whitespace.
178 whitespace_char : '\x20' | '\x09' | '\x0a' | '\x0d' ;
179 whitespace : [ whitespace_char | comment ] + ;
182 The `whitespace_char` production is any nonempty Unicode string consisting of any
183 of the following Unicode characters: `U+0020` (space, `' '`), `U+0009` (tab,
184 `'\t'`), `U+000A` (LF, `'\n'`), `U+000D` (CR, `'\r'`).
186 Rust is a "free-form" language, meaning that all forms of whitespace serve
187 only to separate _tokens_ in the grammar, and have no semantic significance.
189 A Rust program has identical meaning if each whitespace element is replaced
190 with any other legal whitespace element, such as a single space character.
195 simple_token : keyword | unop | binop ;
196 token : simple_token | ident | literal | symbol | whitespace token ;
199 Tokens are primitive productions in the grammar defined by regular
200 (non-recursive) languages. "Simple" tokens are given in [string table
201 production](#string-table-productions) form, and occur in the rest of the
202 grammar as double-quoted strings. Other tokens have exact rules given.
206 The keywords are the following strings:
208 ~~~~ {.text .keyword}
219 self static struct super
225 Each of these keywords has special meaning in its grammar,
226 and all of them are excluded from the `ident` rule.
230 A literal is an expression consisting of a single token, rather than a
231 sequence of tokens, that immediately and directly denotes the value it
232 evaluates to, rather than referring to it by name or some other evaluation
233 rule. A literal is a form of constant expression, so is evaluated (primarily)
237 literal : string_lit | char_lit | byte_string_lit | byte_lit | num_lit ;
240 #### Character and string literals
243 char_lit : '\x27' char_body '\x27' ;
244 string_lit : '"' string_body * '"' | 'r' raw_string ;
246 char_body : non_single_quote
247 | '\x5c' [ '\x27' | common_escape | unicode_escape ] ;
249 string_body : non_double_quote
250 | '\x5c' [ '\x22' | common_escape | unicode_escape ] ;
251 raw_string : '"' raw_string_body '"' | '#' raw_string '#' ;
253 common_escape : '\x5c'
254 | 'n' | 'r' | 't' | '0'
256 unicode_escape : 'u' hex_digit 4
259 hex_digit : 'a' | 'b' | 'c' | 'd' | 'e' | 'f'
260 | 'A' | 'B' | 'C' | 'D' | 'E' | 'F'
262 oct_digit : '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' ;
263 dec_digit : '0' | nonzero_dec ;
264 nonzero_dec: '1' | '2' | '3' | '4'
265 | '5' | '6' | '7' | '8' | '9' ;
268 A _character literal_ is a single Unicode character enclosed within two
269 `U+0027` (single-quote) characters, with the exception of `U+0027` itself,
270 which must be _escaped_ by a preceding U+005C character (`\`).
272 A _string literal_ is a sequence of any Unicode characters enclosed within
273 two `U+0022` (double-quote) characters, with the exception of `U+0022`
274 itself, which must be _escaped_ by a preceding `U+005C` character (`\`),
275 or a _raw string literal_.
277 Some additional _escapes_ are available in either character or non-raw string
278 literals. An escape starts with a `U+005C` (`\`) and continues with one of
281 * An _8-bit codepoint escape_ escape starts with `U+0078` (`x`) and is
282 followed by exactly two _hex digits_. It denotes the Unicode codepoint
283 equal to the provided hex value.
284 * A _16-bit codepoint escape_ starts with `U+0075` (`u`) and is followed
285 by exactly four _hex digits_. It denotes the Unicode codepoint equal to
286 the provided hex value.
287 * A _32-bit codepoint escape_ starts with `U+0055` (`U`) and is followed
288 by exactly eight _hex digits_. It denotes the Unicode codepoint equal to
289 the provided hex value.
290 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
291 (`r`), or `U+0074` (`t`), denoting the unicode values `U+000A` (LF),
292 `U+000D` (CR) or `U+0009` (HT) respectively.
293 * The _backslash escape_ is the character `U+005C` (`\`) which must be
294 escaped in order to denote *itself*.
296 Raw string literals do not process any escapes. They start with the character
297 `U+0072` (`r`), followed by zero or more of the character `U+0023` (`#`) and a
298 `U+0022` (double-quote) character. The _raw string body_ is not defined in the
299 EBNF grammar above: it can contain any sequence of Unicode characters and is
300 terminated only by another `U+0022` (double-quote) character, followed by the
301 same number of `U+0023` (`#`) characters that preceded the opening `U+0022`
302 (double-quote) character.
304 All Unicode characters contained in the raw string body represent themselves,
305 the characters `U+0022` (double-quote) (except when followed by at least as
306 many `U+0023` (`#`) characters as were used to start the raw string literal) or
307 `U+005C` (`\`) do not have any special meaning.
309 Examples for string literals:
312 "foo"; r"foo"; // foo
313 "\"foo\""; r#""foo""#; // "foo"
316 r##"foo #"# bar"##; // foo #"# bar
318 "\x52"; "R"; r"R"; // R
319 "\\x52"; r"\x52"; // \x52
322 #### Byte and byte string literals
325 byte_lit : 'b' '\x27' byte_body '\x27' ;
326 byte_string_lit : 'b' '"' string_body * '"' | 'b' 'r' raw_byte_string ;
328 byte_body : ascii_non_single_quote
329 | '\x5c' [ '\x27' | common_escape ] ;
331 byte_string_body : ascii_non_double_quote
332 | '\x5c' [ '\x22' | common_escape ] ;
333 raw_byte_string : '"' raw_byte_string_body '"' | '#' raw_byte_string '#' ;
337 A _byte literal_ is a single ASCII character (in the `U+0000` to `U+007F` range)
338 enclosed within two `U+0027` (single-quote) characters,
339 with the exception of `U+0027` itself,
340 which must be _escaped_ by a preceding U+005C character (`\`),
341 or a single _escape_.
342 It is equivalent to a `u8` unsigned 8-bit integer _number literal_.
344 A _byte string literal_ is a sequence of ASCII characters and _escapes_
345 enclosed within two `U+0022` (double-quote) characters,
346 with the exception of `U+0022` itself,
347 which must be _escaped_ by a preceding `U+005C` character (`\`),
348 or a _raw byte string literal_.
349 It is equivalent to a `&'static [u8]` borrowed vector of unsigned 8-bit integers.
351 Some additional _escapes_ are available in either byte or non-raw byte string
352 literals. An escape starts with a `U+005C` (`\`) and continues with one of
355 * An _byte escape_ escape starts with `U+0078` (`x`) and is
356 followed by exactly two _hex digits_. It denotes the byte
357 equal to the provided hex value.
358 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
359 (`r`), or `U+0074` (`t`), denoting the bytes values `0x0A` (ASCII LF),
360 `0x0D` (ASCII CR) or `0x09` (ASCII HT) respectively.
361 * The _backslash escape_ is the character `U+005C` (`\`) which must be
362 escaped in order to denote its ASCII encoding `0x5C`.
364 Raw byte string literals do not process any escapes.
365 They start with the character `U+0072` (`r`),
366 followed by `U+0062` (`b`),
367 followed by zero or more of the character `U+0023` (`#`),
368 and a `U+0022` (double-quote) character.
369 The _raw string body_ is not defined in the EBNF grammar above:
370 it can contain any sequence of ASCII characters and is
371 terminated only by another `U+0022` (double-quote) character, followed by the
372 same number of `U+0023` (`#`) characters that preceded the opening `U+0022`
373 (double-quote) character.
374 A raw byte string literal can not contain any non-ASCII byte.
376 All characters contained in the raw string body represent their ASCII encoding,
377 the characters `U+0022` (double-quote) (except when followed by at least as
378 many `U+0023` (`#`) characters as were used to start the raw string literal) or
379 `U+005C` (`\`) do not have any special meaning.
381 Examples for byte string literals:
384 b"foo"; br"foo"; // foo
385 b"\"foo\""; br#""foo""#; // "foo"
388 br##"foo #"# bar"##; // foo #"# bar
390 b"\x52"; b"R"; br"R"; // R
391 b"\\x52"; br"\x52"; // \x52
397 num_lit : nonzero_dec [ dec_digit | '_' ] * num_suffix ?
398 | '0' [ [ dec_digit | '_' ] * num_suffix ?
399 | 'b' [ '1' | '0' | '_' ] + int_suffix ?
400 | 'o' [ oct_digit | '_' ] + int_suffix ?
401 | 'x' [ hex_digit | '_' ] + int_suffix ? ] ;
403 num_suffix : int_suffix | float_suffix ;
405 int_suffix : 'u' int_suffix_size ?
406 | 'i' int_suffix_size ? ;
407 int_suffix_size : [ '8' | '1' '6' | '3' '2' | '6' '4' ] ;
409 float_suffix : [ exponent | '.' dec_lit exponent ? ] ? float_suffix_ty ? ;
410 float_suffix_ty : 'f' [ '3' '2' | '6' '4' ] ;
411 exponent : ['E' | 'e'] ['-' | '+' ] ? dec_lit ;
412 dec_lit : [ dec_digit | '_' ] + ;
415 A _number literal_ is either an _integer literal_ or a _floating-point
416 literal_. The grammar for recognizing the two kinds of literals is mixed,
417 as they are differentiated by suffixes.
419 ##### Integer literals
421 An _integer literal_ has one of four forms:
423 * A _decimal literal_ starts with a *decimal digit* and continues with any
424 mixture of *decimal digits* and _underscores_.
425 * A _hex literal_ starts with the character sequence `U+0030` `U+0078`
426 (`0x`) and continues as any mixture hex digits and underscores.
427 * An _octal literal_ starts with the character sequence `U+0030` `U+006F`
428 (`0o`) and continues as any mixture octal digits and underscores.
429 * A _binary literal_ starts with the character sequence `U+0030` `U+0062`
430 (`0b`) and continues as any mixture binary digits and underscores.
432 An integer literal may be followed (immediately, without any spaces) by an
433 _integer suffix_, which changes the type of the literal. There are two kinds
434 of integer literal suffix:
436 * The `i` and `u` suffixes give the literal type `int` or `uint`,
438 * Each of the signed and unsigned machine types `u8`, `i8`,
439 `u16`, `i16`, `u32`, `i32`, `u64` and `i64`
440 give the literal the corresponding machine type.
442 The type of an _unsuffixed_ integer literal is determined by type inference.
443 If an integer type can be _uniquely_ determined from the surrounding program
444 context, the unsuffixed integer literal has that type. If the program context
445 underconstrains the type, it is considered a static type error;
446 if the program context overconstrains the type,
447 it is also considered a static type error.
449 Examples of integer literals of various forms:
456 0o70_i16; // type i16
457 0b1111_1111_1001_0000_i32; // type i32
460 ##### Floating-point literals
462 A _floating-point literal_ has one of two forms:
464 * Two _decimal literals_ separated by a period
465 character `U+002E` (`.`), with an optional _exponent_ trailing after the
466 second decimal literal.
467 * A single _decimal literal_ followed by an _exponent_.
469 By default, a floating-point literal has a generic type,
470 and, like integer literals, the type must be uniquely determined
472 A floating-point literal may be followed (immediately, without any
473 spaces) by a _floating-point suffix_, which changes the type of the literal.
474 There are two floating-point suffixes: `f32`, and `f64` (the 32-bit and 64-bit
475 floating point types).
477 Examples of floating-point literals of various forms:
480 123.0f64; // type f64
483 12E+99_f64; // type f64
486 ##### Unit and boolean literals
488 The _unit value_, the only value of the type that has the same name, is written as `()`.
489 The two values of the boolean type are written `true` and `false`.
495 | '#' | '[' | ']' | '(' | ')' | '{' | '}'
499 Symbols are a general class of printable [token](#tokens) that play structural
500 roles in a variety of grammar productions. They are catalogued here for
501 completeness as the set of remaining miscellaneous printable tokens that do not
502 otherwise appear as [unary operators](#unary-operator-expressions), [binary
503 operators](#binary-operator-expressions), or [keywords](#keywords).
509 expr_path : [ "::" ] ident [ "::" expr_path_tail ] + ;
510 expr_path_tail : '<' type_expr [ ',' type_expr ] + '>'
513 type_path : ident [ type_path_tail ] + ;
514 type_path_tail : '<' type_expr [ ',' type_expr ] + '>'
518 A _path_ is a sequence of one or more path components _logically_ separated by
519 a namespace qualifier (`::`). If a path consists of only one component, it may
520 refer to either an [item](#items) or a [slot](#memory-slots) in a local
521 control scope. If a path has multiple components, it refers to an item.
523 Every item has a _canonical path_ within its crate, but the path naming an
524 item is only meaningful within a given crate. There is no global namespace
525 across crates; an item's canonical path merely identifies it within the crate.
527 Two examples of simple paths consisting of only identifier components:
534 Path components are usually [identifiers](#identifiers), but the trailing
535 component of a path may be an angle-bracket-enclosed list of type
536 arguments. In [expression](#expressions) context, the type argument list is
537 given after a final (`::`) namespace qualifier in order to disambiguate it
538 from a relational expression involving the less-than symbol (`<`). In type
539 expression context, the final namespace qualifier is omitted.
541 Two examples of paths with type arguments:
544 # struct HashMap<K, V>;
546 # fn id<T>(t: T) -> T { t }
547 type T = HashMap<int,String>; // Type arguments used in a type expression
548 let x = id::<int>(10); // Type arguments used in a call expression
552 Paths can be denoted with various leading qualifiers to change the meaning of
555 * Paths starting with `::` are considered to be global paths where the
556 components of the path start being resolved from the crate root. Each
557 identifier in the path must resolve to an item.
565 ::a::foo(); // call a's foo function
571 * Paths starting with the keyword `super` begin resolution relative to the
572 parent module. Each further identifier must resolve to an item
580 super::a::foo(); // call a's foo function
586 * Paths starting with the keyword `self` begin resolution relative to the
587 current module. Each further identifier must resolve to an item.
599 A number of minor features of Rust are not central enough to have their own
600 syntax, and yet are not implementable as functions. Instead, they are given
601 names, and invoked through a consistent syntax: `name!(...)`. Examples
604 * `format!` : format data into a string
605 * `env!` : look up an environment variable's value at compile time
606 * `file!`: return the path to the file being compiled
607 * `stringify!` : pretty-print the Rust expression given as an argument
608 * `include!` : include the Rust expression in the given file
609 * `include_str!` : include the contents of the given file as a string
610 * `include_bin!` : include the contents of the given file as a binary blob
611 * `error!`, `warn!`, `info!`, `debug!` : provide diagnostic information.
613 All of the above extensions are expressions with values.
618 expr_macro_rules : "macro_rules" '!' ident '(' macro_rule * ')' ;
619 macro_rule : '(' matcher * ')' "=>" '(' transcriber * ')' ';' ;
620 matcher : '(' matcher * ')' | '[' matcher * ']'
621 | '{' matcher * '}' | '$' ident ':' ident
622 | '$' '(' matcher * ')' sep_token? [ '*' | '+' ]
623 | non_special_token ;
624 transcriber : '(' transcriber * ')' | '[' transcriber * ']'
625 | '{' transcriber * '}' | '$' ident
626 | '$' '(' transcriber * ')' sep_token? [ '*' | '+' ]
627 | non_special_token ;
630 User-defined syntax extensions are called "macros",
631 and the `macro_rules` syntax extension defines them.
632 Currently, user-defined macros can expand to expressions, statements, or items.
634 (A `sep_token` is any token other than `*` and `+`.
635 A `non_special_token` is any token other than a delimiter or `$`.)
637 The macro expander looks up macro invocations by name,
638 and tries each macro rule in turn.
639 It transcribes the first successful match.
640 Matching and transcription are closely related to each other,
641 and we will describe them together.
645 The macro expander matches and transcribes every token that does not begin with a `$` literally, including delimiters.
646 For parsing reasons, delimiters must be balanced, but they are otherwise not special.
648 In the matcher, `$` _name_ `:` _designator_ matches the nonterminal in the
649 Rust syntax named by _designator_. Valid designators are `item`, `block`,
650 `stmt`, `pat`, `expr`, `ty` (type), `ident`, `path`, `matchers` (lhs of the `=>` in macro rules),
651 `tt` (rhs of the `=>` in macro rules). In the transcriber, the designator is already known, and so only
652 the name of a matched nonterminal comes after the dollar sign.
654 In both the matcher and transcriber, the Kleene star-like operator indicates repetition.
655 The Kleene star operator consists of `$` and parens, optionally followed by a separator token, followed by `*` or `+`.
656 `*` means zero or more repetitions, `+` means at least one repetition.
657 The parens are not matched or transcribed.
658 On the matcher side, a name is bound to _all_ of the names it
659 matches, in a structure that mimics the structure of the repetition
660 encountered on a successful match. The job of the transcriber is to sort that
663 The rules for transcription of these repetitions are called "Macro By Example".
664 Essentially, one "layer" of repetition is discharged at a time, and all of
665 them must be discharged by the time a name is transcribed. Therefore,
666 `( $( $i:ident ),* ) => ( $i )` is an invalid macro, but
667 `( $( $i:ident ),* ) => ( $( $i:ident ),* )` is acceptable (if trivial).
669 When Macro By Example encounters a repetition, it examines all of the `$`
670 _name_ s that occur in its body. At the "current layer", they all must repeat
671 the same number of times, so
672 ` ( $( $i:ident ),* ; $( $j:ident ),* ) => ( $( ($i,$j) ),* )` is valid if
673 given the argument `(a,b,c ; d,e,f)`, but not `(a,b,c ; d,e)`. The repetition
674 walks through the choices at that layer in lockstep, so the former input
675 transcribes to `( (a,d), (b,e), (c,f) )`.
677 Nested repetitions are allowed.
679 ### Parsing limitations
681 The parser used by the macro system is reasonably powerful, but the parsing of
682 Rust syntax is restricted in two ways:
684 1. The parser will always parse as much as possible. If it attempts to match
685 `$i:expr [ , ]` against `8 [ , ]`, it will attempt to parse `i` as an array
686 index operation and fail. Adding a separator can solve this problem.
687 2. The parser must have eliminated all ambiguity by the time it reaches a `$` _name_ `:` _designator_.
688 This requirement most often affects name-designator pairs when they occur at the beginning of, or immediately after, a `$(...)*`; requiring a distinctive token in front can solve the problem.
690 ## Syntax extensions useful for the macro author
692 * `log_syntax!` : print out the arguments at compile time
693 * `trace_macros!` : supply `true` or `false` to enable or disable macro expansion logging
694 * `stringify!` : turn the identifier argument into a string literal
695 * `concat!` : concatenates a comma-separated list of literals
696 * `concat_idents!` : create a new identifier by concatenating the arguments
698 # Crates and source files
700 Rust is a *compiled* language.
701 Its semantics obey a *phase distinction* between compile-time and run-time.
702 Those semantic rules that have a *static interpretation* govern the success or failure of compilation.
703 We refer to these rules as "static semantics".
704 Semantic rules called "dynamic semantics" govern the behavior of programs at run-time.
705 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.
707 The compilation model centres on artifacts called _crates_.
708 Each compilation processes a single crate in source form, and if successful,
709 produces a single crate in binary form: either an executable or a
710 library.[^cratesourcefile]
712 [^cratesourcefile]: A crate is somewhat analogous to an *assembly* in the
713 ECMA-335 CLI model, a *library* in the SML/NJ Compilation Manager, a *unit*
714 in the Owens and Flatt module system, or a *configuration* in Mesa.
716 A _crate_ is a unit of compilation and linking, as well as versioning, distribution and runtime loading.
717 A crate contains a _tree_ of nested [module](#modules) scopes.
718 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.
720 The Rust compiler is always invoked with a single source file as input, and always produces a single output crate.
721 The processing of that source file may result in other source files being loaded as modules.
722 Source files have the extension `.rs`.
724 A Rust source file describes a module, the name and
725 location of which — in the module tree of the current crate — are defined
726 from outside the source file: either by an explicit `mod_item` in
727 a referencing source file, or by the name of the crate itself.
729 Each source file contains a sequence of zero or more `item` definitions,
730 and may optionally begin with any number of `attributes` that apply to the containing module.
731 Attributes on the anonymous crate module define important metadata that influences
732 the behavior of the compiler.
735 # #![allow(unused_attribute)]
737 #![crate_id = "projx#2.5"]
739 // Additional metadata attributes
740 #![desc = "Project X"]
742 #![comment = "This is a comment on Project X."]
744 // Specify the output type
745 #![crate_type = "lib"]
748 #![warn(non_camel_case_types)]
751 A crate that contains a `main` function can be compiled to an executable.
752 If a `main` function is present, its return type must be [`unit`](#primitive-types) and it must take no arguments.
754 # Items and attributes
756 Crates contain [items](#items),
757 each of which may have some number of [attributes](#attributes) attached to it.
762 item : mod_item | fn_item | type_item | struct_item | enum_item
763 | static_item | trait_item | impl_item | extern_block ;
766 An _item_ is a component of a crate; some module items can be defined in crate
767 files, but most are defined in source files. Items are organized within a
768 crate by a nested set of [modules](#modules). Every crate has a single
769 "outermost" anonymous module; all further items within the crate have
770 [paths](#paths) within the module tree of the crate.
772 Items are entirely determined at compile-time, generally remain fixed during
773 execution, and may reside in read-only memory.
775 There are several kinds of item:
777 * [modules](#modules)
778 * [functions](#functions)
779 * [type definitions](#type-definitions)
780 * [structures](#structures)
781 * [enumerations](#enumerations)
782 * [static items](#static-items)
784 * [implementations](#implementations)
786 Some items form an implicit scope for the declaration of sub-items. In other
787 words, within a function or module, declarations of items can (in many cases)
788 be mixed with the statements, control blocks, and similar artifacts that
789 otherwise compose the item body. The meaning of these scoped items is the same
790 as if the item was declared outside the scope — it is still a static item —
791 except that the item's *path name* within the module namespace is qualified by
792 the name of the enclosing item, or is private to the enclosing item (in the
794 The grammar specifies the exact locations in which sub-item declarations may appear.
798 All items except modules may be *parameterized* by type. Type parameters are
799 given as a comma-separated list of identifiers enclosed in angle brackets
800 (`<...>`), after the name of the item and before its definition.
801 The type parameters of an item are considered "part of the name", not part of the type of the item.
802 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.
803 In practice, the type-inference system can usually infer such argument types from context.
804 There are no general type-parametric types, only type-parametric items.
805 That is, Rust has no notion of type abstraction: there are no first-class "forall" types.
810 mod_item : "mod" ident ( ';' | '{' mod '}' );
811 mod : [ view_item | item ] * ;
814 A module is a container for zero or more [view items](#view-items) and zero or
815 more [items](#items). The view items manage the visibility of the items
816 defined within the module, as well as the visibility of names from outside the
817 module when referenced from inside the module.
819 A _module item_ is a module, surrounded in braces, named, and prefixed with
820 the keyword `mod`. A module item introduces a new, named module into the tree
821 of modules making up a crate. Modules can nest arbitrarily.
823 An example of a module:
827 type Complex = (f64, f64);
828 fn sin(f: f64) -> f64 {
832 fn cos(f: f64) -> f64 {
836 fn tan(f: f64) -> f64 {
843 Modules and types share the same namespace.
844 Declaring a named type that has the same name as a module in scope is forbidden:
845 that is, a type definition, trait, struct, enumeration, or type parameter
846 can't shadow the name of a module in scope, or vice versa.
848 A module without a body is loaded from an external file, by default with the same
849 name as the module, plus the `.rs` extension.
850 When a nested submodule is loaded from an external file,
851 it is loaded from a subdirectory path that mirrors the module hierarchy.
854 // Load the `vec` module from `vec.rs`
858 // Load the `local_data` module from `task/local_data.rs`
863 The directories and files used for loading external file modules can be influenced
864 with the `path` attribute.
867 #[path = "task_files"]
869 // Load the `local_data` module from `task_files/tls.rs`
878 view_item : extern_crate_decl | use_decl ;
881 A view item manages the namespace of a module.
882 View items do not define new items, but rather, simply change other items' visibility.
883 There are several kinds of view item:
885 * [`extern crate` declarations](#extern-crate-declarations)
886 * [`use` declarations](#use-declarations)
888 ##### Extern crate declarations
891 extern_crate_decl : "extern" "crate" ident [ '(' link_attrs ')' ] ? [ '=' string_lit ] ? ;
892 link_attrs : link_attr [ ',' link_attrs ] + ;
893 link_attr : ident '=' literal ;
896 An _`extern crate` declaration_ specifies a dependency on an external crate.
897 The external crate is then bound into the declaring scope as the `ident` provided
898 in the `extern_crate_decl`.
900 The external crate is resolved to a specific `soname` at compile time, and a
901 runtime linkage requirement to that `soname` is passed to the linker for
902 loading at runtime. The `soname` is resolved at compile time by scanning the
903 compiler's library path and matching the optional `crateid` provided as a string literal
904 against the `crateid` attributes that were declared on the external crate when
905 it was compiled. If no `crateid` is provided, a default `name` attribute is
906 assumed, equal to the `ident` given in the `extern_crate_decl`.
908 Four examples of `extern crate` declarations:
913 extern crate std; // equivalent to: extern crate std = "std";
915 extern crate ruststd = "std"; // linking to 'std' under another name
917 extern crate foo = "some/where/rust-foo#foo:1.0"; // a full crate ID for external tools
920 ##### Use declarations
923 use_decl : "pub" ? "use" [ ident '=' path
926 path_glob : ident [ "::" [ path_glob
928 | '{' ident [ ',' ident ] * '}' ;
931 A _use declaration_ creates one or more local name bindings synonymous
932 with some other [path](#paths).
933 Usually a `use` declaration is used to shorten the path required to refer to a
934 module item. These declarations may appear at the top of [modules](#modules) and
937 *Note*: Unlike in many languages,
938 `use` declarations in Rust do *not* declare linkage dependency with external crates.
939 Rather, [`extern crate` declarations](#extern-crate-declarations) declare linkage dependencies.
941 Use declarations support a number of convenient shortcuts:
943 * Rebinding the target name as a new local name, using the syntax `use x = p::q::r;`.
944 * Simultaneously binding a list of paths differing only in their final element,
945 using the glob-like brace syntax `use a::b::{c,d,e,f};`
946 * Binding all paths matching a given prefix, using the asterisk wildcard syntax `use a::b::*;`
948 An example of `use` declarations:
951 use std::iter::range_step;
952 use std::option::{Some, None};
957 // Equivalent to 'std::iter::range_step(0u, 10u, 2u);'
958 range_step(0u, 10u, 2u);
960 // Equivalent to 'foo(vec![std::option::Some(1.0f64),
961 // std::option::None]);'
962 foo(vec![Some(1.0f64), None]);
966 Like items, `use` declarations are private to the containing module, by default.
967 Also like items, a `use` declaration can be public, if qualified by the `pub` keyword.
968 Such a `use` declaration serves to _re-export_ a name.
969 A public `use` declaration can therefore _redirect_ some public name to a different target definition:
970 even a definition with a private canonical path, inside a different module.
971 If a sequence of such redirections form a cycle or cannot be resolved unambiguously,
972 they represent a compile-time error.
974 An example of re-exporting:
979 pub use quux::foo::{bar, baz};
988 In this example, the module `quux` re-exports two public names defined in `foo`.
990 Also note that the paths contained in `use` items are relative to the crate root.
991 So, in the previous example, the `use` refers to `quux::foo::{bar, baz}`, and not simply to `foo::{bar, baz}`.
992 This also means that top-level module declarations should be at the crate root if direct usage
993 of the declared modules within `use` items is desired. It is also possible to use `self` and `super`
994 at the beginning of a `use` item to refer to the current and direct parent modules respectively.
995 All rules regarding accessing declared modules in `use` declarations applies to both module declarations
996 and `extern crate` declarations.
998 An example of what will and will not work for `use` items:
1001 # #![allow(unused_imports)]
1002 use foo::native::start; // good: foo is at the root of the crate
1003 use foo::baz::foobaz; // good: foo is at the root of the crate
1006 extern crate native;
1008 use foo::native::start; // good: foo is at crate root
1009 // use native::start; // bad: native is not at the crate root
1010 use self::baz::foobaz; // good: self refers to module 'foo'
1011 use foo::bar::foobar; // good: foo is at crate root
1018 use super::bar::foobar; // good: super refers to module 'foo'
1028 A _function item_ defines a sequence of [statements](#statements) and an optional final [expression](#expressions), along with a name and a set of parameters.
1029 Functions are declared with the keyword `fn`.
1030 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.
1032 A function may also be copied into a first class *value*, in which case the
1033 value has the corresponding [*function type*](#function-types), and can be
1034 used otherwise exactly as a function item (with a minor additional cost of
1035 calling the function indirectly).
1037 Every control path in a function logically ends with a `return` expression or a
1038 diverging expression. If the outermost block of a function has a
1039 value-producing expression in its final-expression position, that expression
1040 is interpreted as an implicit `return` expression applied to the
1043 An example of a function:
1046 fn add(x: int, y: int) -> int {
1051 As with `let` bindings, function arguments are irrefutable patterns,
1052 so any pattern that is valid in a let binding is also valid as an argument.
1055 fn first((value, _): (int, int)) -> int { value }
1059 #### Generic functions
1061 A _generic function_ allows one or more _parameterized types_ to
1062 appear in its signature. Each type parameter must be explicitly
1063 declared, in an angle-bracket-enclosed, comma-separated list following
1067 fn iter<T>(seq: &[T], f: |T|) {
1068 for elt in seq.iter() { f(elt); }
1070 fn map<T, U>(seq: &[T], f: |T| -> U) -> Vec<U> {
1071 let mut acc = vec![];
1072 for elt in seq.iter() { acc.push(f(elt)); }
1077 Inside the function signature and body, the name of the type parameter
1078 can be used as a type name.
1080 When a generic function is referenced, its type is instantiated based
1081 on the context of the reference. For example, calling the `iter`
1082 function defined above on `[1, 2]` will instantiate type parameter `T`
1083 with `int`, and require the closure parameter to have type
1086 The type parameters can also be explicitly supplied in a trailing
1087 [path](#paths) component after the function name. This might be necessary
1088 if there is not sufficient context to determine the type parameters. For
1089 example, `mem::size_of::<u32>() == 4`.
1091 Since a parameter type is opaque to the generic function, the set of
1092 operations that can be performed on it is limited. Values of parameter
1093 type can only be moved, not copied.
1096 fn id<T>(x: T) -> T { x }
1099 Similarly, [trait](#traits) bounds can be specified for type
1100 parameters to allow methods with that trait to be called on values
1106 Unsafe operations are those that potentially violate the memory-safety guarantees of Rust's static semantics.
1108 The following language level features cannot be used in the safe subset of Rust:
1110 - Dereferencing a [raw pointer](#pointer-types).
1111 - Reading or writing a [mutable static variable](#mutable-statics).
1112 - Calling an unsafe function (including an intrinsic or foreign function).
1114 ##### Unsafe functions
1116 Unsafe functions are functions that are not safe in all contexts and/or for all possible inputs.
1117 Such a function must be prefixed with the keyword `unsafe`.
1121 A block of code can also be prefixed with the `unsafe` keyword, to permit calling `unsafe` functions
1122 or dereferencing raw pointers within a safe function.
1124 When a programmer has sufficient conviction that a sequence of potentially unsafe operations is
1125 actually safe, they can encapsulate that sequence (taken as a whole) within an `unsafe` block. The
1126 compiler will consider uses of such code safe, in the surrounding context.
1128 Unsafe blocks are used to wrap foreign libraries, make direct use of hardware or implement features
1129 not directly present in the language. For example, Rust provides the language features necessary to
1130 implement memory-safe concurrency in the language but the implementation of tasks and message
1131 passing is in the standard library.
1133 Rust's type system is a conservative approximation of the dynamic safety requirements, so in some
1134 cases there is a performance cost to using safe code. For example, a doubly-linked list is not a
1135 tree structure and can only be represented with managed or reference-counted pointers in safe code.
1136 By using `unsafe` blocks to represent the reverse links as raw pointers, it can be implemented with
1137 only owned pointers.
1139 ##### Behavior considered unsafe
1141 This is a list of behavior which is forbidden in all Rust code. Type checking provides the guarantee
1142 that these issues are never caused by safe code. An `unsafe` block or function is responsible for
1143 never invoking this behaviour or exposing an API making it possible for it to occur in safe code.
1146 * Dereferencing a null/dangling raw pointer
1147 * Mutating an immutable value/reference
1148 * Reads of [undef](http://llvm.org/docs/LangRef.html#undefined-values) (uninitialized) memory
1149 * Breaking the [pointer aliasing rules](http://llvm.org/docs/LangRef.html#pointer-aliasing-rules)
1150 with raw pointers (a subset of the rules used by C)
1151 * Invoking undefined behavior via compiler intrinsics:
1152 * Indexing outside of the bounds of an object with `std::ptr::offset` (`offset` intrinsic), with
1153 the exception of one byte past the end which is permitted.
1154 * Using `std::ptr::copy_nonoverlapping_memory` (`memcpy32`/`memcpy64` instrinsics) on
1156 * Invalid values in primitive types, even in private fields/locals:
1157 * Dangling/null pointers in non-raw pointers, or slices
1158 * A value other than `false` (0) or `true` (1) in a `bool`
1159 * A discriminant in an `enum` not included in the type definition
1160 * A value in a `char` which is a surrogate or above `char::MAX`
1161 * non-UTF-8 byte sequences in a `str`
1163 ##### Behaviour not considered unsafe
1165 This is a list of behaviour not considered *unsafe* in Rust terms, but that may be undesired.
1168 * Reading data from private fields (`std::repr`, `format!("{:?}", x)`)
1169 * Leaks due to reference count cycles, even in the global heap
1170 * Exiting without calling destructors
1172 * Accessing/modifying the file system
1173 * Unsigned integer overflow (well-defined as wrapping)
1174 * Signed integer overflow (well-defined as two's complement representation wrapping)
1176 #### Diverging functions
1178 A special kind of function can be declared with a `!` character where the
1179 output slot type would normally be. For example:
1182 fn my_err(s: &str) -> ! {
1188 We call such functions "diverging" because they never return a value to the
1189 caller. Every control path in a diverging function must end with a
1190 `fail!()` or a call to another diverging function on every
1191 control path. The `!` annotation does *not* denote a type. Rather, the result
1192 type of a diverging function is a special type called $\bot$ ("bottom") that
1193 unifies with any type. Rust has no syntax for $\bot$.
1195 It might be necessary to declare a diverging function because as mentioned
1196 previously, the typechecker checks that every control path in a function ends
1197 with a [`return`](#return-expressions) or diverging expression. So, if `my_err`
1198 were declared without the `!` annotation, the following code would not
1202 # fn my_err(s: &str) -> ! { fail!() }
1204 fn f(i: int) -> int {
1209 my_err("Bad number!");
1214 This will not compile without the `!` annotation on `my_err`,
1215 since the `else` branch of the conditional in `f` does not return an `int`,
1216 as required by the signature of `f`.
1217 Adding the `!` annotation to `my_err` informs the typechecker that,
1218 should control ever enter `my_err`, no further type judgments about `f` need to hold,
1219 since control will never resume in any context that relies on those judgments.
1220 Thus the return type on `f` only needs to reflect the `if` branch of the conditional.
1223 #### Extern functions
1225 Extern functions are part of Rust's foreign function interface,
1226 providing the opposite functionality to [external blocks](#external-blocks).
1227 Whereas external blocks allow Rust code to call foreign code,
1228 extern functions with bodies defined in Rust code _can be called by foreign
1229 code_. They are defined in the same way as any other Rust function,
1230 except that they have the `extern` modifier.
1233 // Declares an extern fn, the ABI defaults to "C"
1234 extern fn new_int() -> int { 0 }
1236 // Declares an extern fn with "stdcall" ABI
1237 extern "stdcall" fn new_int_stdcall() -> int { 0 }
1240 Unlike normal functions, extern fns have an `extern "ABI" fn()`.
1241 This is the same type as the functions declared in an extern
1245 # extern fn new_int() -> int { 0 }
1246 let fptr: extern "C" fn() -> int = new_int;
1249 Extern functions may be called directly from Rust code as Rust uses large,
1250 contiguous stack segments like C.
1252 ### Type definitions
1254 A _type definition_ defines a new name for an existing [type](#types). Type
1255 definitions are declared with the keyword `type`. Every value has a single,
1256 specific type; the type-specified aspects of a value include:
1258 * Whether the value is composed of sub-values or is indivisible.
1259 * Whether the value represents textual or numerical information.
1260 * Whether the value represents integral or floating-point information.
1261 * The sequence of memory operations required to access the value.
1262 * The [kind](#type-kinds) of the type.
1264 For example, the type `(u8, u8)` defines the set of immutable values that are composite pairs,
1265 each containing two unsigned 8-bit integers accessed by pattern-matching and laid out in memory with the `x` component preceding the `y` component.
1269 A _structure_ is a nominal [structure type](#structure-types) defined with the keyword `struct`.
1271 An example of a `struct` item and its use:
1274 struct Point {x: int, y: int}
1275 let p = Point {x: 10, y: 11};
1279 A _tuple structure_ is a nominal [tuple type](#tuple-types), also defined with the keyword `struct`.
1283 struct Point(int, int);
1284 let p = Point(10, 11);
1285 let px: int = match p { Point(x, _) => x };
1288 A _unit-like struct_ is a structure without any fields, defined by leaving off the list of fields entirely.
1289 Such types will have a single value, just like the [unit value `()`](#unit-and-boolean-literals) of the unit type.
1294 let c = [Cookie, Cookie, Cookie, Cookie];
1297 By using the `struct_inherit` feature gate, structures may use single inheritance. A Structure may only
1298 inherit from a single other structure, called the _super-struct_. The inheriting structure (sub-struct)
1299 acts as if all fields in the super-struct were present in the sub-struct. Fields declared in a sub-struct
1300 must not have the same name as any field in any (transitive) super-struct. All fields (both declared
1301 and inherited) must be specified in any initializers. Inheritance between structures does not give
1302 subtyping or coercion. The super-struct and sub-struct must be defined in the same crate. The super-struct
1303 must be declared using the `virtual` keyword.
1307 virtual struct Sup { x: int }
1308 struct Sub : Sup { y: int }
1309 let s = Sub {x: 10, y: 11};
1315 An _enumeration_ is a simultaneous definition of a nominal [enumerated type](#enumerated-types) as well as a set of *constructors*,
1316 that can be used to create or pattern-match values of the corresponding enumerated type.
1318 Enumerations are declared with the keyword `enum`.
1320 An example of an `enum` item and its use:
1328 let mut a: Animal = Dog;
1332 Enumeration constructors can have either named or unnamed fields:
1335 # #![feature(struct_variant)]
1339 Cat { name: String, weight: f64 }
1342 let mut a: Animal = Dog("Cocoa".to_string(), 37.2);
1343 a = Cat { name: "Spotty".to_string(), weight: 2.7 };
1347 In this example, `Cat` is a _struct-like enum variant_,
1348 whereas `Dog` is simply called an enum variant.
1353 static_item : "static" ident ':' type '=' expr ';' ;
1356 A *static item* is a named _constant value_ stored in the global data section of a crate.
1357 Immutable static items are stored in the read-only data section.
1358 The constant value bound to a static item is, like all constant values, evaluated at compile time.
1359 Static items have the `static` lifetime, which outlives all other lifetimes in a Rust program.
1360 Static items are declared with the `static` keyword.
1361 A static item must have a _constant expression_ giving its definition.
1363 Static items must be explicitly typed.
1364 The type may be ```bool```, ```char```, a number, or a type derived from those primitive types.
1365 The derived types are references with the `static` lifetime,
1366 fixed-size arrays, tuples, and structs.
1369 static BIT1: uint = 1 << 0;
1370 static BIT2: uint = 1 << 1;
1372 static BITS: [uint, ..2] = [BIT1, BIT2];
1373 static STRING: &'static str = "bitstring";
1375 struct BitsNStrings<'a> {
1376 mybits: [uint, ..2],
1380 static bits_n_strings: BitsNStrings<'static> = BitsNStrings {
1386 #### Mutable statics
1388 If a static item is declared with the ```mut``` keyword, then it is allowed to
1389 be modified by the program. One of Rust's goals is to make concurrency bugs hard
1390 to run into, and this is obviously a very large source of race conditions or
1391 other bugs. For this reason, an ```unsafe``` block is required when either
1392 reading or writing a mutable static variable. Care should be taken to ensure
1393 that modifications to a mutable static are safe with respect to other tasks
1394 running in the same process.
1396 Mutable statics are still very useful, however. They can be used with C
1397 libraries and can also be bound from C libraries (in an ```extern``` block).
1400 # fn atomic_add(_: &mut uint, _: uint) -> uint { 2 }
1402 static mut LEVELS: uint = 0;
1404 // This violates the idea of no shared state, and this doesn't internally
1405 // protect against races, so this function is `unsafe`
1406 unsafe fn bump_levels_unsafe1() -> uint {
1412 // Assuming that we have an atomic_add function which returns the old value,
1413 // this function is "safe" but the meaning of the return value may not be what
1414 // callers expect, so it's still marked as `unsafe`
1415 unsafe fn bump_levels_unsafe2() -> uint {
1416 return atomic_add(&mut LEVELS, 1);
1422 A _trait_ describes a set of method types.
1424 Traits can include default implementations of methods,
1425 written in terms of some unknown [`self` type](#self-types);
1426 the `self` type may either be completely unspecified,
1427 or constrained by some other trait.
1429 Traits are implemented for specific types through separate [implementations](#implementations).
1432 # type Surface = int;
1433 # type BoundingBox = int;
1435 fn draw(&self, Surface);
1436 fn bounding_box(&self) -> BoundingBox;
1440 This defines a trait with two methods.
1441 All values that have [implementations](#implementations) of this trait in scope can have their `draw` and `bounding_box` methods called,
1442 using `value.bounding_box()` [syntax](#method-call-expressions).
1444 Type parameters can be specified for a trait to make it generic.
1445 These appear after the trait name, using the same syntax used in [generic functions](#generic-functions).
1449 fn len(&self) -> uint;
1450 fn elt_at(&self, n: uint) -> T;
1451 fn iter(&self, |T|);
1455 Generic functions may use traits as _bounds_ on their type parameters.
1456 This will have two effects: only types that have the trait may instantiate the parameter,
1457 and within the generic function,
1458 the methods of the trait can be called on values that have the parameter's type.
1462 # type Surface = int;
1463 # trait Shape { fn draw(&self, Surface); }
1464 fn draw_twice<T: Shape>(surface: Surface, sh: T) {
1470 Traits also define an [object type](#object-types) with the same name as the trait.
1471 Values of this type are created by [casting](#type-cast-expressions) pointer values
1472 (pointing to a type for which an implementation of the given trait is in scope)
1473 to pointers to the trait name, used as a type.
1477 # impl Shape for int { }
1478 # let mycircle = 0i;
1479 let myshape: Box<Shape> = box mycircle as Box<Shape>;
1482 The resulting value is a box containing the value that was cast,
1483 along with information that identifies the methods of the implementation that was used.
1484 Values with a trait type can have [methods called](#method-call-expressions) on them,
1485 for any method in the trait,
1486 and can be used to instantiate type parameters that are bounded by the trait.
1488 Trait methods may be static,
1489 which means that they lack a `self` argument.
1490 This means that they can only be called with function call syntax (`f(x)`)
1491 and not method call syntax (`obj.f()`).
1492 The way to refer to the name of a static method is to qualify it with the trait name,
1493 treating the trait name like a module.
1498 fn from_int(n: int) -> Self;
1501 fn from_int(n: int) -> f64 { n as f64 }
1503 let x: f64 = Num::from_int(42);
1506 Traits may inherit from other traits. For example, in
1509 trait Shape { fn area() -> f64; }
1510 trait Circle : Shape { fn radius() -> f64; }
1513 the syntax `Circle : Shape` means that types that implement `Circle` must also have an implementation for `Shape`.
1514 Multiple supertraits are separated by `+`, `trait Circle : Shape + PartialEq { }`.
1515 In an implementation of `Circle` for a given type `T`, methods can refer to `Shape` methods,
1516 since the typechecker checks that any type with an implementation of `Circle` also has an implementation of `Shape`.
1518 In type-parameterized functions,
1519 methods of the supertrait may be called on values of subtrait-bound type parameters.
1520 Referring to the previous example of `trait Circle : Shape`:
1523 # trait Shape { fn area(&self) -> f64; }
1524 # trait Circle : Shape { fn radius(&self) -> f64; }
1525 fn radius_times_area<T: Circle>(c: T) -> f64 {
1526 // `c` is both a Circle and a Shape
1527 c.radius() * c.area()
1531 Likewise, supertrait methods may also be called on trait objects.
1534 # trait Shape { fn area(&self) -> f64; }
1535 # trait Circle : Shape { fn radius(&self) -> f64; }
1536 # impl Shape for int { fn area(&self) -> f64 { 0.0 } }
1537 # impl Circle for int { fn radius(&self) -> f64 { 0.0 } }
1539 let mycircle: Circle = ~mycircle as ~Circle;
1540 let nonsense = mycircle.radius() * mycircle.area();
1545 An _implementation_ is an item that implements a [trait](#traits) for a specific type.
1547 Implementations are defined with the keyword `impl`.
1550 # struct Point {x: f64, y: f64};
1551 # type Surface = int;
1552 # struct BoundingBox {x: f64, y: f64, width: f64, height: f64};
1553 # trait Shape { fn draw(&self, Surface); fn bounding_box(&self) -> BoundingBox; }
1554 # fn do_draw_circle(s: Surface, c: Circle) { }
1560 impl Shape for Circle {
1561 fn draw(&self, s: Surface) { do_draw_circle(s, *self); }
1562 fn bounding_box(&self) -> BoundingBox {
1563 let r = self.radius;
1564 BoundingBox{x: self.center.x - r, y: self.center.y - r,
1565 width: 2.0 * r, height: 2.0 * r}
1570 It is possible to define an implementation without referring to a trait.
1571 The methods in such an implementation can only be used
1572 as direct calls on the values of the type that the implementation targets.
1573 In such an implementation, the trait type and `for` after `impl` are omitted.
1574 Such implementations are limited to nominal types (enums, structs),
1575 and the implementation must appear in the same module or a sub-module as the `self` type.
1577 When a trait _is_ specified in an `impl`,
1578 all methods declared as part of the trait must be implemented,
1579 with matching types and type parameter counts.
1581 An implementation can take type parameters,
1582 which can be different from the type parameters taken by the trait it implements.
1583 Implementation parameters are written after the `impl` keyword.
1587 impl<T> Seq<T> for Vec<T> {
1590 impl Seq<bool> for u32 {
1591 /* Treat the integer as a sequence of bits */
1598 extern_block_item : "extern" '{' extern_block '}' ;
1599 extern_block : [ foreign_fn ] * ;
1602 External blocks form the basis for Rust's foreign function interface.
1603 Declarations in an external block describe symbols
1604 in external, non-Rust libraries.
1606 Functions within external blocks
1607 are declared in the same way as other Rust functions,
1608 with the exception that they may not have a body
1609 and are instead terminated by a semicolon.
1613 use libc::{c_char, FILE};
1616 fn fopen(filename: *const c_char, mode: *const c_char) -> *mut FILE;
1621 Functions within external blocks may be called by Rust code,
1622 just like functions defined in Rust.
1623 The Rust compiler automatically translates
1624 between the Rust ABI and the foreign ABI.
1626 A number of [attributes](#attributes) control the behavior of external
1629 By default external blocks assume that the library they are calling
1630 uses the standard C "cdecl" ABI. Other ABIs may be specified using
1631 an `abi` string, as shown here:
1634 // Interface to the Windows API
1635 extern "stdcall" { }
1638 The `link` attribute allows the name of the library to be specified. When
1639 specified the compiler will attempt to link against the native library of the
1643 #[link(name = "crypto")]
1647 The type of a function declared in an extern block is `extern "abi" fn(A1,
1648 ..., An) -> R`, where `A1...An` are the declared types of its arguments and
1649 `R` is the declared return type.
1651 ## Visibility and Privacy
1653 These two terms are often used interchangeably, and what they are attempting to
1654 convey is the answer to the question "Can this item be used at this location?"
1656 Rust's name resolution operates on a global hierarchy of namespaces. Each level
1657 in the hierarchy can be thought of as some item. The items are one of those
1658 mentioned above, but also include external crates. Declaring or defining a new
1659 module can be thought of as inserting a new tree into the hierarchy at the
1660 location of the definition.
1662 To control whether interfaces can be used across modules, Rust checks each use
1663 of an item to see whether it should be allowed or not. This is where privacy
1664 warnings are generated, or otherwise "you used a private item of another module
1665 and weren't allowed to."
1667 By default, everything in rust is *private*, with one exception. Enum variants
1668 in a `pub` enum are also public by default. You are allowed to alter this
1669 default visibility with the `priv` keyword. When an item is declared as `pub`,
1670 it can be thought of as being accessible to the outside world. For example:
1674 // Declare a private struct
1677 // Declare a public struct with a private field
1682 // Declare a public enum with two public variants
1684 PubliclyAccessibleState,
1685 PubliclyAccessibleState2,
1689 With the notion of an item being either public or private, Rust allows item
1690 accesses in two cases:
1692 1. If an item is public, then it can be used externally through any of its
1694 2. If an item is private, it may be accessed by the current module and its
1697 These two cases are surprisingly powerful for creating module hierarchies
1698 exposing public APIs while hiding internal implementation details. To help
1699 explain, here's a few use cases and what they would entail.
1701 * A library developer needs to expose functionality to crates which link against
1702 their library. As a consequence of the first case, this means that anything
1703 which is usable externally must be `pub` from the root down to the destination
1704 item. Any private item in the chain will disallow external accesses.
1706 * A crate needs a global available "helper module" to itself, but it doesn't
1707 want to expose the helper module as a public API. To accomplish this, the root
1708 of the crate's hierarchy would have a private module which then internally has
1709 a "public api". Because the entire crate is a descendant of the root, then the
1710 entire local crate can access this private module through the second case.
1712 * When writing unit tests for a module, it's often a common idiom to have an
1713 immediate child of the module to-be-tested named `mod test`. This module could
1714 access any items of the parent module through the second case, meaning that
1715 internal implementation details could also be seamlessly tested from the child
1718 In the second case, it mentions that a private item "can be accessed" by the
1719 current module and its descendants, but the exact meaning of accessing an item
1720 depends on what the item is. Accessing a module, for example, would mean looking
1721 inside of it (to import more items). On the other hand, accessing a function
1722 would mean that it is invoked. Additionally, path expressions and import
1723 statements are considered to access an item in the sense that the
1724 import/expression is only valid if the destination is in the current visibility
1727 Here's an example of a program which exemplifies the three cases outlined above.
1730 // This module is private, meaning that no external crate can access this
1731 // module. Because it is private at the root of this current crate, however, any
1732 // module in the crate may access any publicly visible item in this module.
1733 mod crate_helper_module {
1735 // This function can be used by anything in the current crate
1736 pub fn crate_helper() {}
1738 // This function *cannot* be used by anything else in the crate. It is not
1739 // publicly visible outside of the `crate_helper_module`, so only this
1740 // current module and its descendants may access it.
1741 fn implementation_detail() {}
1744 // This function is "public to the root" meaning that it's available to external
1745 // crates linking against this one.
1746 pub fn public_api() {}
1748 // Similarly to 'public_api', this module is public so external crates may look
1751 use crate_helper_module;
1753 pub fn my_method() {
1754 // Any item in the local crate may invoke the helper module's public
1755 // interface through a combination of the two rules above.
1756 crate_helper_module::crate_helper();
1759 // This function is hidden to any module which is not a descendant of
1761 fn my_implementation() {}
1767 fn test_my_implementation() {
1768 // Because this module is a descendant of `submodule`, it's allowed
1769 // to access private items inside of `submodule` without a privacy
1771 super::my_implementation();
1779 For a rust program to pass the privacy checking pass, all paths must be valid
1780 accesses given the two rules above. This includes all use statements,
1781 expressions, types, etc.
1783 ### Re-exporting and Visibility
1785 Rust allows publicly re-exporting items through a `pub use` directive. Because
1786 this is a public directive, this allows the item to be used in the current
1787 module through the rules above. It essentially allows public access into the
1788 re-exported item. For example, this program is valid:
1791 pub use api = self::implementation;
1793 mod implementation {
1800 This means that any external crate referencing `implementation::f` would receive
1801 a privacy violation, while the path `api::f` would be allowed.
1803 When re-exporting a private item, it can be thought of as allowing the "privacy
1804 chain" being short-circuited through the reexport instead of passing through the
1805 namespace hierarchy as it normally would.
1807 ### Glob imports and Visibility
1809 Currently glob imports are considered an "experimental" language feature. For
1810 sanity purpose along with helping the implementation, glob imports will only
1811 import public items from their destination, not private items.
1813 > **Note:** This is subject to change, glob exports may be removed entirely or
1814 > they could possibly import private items for a privacy error to later be
1815 > issued if the item is used.
1820 attribute : '#' '!' ? '[' meta_item ']' ;
1821 meta_item : ident [ '=' literal
1822 | '(' meta_seq ')' ] ? ;
1823 meta_seq : meta_item [ ',' meta_seq ] ? ;
1826 Static entities in Rust — crates, modules and items — may have _attributes_
1827 applied to them. Attributes in Rust are modeled on Attributes in ECMA-335,
1828 with the syntax coming from ECMA-334 (C#). An attribute is a general,
1829 free-form metadatum that is interpreted according to name, convention, and
1830 language and compiler version. Attributes may appear as any of:
1832 * A single identifier, the attribute name
1833 * An identifier followed by the equals sign '=' and a literal, providing a
1835 * An identifier followed by a parenthesized list of sub-attribute arguments
1837 Attributes with a bang ("!") after the hash ("#") apply to the item that the
1838 attribute is declared within. Attributes that do not have a bang after the
1839 hash apply to the item that follows the attribute.
1841 An example of attributes:
1844 // General metadata applied to the enclosing module or crate.
1847 // A function marked as a unit test
1853 // A conditionally-compiled module
1854 #[cfg(target_os="linux")]
1859 // A lint attribute used to suppress a warning/error
1860 #[allow(non_camel_case_types)]
1864 > **Note:** At some point in the future, the compiler will distinguish between
1865 > language-reserved and user-available attributes. Until then, there is
1866 > effectively no difference between an attribute handled by a loadable syntax
1867 > extension and the compiler.
1869 ### Crate-only attributes
1871 - `crate_id` - specify the this crate's crate ID.
1872 - `crate_type` - see [linkage](#linkage).
1873 - `feature` - see [compiler features](#compiler-features).
1874 - `no_main` - disable emitting the `main` symbol. Useful when some other
1875 object being linked to defines `main`.
1876 - `no_start` - disable linking to the `native` crate, which specifies the
1877 "start" language item.
1878 - `no_std` - disable linking to the `std` crate.
1879 - `no_builtins` - disable optimizing certain code patterns to invocations of
1880 library functions that are assumed to exist
1882 ### Module-only attributes
1884 - `macro_escape` - macros defined in this module will be visible in the
1885 module's parent, after this module has been included.
1886 - `no_implicit_prelude` - disable injecting `use std::prelude::*` in this
1888 - `path` - specifies the file to load the module from. `#[path="foo.rs"] mod
1889 bar;` is equivalent to `mod bar { /* contents of foo.rs */ }`. The path is
1890 taken relative to the directory that the current module is in.
1892 ### Function-only attributes
1894 - `plugin_registrar` - mark this function as the registration point for
1895 compiler plugins, such as loadable syntax extensions.
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 - `start` - indicates that this function should be used as the entry point,
1899 overriding the "start" language item. See the "start" [language
1900 item](#language-items) for more details.
1902 ### Static-only attributes
1904 - `thread_local` - on a `static mut`, this signals that the value of this
1905 static may change depending on the current thread. The exact consequences of
1906 this are implementation-defined.
1910 On an `extern` block, the following attributes are interpreted:
1912 - `link_args` - specify arguments to the linker, rather than just the library
1913 name and type. This is feature gated and the exact behavior is
1914 implementation-defined (due to variety of linker invocation syntax).
1915 - `link` - indicate that a native library should be linked to for the
1916 declarations in this block to be linked correctly. See [external
1917 blocks](#external-blocks)
1919 On declarations inside an `extern` block, the following attributes are
1922 - `link_name` - the name of the symbol that this function or static should be
1924 - `linkage` - on a static, this specifies the [linkage
1925 type](http://llvm.org/docs/LangRef.html#linkage-types).
1927 ### Miscellaneous attributes
1929 - `link_section` - on statics and functions, this specifies the section of the
1930 object file that this item's contents will be placed into.
1931 - `macro_export` - export a macro for cross-crate usage.
1932 - `no_mangle` - on any item, do not apply the standard name mangling. Set the
1933 symbol for this item to its identifier.
1934 - `packed` - on structs or enums, eliminate any padding that would be used to
1936 - `repr` - on C-like enums, this sets the underlying type used for
1937 representation. Useful for FFI. Takes one argument, which is the primitive
1938 type this enum should be represented for, or `C`, which specifies that it
1939 should be the default `enum` size of the C ABI for that platform. Note that
1940 enum representation in C is undefined, and this may be incorrect when the C
1941 code is compiled with certain flags.
1942 - `simd` - on certain tuple structs, derive the arithmetic operators, which
1943 lower to the target's SIMD instructions, if any; the `simd` feature gate
1944 is necessary to use this attribute.
1945 - `static_assert` - on statics whose type is `bool`, terminates compilation
1946 with an error if it is not initialized to `true`.
1947 - `unsafe_destructor` - allow implementations of the "drop" language item
1948 where the type it is implemented for does not implement the "send" language
1949 item; the `unsafe_destructor` feature gate is needed to use this attribute
1950 - `unsafe_no_drop_flag` - on structs, remove the flag that prevents
1951 destructors from being run twice. Destructors might be run multiple times on
1952 the same object with this attribute.
1954 ### Conditional compilation
1956 Sometimes one wants to have different compiler outputs from the same code,
1957 depending on build target, such as targeted operating system, or to enable
1960 There are two kinds of configuration options, one that is either defined or not
1961 (`#[cfg(foo)]`), and the other that contains a string that can be checked
1962 against (`#[cfg(bar = "baz")]` (currently only compiler-defined configuration
1963 options can have the latter form).
1966 // The function is only included in the build when compiling for OSX
1967 #[cfg(target_os = "macos")]
1972 // This function is only included when either foo or bar is defined
1975 fn needs_foo_or_bar() {
1979 // This function is only included when compiling for a unixish OS with a 32-bit
1981 #[cfg(unix, target_word_size = "32")]
1982 fn on_32bit_unix() {
1987 This illustrates some conditional compilation can be achieved using the
1988 `#[cfg(...)]` attribute. Note that `#[cfg(foo, bar)]` is a condition that needs
1989 both `foo` and `bar` to be defined while `#[cfg(foo)] #[cfg(bar)]` only needs
1990 one of `foo` and `bar` to be defined (this resembles in the disjunctive normal
1991 form). Additionally, one can reverse a condition by enclosing it in a
1992 `not(...)`, like e. g. `#[cfg(not(target_os = "win32"))]`.
1994 The following configurations must be defined by the implementation:
1996 * `target_arch = "..."`. Target CPU architecture, such as `"x86"`, `"x86_64"`
1997 `"mips"`, or `"arm"`.
1998 * `target_endian = "..."`. Endianness of the target CPU, either `"little"` or
2000 * `target_family = "..."`. Operating system family of the target, e. g.
2001 `"unix"` or `"windows"`. The value of this configuration option is defined as
2002 a configuration itself, like `unix` or `windows`.
2003 * `target_os = "..."`. Operating system of the target, examples include
2004 `"win32"`, `"macos"`, `"linux"`, `"android"` or `"freebsd"`.
2005 * `target_word_size = "..."`. Target word size in bits. This is set to `"32"`
2006 for targets with 32-bit pointers, and likewise set to `"64"` for 64-bit
2008 * `unix`. See `target_family`.
2009 * `windows`. See `target_family`.
2011 ### Lint check attributes
2013 A lint check names a potentially undesirable coding pattern, such as
2014 unreachable code or omitted documentation, for the static entity to
2015 which the attribute applies.
2017 For any lint check `C`:
2019 * `warn(C)` warns about violations of `C` but continues compilation,
2020 * `deny(C)` signals an error after encountering a violation of `C`,
2021 * `allow(C)` overrides the check for `C` so that violations will go
2023 * `forbid(C)` is the same as `deny(C)`, but also forbids changing the lint
2026 The lint checks supported by the compiler can be found via `rustc -W help`,
2027 along with their default settings.
2031 // Missing documentation is ignored here
2032 #[allow(missing_doc)]
2033 pub fn undocumented_one() -> int { 1 }
2035 // Missing documentation signals a warning here
2036 #[warn(missing_doc)]
2037 pub fn undocumented_too() -> int { 2 }
2039 // Missing documentation signals an error here
2040 #[deny(missing_doc)]
2041 pub fn undocumented_end() -> int { 3 }
2045 This example shows how one can use `allow` and `warn` to toggle
2046 a particular check on and off.
2049 #[warn(missing_doc)]
2051 #[allow(missing_doc)]
2053 // Missing documentation is ignored here
2054 pub fn undocumented_one() -> int { 1 }
2056 // Missing documentation signals a warning here,
2057 // despite the allow above.
2058 #[warn(missing_doc)]
2059 pub fn undocumented_two() -> int { 2 }
2062 // Missing documentation signals a warning here
2063 pub fn undocumented_too() -> int { 3 }
2067 This example shows how one can use `forbid` to disallow uses
2068 of `allow` for that lint check.
2071 #[forbid(missing_doc)]
2073 // Attempting to toggle warning signals an error here
2074 #[allow(missing_doc)]
2076 pub fn undocumented_too() -> int { 2 }
2082 Some primitive Rust operations are defined in Rust code, rather than being
2083 implemented directly in C or assembly language. The definitions of these
2084 operations have to be easy for the compiler to find. The `lang` attribute
2085 makes it possible to declare these operations. For example, the `str` module
2086 in the Rust standard library defines the string equality function:
2090 pub fn eq_slice(a: &str, b: &str) -> bool {
2095 The name `str_eq` has a special meaning to the Rust compiler,
2096 and the presence of this definition means that it will use this definition
2097 when generating calls to the string equality function.
2099 A complete list of the built-in language items follows:
2101 #### Built-in Traits
2104 : Able to be sent across task boundaries.
2106 : Has a size known at compile time.
2108 : Types that do not move ownership when used by-value.
2110 : Able to be safely shared between tasks when aliased.
2116 These language items are traits:
2119 : Elements can be added (for example, integers and floats).
2121 : Elements can be subtracted.
2123 : Elements can be multiplied.
2125 : Elements have a division operation.
2127 : Elements have a remainder operation.
2129 : Elements can be negated arithmetically.
2131 : Elements can be negated logically.
2133 : Elements have an exclusive-or operation.
2135 : Elements have a bitwise `and` operation.
2137 : Elements have a bitwise `or` operation.
2139 : Elements have a left shift operation.
2141 : Elements have a right shift operation.
2143 : Elements can be indexed.
2145 : Elements can be compared for equality.
2147 : Elements have a partial ordering.
2149 : `*` can be applied, yielding a reference to another type
2151 : `*` can be applied, yielding a mutable reference to another type
2154 These are functions:
2157 : Compare two strings (`&str`) for equality.
2159 : Return a new unique string
2160 containing a copy of the contents of a unique string.
2165 : A type whose contents can be mutated through an immutable reference
2167 : The type returned by the `type_id` intrinsic.
2171 These types help drive the compiler's analysis
2174 : The type parameter should be considered covariant
2175 * `contravariant_type`
2176 : The type parameter should be considered contravariant
2178 : The type parameter should be considered invariant
2179 * `covariant_lifetime`
2180 : The lifetime parameter should be considered covariant
2181 * `contravariant_lifetime`
2182 : The lifetime parameter should be considered contravariant
2183 * `invariant_lifetime`
2184 : The lifetime parameter should be considered invariant
2186 : This type does not implement "send", even if eligible
2188 : This type does not implement "copy", even if eligible
2190 : This type does not implement "share", even if eligible
2192 : This type implements "managed"
2195 : Abort the program with an error.
2196 * `fail_bounds_check`
2197 : Abort the program with a bounds check error.
2199 : Allocate memory on the exchange heap.
2201 : Free memory that was allocated on the exchange heap.
2203 : Allocate memory on the managed heap.
2205 : Free memory that was allocated on the managed heap.
2207 > **Note:** This list is likely to become out of date. We should auto-generate it
2208 > from `librustc/middle/lang_items.rs`.
2210 ### Inline attributes
2212 The inline attribute is used to suggest to the compiler to perform an inline
2213 expansion and place a copy of the function or static in the caller rather than
2214 generating code to call the function or access the static where it is defined.
2216 The compiler automatically inlines functions based on internal heuristics.
2217 Incorrectly inlining functions can actually making the program slower, so it
2218 should be used with care.
2220 Immutable statics are always considered inlineable
2221 unless marked with `#[inline(never)]`.
2223 whether two different inlineable statics
2224 have the same memory address.
2226 the compiler is free
2227 to collapse duplicate inlineable statics together.
2229 `#[inline]` and `#[inline(always)]` always causes the function to be serialized
2230 into crate metadata to allow cross-crate inlining.
2232 There are three different types of inline attributes:
2234 * `#[inline]` hints the compiler to perform an inline expansion.
2235 * `#[inline(always)]` asks the compiler to always perform an inline expansion.
2236 * `#[inline(never)]` asks the compiler to never perform an inline expansion.
2240 The `deriving` attribute allows certain traits to be automatically
2241 implemented for data structures. For example, the following will
2242 create an `impl` for the `PartialEq` and `Clone` traits for `Foo`, the type
2243 parameter `T` will be given the `PartialEq` or `Clone` constraints for the
2247 #[deriving(PartialEq, Clone)]
2254 The generated `impl` for `PartialEq` is equivalent to
2257 # struct Foo<T> { a: int, b: T }
2258 impl<T: PartialEq> PartialEq for Foo<T> {
2259 fn eq(&self, other: &Foo<T>) -> bool {
2260 self.a == other.a && self.b == other.b
2263 fn ne(&self, other: &Foo<T>) -> bool {
2264 self.a != other.a || self.b != other.b
2269 Supported traits for `deriving` are:
2271 * Comparison traits: `PartialEq`, `Eq`, `PartialOrd`, `Ord`.
2272 * Serialization: `Encodable`, `Decodable`. These require `serialize`.
2273 * `Clone`, to create `T` from `&T` via a copy.
2274 * `Hash`, to iterate over the bytes in a data type.
2275 * `Rand`, to create a random instance of a data type.
2276 * `Default`, to create an empty instance of a data type.
2277 * `Zero`, to create a zero instance of a numeric data type.
2278 * `FromPrimitive`, to create an instance from a numeric primitive.
2279 * `Show`, to format a value using the `{}` formatter.
2283 One can indicate the stability of an API using the following attributes:
2285 * `deprecated`: This item should no longer be used, e.g. it has been
2286 replaced. No guarantee of backwards-compatibility.
2287 * `experimental`: This item was only recently introduced or is
2288 otherwise in a state of flux. It may change significantly, or even
2289 be removed. No guarantee of backwards-compatibility.
2290 * `unstable`: This item is still under development, but requires more
2291 testing to be considered stable. No guarantee of backwards-compatibility.
2292 * `stable`: This item is considered stable, and will not change
2293 significantly. Guarantee of backwards-compatibility.
2294 * `frozen`: This item is very stable, and is unlikely to
2295 change. Guarantee of backwards-compatibility.
2296 * `locked`: This item will never change unless a serious bug is
2297 found. Guarantee of backwards-compatibility.
2299 These levels are directly inspired by
2300 [Node.js' "stability index"](http://nodejs.org/api/documentation.html).
2302 Stability levels are inherited, so an items's stability attribute is the
2303 default stability for everything nested underneath it.
2305 There are lints for disallowing items marked with certain levels: `deprecated`,
2306 `experimental` and `unstable`. For now, only `deprecated` warns by default, but
2307 this will change once the standard library has been stabilized.
2308 Stability levels are meant to be promises at the crate
2309 level, so these lints only apply when referencing
2310 items from an _external_ crate, not to items defined within the
2311 current crate. Items with no stability level are considered
2312 to be unstable for the purposes of the lint. One can give an optional
2313 string that will be displayed when the lint flags the use of an item.
2315 For example, if we define one crate called `stability_levels`:
2318 #[deprecated="replaced by `best`"]
2320 // delete everything
2324 // delete fewer things
2333 then the lints will work as follows for a client crate:
2337 extern crate stability_levels;
2338 use stability_levels::{bad, better, best};
2341 bad(); // "warning: use of deprecated item: replaced by `best`"
2343 better(); // "warning: use of unmarked item"
2345 best(); // no warning
2349 > **Note:** Currently these are only checked when applied to
2350 > individual functions, structs, methods and enum variants, *not* to
2351 > entire modules, traits, impls or enums themselves.
2353 ### Compiler Features
2355 Certain aspects of Rust may be implemented in the compiler, but they're not
2356 necessarily ready for every-day use. These features are often of "prototype
2357 quality" or "almost production ready", but may not be stable enough to be
2358 considered a full-fledged language feature.
2360 For this reason, Rust recognizes a special crate-level attribute of the form:
2363 #![feature(feature1, feature2, feature3)]
2366 This directive informs the compiler that the feature list: `feature1`,
2367 `feature2`, and `feature3` should all be enabled. This is only recognized at a
2368 crate-level, not at a module-level. Without this directive, all features are
2369 considered off, and using the features will result in a compiler error.
2371 The currently implemented features of the reference compiler are:
2373 * `macro_rules` - The definition of new macros. This does not encompass
2374 macro-invocation, that is always enabled by default, this only
2375 covers the definition of new macros. There are currently
2376 various problems with invoking macros, how they interact with
2377 their environment, and possibly how they are used outside of
2378 location in which they are defined. Macro definitions are
2379 likely to change slightly in the future, so they are currently
2380 hidden behind this feature.
2382 * `globs` - Importing everything in a module through `*`. This is currently a
2383 large source of bugs in name resolution for Rust, and it's not clear
2384 whether this will continue as a feature or not. For these reasons,
2385 the glob import statement has been hidden behind this feature flag.
2387 * `struct_variant` - Structural enum variants (those with named fields). It is
2388 currently unknown whether this style of enum variant is as
2389 fully supported as the tuple-forms, and it's not certain
2390 that this style of variant should remain in the language.
2391 For now this style of variant is hidden behind a feature
2394 * `once_fns` - Onceness guarantees a closure is only executed once. Defining a
2395 closure as `once` is unlikely to be supported going forward. So
2396 they are hidden behind this feature until they are to be removed.
2398 * `managed_boxes` - Usage of `@` pointers is gated due to many
2399 planned changes to this feature. In the past, this has meant
2400 "a GC pointer", but the current implementation uses
2401 reference counting and will likely change drastically over
2402 time. Additionally, the `@` syntax will no longer be used to
2405 * `asm` - The `asm!` macro provides a means for inline assembly. This is often
2406 useful, but the exact syntax for this feature along with its semantics
2407 are likely to change, so this macro usage must be opted into.
2409 * `non_ascii_idents` - The compiler supports the use of non-ascii identifiers,
2410 but the implementation is a little rough around the
2411 edges, so this can be seen as an experimental feature for
2412 now until the specification of identifiers is fully
2415 * `thread_local` - The usage of the `#[thread_local]` attribute is experimental
2416 and should be seen as unstable. This attribute is used to
2417 declare a `static` as being unique per-thread leveraging
2418 LLVM's implementation which works in concert with the kernel
2419 loader and dynamic linker. This is not necessarily available
2420 on all platforms, and usage of it is discouraged (rust
2421 focuses more on task-local data instead of thread-local
2424 * `link_args` - This attribute is used to specify custom flags to the linker,
2425 but usage is strongly discouraged. The compiler's usage of the
2426 system linker is not guaranteed to continue in the future, and
2427 if the system linker is not used then specifying custom flags
2428 doesn't have much meaning.
2430 If a feature is promoted to a language feature, then all existing programs will
2431 start to receive compilation warnings about #[feature] directives which enabled
2432 the new feature (because the directive is no longer necessary). However, if
2433 a feature is decided to be removed from the language, errors will be issued (if
2434 there isn't a parser error first). The directive in this case is no longer
2435 necessary, and it's likely that existing code will break if the feature isn't
2438 If a unknown feature is found in a directive, it results in a compiler error. An
2439 unknown feature is one which has never been recognized by the compiler.
2441 # Statements and expressions
2443 Rust is _primarily_ an expression language. This means that most forms of
2444 value-producing or effect-causing evaluation are directed by the uniform
2445 syntax category of _expressions_. Each kind of expression can typically _nest_
2446 within each other kind of expression, and rules for evaluation of expressions
2447 involve specifying both the value produced by the expression and the order in
2448 which its sub-expressions are themselves evaluated.
2450 In contrast, statements in Rust serve _mostly_ to contain and explicitly
2451 sequence expression evaluation.
2455 A _statement_ is a component of a block, which is in turn a component of an
2456 outer [expression](#expressions) or [function](#functions).
2458 Rust has two kinds of statement:
2459 [declaration statements](#declaration-statements) and
2460 [expression statements](#expression-statements).
2462 ### Declaration statements
2464 A _declaration statement_ is one that introduces one or more *names* into the enclosing statement block.
2465 The declared names may denote new slots or new items.
2467 #### Item declarations
2469 An _item declaration statement_ has a syntactic form identical to an
2470 [item](#items) declaration within a module. Declaring an item — a function,
2471 enumeration, structure, type, static, trait, implementation or module — locally
2472 within a statement block is simply a way of restricting its scope to a narrow
2473 region containing all of its uses; it is otherwise identical in meaning to
2474 declaring the item outside the statement block.
2476 Note: there is no implicit capture of the function's dynamic environment when
2477 declaring a function-local item.
2479 #### Slot declarations
2482 let_decl : "let" pat [':' type ] ? [ init ] ? ';' ;
2483 init : [ '=' ] expr ;
2486 A _slot declaration_ introduces a new set of slots, given by a pattern.
2487 The pattern may be followed by a type annotation, and/or an initializer expression.
2488 When no type annotation is given, the compiler will infer the type,
2489 or signal an error if insufficient type information is available for definite inference.
2490 Any slots introduced by a slot declaration are visible from the point of declaration until the end of the enclosing block scope.
2492 ### Expression statements
2494 An _expression statement_ is one that evaluates an [expression](#expressions)
2495 and ignores its result.
2496 The type of an expression statement `e;` is always `()`, regardless of the type of `e`.
2497 As a rule, an expression statement's purpose is to trigger the effects of evaluating its expression.
2501 An expression may have two roles: it always produces a *value*, and it may have *effects*
2502 (otherwise known as "side effects").
2503 An expression *evaluates to* a value, and has effects during *evaluation*.
2504 Many expressions contain sub-expressions (operands).
2505 The meaning of each kind of expression dictates several things:
2506 * Whether or not to evaluate the sub-expressions when evaluating the expression
2507 * The order in which to evaluate the sub-expressions
2508 * How to combine the sub-expressions' values to obtain the value of the expression.
2510 In this way, the structure of expressions dictates the structure of execution.
2511 Blocks are just another kind of expression,
2512 so blocks, statements, expressions, and blocks again can recursively nest inside each other
2513 to an arbitrary depth.
2515 #### Lvalues, rvalues and temporaries
2517 Expressions are divided into two main categories: _lvalues_ and _rvalues_.
2518 Likewise within each expression, sub-expressions may occur in _lvalue context_ or _rvalue context_.
2519 The evaluation of an expression depends both on its own category and the context it occurs within.
2521 An lvalue is an expression that represents a memory location. These
2522 expressions are [paths](#path-expressions) (which refer to local
2523 variables, function and method arguments, or static variables),
2524 dereferences (`*expr`), [indexing expressions](#index-expressions)
2525 (`expr[expr]`), and [field references](#field-expressions) (`expr.f`).
2526 All other expressions are rvalues.
2528 The left operand of an [assignment](#assignment-expressions) or
2529 [compound-assignment](#compound-assignment-expressions) expression is an lvalue context,
2530 as is the single operand of a unary [borrow](#unary-operator-expressions).
2531 All other expression contexts are rvalue contexts.
2533 When an lvalue is evaluated in an _lvalue context_, it denotes a memory location;
2534 when evaluated in an _rvalue context_, it denotes the value held _in_ that memory location.
2536 When an rvalue is used in lvalue context, a temporary un-named lvalue is created and used instead.
2537 A temporary's lifetime equals the largest lifetime of any reference that points to it.
2539 #### Moved and copied types
2541 When a [local variable](#memory-slots) is used
2542 as an [rvalue](#lvalues,-rvalues-and-temporaries)
2543 the variable will either be moved or copied, depending on its type.
2544 For types that contain [owning pointers](#pointer-types)
2545 or values that implement the special trait `Drop`,
2546 the variable is moved.
2547 All other types are copied.
2549 ### Literal expressions
2551 A _literal expression_ consists of one of the [literal](#literals)
2552 forms described earlier. It directly describes a number, character,
2553 string, boolean value, or the unit value.
2557 "hello"; // string type
2558 '5'; // character type
2562 ### Path expressions
2564 A [path](#paths) used as an expression context denotes either a local variable or an item.
2565 Path expressions are [lvalues](#lvalues,-rvalues-and-temporaries).
2567 ### Tuple expressions
2569 Tuples are written by enclosing one or more comma-separated
2570 expressions in parentheses. They are used to create [tuple-typed](#tuple-types)
2579 ### Structure expressions
2582 struct_expr : expr_path '{' ident ':' expr
2583 [ ',' ident ':' expr ] *
2586 [ ',' expr ] * ')' |
2590 There are several forms of structure expressions.
2591 A _structure expression_ consists of the [path](#paths) of a [structure item](#structures),
2592 followed by a brace-enclosed list of one or more comma-separated name-value pairs,
2593 providing the field values of a new instance of the structure.
2594 A field name can be any identifier, and is separated from its value expression by a colon.
2595 The location denoted by a structure field is mutable if and only if the enclosing structure is mutable.
2597 A _tuple structure expression_ consists of the [path](#paths) of a [structure item](#structures),
2598 followed by a parenthesized list of one or more comma-separated expressions
2599 (in other words, the path of a structure item followed by a tuple expression).
2600 The structure item must be a tuple structure item.
2602 A _unit-like structure expression_ consists only of the [path](#paths) of a [structure item](#structures).
2604 The following are examples of structure expressions:
2607 # struct Point { x: f64, y: f64 }
2608 # struct TuplePoint(f64, f64);
2609 # mod game { pub struct User<'a> { pub name: &'a str, pub age: uint, pub score: uint } }
2610 # struct Cookie; fn some_fn<T>(t: T) {}
2611 Point {x: 10.0, y: 20.0};
2612 TuplePoint(10.0, 20.0);
2613 let u = game::User {name: "Joe", age: 35, score: 100_000};
2614 some_fn::<Cookie>(Cookie);
2617 A structure expression forms a new value of the named structure type.
2618 Note that for a given *unit-like* structure type, this will always be the same value.
2620 A structure expression can terminate with the syntax `..` followed by an expression to denote a functional update.
2621 The expression following `..` (the base) must have the same structure type as the new structure type being formed.
2622 The entire expression denotes the result of constructing a new structure
2623 (with the same type as the base expression)
2624 with the given values for the fields that were explicitly specified
2625 and the values in the base expression for all other fields.
2628 # struct Point3d { x: int, y: int, z: int }
2629 let base = Point3d {x: 1, y: 2, z: 3};
2630 Point3d {y: 0, z: 10, .. base};
2633 ### Block expressions
2636 block_expr : '{' [ view_item ] *
2637 [ stmt ';' | item ] *
2641 A _block expression_ is similar to a module in terms of the declarations that
2642 are possible. Each block conceptually introduces a new namespace scope. View
2643 items can bring new names into scopes and declared items are in scope for only
2646 A block will execute each statement sequentially, and then execute the
2647 expression (if given). If the final expression is omitted, the type and return
2648 value of the block are `()`, but if it is provided, the type and return value
2649 of the block are that of the expression itself.
2651 ### Method-call expressions
2654 method_call_expr : expr '.' ident paren_expr_list ;
2657 A _method call_ consists of an expression followed by a single dot, an identifier, and a parenthesized expression-list.
2658 Method calls are resolved to methods on specific traits,
2659 either statically dispatching to a method if the exact `self`-type of the left-hand-side is known,
2660 or dynamically dispatching if the left-hand-side expression is an indirect [object type](#object-types).
2662 ### Field expressions
2665 field_expr : expr '.' ident ;
2668 A _field expression_ consists of an expression followed by a single dot and an identifier,
2669 when not immediately followed by a parenthesized expression-list (the latter is a [method call expression](#method-call-expressions)).
2670 A field expression denotes a field of a [structure](#structure-types).
2672 ~~~~ {.ignore .field}
2675 (Struct {a: 10, b: 20}).a;
2678 A field access is an [lvalue](#lvalues,-rvalues-and-temporaries) referring to the value of that field.
2679 When the type providing the field inherits mutabilty, it can be [assigned](#assignment-expressions) to.
2681 Also, if the type of the expression to the left of the dot is a pointer,
2682 it is automatically dereferenced to make the field access possible.
2684 ### Vector expressions
2687 vec_expr : '[' "mut" ? vec_elems? ']' ;
2689 vec_elems : [expr [',' expr]*] | [expr ',' ".." expr] ;
2692 A [_vector_](#vector-types) _expression_ is written by enclosing zero or
2693 more comma-separated expressions of uniform type in square brackets.
2695 In the `[expr ',' ".." expr]` form, the expression after the `".."`
2696 must be a constant expression that can be evaluated at compile time, such
2697 as a [literal](#literals) or a [static item](#static-items).
2701 ["a", "b", "c", "d"];
2702 [0i, ..128]; // vector with 128 zeros
2703 [0u8, 0u8, 0u8, 0u8];
2706 ### Index expressions
2709 idx_expr : expr '[' expr ']' ;
2712 [Vector](#vector-types)-typed expressions can be indexed by writing a
2713 square-bracket-enclosed expression (the index) after them. When the
2714 vector is mutable, the resulting [lvalue](#lvalues,-rvalues-and-temporaries) can be assigned to.
2716 Indices are zero-based, and may be of any integral type. Vector access
2717 is bounds-checked at run-time. When the check fails, it will put the
2718 task in a _failing state_.
2722 # task::spawn(proc() {
2725 (["a", "b"])[10]; // fails
2730 ### Unary operator expressions
2732 Rust defines six symbolic unary operators.
2733 They are all written as prefix operators,
2734 before the expression they apply to.
2737 : Negation. May only be applied to numeric types.
2739 : Dereference. When applied to a [pointer](#pointer-types) it denotes the pointed-to location.
2740 For pointers to mutable locations, the resulting [lvalue](#lvalues,-rvalues-and-temporaries) can be assigned to.
2741 On non-pointer types, it calls the `deref` method of the `std::ops::Deref` trait, or the
2742 `deref_mut` method of the `std::ops::DerefMut` trait (if implemented by the type and required
2743 for an outer expression that will or could mutate the dereference), and produces the
2744 result of dereferencing the `&` or `&mut` borrowed pointer returned from the overload method.
2747 : Logical negation. On the boolean type, this flips between `true` and
2748 `false`. On integer types, this inverts the individual bits in the
2749 two's complement representation of the value.
2751 : [Boxing](#pointer-types) operators. Allocate a box to hold the value they are applied to,
2752 and store the value in it. `box` creates an owned box.
2754 : Borrow operator. Returns a reference, pointing to its operand.
2755 The operand of a borrow is statically proven to outlive the resulting pointer.
2756 If the borrow-checker cannot prove this, it is a compilation error.
2758 ### Binary operator expressions
2761 binop_expr : expr binop expr ;
2764 Binary operators expressions are given in terms of
2765 [operator precedence](#operator-precedence).
2767 #### Arithmetic operators
2769 Binary arithmetic expressions are syntactic sugar for calls to built-in traits,
2770 defined in the `std::ops` module of the `std` library.
2771 This means that arithmetic operators can be overridden for user-defined types.
2772 The default meaning of the operators on standard types is given here.
2775 : Addition and vector/string concatenation.
2776 Calls the `add` method on the `std::ops::Add` trait.
2779 Calls the `sub` method on the `std::ops::Sub` trait.
2782 Calls the `mul` method on the `std::ops::Mul` trait.
2785 Calls the `div` method on the `std::ops::Div` trait.
2788 Calls the `rem` method on the `std::ops::Rem` trait.
2790 #### Bitwise operators
2792 Like the [arithmetic operators](#arithmetic-operators), bitwise operators
2793 are syntactic sugar for calls to methods of built-in traits.
2794 This means that bitwise operators can be overridden for user-defined types.
2795 The default meaning of the operators on standard types is given here.
2799 Calls the `bitand` method of the `std::ops::BitAnd` trait.
2802 Calls the `bitor` method of the `std::ops::BitOr` trait.
2805 Calls the `bitxor` method of the `std::ops::BitXor` trait.
2807 : Logical left shift.
2808 Calls the `shl` method of the `std::ops::Shl` trait.
2810 : Logical right shift.
2811 Calls the `shr` method of the `std::ops::Shr` trait.
2813 #### Lazy boolean operators
2815 The operators `||` and `&&` may be applied to operands of boolean type.
2816 The `||` operator denotes logical 'or', and the `&&` operator denotes logical 'and'.
2817 They differ from `|` and `&` in that the right-hand operand is only evaluated
2818 when the left-hand operand does not already determine the result of the expression.
2819 That is, `||` only evaluates its right-hand operand
2820 when the left-hand operand evaluates to `false`, and `&&` only when it evaluates to `true`.
2822 #### Comparison operators
2824 Comparison operators are, like the [arithmetic operators](#arithmetic-operators),
2825 and [bitwise operators](#bitwise-operators),
2826 syntactic sugar for calls to built-in traits.
2827 This means that comparison operators can be overridden for user-defined types.
2828 The default meaning of the operators on standard types is given here.
2832 Calls the `eq` method on the `std::cmp::PartialEq` trait.
2835 Calls the `ne` method on the `std::cmp::PartialEq` trait.
2838 Calls the `lt` method on the `std::cmp::PartialOrd` trait.
2841 Calls the `gt` method on the `std::cmp::PartialOrd` trait.
2843 : Less than or equal.
2844 Calls the `le` method on the `std::cmp::PartialOrd` trait.
2846 : Greater than or equal.
2847 Calls the `ge` method on the `std::cmp::PartialOrd` trait.
2849 #### Type cast expressions
2851 A type cast expression is denoted with the binary operator `as`.
2853 Executing an `as` expression casts the value on the left-hand side to the type
2854 on the right-hand side.
2856 A numeric value can be cast to any numeric type.
2857 A raw pointer value can be cast to or from any integral type or raw pointer type.
2858 Any other cast is unsupported and will fail to compile.
2860 An example of an `as` expression:
2863 # fn sum(v: &[f64]) -> f64 { 0.0 }
2864 # fn len(v: &[f64]) -> int { 0 }
2866 fn avg(v: &[f64]) -> f64 {
2867 let sum: f64 = sum(v);
2868 let sz: f64 = len(v) as f64;
2873 #### Assignment expressions
2875 An _assignment expression_ consists of an [lvalue](#lvalues,-rvalues-and-temporaries) expression followed by an
2876 equals sign (`=`) and an [rvalue](#lvalues,-rvalues-and-temporaries) expression.
2878 Evaluating an assignment expression [either copies or moves](#moved-and-copied-types) its right-hand operand to its left-hand operand.
2887 #### Compound assignment expressions
2889 The `+`, `-`, `*`, `/`, `%`, `&`, `|`, `^`, `<<`, and `>>`
2890 operators may be composed with the `=` operator. The expression `lval
2891 OP= val` is equivalent to `lval = lval OP val`. For example, `x = x +
2892 1` may be written as `x += 1`.
2894 Any such expression always has the [`unit`](#primitive-types) type.
2896 #### Operator precedence
2898 The precedence of Rust binary operators is ordered as follows, going
2899 from strong to weak:
2901 ~~~~ {.text .precedence}
2916 Operators at the same precedence level are evaluated left-to-right. [Unary operators](#unary-operator-expressions)
2917 have the same precedence level and it is stronger than any of the binary operators'.
2919 ### Grouped expressions
2921 An expression enclosed in parentheses evaluates to the result of the enclosed
2922 expression. Parentheses can be used to explicitly specify evaluation order
2923 within an expression.
2926 paren_expr : '(' expr ')' ;
2929 An example of a parenthesized expression:
2932 let x: int = (2 + 3) * 4;
2936 ### Call expressions
2939 expr_list : [ expr [ ',' expr ]* ] ? ;
2940 paren_expr_list : '(' expr_list ')' ;
2941 call_expr : expr paren_expr_list ;
2944 A _call expression_ invokes a function, providing zero or more input slots and
2945 an optional reference slot to serve as the function's output, bound to the
2946 `lval` on the right hand side of the call. If the function eventually returns,
2947 then the expression completes.
2949 Some examples of call expressions:
2952 # use std::from_str::FromStr;
2953 # fn add(x: int, y: int) -> int { 0 }
2955 let x: int = add(1, 2);
2956 let pi: Option<f32> = FromStr::from_str("3.14");
2959 ### Lambda expressions
2962 ident_list : [ ident [ ',' ident ]* ] ? ;
2963 lambda_expr : '|' ident_list '|' expr ;
2966 A _lambda expression_ (sometimes called an "anonymous function expression") defines a function and denotes it as a value,
2967 in a single expression.
2968 A lambda expression is a pipe-symbol-delimited (`|`) list of identifiers followed by an expression.
2970 A lambda expression denotes a function that maps a list of parameters (`ident_list`)
2971 onto the expression that follows the `ident_list`.
2972 The identifiers in the `ident_list` are the parameters to the function.
2973 These parameters' types need not be specified, as the compiler infers them from context.
2975 Lambda expressions are most useful when passing functions as arguments to other functions,
2976 as an abbreviation for defining and capturing a separate function.
2978 Significantly, lambda expressions _capture their environment_,
2979 which regular [function definitions](#functions) do not.
2980 The exact type of capture depends on the [function type](#function-types) inferred for the lambda expression.
2981 In the simplest and least-expensive form (analogous to a ```|| { }``` expression),
2982 the lambda expression captures its environment by reference,
2983 effectively borrowing pointers to all outer variables mentioned inside the function.
2984 Alternately, the compiler may infer that a lambda expression should copy or move values (depending on their type.)
2985 from the environment into the lambda expression's captured environment.
2987 In this example, we define a function `ten_times` that takes a higher-order function argument,
2988 and call it with a lambda expression as an argument.
2991 fn ten_times(f: |int|) {
2999 ten_times(|j| println!("hello, {}", j));
3005 while_expr : "while" no_struct_literal_expr '{' block '}' ;
3008 A `while` loop begins by evaluating the boolean loop conditional expression.
3009 If the loop conditional expression evaluates to `true`, the loop body block
3010 executes and control returns to the loop conditional expression. If the loop
3011 conditional expression evaluates to `false`, the `while` expression completes.
3026 A `loop` expression denotes an infinite loop.
3029 loop_expr : [ lifetime ':' ] "loop" '{' block '}';
3032 A `loop` expression may optionally have a _label_.
3033 If a label is present,
3034 then labeled `break` and `continue` expressions nested within this loop may exit out of this loop or return control to its head.
3035 See [Break expressions](#break-expressions) and [Continue expressions](#continue-expressions).
3037 ### Break expressions
3040 break_expr : "break" [ lifetime ];
3043 A `break` expression has an optional _label_.
3044 If the label is absent, then executing a `break` expression immediately terminates the innermost loop enclosing it.
3045 It is only permitted in the body of a loop.
3046 If the label is present, then `break foo` terminates the loop with label `foo`,
3047 which need not be the innermost label enclosing the `break` expression,
3048 but must enclose it.
3050 ### Continue expressions
3053 continue_expr : "continue" [ lifetime ];
3056 A `continue` expression has an optional _label_.
3057 If the label is absent,
3058 then executing a `continue` expression immediately terminates the current iteration of the innermost loop enclosing it,
3059 returning control to the loop *head*.
3060 In the case of a `while` loop,
3061 the head is the conditional expression controlling the loop.
3062 In the case of a `for` loop, the head is the call-expression controlling the loop.
3063 If the label is present, then `continue foo` returns control to the head of the loop with label `foo`,
3064 which need not be the innermost label enclosing the `break` expression,
3065 but must enclose it.
3067 A `continue` expression is only permitted in the body of a loop.
3072 for_expr : "for" pat "in" no_struct_literal_expr '{' block '}' ;
3075 A `for` expression is a syntactic construct for looping over elements
3076 provided by an implementation of `std::iter::Iterator`.
3078 An example of a for loop over the contents of a vector:
3082 # fn bar(f: Foo) { }
3087 let v: &[Foo] = &[a, b, c];
3094 An example of a for loop over a series of integers:
3097 # fn bar(b:uint) { }
3098 for i in range(0u, 256) {
3106 if_expr : "if" no_struct_literal_expr '{' block '}'
3109 else_tail : "else" [ if_expr
3113 An `if` expression is a conditional branch in program control. The form of
3114 an `if` expression is a condition expression, followed by a consequent
3115 block, any number of `else if` conditions and blocks, and an optional
3116 trailing `else` block. The condition expressions must have type
3117 `bool`. If a condition expression evaluates to `true`, the
3118 consequent block is executed and any subsequent `else if` or `else`
3119 block is skipped. If a condition expression evaluates to `false`, the
3120 consequent block is skipped and any subsequent `else if` condition is
3121 evaluated. If all `if` and `else if` conditions evaluate to `false`
3122 then any `else` block is executed.
3124 ### Match expressions
3127 match_expr : "match" no_struct_literal_expr '{' match_arm * '}' ;
3129 match_arm : attribute * match_pat "=>" [ expr "," | '{' block '}' ] ;
3131 match_pat : pat [ '|' pat ] * [ "if" expr ] ? ;
3134 A `match` expression branches on a *pattern*. The exact form of matching that
3135 occurs depends on the pattern. Patterns consist of some combination of
3136 literals, destructured vectors or enum constructors, structures and
3137 tuples, variable binding specifications, wildcards (`..`), and placeholders
3138 (`_`). A `match` expression has a *head expression*, which is the value to
3139 compare to the patterns. The type of the patterns must equal the type of the
3142 In a pattern whose head expression has an `enum` type, a placeholder (`_`)
3143 stands for a *single* data field, whereas a wildcard `..` stands for *all* the
3144 fields of a particular variant. For example:
3147 enum List<X> { Nil, Cons(X, Box<List<X>>) }
3149 let x: List<int> = Cons(10, box Cons(11, box Nil));
3152 Cons(_, box Nil) => fail!("singleton list"),
3154 Nil => fail!("empty list")
3158 The first pattern matches lists constructed by applying `Cons` to any head
3159 value, and a tail value of `box Nil`. The second pattern matches _any_ list
3160 constructed with `Cons`, ignoring the values of its arguments. The difference
3161 between `_` and `..` is that the pattern `C(_)` is only type-correct if `C` has
3162 exactly one argument, while the pattern `C(..)` is type-correct for any enum
3163 variant `C`, regardless of how many arguments `C` has.
3165 Used inside a vector pattern, `..` stands for any number of elements. This
3166 wildcard can be used at most once for a given vector, which implies that it
3167 cannot be used to specifically match elements that are at an unknown distance
3168 from both ends of a vector, like `[.., 42, ..]`. If followed by a variable name,
3169 it will bind the corresponding slice to the variable. Example:
3172 fn is_symmetric(list: &[uint]) -> bool {
3175 [x, ..inside, y] if x == y => is_symmetric(inside),
3181 let sym = &[0, 1, 4, 2, 4, 1, 0];
3182 let not_sym = &[0, 1, 7, 2, 4, 1, 0];
3183 assert!(is_symmetric(sym));
3184 assert!(!is_symmetric(not_sym));
3188 A `match` behaves differently depending on whether or not the head expression
3189 is an [lvalue or an rvalue](#lvalues,-rvalues-and-temporaries).
3190 If the head expression is an rvalue, it is
3191 first evaluated into a temporary location, and the resulting value
3192 is sequentially compared to the patterns in the arms until a match
3193 is found. The first arm with a matching pattern is chosen as the branch target
3194 of the `match`, any variables bound by the pattern are assigned to local
3195 variables in the arm's block, and control enters the block.
3197 When the head expression is an lvalue, the match does not allocate a
3198 temporary location (however, a by-value binding may copy or move from
3199 the lvalue). When possible, it is preferable to match on lvalues, as the
3200 lifetime of these matches inherits the lifetime of the lvalue, rather
3201 than being restricted to the inside of the match.
3203 An example of a `match` expression:
3206 # fn process_pair(a: int, b: int) { }
3207 # fn process_ten() { }
3209 enum List<X> { Nil, Cons(X, Box<List<X>>) }
3211 let x: List<int> = Cons(10, box Cons(11, box Nil));
3214 Cons(a, box Cons(b, _)) => {
3229 Patterns that bind variables
3230 default to binding to a copy or move of the matched value
3231 (depending on the matched value's type).
3232 This can be changed to bind to a reference by
3233 using the `ref` keyword,
3234 or to a mutable reference using `ref mut`.
3236 Subpatterns can also be bound to variables by the use of the syntax
3237 `variable @ subpattern`.
3241 enum List { Nil, Cons(uint, Box<List>) }
3243 fn is_sorted(list: &List) -> bool {
3245 Nil | Cons(_, box Nil) => true,
3246 Cons(x, ref r @ box Cons(y, _)) => (x <= y) && is_sorted(*r)
3251 let a = Cons(6, box Cons(7, box Cons(42, box Nil)));
3252 assert!(is_sorted(&a));
3257 Patterns can also dereference pointers by using the `&`,
3258 `box` or `@` symbols, as appropriate. For example, these two matches
3259 on `x: &int` are equivalent:
3263 let y = match *x { 0 => "zero", _ => "some" };
3264 let z = match x { &0 => "zero", _ => "some" };
3269 A pattern that's just an identifier, like `Nil` in the previous example,
3270 could either refer to an enum variant that's in scope, or bind a new variable.
3271 The compiler resolves this ambiguity by forbidding variable bindings that occur
3272 in `match` patterns from shadowing names of variants that are in scope.
3273 For example, wherever `List` is in scope,
3274 a `match` pattern would not be able to bind `Nil` as a new name.
3275 The compiler interprets a variable pattern `x` as a binding _only_ if there is
3276 no variant named `x` in scope.
3277 A convention you can use to avoid conflicts is simply to name variants with
3278 upper-case letters, and local variables with lower-case letters.
3280 Multiple match patterns may be joined with the `|` operator.
3281 A range of values may be specified with `..`.
3287 let message = match x {
3288 0 | 1 => "not many",
3294 Range patterns only work on scalar types
3295 (like integers and characters; not like vectors and structs, which have sub-components).
3296 A range pattern may not be a sub-range of another range pattern inside the same `match`.
3298 Finally, match patterns can accept *pattern guards* to further refine the
3299 criteria for matching a case. Pattern guards appear after the pattern and
3300 consist of a bool-typed expression following the `if` keyword. A pattern
3301 guard may refer to the variables bound within the pattern they follow.
3304 # let maybe_digit = Some(0);
3305 # fn process_digit(i: int) { }
3306 # fn process_other(i: int) { }
3308 let message = match maybe_digit {
3309 Some(x) if x < 10 => process_digit(x),
3310 Some(x) => process_other(x),
3315 ### Return expressions
3318 return_expr : "return" expr ? ;
3321 Return expressions are denoted with the keyword `return`. Evaluating a `return`
3322 expression moves its argument into the output slot of the current
3323 function, destroys the current function activation frame, and transfers
3324 control to the caller frame.
3326 An example of a `return` expression:
3329 fn max(a: int, b: int) -> int {
3341 Every slot, item and value in a Rust program has a type. The _type_ of a *value*
3342 defines the interpretation of the memory holding it.
3344 Built-in types and type-constructors are tightly integrated into the language,
3345 in nontrivial ways that are not possible to emulate in user-defined
3346 types. User-defined types have limited capabilities.
3350 The primitive types are the following:
3352 * The "unit" type `()`, having the single "unit" value `()` (occasionally called
3354 * The boolean type `bool` with values `true` and `false`.
3355 * The machine types.
3356 * The machine-dependent integer and floating-point types.
3358 [^unittype]: The "unit" value `()` is *not* a sentinel "null pointer" value for
3359 reference slots; the "unit" type is the implicit return type from functions
3360 otherwise lacking a return type, and can be used in other contexts (such as
3361 message-sending or type-parametric code) as a zero-size type.]
3365 The machine types are the following:
3367 * The unsigned word types `u8`, `u16`, `u32` and `u64`, with values drawn from
3368 the integer intervals [0, 2^8 - 1], [0, 2^16 - 1], [0, 2^32 - 1] and
3369 [0, 2^64 - 1] respectively.
3371 * The signed two's complement word types `i8`, `i16`, `i32` and `i64`, with
3372 values drawn from the integer intervals [-(2^(7)), 2^7 - 1],
3373 [-(2^(15)), 2^15 - 1], [-(2^(31)), 2^31 - 1], [-(2^(63)), 2^63 - 1]
3376 * The IEEE 754-2008 `binary32` and `binary64` floating-point types: `f32` and
3377 `f64`, respectively.
3379 #### Machine-dependent integer types
3381 The Rust type `uint` [^rustuint] is an
3382 unsigned integer type with target-machine-dependent size. Its size, in
3383 bits, is equal to the number of bits required to hold any memory address on
3386 The Rust type `int` [^rustint] is a
3387 two's complement signed integer type with target-machine-dependent size. Its
3388 size, in bits, is equal to the size of the rust type `uint` on the same target
3391 [^rustuint]: A Rust `uint` is analogous to a C99 `uintptr_t`.
3392 [^rustint]: A Rust `int` is analogous to a C99 `intptr_t`.
3396 The types `char` and `str` hold textual data.
3398 A value of type `char` is a [Unicode scalar value](
3399 http://www.unicode.org/glossary/#unicode_scalar_value)
3400 (ie. a code point that is not a surrogate),
3401 represented as a 32-bit unsigned word in the 0x0000 to 0xD7FF
3402 or 0xE000 to 0x10FFFF range.
3403 A `[char]` vector is effectively an UCS-4 / UTF-32 string.
3405 A value of type `str` is a Unicode string,
3406 represented as a vector of 8-bit unsigned bytes holding a sequence of UTF-8 codepoints.
3407 Since `str` is of unknown size, it is not a _first class_ type,
3408 but can only be instantiated through a pointer type,
3409 such as `&str` or `String`.
3413 A tuple *type* is a heterogeneous product of other types, called the *elements*
3414 of the tuple. It has no nominal name and is instead structurally typed.
3416 Tuple types and values are denoted by listing the types or values of their
3417 elements, respectively, in a parenthesized, comma-separated
3420 Because tuple elements don't have a name, they can only be accessed by pattern-matching.
3422 The members of a tuple are laid out in memory contiguously, in
3423 order specified by the tuple type.
3425 An example of a tuple type and its use:
3428 type Pair<'a> = (int, &'a str);
3429 let p: Pair<'static> = (10, "hello");
3431 assert!(b != "world");
3436 The vector type constructor represents a homogeneous array of values of a given type.
3437 A vector has a fixed size.
3438 (Operations like `vec.push` operate solely on owned vectors.)
3439 A vector type can be annotated with a _definite_ size, such as `[int, ..10]`.
3440 Such a definite-sized vector type is a first-class type, since its size is known statically.
3441 A vector without such a size is said to be of _indefinite_ size,
3442 and is therefore not a _first-class_ type.
3443 An indefinite-size vector can only be instantiated through a pointer type,
3444 such as `&[T]` or `Vec<T>`.
3445 The kind of a vector type depends on the kind of its element type,
3446 as with other simple structural types.
3448 Expressions producing vectors of definite size cannot be evaluated in a
3449 context expecting a vector of indefinite size; one must copy the
3450 definite-sized vector contents into a distinct vector of indefinite size.
3452 An example of a vector type and its use:
3455 let v: &[int] = &[7, 5, 3];
3460 All in-bounds elements of a vector are always initialized,
3461 and access to a vector is always bounds-checked.
3465 A `struct` *type* is a heterogeneous product of other types, called the *fields*
3466 of the type.[^structtype]
3468 [^structtype]: `struct` types are analogous `struct` types in C,
3469 the *record* types of the ML family,
3470 or the *structure* types of the Lisp family.
3472 New instances of a `struct` can be constructed with a [struct expression](#structure-expressions).
3474 The memory layout of a `struct` is undefined by default to allow for compiler optimziations like
3475 field reordering, but it can be fixed with the `#[repr(...)]` attribute.
3476 In either case, fields may be given in any order in a corresponding struct *expression*;
3477 the resulting `struct` value will always have the same memory layout.
3479 The fields of a `struct` may be qualified by [visibility modifiers](#re-exporting-and-visibility),
3480 to allow access to data in a structure outside a module.
3482 A _tuple struct_ type is just like a structure type, except that the fields are anonymous.
3484 A _unit-like struct_ type is like a structure type, except that it has no fields.
3485 The one value constructed by the associated [structure expression](#structure-expressions)
3486 is the only value that inhabits such a type.
3488 ### Enumerated types
3490 An *enumerated type* is a nominal, heterogeneous disjoint union type,
3491 denoted by the name of an [`enum` item](#enumerations). [^enumtype]
3493 [^enumtype]: The `enum` type is analogous to a `data` constructor declaration in
3494 ML, or a *pick ADT* in Limbo.
3496 An [`enum` item](#enumerations) declares both the type and a number of *variant constructors*,
3497 each of which is independently named and takes an optional tuple of arguments.
3499 New instances of an `enum` can be constructed by calling one of the variant constructors,
3500 in a [call expression](#call-expressions).
3502 Any `enum` value consumes as much memory as the largest variant constructor for its corresponding `enum` type.
3504 Enum types cannot be denoted *structurally* as types,
3505 but must be denoted by named reference to an [`enum` item](#enumerations).
3509 Nominal types — [enumerations](#enumerated-types) and [structures](#structure-types) — may be recursive.
3510 That is, each `enum` constructor or `struct` field may refer, directly or indirectly, to the enclosing `enum` or `struct` type itself.
3511 Such recursion has restrictions:
3513 * Recursive types must include a nominal type in the recursion
3514 (not mere [type definitions](#type-definitions),
3515 or other structural types such as [vectors](#vector-types) or [tuples](#tuple-types)).
3516 * A recursive `enum` item must have at least one non-recursive constructor
3517 (in order to give the recursion a basis case).
3518 * The size of a recursive type must be finite;
3519 in other words the recursive fields of the type must be [pointer types](#pointer-types).
3520 * Recursive type definitions can cross module boundaries, but not module *visibility* boundaries,
3521 or crate boundaries (in order to simplify the module system and type checker).
3523 An example of a *recursive* type and its use:
3528 Cons(T, Box<List<T>>)
3531 let a: List<int> = Cons(7, box Cons(13, box Nil));
3536 All pointers in Rust are explicit first-class values.
3537 They can be copied, stored into data structures, and returned from functions.
3538 There are four varieties of pointer in Rust:
3540 * Owning pointers (`Box`)
3541 : These point to owned heap allocations (or "boxes") in the shared, inter-task heap.
3542 Each owned box has a single owning pointer; pointer and pointee retain a 1:1 relationship at all times.
3543 Owning pointers are written `Box<content>`,
3544 for example `Box<int>` means an owning pointer to an owned box containing an integer.
3545 Copying an owned box is a "deep" operation:
3546 it involves allocating a new owned box and copying the contents of the old box into the new box.
3547 Releasing an owning pointer immediately releases its corresponding owned box.
3550 : These point to memory _owned by some other value_.
3551 References arise by (automatic) conversion from owning pointers, managed pointers,
3552 or by applying the borrowing operator `&` to some other value,
3553 including [lvalues, rvalues or temporaries](#lvalues,-rvalues-and-temporaries).
3554 References are written `&content`, or in some cases `&'f content` for some lifetime-variable `f`,
3555 for example `&int` means a reference to an integer.
3556 Copying a reference is a "shallow" operation:
3557 it involves only copying the pointer itself.
3558 Releasing a reference typically has no effect on the value it points to,
3559 with the exception of temporary values,
3560 which are released when the last reference to them is released.
3562 * Raw pointers (`*`)
3563 : Raw pointers are pointers without safety or liveness guarantees.
3564 Raw pointers are written as `*const T` or `*mut T`,
3565 for example `*const int` means a raw pointer to an integer.
3566 Copying or dropping a raw pointer has no effect on the lifecycle of any
3567 other value. Dereferencing a raw pointer or converting it to any other
3568 pointer type is an [`unsafe` operation](#unsafe-functions).
3569 Raw pointers are generally discouraged in Rust code;
3570 they exist to support interoperability with foreign code,
3571 and writing performance-critical or low-level functions.
3575 The function type constructor `fn` forms new function types.
3576 A function type consists of a possibly-empty set of function-type modifiers
3577 (such as `unsafe` or `extern`), a sequence of input types and an output type.
3579 An example of a `fn` type:
3582 fn add(x: int, y: int) -> int {
3586 let mut x = add(5,7);
3588 type Binop<'a> = |int,int|: 'a -> int;
3589 let bo: Binop = add;
3595 ~~~~ {.ebnf .notation}
3596 closure_type := [ 'unsafe' ] [ '<' lifetime-list '>' ] '|' arg-list '|'
3597 [ ':' bound-list ] [ '->' type ]
3598 procedure_type := 'proc' [ '<' lifetime-list '>' ] '(' arg-list ')'
3599 [ ':' bound-list ] [ '->' type ]
3600 lifetime-list := lifetime | lifetime ',' lifetime-list
3601 arg-list := ident ':' type | ident ':' type ',' arg-list
3602 bound-list := bound | bound '+' bound-list
3603 bound := path | lifetime
3606 The type of a closure mapping an input of type `A` to an output of type `B` is
3607 `|A| -> B`. A closure with no arguments or return values has type `||`.
3608 Similarly, a procedure mapping `A` to `B` is `proc(A) -> B` and a no-argument
3609 and no-return value closure has type `proc()`.
3611 An example of creating and calling a closure:
3614 let captured_var = 10i;
3616 let closure_no_args = || println!("captured_var={}", captured_var);
3618 let closure_args = |arg: int| -> int {
3619 println!("captured_var={}, arg={}", captured_var, arg);
3620 arg // Note lack of semicolon after 'arg'
3623 fn call_closure(c1: ||, c2: |int| -> int) {
3628 call_closure(closure_no_args, closure_args);
3632 Unlike closures, procedures may only be invoked once, but own their
3633 environment, and are allowed to move out of their environment. Procedures are
3634 allocated on the heap (unlike closures). An example of creating and calling a
3638 let string = "Hello".to_string();
3640 // Creates a new procedure, passing it to the `spawn` function.
3642 println!("{} world!", string);
3645 // the variable `string` has been moved into the previous procedure, so it is
3646 // no longer usable.
3649 // Create an invoke a procedure. Note that the procedure is *moved* when
3650 // invoked, so it cannot be invoked again.
3651 let f = proc(n: int) { n + 22 };
3652 println!("answer: {}", f(20));
3658 Every trait item (see [traits](#traits)) defines a type with the same name as the trait.
3659 This type is called the _object type_ of the trait.
3660 Object types permit "late binding" of methods, dispatched using _virtual method tables_ ("vtables").
3661 Whereas most calls to trait methods are "early bound" (statically resolved) to specific implementations at compile time,
3662 a call to a method on an object type is only resolved to a vtable entry at compile time.
3663 The actual implementation for each vtable entry can vary on an object-by-object basis.
3665 Given a pointer-typed expression `E` of type `&T` or `Box<T>`, where `T` implements trait `R`,
3666 casting `E` to the corresponding pointer type `&R` or `Box<R>` results in a value of the _object type_ `R`.
3667 This result is represented as a pair of pointers:
3668 the vtable pointer for the `T` implementation of `R`, and the pointer value of `E`.
3670 An example of an object type:
3674 fn to_string(&self) -> String;
3677 impl Printable for int {
3678 fn to_string(&self) -> String { self.to_str() }
3681 fn print(a: Box<Printable>) {
3682 println!("{}", a.to_string());
3686 print(box 10i as Box<Printable>);
3690 In this example, the trait `Printable` occurs as an object type in both the type signature of `print`,
3691 and the cast expression in `main`.
3695 Within the body of an item that has type parameter declarations, the names of its type parameters are types:
3698 fn map<A: Clone, B: Clone>(f: |A| -> B, xs: &[A]) -> Vec<B> {
3702 let first: B = f(xs[0].clone());
3703 let rest: Vec<B> = map(f, xs.slice(1, xs.len()));
3704 return vec![first].append(rest.as_slice());
3708 Here, `first` has type `B`, referring to `map`'s `B` type parameter;
3709 and `rest` has type `Vec<B>`, a vector type with element type `B`.
3713 The special type `self` has a meaning within methods inside an
3714 impl item. It refers to the type of the implicit `self` argument. For
3719 fn make_string(&self) -> String;
3722 impl Printable for String {
3723 fn make_string(&self) -> String {
3729 `self` refers to the value of type `String` that is the receiver for a
3730 call to the method `make_string`.
3734 Types in Rust are categorized into kinds, based on various properties of the components of the type.
3738 : Types of this kind can be safely sent between tasks.
3739 This kind includes scalars, owning pointers, owned closures, and
3740 structural types containing only other owned types.
3741 All `Send` types are `'static`.
3743 : Types of this kind consist of "Plain Old Data"
3744 which can be copied by simply moving bits.
3745 All values of this kind can be implicitly copied.
3746 This kind includes scalars and immutable references,
3747 as well as structural types containing other `Copy` types.
3749 : Types of this kind do not contain any references (except for
3750 references with the `static` lifetime, which are allowed).
3751 This can be a useful guarantee for code
3752 that breaks borrowing assumptions
3753 using [`unsafe` operations](#unsafe-functions).
3755 : This is not strictly a kind,
3756 but its presence interacts with kinds:
3757 the `Drop` trait provides a single method `drop`
3758 that takes no parameters,
3759 and is run when values of the type are dropped.
3760 Such a method is called a "destructor",
3761 and are always executed in "top-down" order:
3762 a value is completely destroyed
3763 before any of the values it owns run their destructors.
3764 Only `Send` types can implement `Drop`.
3767 : Types with destructors, closure environments,
3768 and various other _non-first-class_ types,
3769 are not copyable at all.
3770 Such types can usually only be accessed through pointers,
3771 or in some cases, moved between mutable locations.
3773 Kinds can be supplied as _bounds_ on type parameters, like traits,
3774 in which case the parameter is constrained to types satisfying that kind.
3776 By default, type parameters do not carry any assumed kind-bounds at all.
3777 When instantiating a type parameter,
3778 the kind bounds on the parameter are checked
3779 to be the same or narrower than the kind
3780 of the type that it is instantiated with.
3782 Sending operations are not part of the Rust language,
3783 but are implemented in the library.
3784 Generic functions that send values
3785 bound the kind of these values to sendable.
3787 # Memory and concurrency models
3789 Rust has a memory model centered around concurrently-executing _tasks_. Thus
3790 its memory model and its concurrency model are best discussed simultaneously,
3791 as parts of each only make sense when considered from the perspective of the
3794 When reading about the memory model, keep in mind that it is partitioned in
3795 order to support tasks; and when reading about tasks, keep in mind that their
3796 isolation and communication mechanisms are only possible due to the ownership
3797 and lifetime semantics of the memory model.
3801 A Rust program's memory consists of a static set of *items*, a set of
3802 [tasks](#tasks) each with its own *stack*, and a *heap*. Immutable portions of
3803 the heap may be shared between tasks, mutable portions may not.
3805 Allocations in the stack consist of *slots*, and allocations in the heap
3808 ### Memory allocation and lifetime
3810 The _items_ of a program are those functions, modules and types
3811 that have their value calculated at compile-time and stored uniquely in the
3812 memory image of the rust process. Items are neither dynamically allocated nor
3815 A task's _stack_ consists of activation frames automatically allocated on
3816 entry to each function as the task executes. A stack allocation is reclaimed
3817 when control leaves the frame containing it.
3819 The _heap_ is a general term that describes two separate sets of boxes:
3820 managed boxes — which may be subject to garbage collection — and owned
3821 boxes. The lifetime of an allocation in the heap depends on the lifetime of
3822 the box values pointing to it. Since box values may themselves be passed in
3823 and out of frames, or stored in the heap, heap allocations may outlive the
3824 frame they are allocated within.
3826 ### Memory ownership
3828 A task owns all memory it can *safely* reach through local variables,
3829 as well as managed, owned boxes and references.
3831 When a task sends a value that has the `Send` trait to another task,
3832 it loses ownership of the value sent and can no longer refer to it.
3833 This is statically guaranteed by the combined use of "move semantics",
3834 and the compiler-checked _meaning_ of the `Send` trait:
3835 it is only instantiated for (transitively) sendable kinds of data constructor and pointers,
3836 never including managed boxes or references.
3838 When a stack frame is exited, its local allocations are all released, and its
3839 references to boxes (both managed and owned) are dropped.
3841 A managed box may (in the case of a recursive, mutable managed type) be cyclic;
3842 in this case the release of memory inside the managed structure may be deferred
3843 until task-local garbage collection can reclaim it. Code can ensure no such
3844 delayed deallocation occurs by restricting itself to owned boxes and similar
3845 unmanaged kinds of data.
3847 When a task finishes, its stack is necessarily empty and it therefore has no
3848 references to any boxes; the remainder of its heap is immediately freed.
3852 A task's stack contains slots.
3854 A _slot_ is a component of a stack frame, either a function parameter,
3855 a [temporary](#lvalues,-rvalues-and-temporaries), or a local variable.
3857 A _local variable_ (or *stack-local* allocation) holds a value directly,
3858 allocated within the stack's memory. The value is a part of the stack frame.
3860 Local variables are immutable unless declared otherwise like: `let mut x = ...`.
3862 Function parameters are immutable unless declared with `mut`. The
3863 `mut` keyword applies only to the following parameter (so `|mut x, y|`
3864 and `fn f(mut x: Box<int>, y: Box<int>)` declare one mutable variable `x` and
3865 one immutable variable `y`).
3867 Methods that take either `self` or `~self` can optionally place them in a
3868 mutable slot by prefixing them with `mut` (similar to regular arguments):
3872 fn change(mut self) -> Self;
3873 fn modify(mut ~self) -> Box<Self>;
3877 Local variables are not initialized when allocated; the entire frame worth of
3878 local variables are allocated at once, on frame-entry, in an uninitialized
3879 state. Subsequent statements within a function may or may not initialize the
3880 local variables. Local variables can be used only after they have been
3881 initialized; this is enforced by the compiler.
3885 An _owned box_ is a reference to a heap allocation holding another value, which is constructed
3886 by the prefix operator `box`. When the standard library is in use, the type of an owned box is
3887 `std::owned::Box<T>`.
3889 An example of an owned box type and value:
3893 let x: Box<int> = box 10;
3896 Owned box values exist in 1:1 correspondence with their heap allocation
3897 copying an owned box value makes a shallow copy of the pointer
3898 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.
3901 let x: Box<int> = box 10;
3903 // attempting to use `x` will result in an error here
3910 An executing Rust program consists of a tree of tasks.
3911 A Rust _task_ consists of an entry function, a stack,
3912 a set of outgoing communication channels and incoming communication ports,
3913 and ownership of some portion of the heap of a single operating-system process.
3914 (We expect that many programs will not use channels and ports directly,
3915 but will instead use higher-level abstractions provided in standard libraries,
3918 Multiple Rust tasks may coexist in a single operating-system process.
3919 The runtime scheduler maps tasks to a certain number of operating-system threads.
3920 By default, the scheduler chooses the number of threads based on
3921 the number of concurrent physical CPUs detected at startup.
3922 It's also possible to override this choice at runtime.
3923 When the number of tasks exceeds the number of threads — which is likely —
3924 the scheduler multiplexes the tasks onto threads.[^mnscheduler]
3926 [^mnscheduler]: This is an M:N scheduler, which is known to give suboptimal
3927 results for CPU-bound concurrency problems. In such cases, running with the
3928 same number of threads and tasks can yield better results. Rust has M:N
3929 scheduling in order to support very large numbers of tasks in contexts where
3930 threads are too resource-intensive to use in large number. The cost of
3931 threads varies substantially per operating system, and is sometimes quite
3932 low, so this flexibility is not always worth exploiting.
3934 ### Communication between tasks
3936 Rust tasks are isolated and generally unable to interfere with one another's memory directly,
3937 except through [`unsafe` code](#unsafe-functions).
3938 All contact between tasks is mediated by safe forms of ownership transfer,
3939 and data races on memory are prohibited by the type system.
3941 Inter-task communication and co-ordination facilities are provided in the standard library.
3944 - synchronous and asynchronous communication channels with various communication topologies
3945 - read-only and read-write shared variables with various safe mutual exclusion patterns
3946 - simple locks and semaphores
3948 When such facilities carry values, the values are restricted to the [`Send` type-kind](#type-kinds).
3949 Restricting communication interfaces to this kind ensures that no references or managed pointers move between tasks.
3950 Thus access to an entire data structure can be mediated through its owning "root" value;
3951 no further locking or copying is required to avoid data races within the substructure of such a value.
3955 The _lifecycle_ of a task consists of a finite set of states and events
3956 that cause transitions between the states. The lifecycle states of a task are:
3963 A task begins its lifecycle — once it has been spawned — in the *running*
3964 state. In this state it executes the statements of its entry function, and any
3965 functions called by the entry function.
3967 A task may transition from the *running* state to the *blocked*
3968 state any time it makes a blocking communication call. When the
3969 call can be completed — when a message arrives at a sender, or a
3970 buffer opens to receive a message — then the blocked task will
3971 unblock and transition back to *running*.
3973 A task may transition to the *failing* state at any time, due being
3974 killed by some external event or internally, from the evaluation of a
3975 `fail!()` macro. Once *failing*, a task unwinds its stack and
3976 transitions to the *dead* state. Unwinding the stack of a task is done by
3977 the task itself, on its own control stack. If a value with a destructor is
3978 freed during unwinding, the code for the destructor is run, also on the task's
3979 control stack. Running the destructor code causes a temporary transition to a
3980 *running* state, and allows the destructor code to cause any subsequent
3981 state transitions. The original task of unwinding and failing thereby may
3982 suspend temporarily, and may involve (recursive) unwinding of the stack of a
3983 failed destructor. Nonetheless, the outermost unwinding activity will continue
3984 until the stack is unwound and the task transitions to the *dead*
3985 state. There is no way to "recover" from task failure. Once a task has
3986 temporarily suspended its unwinding in the *failing* state, failure
3987 occurring from within this destructor results in *hard* failure.
3988 A hard failure currently results in the process aborting.
3990 A task in the *dead* state cannot transition to other states; it exists
3991 only to have its termination status inspected by other tasks, and/or to await
3992 reclamation when the last reference to it drops.
3996 The currently scheduled task is given a finite *time slice* in which to
3997 execute, after which it is *descheduled* at a loop-edge or similar
3998 preemption point, and another task within is scheduled, pseudo-randomly.
4000 An executing task can yield control at any time, by making a library call to
4001 `std::task::yield`, which deschedules it immediately. Entering any other
4002 non-executing state (blocked, dead) similarly deschedules the task.
4004 # Runtime services, linkage and debugging
4006 The Rust _runtime_ is a relatively compact collection of C++ and Rust code
4007 that provides fundamental services and datatypes to all Rust tasks at
4008 run-time. It is smaller and simpler than many modern language runtimes. It is
4009 tightly integrated into the language's execution model of memory, tasks,
4010 communication and logging.
4012 > **Note:** The runtime library will merge with the `std` library in future versions of Rust.
4014 ### Memory allocation
4016 The runtime memory-management system is based on a _service-provider interface_,
4017 through which the runtime requests blocks of memory from its environment
4018 and releases them back to its environment when they are no longer needed.
4019 The default implementation of the service-provider interface
4020 consists of the C runtime functions `malloc` and `free`.
4022 The runtime memory-management system, in turn, supplies Rust tasks with
4023 facilities for allocating releasing stacks, as well as allocating and freeing
4028 The runtime provides C and Rust code to assist with various built-in types,
4029 such as vectors, strings, and the low level communication system (ports,
4032 Support for other built-in types such as simple types, tuples and
4033 enums is open-coded by the Rust compiler.
4035 ### Task scheduling and communication
4037 The runtime provides code to manage inter-task communication. This includes
4038 the system of task-lifecycle state transitions depending on the contents of
4039 queues, as well as code to copy values between queues and their recipients and
4040 to serialize values for transmission over operating-system inter-process
4041 communication facilities.
4045 The Rust compiler supports various methods to link crates together both
4046 statically and dynamically. This section will explore the various methods to
4047 link Rust crates together, and more information about native libraries can be
4048 found in the [ffi tutorial][ffi].
4050 In one session of compilation, the compiler can generate multiple artifacts
4051 through the usage of either command line flags or the `crate_type` attribute.
4052 If one or more command line flag is specified, all `crate_type` attributes will
4053 be ignored in favor of only building the artifacts specified by command line.
4055 * `--crate-type=bin`, `#[crate_type = "bin"]` - A runnable executable will be
4056 produced. This requires that there is a `main` function in the crate which
4057 will be run when the program begins executing. This will link in all Rust and
4058 native dependencies, producing a distributable binary.
4060 * `--crate-type=lib`, `#[crate_type = "lib"]` - A Rust library will be produced.
4061 This is an ambiguous concept as to what exactly is produced because a library
4062 can manifest itself in several forms. The purpose of this generic `lib` option
4063 is to generate the "compiler recommended" style of library. The output library
4064 will always be usable by rustc, but the actual type of library may change from
4065 time-to-time. The remaining output types are all different flavors of
4066 libraries, and the `lib` type can be seen as an alias for one of them (but the
4067 actual one is compiler-defined).
4069 * `--crate-type=dylib`, `#[crate_type = "dylib"]` - A dynamic Rust library will
4070 be produced. This is different from the `lib` output type in that this forces
4071 dynamic library generation. The resulting dynamic library can be used as a
4072 dependency for other libraries and/or executables. This output type will
4073 create `*.so` files on linux, `*.dylib` files on osx, and `*.dll` files on
4076 * `--crate-type=staticlib`, `#[crate_type = "staticlib"]` - A static system
4077 library will be produced. This is different from other library outputs in that
4078 the Rust compiler will never attempt to link to `staticlib` outputs. The
4079 purpose of this output type is to create a static library containing all of
4080 the local crate's code along with all upstream dependencies. The static
4081 library is actually a `*.a` archive on linux and osx and a `*.lib` file on
4082 windows. This format is recommended for use in situtations such as linking
4083 Rust code into an existing non-Rust application because it will not have
4084 dynamic dependencies on other Rust code.
4086 * `--crate-type=rlib`, `#[crate_type = "rlib"]` - A "Rust library" file will be
4087 produced. This is used as an intermediate artifact and can be thought of as a
4088 "static Rust library". These `rlib` files, unlike `staticlib` files, are
4089 interpreted by the Rust compiler in future linkage. This essentially means
4090 that `rustc` will look for metadata in `rlib` files like it looks for metadata
4091 in dynamic libraries. This form of output is used to produce statically linked
4092 executables as well as `staticlib` outputs.
4094 Note that these outputs are stackable in the sense that if multiple are
4095 specified, then the compiler will produce each form of output at once without
4096 having to recompile. However, this only applies for outputs specified by the same
4097 method. If only `crate_type` attributes are specified, then they will all be
4098 built, but if one or more `--crate-type` command line flag is specified,
4099 then only those outputs will be built.
4101 With all these different kinds of outputs, if crate A depends on crate B, then
4102 the compiler could find B in various different forms throughout the system. The
4103 only forms looked for by the compiler, however, are the `rlib` format and the
4104 dynamic library format. With these two options for a dependent library, the
4105 compiler must at some point make a choice between these two formats. With this
4106 in mind, the compiler follows these rules when determining what format of
4107 dependencies will be used:
4109 1. If a static library is being produced, all upstream dependencies are
4110 required to be available in `rlib` formats. This requirement stems from the
4111 reason that a dynamic library cannot be converted into a static format.
4113 Note that it is impossible to link in native dynamic dependencies to a static
4114 library, and in this case warnings will be printed about all unlinked native
4115 dynamic dependencies.
4117 2. If an `rlib` file is being produced, then there are no restrictions on what
4118 format the upstream dependencies are available in. It is simply required that
4119 all upstream dependencies be available for reading metadata from.
4121 The reason for this is that `rlib` files do not contain any of their upstream
4122 dependencies. It wouldn't be very efficient for all `rlib` files to contain a
4123 copy of `libstd.rlib`!
4125 3. If an executable is being produced and the `-C prefer-dynamic` flag is not
4126 specified, then dependencies are first attempted to be found in the `rlib`
4127 format. If some dependencies are not available in an rlib format, then
4128 dynamic linking is attempted (see below).
4130 4. If a dynamic library or an executable that is being dynamically linked is
4131 being produced, then the compiler will attempt to reconcile the available
4132 dependencies in either the rlib or dylib format to create a final product.
4134 A major goal of the compiler is to ensure that a library never appears more
4135 than once in any artifact. For example, if dynamic libraries B and C were
4136 each statically linked to library A, then a crate could not link to B and C
4137 together because there would be two copies of A. The compiler allows mixing
4138 the rlib and dylib formats, but this restriction must be satisfied.
4140 The compiler currently implements no method of hinting what format a library
4141 should be linked with. When dynamically linking, the compiler will attempt to
4142 maximize dynamic dependencies while still allowing some dependencies to be
4143 linked in via an rlib.
4145 For most situations, having all libraries available as a dylib is recommended
4146 if dynamically linking. For other situations, the compiler will emit a
4147 warning if it is unable to determine which formats to link each library with.
4149 In general, `--crate-type=bin` or `--crate-type=lib` should be sufficient for
4150 all compilation needs, and the other options are just available if more
4151 fine-grained control is desired over the output format of a Rust crate.
4155 The runtime contains a system for directing [logging
4156 expressions](#logging-expressions) to a logging console and/or internal logging
4157 buffers. Logging can be enabled per module.
4159 Logging output is enabled by setting the `RUST_LOG` environment
4160 variable. `RUST_LOG` accepts a logging specification made up of a
4161 comma-separated list of paths, with optional log levels. For each
4162 module containing log expressions, if `RUST_LOG` contains the path to
4163 that module or a parent of that module, then logs of the appropriate
4164 level will be output to the console.
4166 The path to a module consists of the crate name, any parent modules,
4167 then the module itself, all separated by double colons (`::`). The
4168 optional log level can be appended to the module path with an equals
4169 sign (`=`) followed by the log level, from 1 to 4, inclusive. Level 1
4170 is the error level, 2 is warning, 3 info, and 4 debug. You can also
4171 use the symbolic constants `error`, `warn`, `info`, and `debug`. Any
4172 logs less than or equal to the specified level will be output. If not
4173 specified then log level 4 is assumed. Debug messages can be omitted
4174 by passing `--cfg ndebug` to `rustc`.
4176 As an example, to see all the logs generated by the compiler, you would set
4177 `RUST_LOG` to `rustc`, which is the crate name (as specified in its `crate_id`
4178 [attribute](#attributes)). To narrow down the logs to just crate resolution,
4179 you would set it to `rustc::metadata::creader`. To see just error logging
4182 Note that when compiling source files that don't specify a
4183 crate name the crate is given a default name that matches the source file,
4184 with the extension removed. In that case, to turn on logging for a program
4185 compiled from, e.g. `helloworld.rs`, `RUST_LOG` should be set to `helloworld`.
4187 #### Logging Expressions
4189 Rust provides several macros to log information. Here's a simple Rust program
4190 that demonstrates all four of them:
4194 #[phase(plugin, link)] extern crate log;
4197 error!("This is an error log")
4198 warn!("This is a warn log")
4199 info!("this is an info log")
4200 debug!("This is a debug log")
4204 These four log levels correspond to levels 1-4, as controlled by `RUST_LOG`:
4207 $ RUST_LOG=rust=3 ./rust
4208 This is an error log
4213 # Appendix: Rationales and design tradeoffs
4217 # Appendix: Influences and further references
4221 > The essential problem that must be solved in making a fault-tolerant
4222 > software system is therefore that of fault-isolation. Different programmers
4223 > will write different modules, some modules will be correct, others will have
4224 > errors. We do not want the errors in one module to adversely affect the
4225 > behaviour of a module which does not have any errors.
4227 > — Joe Armstrong
4229 > In our approach, all data is private to some process, and processes can
4230 > only communicate through communications channels. *Security*, as used
4231 > in this paper, is the property which guarantees that processes in a system
4232 > cannot affect each other except by explicit communication.
4234 > When security is absent, nothing which can be proven about a single module
4235 > in isolation can be guaranteed to hold when that module is embedded in a
4238 > — Robert Strom and Shaula Yemini
4240 > Concurrent and applicative programming complement each other. The
4241 > ability to send messages on channels provides I/O without side effects,
4242 > while the avoidance of shared data helps keep concurrent processes from
4247 Rust is not a particularly original language. It may however appear unusual
4248 by contemporary standards, as its design elements are drawn from a number of
4249 "historical" languages that have, with a few exceptions, fallen out of
4250 favour. Five prominent lineages contribute the most, though their influences
4251 have come and gone during the course of Rust's development:
4253 * The NIL (1981) and Hermes (1990) family. These languages were developed by
4254 Robert Strom, Shaula Yemini, David Bacon and others in their group at IBM
4255 Watson Research Center (Yorktown Heights, NY, USA).
4257 * The Erlang (1987) language, developed by Joe Armstrong, Robert Virding, Claes
4258 Wikström, Mike Williams and others in their group at the Ericsson Computer
4259 Science Laboratory (Älvsjö, Stockholm, Sweden) .
4261 * The Sather (1990) language, developed by Stephen Omohundro, Chu-Cheow Lim,
4262 Heinz Schmidt and others in their group at The International Computer
4263 Science Institute of the University of California, Berkeley (Berkeley, CA,
4266 * The Newsqueak (1988), Alef (1995), and Limbo (1996) family. These
4267 languages were developed by Rob Pike, Phil Winterbottom, Sean Dorward and
4268 others in their group at Bell Labs Computing Sciences Research Center
4269 (Murray Hill, NJ, USA).
4271 * The Napier (1985) and Napier88 (1988) family. These languages were
4272 developed by Malcolm Atkinson, Ron Morrison and others in their group at
4273 the University of St. Andrews (St. Andrews, Fife, UK).
4275 Additional specific influences can be seen from the following languages:
4277 * The structural algebraic types and compilation manager of SML.
4278 * The attribute and assembly systems of C#.
4279 * The references and deterministic destructor system of C++.
4280 * The memory region systems of the ML Kit and Cyclone.
4281 * The typeclass system of Haskell.
4282 * The lexical identifier rule of Python.
4283 * The block syntax of Ruby.
4285 [ffi]: guide-ffi.html