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] or [extra]
21 libraries 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
27 [extra]: extra/index.html
31 Rust is a work in progress. The language continues to evolve as the design
32 shifts and is fleshed out in working code. Certain parts work, certain parts
33 do not, certain parts will be removed or changed.
35 This manual is a snapshot written in the present tense. All features described
36 exist in working code unless otherwise noted, but some are quite primitive or
37 remain to be further modified by planned work. Some may be temporary. It is a
38 *draft*, and we ask that you not take anything you read here as final.
40 If you have suggestions to make, please try to focus them on *reductions* to
41 the language: possible features that can be combined or omitted. We aim to
42 keep the size and complexity of the language under control.
44 > **Note:** The grammar for Rust given in this document is rough and
45 > very incomplete; only a modest number of sections have accompanying grammar
46 > rules. Formalizing the grammar accepted by the Rust parser is ongoing work,
47 > but future versions of this document will contain a complete
48 > grammar. Moreover, we hope that this grammar will be extracted and verified
49 > as LL(1) by an automated grammar-analysis tool, and further tested against the
50 > Rust sources. Preliminary versions of this automation exist, but are not yet
55 Rust's grammar is defined over Unicode codepoints, each conventionally
56 denoted `U+XXXX`, for 4 or more hexadecimal digits `X`. _Most_ of Rust's
57 grammar is confined to the ASCII range of Unicode, and is described in this
58 document by a dialect of Extended Backus-Naur Form (EBNF), specifically a
59 dialect of EBNF supported by common automated LL(k) parsing tools such as
60 `llgen`, rather than the dialect given in ISO 14977. The dialect can be
61 defined self-referentially as follows:
63 ~~~~ {.notrust .ebnf .notation}
65 rule : nonterminal ':' productionrule ';' ;
66 productionrule : production [ '|' production ] * ;
68 term : element repeats ;
69 element : LITERAL | IDENTIFIER | '[' productionrule ']' ;
70 repeats : [ '*' | '+' ] NUMBER ? | NUMBER ? | '?' ;
75 - Whitespace in the grammar is ignored.
76 - Square brackets are used to group rules.
77 - `LITERAL` is a single printable ASCII character, or an escaped hexadecimal
78 ASCII code of the form `\xQQ`, in single quotes, denoting the corresponding
79 Unicode codepoint `U+00QQ`.
80 - `IDENTIFIER` is a nonempty string of ASCII letters and underscores.
81 - The `repeat` forms apply to the adjacent `element`, and are as follows:
82 - `?` means zero or one repetition
83 - `*` means zero or more repetitions
84 - `+` means one or more repetitions
85 - NUMBER trailing a repeat symbol gives a maximum repetition count
86 - NUMBER on its own gives an exact repetition count
88 This EBNF dialect should hopefully be familiar to many readers.
90 ## Unicode productions
92 A few productions in Rust's grammar permit Unicode codepoints outside the ASCII range.
93 We define these productions in terms of character properties specified in the Unicode standard,
94 rather than in terms of ASCII-range codepoints.
95 The section [Special Unicode Productions](#special-unicode-productions) lists these productions.
97 ## String table productions
99 Some rules in the grammar -- notably [unary
100 operators](#unary-operator-expressions), [binary
101 operators](#binary-operator-expressions), and [keywords](#keywords) --
102 are given in a simplified form: as a listing of a table of unquoted,
103 printable whitespace-separated strings. These cases form a subset of
104 the rules regarding the [token](#tokens) rule, and are assumed to be
105 the result of a lexical-analysis phase feeding the parser, driven by a
106 DFA, operating over the disjunction of all such string table entries.
108 When such a string enclosed in double-quotes (`"`) occurs inside the
109 grammar, it is an implicit reference to a single member of such a string table
110 production. See [tokens](#tokens) for more information.
116 Rust input is interpreted as a sequence of Unicode codepoints encoded in UTF-8,
117 normalized to Unicode normalization form NFKC.
118 Most Rust grammar rules are defined in terms of printable ASCII-range codepoints,
119 but a small number are defined in terms of Unicode properties or explicit codepoint lists.
120 ^[Substitute definitions for the special Unicode productions are provided to the grammar verifier, restricted to ASCII range, when verifying the grammar in this document.]
122 ## Special Unicode Productions
124 The following productions in the Rust grammar are defined in terms of Unicode properties:
125 `ident`, `non_null`, `non_star`, `non_eol`, `non_slash_or_star`, `non_single_quote` and `non_double_quote`.
129 The `ident` production is any nonempty Unicode string of the following form:
131 - The first character has property `XID_start`
132 - The remaining characters have property `XID_continue`
134 that does _not_ occur in the set of [keywords](#keywords).
136 Note: `XID_start` and `XID_continue` as character properties cover the
137 character ranges used to form the more familiar C and Java language-family
140 ### Delimiter-restricted productions
142 Some productions are defined by exclusion of particular Unicode characters:
144 - `non_null` is any single Unicode character aside from `U+0000` (null)
145 - `non_eol` is `non_null` restricted to exclude `U+000A` (`'\n'`)
146 - `non_star` is `non_null` restricted to exclude `U+002A` (`*`)
147 - `non_slash_or_star` is `non_null` restricted to exclude `U+002F` (`/`) and `U+002A` (`*`)
148 - `non_single_quote` is `non_null` restricted to exclude `U+0027` (`'`)
149 - `non_double_quote` is `non_null` restricted to exclude `U+0022` (`"`)
153 ~~~~ {.notrust .ebnf .gram}
154 comment : block_comment | line_comment ;
155 block_comment : "/*" block_comment_body * '*' + '/' ;
156 block_comment_body : [block_comment | character] * ;
157 line_comment : "//" non_eol * ;
160 Comments in Rust code follow the general C++ style of line and block-comment forms,
161 with no nesting of block-comment delimiters.
163 Line comments beginning with exactly _three_ slashes (`///`), and block
164 comments beginning with a exactly one repeated asterisk in the block-open
165 sequence (`/**`), are interpreted as a special syntax for `doc`
166 [attributes](#attributes). That is, they are equivalent to writing
167 `#[doc="..."]` around the body of the comment (this includes the comment
168 characters themselves, ie `/// Foo` turns into `#[doc="/// Foo"]`).
170 Non-doc comments are interpreted as a form of whitespace.
174 ~~~~ {.notrust .ebnf .gram}
175 whitespace_char : '\x20' | '\x09' | '\x0a' | '\x0d' ;
176 whitespace : [ whitespace_char | comment ] + ;
179 The `whitespace_char` production is any nonempty Unicode string consisting of any
180 of the following Unicode characters: `U+0020` (space, `' '`), `U+0009` (tab,
181 `'\t'`), `U+000A` (LF, `'\n'`), `U+000D` (CR, `'\r'`).
183 Rust is a "free-form" language, meaning that all forms of whitespace serve
184 only to separate _tokens_ in the grammar, and have no semantic significance.
186 A Rust program has identical meaning if each whitespace element is replaced
187 with any other legal whitespace element, such as a single space character.
191 ~~~~ {.notrust .ebnf .gram}
192 simple_token : keyword | unop | binop ;
193 token : simple_token | ident | literal | symbol | whitespace token ;
196 Tokens are primitive productions in the grammar defined by regular
197 (non-recursive) languages. "Simple" tokens are given in [string table
198 production](#string-table-productions) form, and occur in the rest of the
199 grammar as double-quoted strings. Other tokens have exact rules given.
203 The keywords are the following strings:
205 ~~~~ {.notrust .keyword}
217 self static struct super
223 Each of these keywords has special meaning in its grammar,
224 and all of them are excluded from the `ident` rule.
228 A literal is an expression consisting of a single token, rather than a
229 sequence of tokens, that immediately and directly denotes the value it
230 evaluates to, rather than referring to it by name or some other evaluation
231 rule. A literal is a form of constant expression, so is evaluated (primarily)
234 ~~~~ {.notrust .ebnf .gram}
235 literal : string_lit | char_lit | num_lit ;
238 #### Character and string literals
240 ~~~~ {.notrust .ebnf .gram}
241 char_lit : '\x27' char_body '\x27' ;
242 string_lit : '"' string_body * '"' | 'r' raw_string ;
244 char_body : non_single_quote
245 | '\x5c' [ '\x27' | common_escape ] ;
247 string_body : non_double_quote
248 | '\x5c' [ '\x22' | common_escape ] ;
249 raw_string : '"' raw_string_body '"' | '#' raw_string '#' ;
251 common_escape : '\x5c'
252 | 'n' | 'r' | 't' | '0'
257 hex_digit : 'a' | 'b' | 'c' | 'd' | 'e' | 'f'
258 | 'A' | 'B' | 'C' | 'D' | 'E' | 'F'
260 oct_digit : '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' ;
261 dec_digit : '0' | nonzero_dec ;
262 nonzero_dec: '1' | '2' | '3' | '4'
263 | '5' | '6' | '7' | '8' | '9' ;
266 A _character literal_ is a single Unicode character enclosed within two
267 `U+0027` (single-quote) characters, with the exception of `U+0027` itself,
268 which must be _escaped_ by a preceding U+005C character (`\`).
270 A _string literal_ is a sequence of any Unicode characters enclosed within
271 two `U+0022` (double-quote) characters, with the exception of `U+0022`
272 itself, which must be _escaped_ by a preceding `U+005C` character (`\`),
273 or a _raw string literal_.
275 Some additional _escapes_ are available in either character or non-raw string
276 literals. An escape starts with a `U+005C` (`\`) and continues with one of
279 * An _8-bit codepoint escape_ escape starts with `U+0078` (`x`) and is
280 followed by exactly two _hex digits_. It denotes the Unicode codepoint
281 equal to the provided hex value.
282 * A _16-bit codepoint escape_ starts with `U+0075` (`u`) and is followed
283 by exactly four _hex digits_. It denotes the Unicode codepoint equal to
284 the provided hex value.
285 * A _32-bit codepoint escape_ starts with `U+0055` (`U`) and is followed
286 by exactly eight _hex digits_. It denotes the Unicode codepoint equal to
287 the provided hex value.
288 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
289 (`r`), or `U+0074` (`t`), denoting the unicode values `U+000A` (LF),
290 `U+000D` (CR) or `U+0009` (HT) respectively.
291 * The _backslash escape_ is the character `U+005C` (`\`) which must be
292 escaped in order to denote *itself*.
294 Raw string literals do not process any escapes. They start with the character
295 `U+0072` (`r`), followed zero or more of the character `U+0023` (`#`) and a
296 `U+0022` (double-quote) character. The _raw string body_ is not defined in the
297 EBNF grammar above: it can contain any sequence of Unicode characters and is
298 terminated only by another `U+0022` (double-quote) character, followed by the
299 same number of `U+0023` (`#`) characters that preceeded the opening `U+0022`
300 (double-quote) character.
302 All Unicode characters contained in the raw string body represent themselves,
303 the characters `U+0022` (double-quote) (except when followed by at least as
304 many `U+0023` (`#`) characters as were used to start the raw string literal) or
305 `U+005C` (`\`) do not have any special meaning.
307 Examples for string literals:
310 "foo"; r"foo"; // foo
311 "\"foo\""; r#""foo""#; // "foo"
314 r##"foo #"# bar"##; // foo #"# bar
316 "\x52"; "R"; r"R"; // R
317 "\\x52"; r"\x52"; // \x52
322 ~~~~ {.notrust .ebnf .gram}
323 num_lit : nonzero_dec [ dec_digit | '_' ] * num_suffix ?
324 | '0' [ [ dec_digit | '_' ] * num_suffix ?
325 | 'b' [ '1' | '0' | '_' ] + int_suffix ?
326 | 'o' [ oct_digit | '_' ] + int_suffix ?
327 | 'x' [ hex_digit | '_' ] + int_suffix ? ] ;
329 num_suffix : int_suffix | float_suffix ;
331 int_suffix : 'u' int_suffix_size ?
332 | 'i' int_suffix_size ? ;
333 int_suffix_size : [ '8' | '1' '6' | '3' '2' | '6' '4' ] ;
335 float_suffix : [ exponent | '.' dec_lit exponent ? ] ? float_suffix_ty ? ;
336 float_suffix_ty : 'f' [ '3' '2' | '6' '4' ] ;
337 exponent : ['E' | 'e'] ['-' | '+' ] ? dec_lit ;
338 dec_lit : [ dec_digit | '_' ] + ;
341 A _number literal_ is either an _integer literal_ or a _floating-point
342 literal_. The grammar for recognizing the two kinds of literals is mixed,
343 as they are differentiated by suffixes.
345 ##### Integer literals
347 An _integer literal_ has one of four forms:
349 * A _decimal literal_ starts with a *decimal digit* and continues with any
350 mixture of *decimal digits* and _underscores_.
351 * A _hex literal_ starts with the character sequence `U+0030` `U+0078`
352 (`0x`) and continues as any mixture hex digits and underscores.
353 * An _octal literal_ starts with the character sequence `U+0030` `U+006F`
354 (`0o`) and continues as any mixture octal digits and underscores.
355 * A _binary literal_ starts with the character sequence `U+0030` `U+0062`
356 (`0b`) and continues as any mixture binary digits and underscores.
358 An integer literal may be followed (immediately, without any spaces) by an
359 _integer suffix_, which changes the type of the literal. There are two kinds
360 of integer literal suffix:
362 * The `i` and `u` suffixes give the literal type `int` or `uint`,
364 * Each of the signed and unsigned machine types `u8`, `i8`,
365 `u16`, `i16`, `u32`, `i32`, `u64` and `i64`
366 give the literal the corresponding machine type.
368 The type of an _unsuffixed_ integer literal is determined by type inference.
369 If a integer type can be _uniquely_ determined from the surrounding program
370 context, the unsuffixed integer literal has that type. If the program context
371 underconstrains the type, the unsuffixed integer literal's type is `int`; if
372 the program context overconstrains the type, it is considered a static type
375 Examples of integer literals of various forms:
378 123; 0xff00; // type determined by program context
379 // defaults to int in absence of type
385 0o70_i16; // type i16
386 0b1111_1111_1001_0000_i32; // type i32
389 ##### Floating-point literals
391 A _floating-point literal_ has one of two forms:
393 * Two _decimal literals_ separated by a period
394 character `U+002E` (`.`), with an optional _exponent_ trailing after the
395 second decimal literal.
396 * A single _decimal literal_ followed by an _exponent_.
398 By default, a floating-point literal has a generic type, but will fall back to
399 `f64`. A floating-point literal may be followed (immediately, without any
400 spaces) by a _floating-point suffix_, which changes the type of the literal.
401 There are two floating-point suffixes: `f32`, and `f64` (the 32-bit and 64-bit
402 floating point types).
404 Examples of floating-point literals of various forms:
410 12E+99_f64; // type f64
413 ##### Unit and boolean literals
415 The _unit value_, the only value of the type that has the same name, is written as `()`.
416 The two values of the boolean type are written `true` and `false`.
420 ~~~~ {.notrust .ebnf .gram}
422 | '#' | '[' | ']' | '(' | ')' | '{' | '}'
426 Symbols are a general class of printable [token](#tokens) that play structural
427 roles in a variety of grammar productions. They are catalogued here for
428 completeness as the set of remaining miscellaneous printable tokens that do not
429 otherwise appear as [unary operators](#unary-operator-expressions), [binary
430 operators](#binary-operator-expressions), or [keywords](#keywords).
435 ~~~~ {.notrust .ebnf .gram}
436 expr_path : ident [ "::" expr_path_tail ] + ;
437 expr_path_tail : '<' type_expr [ ',' type_expr ] + '>'
440 type_path : ident [ type_path_tail ] + ;
441 type_path_tail : '<' type_expr [ ',' type_expr ] + '>'
445 A _path_ is a sequence of one or more path components _logically_ separated by
446 a namespace qualifier (`::`). If a path consists of only one component, it may
447 refer to either an [item](#items) or a [slot](#memory-slots) in a local
448 control scope. If a path has multiple components, it refers to an item.
450 Every item has a _canonical path_ within its crate, but the path naming an
451 item is only meaningful within a given crate. There is no global namespace
452 across crates; an item's canonical path merely identifies it within the crate.
454 Two examples of simple paths consisting of only identifier components:
461 Path components are usually [identifiers](#identifiers), but the trailing
462 component of a path may be an angle-bracket-enclosed list of type
463 arguments. In [expression](#expressions) context, the type argument list is
464 given after a final (`::`) namespace qualifier in order to disambiguate it
465 from a relational expression involving the less-than symbol (`<`). In type
466 expression context, the final namespace qualifier is omitted.
468 Two examples of paths with type arguments:
471 # struct HashMap<K, V>;
473 # fn id<T>(t: T) -> T { t }
474 type T = HashMap<int,~str>; // Type arguments used in a type expression
475 let x = id::<int>(10); // Type arguments used in a call expression
481 A number of minor features of Rust are not central enough to have their own
482 syntax, and yet are not implementable as functions. Instead, they are given
483 names, and invoked through a consistent syntax: `name!(...)`. Examples
486 * `format!` : format data into a string
487 * `env!` : look up an environment variable's value at compile time
488 * `file!`: return the path to the file being compiled
489 * `stringify!` : pretty-print the Rust expression given as an argument
490 * `include!` : include the Rust expression in the given file
491 * `include_str!` : include the contents of the given file as a string
492 * `include_bin!` : include the contents of the given file as a binary blob
493 * `error!`, `warn!`, `info!`, `debug!` : provide diagnostic information.
495 All of the above extensions are expressions with values.
499 ~~~~ {.notrust .ebnf .gram}
500 expr_macro_rules : "macro_rules" '!' ident '(' macro_rule * ')' ;
501 macro_rule : '(' matcher * ')' "=>" '(' transcriber * ')' ';' ;
502 matcher : '(' matcher * ')' | '[' matcher * ']'
503 | '{' matcher * '}' | '$' ident ':' ident
504 | '$' '(' matcher * ')' sep_token? [ '*' | '+' ]
505 | non_special_token ;
506 transcriber : '(' transcriber * ')' | '[' transcriber * ']'
507 | '{' transcriber * '}' | '$' ident
508 | '$' '(' transcriber * ')' sep_token? [ '*' | '+' ]
509 | non_special_token ;
512 User-defined syntax extensions are called "macros",
513 and the `macro_rules` syntax extension defines them.
514 Currently, user-defined macros can expand to expressions, statements, or items.
516 (A `sep_token` is any token other than `*` and `+`.
517 A `non_special_token` is any token other than a delimiter or `$`.)
519 The macro expander looks up macro invocations by name,
520 and tries each macro rule in turn.
521 It transcribes the first successful match.
522 Matching and transcription are closely related to each other,
523 and we will describe them together.
527 The macro expander matches and transcribes every token that does not begin with a `$` literally, including delimiters.
528 For parsing reasons, delimiters must be balanced, but they are otherwise not special.
530 In the matcher, `$` _name_ `:` _designator_ matches the nonterminal in the
531 Rust syntax named by _designator_. Valid designators are `item`, `block`,
532 `stmt`, `pat`, `expr`, `ty` (type), `ident`, `path`, `matchers` (lhs of the `=>` in macro rules),
533 `tt` (rhs of the `=>` in macro rules). In the transcriber, the designator is already known, and so only
534 the name of a matched nonterminal comes after the dollar sign.
536 In both the matcher and transcriber, the Kleene star-like operator indicates repetition.
537 The Kleene star operator consists of `$` and parens, optionally followed by a separator token, followed by `*` or `+`.
538 `*` means zero or more repetitions, `+` means at least one repetition.
539 The parens are not matched or transcribed.
540 On the matcher side, a name is bound to _all_ of the names it
541 matches, in a structure that mimics the structure of the repetition
542 encountered on a successful match. The job of the transcriber is to sort that
545 The rules for transcription of these repetitions are called "Macro By Example".
546 Essentially, one "layer" of repetition is discharged at a time, and all of
547 them must be discharged by the time a name is transcribed. Therefore,
548 `( $( $i:ident ),* ) => ( $i )` is an invalid macro, but
549 `( $( $i:ident ),* ) => ( $( $i:ident ),* )` is acceptable (if trivial).
551 When Macro By Example encounters a repetition, it examines all of the `$`
552 _name_ s that occur in its body. At the "current layer", they all must repeat
553 the same number of times, so
554 ` ( $( $i:ident ),* ; $( $j:ident ),* ) => ( $( ($i,$j) ),* )` is valid if
555 given the argument `(a,b,c ; d,e,f)`, but not `(a,b,c ; d,e)`. The repetition
556 walks through the choices at that layer in lockstep, so the former input
557 transcribes to `( (a,d), (b,e), (c,f) )`.
559 Nested repetitions are allowed.
561 ### Parsing limitations
563 The parser used by the macro system is reasonably powerful, but the parsing of
564 Rust syntax is restricted in two ways:
566 1. The parser will always parse as much as possible. If it attempts to match
567 `$i:expr [ , ]` against `8 [ , ]`, it will attempt to parse `i` as an array
568 index operation and fail. Adding a separator can solve this problem.
569 2. The parser must have eliminated all ambiguity by the time it reaches a `$` _name_ `:` _designator_.
570 This requirement most often affects name-designator pairs when they occur at the beginning of, or immediately after, a `$(...)*`; requiring a distinctive token in front can solve the problem.
572 ## Syntax extensions useful for the macro author
574 * `log_syntax!` : print out the arguments at compile time
575 * `trace_macros!` : supply `true` or `false` to enable or disable macro expansion logging
576 * `stringify!` : turn the identifier argument into a string literal
577 * `concat!` : concatenates a comma-separated list of literals
578 * `concat_idents!` : create a new identifier by concatenating the arguments
580 # Crates and source files
582 Rust is a *compiled* language.
583 Its semantics obey a *phase distinction* between compile-time and run-time.
584 Those semantic rules that have a *static interpretation* govern the success or failure of compilation.
585 We refer to these rules as "static semantics".
586 Semantic rules called "dynamic semantics" govern the behavior of programs at run-time.
587 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.
589 The compilation model centres on artifacts called _crates_.
590 Each compilation processes a single crate in source form, and if successful, produces a single crate in binary form: either an executable or a library.^[A crate is somewhat
591 analogous to an *assembly* in the ECMA-335 CLI model, a *library* in the
592 SML/NJ Compilation Manager, a *unit* in the Owens and Flatt module system,
593 or a *configuration* in Mesa.]
595 A _crate_ is a unit of compilation and linking, as well as versioning, distribution and runtime loading.
596 A crate contains a _tree_ of nested [module](#modules) scopes.
597 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.
599 The Rust compiler is always invoked with a single source file as input, and always produces a single output crate.
600 The processing of that source file may result in other source files being loaded as modules.
601 Source files have the extension `.rs`.
603 A Rust source file describes a module, the name and
604 location of which -- in the module tree of the current crate -- are defined
605 from outside the source file: either by an explicit `mod_item` in
606 a referencing source file, or by the name of the crate itself.
608 Each source file contains a sequence of zero or more `item` definitions,
609 and may optionally begin with any number of `attributes` that apply to the containing module.
610 Attributes on the anonymous crate module define important metadata that influences
611 the behavior of the compiler.
615 #![crate_id = "projx#2.5"]
617 // Additional metadata attributes
618 #![desc = "Project X"]
620 #![comment = "This is a comment on Project X."]
622 // Specify the output type
623 #![crate_type = "lib"]
626 #![warn(non_camel_case_types)]
629 A crate that contains a `main` function can be compiled to an executable.
630 If a `main` function is present, its return type must be [`unit`](#primitive-types) and it must take no arguments.
632 # Items and attributes
634 Crates contain [items](#items),
635 each of which may have some number of [attributes](#attributes) attached to it.
639 ~~~~ {.notrust .ebnf .gram}
640 item : mod_item | fn_item | type_item | struct_item | enum_item
641 | static_item | trait_item | impl_item | extern_block ;
644 An _item_ is a component of a crate; some module items can be defined in crate
645 files, but most are defined in source files. Items are organized within a
646 crate by a nested set of [modules](#modules). Every crate has a single
647 "outermost" anonymous module; all further items within the crate have
648 [paths](#paths) within the module tree of the crate.
650 Items are entirely determined at compile-time, generally remain fixed during
651 execution, and may reside in read-only memory.
653 There are several kinds of item:
655 * [modules](#modules)
656 * [functions](#functions)
657 * [type definitions](#type-definitions)
658 * [structures](#structures)
659 * [enumerations](#enumerations)
660 * [static items](#static-items)
662 * [implementations](#implementations)
664 Some items form an implicit scope for the declaration of sub-items. In other
665 words, within a function or module, declarations of items can (in many cases)
666 be mixed with the statements, control blocks, and similar artifacts that
667 otherwise compose the item body. The meaning of these scoped items is the same
668 as if the item was declared outside the scope -- it is still a static item --
669 except that the item's *path name* within the module namespace is qualified by
670 the name of the enclosing item, or is private to the enclosing item (in the
672 The grammar specifies the exact locations in which sub-item declarations may appear.
676 All items except modules may be *parameterized* by type. Type parameters are
677 given as a comma-separated list of identifiers enclosed in angle brackets
678 (`<...>`), after the name of the item and before its definition.
679 The type parameters of an item are considered "part of the name", not part of the type of the item.
680 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.
681 In practice, the type-inference system can usually infer such argument types from context.
682 There are no general type-parametric types, only type-parametric items.
683 That is, Rust has no notion of type abstraction: there are no first-class "forall" types.
687 ~~~~ {.notrust .ebnf .gram}
688 mod_item : "mod" ident ( ';' | '{' mod '}' );
689 mod : [ view_item | item ] * ;
692 A module is a container for zero or more [view items](#view-items) and zero or
693 more [items](#items). The view items manage the visibility of the items
694 defined within the module, as well as the visibility of names from outside the
695 module when referenced from inside the module.
697 A _module item_ is a module, surrounded in braces, named, and prefixed with
698 the keyword `mod`. A module item introduces a new, named module into the tree
699 of modules making up a crate. Modules can nest arbitrarily.
701 An example of a module:
705 type Complex = (f64, f64);
706 fn sin(f: f64) -> f64 {
710 fn cos(f: f64) -> f64 {
714 fn tan(f: f64) -> f64 {
721 Modules and types share the same namespace.
722 Declaring a named type that has the same name as a module in scope is forbidden:
723 that is, a type definition, trait, struct, enumeration, or type parameter
724 can't shadow the name of a module in scope, or vice versa.
726 A module without a body is loaded from an external file, by default with the same
727 name as the module, plus the `.rs` extension.
728 When a nested submodule is loaded from an external file,
729 it is loaded from a subdirectory path that mirrors the module hierarchy.
732 // Load the `vec` module from `vec.rs`
736 // Load the `local_data` module from `task/local_data.rs`
741 The directories and files used for loading external file modules can be influenced
742 with the `path` attribute.
745 #[path = "task_files"]
747 // Load the `local_data` module from `task_files/tls.rs`
755 ~~~~ {.notrust .ebnf .gram}
756 view_item : extern_crate_decl | use_decl ;
759 A view item manages the namespace of a module.
760 View items do not define new items, but rather, simply change other items' visibility.
761 There are several kinds of view item:
763 * [`extern crate` declarations](#extern-crate-declarations)
764 * [`use` declarations](#use-declarations)
766 ##### Extern crate declarations
768 ~~~~ {.notrust .ebnf .gram}
769 extern_crate_decl : "extern" "crate" ident [ '(' link_attrs ')' ] ? [ '=' string_lit ] ? ;
770 link_attrs : link_attr [ ',' link_attrs ] + ;
771 link_attr : ident '=' literal ;
774 An _`extern crate` declaration_ specifies a dependency on an external crate.
775 The external crate is then bound into the declaring scope as the `ident` provided
776 in the `extern_crate_decl`.
778 The external crate is resolved to a specific `soname` at compile time, and a
779 runtime linkage requirement to that `soname` is passed to the linker for
780 loading at runtime. The `soname` is resolved at compile time by scanning the
781 compiler's library path and matching the optional `crateid` provided as a string literal
782 against the `crateid` attributes that were declared on the external crate when
783 it was compiled. If no `crateid` is provided, a default `name` attribute is
784 assumed, equal to the `ident` given in the `extern_crate_decl`.
786 Four examples of `extern crate` declarations:
791 extern crate std; // equivalent to: extern crate std = "std";
793 extern crate ruststd = "std"; // linking to 'std' under another name
795 extern crate foo = "some/where/rust-foo#foo:1.0"; // a full crate ID for external tools
798 ##### Use declarations
800 ~~~~ {.notrust .ebnf .gram}
801 use_decl : "pub" ? "use" ident [ '=' path
804 path_glob : ident [ "::" path_glob ] ?
806 | '{' ident [ ',' ident ] * '}' ;
809 A _use declaration_ creates one or more local name bindings synonymous
810 with some other [path](#paths).
811 Usually a `use` declaration is used to shorten the path required to refer to a
812 module item. These declarations may appear at the top of [modules](#modules) and
815 *Note*: Unlike in many languages,
816 `use` declarations in Rust do *not* declare linkage dependency with external crates.
817 Rather, [`extern crate` declarations](#extern-crate-declarations) declare linkage dependencies.
819 Use declarations support a number of convenient shortcuts:
821 * Rebinding the target name as a new local name, using the syntax `use x = p::q::r;`.
822 * Simultaneously binding a list of paths differing only in their final element,
823 using the glob-like brace syntax `use a::b::{c,d,e,f};`
824 * Binding all paths matching a given prefix, using the asterisk wildcard syntax `use a::b::*;`
826 An example of `use` declarations:
829 use std::iter::range_step;
830 use std::option::{Some, None};
835 // Equivalent to 'std::iter::range_step(0, 10, 2);'
836 range_step(0, 10, 2);
838 // Equivalent to 'foo(~[std::option::Some(1.0), std::option::None]);'
839 foo(~[Some(1.0), None]);
843 Like items, `use` declarations are private to the containing module, by default.
844 Also like items, a `use` declaration can be public, if qualified by the `pub` keyword.
845 Such a `use` declaration serves to _re-export_ a name.
846 A public `use` declaration can therefore _redirect_ some public name to a different target definition:
847 even a definition with a private canonical path, inside a different module.
848 If a sequence of such redirections form a cycle or cannot be resolved unambiguously,
849 they represent a compile-time error.
851 An example of re-exporting:
856 pub use quux::foo::*;
865 In this example, the module `quux` re-exports all of the public names defined in `foo`.
867 Also note that the paths contained in `use` items are relative to the crate root.
868 So, in the previous example, the `use` refers to `quux::foo::*`, and not simply to `foo::*`.
869 This also means that top-level module declarations should be at the crate root if direct usage
870 of the declared modules within `use` items is desired. It is also possible to use `self` and `super`
871 at the beginning of a `use` item to refer to the current and direct parent modules respectively.
872 All rules regarding accessing declared modules in `use` declarations applies to both module declarations
873 and `extern crate` declarations.
875 An example of what will and will not work for `use` items:
878 # #[allow(unused_imports)];
879 use foo::native::start; // good: foo is at the root of the crate
880 use foo::baz::foobaz; // good: foo is at the root of the crate
885 use foo::native::start; // good: foo is at crate root
886 // use native::start; // bad: native is not at the crate root
887 use self::baz::foobaz; // good: self refers to module 'foo'
888 use foo::bar::foobar; // good: foo is at crate root
895 use super::bar::foobar; // good: super refers to module 'foo'
905 A _function item_ defines a sequence of [statements](#statements) and an optional final [expression](#expressions), along with a name and a set of parameters.
906 Functions are declared with the keyword `fn`.
907 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.
909 A function may also be copied into a first class *value*, in which case the
910 value has the corresponding [*function type*](#function-types), and can be
911 used otherwise exactly as a function item (with a minor additional cost of
912 calling the function indirectly).
914 Every control path in a function logically ends with a `return` expression or a
915 diverging expression. If the outermost block of a function has a
916 value-producing expression in its final-expression position, that expression
917 is interpreted as an implicit `return` expression applied to the
920 An example of a function:
923 fn add(x: int, y: int) -> int {
928 As with `let` bindings, function arguments are irrefutable patterns,
929 so any pattern that is valid in a let binding is also valid as an argument.
932 fn first((value, _): (int, int)) -> int { value }
936 #### Generic functions
938 A _generic function_ allows one or more _parameterized types_ to
939 appear in its signature. Each type parameter must be explicitly
940 declared, in an angle-bracket-enclosed, comma-separated list following
944 fn iter<T>(seq: &[T], f: |T|) {
945 for elt in seq.iter() { f(elt); }
947 fn map<T, U>(seq: &[T], f: |T| -> U) -> ~[U] {
949 for elt in seq.iter() { acc.push(f(elt)); }
954 Inside the function signature and body, the name of the type parameter
955 can be used as a type name.
957 When a generic function is referenced, its type is instantiated based
958 on the context of the reference. For example, calling the `iter`
959 function defined above on `[1, 2]` will instantiate type parameter `T`
960 with `int`, and require the closure parameter to have type
963 The type parameters can also be explicitly supplied in a trailing
964 [path](#paths) component after the function name. This might be necessary
965 if there is not sufficient context to determine the type parameters. For
966 example, `mem::size_of::<u32>() == 4`.
968 Since a parameter type is opaque to the generic function, the set of
969 operations that can be performed on it is limited. Values of parameter
970 type can only be moved, not copied.
973 fn id<T>(x: T) -> T { x }
976 Similarly, [trait](#traits) bounds can be specified for type
977 parameters to allow methods with that trait to be called on values
983 Unsafe operations are those that potentially violate the memory-safety guarantees of Rust's static semantics.
985 The following language level features cannot be used in the safe subset of Rust:
987 - Dereferencing a [raw pointer](#pointer-types).
988 - Calling an unsafe function (including an intrinsic or foreign function).
990 ##### Unsafe functions
992 Unsafe functions are functions that are not safe in all contexts and/or for all possible inputs.
993 Such a function must be prefixed with the keyword `unsafe`.
997 A block of code can also be prefixed with the `unsafe` keyword, to permit calling `unsafe` functions
998 or dereferencing raw pointers within a safe function.
1000 When a programmer has sufficient conviction that a sequence of potentially unsafe operations is
1001 actually safe, they can encapsulate that sequence (taken as a whole) within an `unsafe` block. The
1002 compiler will consider uses of such code safe, in the surrounding context.
1004 Unsafe blocks are used to wrap foreign libraries, make direct use of hardware or implement features
1005 not directly present in the language. For example, Rust provides the language features necessary to
1006 implement memory-safe concurrency in the language but the implementation of tasks and message
1007 passing is in the standard library.
1009 Rust's type system is a conservative approximation of the dynamic safety requirements, so in some
1010 cases there is a performance cost to using safe code. For example, a doubly-linked list is not a
1011 tree structure and can only be represented with managed or reference-counted pointers in safe code.
1012 By using `unsafe` blocks to represent the reverse links as raw pointers, it can be implemented with
1013 only owned pointers.
1015 ##### Behavior considered unsafe
1017 This is a list of behavior which is forbidden in all Rust code. Type checking provides the guarantee
1018 that these issues are never caused by safe code. An `unsafe` block or function is responsible for
1019 never invoking this behaviour or exposing an API making it possible for it to occur in safe code.
1022 * Dereferencing a null/dangling raw pointer
1023 * Mutating an immutable value/reference
1024 * Reads of [undef](http://llvm.org/docs/LangRef.html#undefined-values) (uninitialized) memory
1025 * Breaking the [pointer aliasing rules](http://llvm.org/docs/LangRef.html#pointer-aliasing-rules)
1026 with raw pointers (a subset of the rules used by C)
1027 * Invoking undefined behavior via compiler intrinsics:
1028 * Indexing outside of the bounds of an object with `std::ptr::offset` (`offset` intrinsic), with
1029 the exception of one byte past the end which is permitted.
1030 * Using `std::ptr::copy_nonoverlapping_memory` (`memcpy32`/`memcpy64` instrinsics) on
1032 * Invalid values in primitive types, even in private fields/locals:
1033 * Dangling/null pointers in non-raw pointers, or slices
1034 * A value other than `false` (0) or `true` (1) in a `bool`
1035 * A discriminant in an `enum` not included in the type definition
1036 * A value in a `char` which is a surrogate or above `char::MAX`
1037 * non-UTF-8 byte sequences in a `str`
1039 ##### Behaviour not considered unsafe
1041 This is a list of behaviour not considered *unsafe* in Rust terms, but that may be undesired.
1044 * Reading data from private fields (`std::repr`, `format!("{:?}", x)`)
1045 * Leaks due to reference count cycles, even in the global heap
1046 * Exiting without calling destructors
1048 * Accessing/modifying the file system
1049 * Unsigned integer overflow (well-defined as wrapping)
1050 * Signed integer overflow (well-defined as two's complement representation wrapping)
1052 #### Diverging functions
1054 A special kind of function can be declared with a `!` character where the
1055 output slot type would normally be. For example:
1058 fn my_err(s: &str) -> ! {
1064 We call such functions "diverging" because they never return a value to the
1065 caller. Every control path in a diverging function must end with a
1066 `fail!()` or a call to another diverging function on every
1067 control path. The `!` annotation does *not* denote a type. Rather, the result
1068 type of a diverging function is a special type called $\bot$ ("bottom") that
1069 unifies with any type. Rust has no syntax for $\bot$.
1071 It might be necessary to declare a diverging function because as mentioned
1072 previously, the typechecker checks that every control path in a function ends
1073 with a [`return`](#return-expressions) or diverging expression. So, if `my_err`
1074 were declared without the `!` annotation, the following code would not
1078 # fn my_err(s: &str) -> ! { fail!() }
1080 fn f(i: int) -> int {
1085 my_err("Bad number!");
1090 This will not compile without the `!` annotation on `my_err`,
1091 since the `else` branch of the conditional in `f` does not return an `int`,
1092 as required by the signature of `f`.
1093 Adding the `!` annotation to `my_err` informs the typechecker that,
1094 should control ever enter `my_err`, no further type judgments about `f` need to hold,
1095 since control will never resume in any context that relies on those judgments.
1096 Thus the return type on `f` only needs to reflect the `if` branch of the conditional.
1099 #### Extern functions
1101 Extern functions are part of Rust's foreign function interface,
1102 providing the opposite functionality to [external blocks](#external-blocks).
1103 Whereas external blocks allow Rust code to call foreign code,
1104 extern functions with bodies defined in Rust code _can be called by foreign
1105 code_. They are defined in the same way as any other Rust function,
1106 except that they have the `extern` modifier.
1109 // Declares an extern fn, the ABI defaults to "C"
1110 extern fn new_vec() -> ~[int] { ~[] }
1112 // Declares an extern fn with "stdcall" ABI
1113 extern "stdcall" fn new_vec_stdcall() -> ~[int] { ~[] }
1116 Unlike normal functions, extern fns have an `extern "ABI" fn()`.
1117 This is the same type as the functions declared in an extern
1121 # extern fn new_vec() -> ~[int] { ~[] }
1122 let fptr: extern "C" fn() -> ~[int] = new_vec;
1125 Extern functions may be called directly from Rust code as Rust uses large,
1126 contiguous stack segments like C.
1128 ### Type definitions
1130 A _type definition_ defines a new name for an existing [type](#types). Type
1131 definitions are declared with the keyword `type`. Every value has a single,
1132 specific type; the type-specified aspects of a value include:
1134 * Whether the value is composed of sub-values or is indivisible.
1135 * Whether the value represents textual or numerical information.
1136 * Whether the value represents integral or floating-point information.
1137 * The sequence of memory operations required to access the value.
1138 * The [kind](#type-kinds) of the type.
1140 For example, the type `(u8, u8)` defines the set of immutable values that are composite pairs,
1141 each containing two unsigned 8-bit integers accessed by pattern-matching and laid out in memory with the `x` component preceding the `y` component.
1145 A _structure_ is a nominal [structure type](#structure-types) defined with the keyword `struct`.
1147 An example of a `struct` item and its use:
1150 struct Point {x: int, y: int}
1151 let p = Point {x: 10, y: 11};
1155 A _tuple structure_ is a nominal [tuple type](#tuple-types), also defined with the keyword `struct`.
1159 struct Point(int, int);
1160 let p = Point(10, 11);
1161 let px: int = match p { Point(x, _) => x };
1164 A _unit-like struct_ is a structure without any fields, defined by leaving off the list of fields entirely.
1165 Such types will have a single value, just like the [unit value `()`](#unit-and-boolean-literals) of the unit type.
1170 let c = [Cookie, Cookie, Cookie, Cookie];
1175 An _enumeration_ is a simultaneous definition of a nominal [enumerated type](#enumerated-types) as well as a set of *constructors*,
1176 that can be used to create or pattern-match values of the corresponding enumerated type.
1178 Enumerations are declared with the keyword `enum`.
1180 An example of an `enum` item and its use:
1188 let mut a: Animal = Dog;
1192 Enumeration constructors can have either named or unnamed fields:
1197 Cat { name: ~str, weight: f64 }
1200 let mut a: Animal = Dog(~"Cocoa", 37.2);
1201 a = Cat{ name: ~"Spotty", weight: 2.7 };
1204 In this example, `Cat` is a _struct-like enum variant_,
1205 whereas `Dog` is simply called an enum variant.
1209 ~~~~ {.notrust .ebnf .gram}
1210 static_item : "static" ident ':' type '=' expr ';' ;
1213 A *static item* is a named _constant value_ stored in the global data section of a crate.
1214 Immutable static items are stored in the read-only data section.
1215 The constant value bound to a static item is, like all constant values, evaluated at compile time.
1216 Static items have the `static` lifetime, which outlives all other lifetimes in a Rust program.
1217 Static items are declared with the `static` keyword.
1218 A static item must have a _constant expression_ giving its definition.
1220 Static items must be explicitly typed.
1221 The type may be ```bool```, ```char```, a number, or a type derived from those primitive types.
1222 The derived types are references with the `static` lifetime,
1223 fixed-size arrays, tuples, and structs.
1226 static BIT1: uint = 1 << 0;
1227 static BIT2: uint = 1 << 1;
1229 static BITS: [uint, ..2] = [BIT1, BIT2];
1230 static STRING: &'static str = "bitstring";
1232 struct BitsNStrings<'a> {
1233 mybits: [uint, ..2],
1237 static bits_n_strings: BitsNStrings<'static> = BitsNStrings {
1243 #### Mutable statics
1245 If a static item is declared with the ```mut``` keyword, then it is allowed to
1246 be modified by the program. One of Rust's goals is to make concurrency bugs hard
1247 to run into, and this is obviously a very large source of race conditions or
1248 other bugs. For this reason, an ```unsafe``` block is required when either
1249 reading or writing a mutable static variable. Care should be taken to ensure
1250 that modifications to a mutable static are safe with respect to other tasks
1251 running in the same process.
1253 Mutable statics are still very useful, however. They can be used with C
1254 libraries and can also be bound from C libraries (in an ```extern``` block).
1257 # fn atomic_add(_: &mut uint, _: uint) -> uint { 2 }
1259 static mut LEVELS: uint = 0;
1261 // This violates the idea of no shared state, and this doesn't internally
1262 // protect against races, so this function is `unsafe`
1263 unsafe fn bump_levels_unsafe1() -> uint {
1269 // Assuming that we have an atomic_add function which returns the old value,
1270 // this function is "safe" but the meaning of the return value may not be what
1271 // callers expect, so it's still marked as `unsafe`
1272 unsafe fn bump_levels_unsafe2() -> uint {
1273 return atomic_add(&mut LEVELS, 1);
1279 A _trait_ describes a set of method types.
1281 Traits can include default implementations of methods,
1282 written in terms of some unknown [`self` type](#self-types);
1283 the `self` type may either be completely unspecified,
1284 or constrained by some other trait.
1286 Traits are implemented for specific types through separate [implementations](#implementations).
1289 # type Surface = int;
1290 # type BoundingBox = int;
1293 fn draw(&self, Surface);
1294 fn bounding_box(&self) -> BoundingBox;
1298 This defines a trait with two methods.
1299 All values that have [implementations](#implementations) of this trait in scope can have their `draw` and `bounding_box` methods called,
1300 using `value.bounding_box()` [syntax](#method-call-expressions).
1302 Type parameters can be specified for a trait to make it generic.
1303 These appear after the trait name, using the same syntax used in [generic functions](#generic-functions).
1307 fn len(&self) -> uint;
1308 fn elt_at(&self, n: uint) -> T;
1309 fn iter(&self, |T|);
1313 Generic functions may use traits as _bounds_ on their type parameters.
1314 This will have two effects: only types that have the trait may instantiate the parameter,
1315 and within the generic function,
1316 the methods of the trait can be called on values that have the parameter's type.
1320 # type Surface = int;
1321 # trait Shape { fn draw(&self, Surface); }
1323 fn draw_twice<T: Shape>(surface: Surface, sh: T) {
1329 Traits also define an [object type](#object-types) with the same name as the trait.
1330 Values of this type are created by [casting](#type-cast-expressions) pointer values
1331 (pointing to a type for which an implementation of the given trait is in scope)
1332 to pointers to the trait name, used as a type.
1336 # impl Shape for int { }
1339 let myshape: ~Shape = ~mycircle as ~Shape;
1342 The resulting value is a managed box containing the value that was cast,
1343 along with information that identifies the methods of the implementation that was used.
1344 Values with a trait type can have [methods called](#method-call-expressions) on them,
1345 for any method in the trait,
1346 and can be used to instantiate type parameters that are bounded by the trait.
1348 Trait methods may be static,
1349 which means that they lack a `self` argument.
1350 This means that they can only be called with function call syntax (`f(x)`)
1351 and not method call syntax (`obj.f()`).
1352 The way to refer to the name of a static method is to qualify it with the trait name,
1353 treating the trait name like a module.
1358 fn from_int(n: int) -> Self;
1361 fn from_int(n: int) -> f64 { n as f64 }
1363 let x: f64 = Num::from_int(42);
1366 Traits may inherit from other traits. For example, in
1369 trait Shape { fn area() -> f64; }
1370 trait Circle : Shape { fn radius() -> f64; }
1373 the syntax `Circle : Shape` means that types that implement `Circle` must also have an implementation for `Shape`.
1374 Multiple supertraits are separated by spaces, `trait Circle : Shape Eq { }`.
1375 In an implementation of `Circle` for a given type `T`, methods can refer to `Shape` methods,
1376 since the typechecker checks that any type with an implementation of `Circle` also has an implementation of `Shape`.
1378 In type-parameterized functions,
1379 methods of the supertrait may be called on values of subtrait-bound type parameters.
1380 Referring to the previous example of `trait Circle : Shape`:
1383 # trait Shape { fn area(&self) -> f64; }
1384 # trait Circle : Shape { fn radius(&self) -> f64; }
1385 fn radius_times_area<T: Circle>(c: T) -> f64 {
1386 // `c` is both a Circle and a Shape
1387 c.radius() * c.area()
1391 Likewise, supertrait methods may also be called on trait objects.
1394 # trait Shape { fn area(&self) -> f64; }
1395 # trait Circle : Shape { fn radius(&self) -> f64; }
1396 # impl Shape for int { fn area(&self) -> f64 { 0.0 } }
1397 # impl Circle for int { fn radius(&self) -> f64 { 0.0 } }
1400 let mycircle: Circle = ~mycircle as ~Circle;
1401 let nonsense = mycircle.radius() * mycircle.area();
1406 An _implementation_ is an item that implements a [trait](#traits) for a specific type.
1408 Implementations are defined with the keyword `impl`.
1411 # struct Point {x: f64, y: f64};
1412 # type Surface = int;
1413 # struct BoundingBox {x: f64, y: f64, width: f64, height: f64};
1414 # trait Shape { fn draw(&self, Surface); fn bounding_box(&self) -> BoundingBox; }
1415 # fn do_draw_circle(s: Surface, c: Circle) { }
1422 impl Shape for Circle {
1423 fn draw(&self, s: Surface) { do_draw_circle(s, *self); }
1424 fn bounding_box(&self) -> BoundingBox {
1425 let r = self.radius;
1426 BoundingBox{x: self.center.x - r, y: self.center.y - r,
1427 width: 2.0 * r, height: 2.0 * r}
1432 It is possible to define an implementation without referring to a trait.
1433 The methods in such an implementation can only be used
1434 as direct calls on the values of the type that the implementation targets.
1435 In such an implementation, the trait type and `for` after `impl` are omitted.
1436 Such implementations are limited to nominal types (enums, structs),
1437 and the implementation must appear in the same module or a sub-module as the `self` type.
1439 When a trait _is_ specified in an `impl`,
1440 all methods declared as part of the trait must be implemented,
1441 with matching types and type parameter counts.
1443 An implementation can take type parameters,
1444 which can be different from the type parameters taken by the trait it implements.
1445 Implementation parameters are written after the `impl` keyword.
1450 impl<T> Seq<T> for ~[T] {
1453 impl Seq<bool> for u32 {
1454 /* Treat the integer as a sequence of bits */
1460 ~~~~ {.notrust .ebnf .gram}
1461 extern_block_item : "extern" '{' extern_block '}' ;
1462 extern_block : [ foreign_fn ] * ;
1465 External blocks form the basis for Rust's foreign function interface.
1466 Declarations in an external block describe symbols
1467 in external, non-Rust libraries.
1469 Functions within external blocks
1470 are declared in the same way as other Rust functions,
1471 with the exception that they may not have a body
1472 and are instead terminated by a semicolon.
1475 # use std::libc::{c_char, FILE};
1478 fn fopen(filename: *c_char, mode: *c_char) -> *FILE;
1482 Functions within external blocks may be called by Rust code,
1483 just like functions defined in Rust.
1484 The Rust compiler automatically translates
1485 between the Rust ABI and the foreign ABI.
1487 A number of [attributes](#attributes) control the behavior of external
1490 By default external blocks assume that the library they are calling
1491 uses the standard C "cdecl" ABI. Other ABIs may be specified using
1492 an `abi` string, as shown here:
1495 // Interface to the Windows API
1496 extern "stdcall" { }
1499 The `link` attribute allows the name of the library to be specified. When
1500 specified the compiler will attempt to link against the native library of the
1504 #[link(name = "crypto")]
1508 The type of a function declared in an extern block is `extern "abi" fn(A1,
1509 ..., An) -> R`, where `A1...An` are the declared types of its arguments and
1510 `R` is the declared return type.
1512 ## Visibility and Privacy
1514 These two terms are often used interchangeably, and what they are attempting to
1515 convey is the answer to the question "Can this item be used at this location?"
1517 Rust's name resolution operates on a global hierarchy of namespaces. Each level
1518 in the hierarchy can be thought of as some item. The items are one of those
1519 mentioned above, but also include external crates. Declaring or defining a new
1520 module can be thought of as inserting a new tree into the hierarchy at the
1521 location of the definition.
1523 To control whether interfaces can be used across modules, Rust checks each use
1524 of an item to see whether it should be allowed or not. This is where privacy
1525 warnings are generated, or otherwise "you used a private item of another module
1526 and weren't allowed to."
1528 By default, everything in rust is *private*, with one exception. Enum variants
1529 in a `pub` enum are also public by default. You are allowed to alter this
1530 default visibility with the `priv` keyword. When an item is declared as `pub`,
1531 it can be thought of as being accessible to the outside world. For example:
1535 // Declare a private struct
1538 // Declare a public struct with a private field
1543 // Declare a public enum with public and private variants
1545 PubliclyAccessibleState,
1546 priv PrivatelyAccessibleState
1550 With the notion of an item being either public or private, Rust allows item
1551 accesses in two cases:
1553 1. If an item is public, then it can be used externally through any of its
1555 2. If an item is private, it may be accessed by the current module and its
1558 These two cases are surprisingly powerful for creating module hierarchies
1559 exposing public APIs while hiding internal implementation details. To help
1560 explain, here's a few use cases and what they would entail.
1562 * A library developer needs to expose functionality to crates which link against
1563 their library. As a consequence of the first case, this means that anything
1564 which is usable externally must be `pub` from the root down to the destination
1565 item. Any private item in the chain will disallow external accesses.
1567 * A crate needs a global available "helper module" to itself, but it doesn't
1568 want to expose the helper module as a public API. To accomplish this, the root
1569 of the crate's hierarchy would have a private module which then internally has
1570 a "public api". Because the entire crate is a descendant of the root, then the
1571 entire local crate can access this private module through the second case.
1573 * When writing unit tests for a module, it's often a common idiom to have an
1574 immediate child of the module to-be-tested named `mod test`. This module could
1575 access any items of the parent module through the second case, meaning that
1576 internal implementation details could also be seamlessly tested from the child
1579 In the second case, it mentions that a private item "can be accessed" by the
1580 current module and its descendants, but the exact meaning of accessing an item
1581 depends on what the item is. Accessing a module, for example, would mean looking
1582 inside of it (to import more items). On the other hand, accessing a function
1583 would mean that it is invoked. Additionally, path expressions and import
1584 statements are considered to access an item in the sense that the
1585 import/expression is only valid if the destination is in the current visibility
1588 Here's an example of a program which exemplifies the three cases outlined above.
1591 // This module is private, meaning that no external crate can access this
1592 // module. Because it is private at the root of this current crate, however, any
1593 // module in the crate may access any publicly visible item in this module.
1594 mod crate_helper_module {
1596 // This function can be used by anything in the current crate
1597 pub fn crate_helper() {}
1599 // This function *cannot* be used by anything else in the crate. It is not
1600 // publicly visible outside of the `crate_helper_module`, so only this
1601 // current module and its descendants may access it.
1602 fn implementation_detail() {}
1605 // This function is "public to the root" meaning that it's available to external
1606 // crates linking against this one.
1607 pub fn public_api() {}
1609 // Similarly to 'public_api', this module is public so external crates may look
1612 use crate_helper_module;
1614 pub fn my_method() {
1615 // Any item in the local crate may invoke the helper module's public
1616 // interface through a combination of the two rules above.
1617 crate_helper_module::crate_helper();
1620 // This function is hidden to any module which is not a descendant of
1622 fn my_implementation() {}
1628 fn test_my_implementation() {
1629 // Because this module is a descendant of `submodule`, it's allowed
1630 // to access private items inside of `submodule` without a privacy
1632 super::my_implementation();
1640 For a rust program to pass the privacy checking pass, all paths must be valid
1641 accesses given the two rules above. This includes all use statements,
1642 expressions, types, etc.
1644 ### Re-exporting and Visibility
1646 Rust allows publicly re-exporting items through a `pub use` directive. Because
1647 this is a public directive, this allows the item to be used in the current
1648 module through the rules above. It essentially allows public access into the
1649 re-exported item. For example, this program is valid:
1652 pub use api = self::implementation;
1654 mod implementation {
1661 This means that any external crate referencing `implementation::f` would receive
1662 a privacy violation, while the path `api::f` would be allowed.
1664 When re-exporting a private item, it can be thought of as allowing the "privacy
1665 chain" being short-circuited through the reexport instead of passing through the
1666 namespace hierarchy as it normally would.
1668 ### Glob imports and Visibility
1670 Currently glob imports are considered an "experimental" language feature. For
1671 sanity purpose along with helping the implementation, glob imports will only
1672 import public items from their destination, not private items.
1674 > **Note:** This is subject to change, glob exports may be removed entirely or
1675 > they could possibly import private items for a privacy error to later be
1676 > issued if the item is used.
1680 ~~~~ {.notrust .ebnf .gram}
1681 attribute : '#' '!' ? '[' attr_list ']' ;
1682 attr_list : attr [ ',' attr_list ]* ;
1683 attr : ident [ '=' literal
1684 | '(' attr_list ')' ] ? ;
1687 Static entities in Rust -- crates, modules and items -- may have _attributes_
1688 applied to them. Attributes in Rust are modeled on Attributes in ECMA-335, C#.
1689 An attribute is a general, free-form metadatum that is interpreted according
1690 to name, convention, and language and compiler version. Attributes may appear
1693 * A single identifier, the attribute name
1694 * An identifier followed by the equals sign '=' and a literal, providing a
1696 * An identifier followed by a parenthesized list of sub-attribute arguments
1698 Attributes with a bang ("!") after the hash ("#") apply to the item that the
1699 attribute is declared within. Attributes that do not have a bang after the
1700 hash apply to the item that follows the attribute.
1702 An example of attributes:
1705 // General metadata applied to the enclosing module or crate.
1708 // A function marked as a unit test
1714 // A conditionally-compiled module
1715 #[cfg(target_os="linux")]
1720 // A lint attribute used to suppress a warning/error
1721 #[allow(non_camel_case_types)]
1722 pub type int8_t = i8;
1725 > **Note:** At some point in the future, the compiler will distinguish between
1726 > language-reserved and user-available attributes. Until then, there is
1727 > effectively no difference between an attribute handled by a loadable syntax
1728 > extension and the compiler.
1730 ### Crate-only attributes
1732 - `crate_id` - specify the this crate's crate ID.
1733 - `crate_type` - see [linkage](#linkage).
1734 - `feature` - see [compiler features](#compiler-features).
1735 - `no_main` - disable emitting the `main` symbol. Useful when some other
1736 object being linked to defines `main`.
1737 - `no_start` - disable linking to the `native` crate, which specifies the
1738 "start" language item.
1739 - `no_std` - disable linking to the `std` crate.
1741 ### Module-only attributes
1743 - `macro_escape` - macros defined in this module will be visible in the
1744 module's parent, after this module has been included.
1745 - `no_implicit_prelude` - disable injecting `use std::prelude::*` in this
1747 - `path` - specifies the file to load the module from. `#[path="foo.rs"] mod
1748 bar;` is equivalent to `mod bar { /* contents of foo.rs */ }`. The path is
1749 taken relative to the directory that the current module is in.
1751 ### Function-only attributes
1753 - `macro_registrar` - when using loadable syntax extensions, mark this
1754 function as the registration point for the current crate's syntax
1756 - `main` - indicates that this function should be passed to the entry point,
1757 rather than the function in the crate root named `main`.
1758 - `start` - indicates that this function should be used as the entry point,
1759 overriding the "start" language item. See the "start" [language
1760 item](#language-items) for more details.
1762 ### Static-only attributes
1764 - `address_insignificant` - references to this static may alias with
1765 references to other statics, potentially of unrelated type.
1766 - `thread_local` - on a `static mut`, this signals that the value of this
1767 static may change depending on the current thread. The exact consequences of
1768 this are implementation-defined.
1772 On an `extern` block, the following attributes are interpreted:
1774 - `link_args` - specify arguments to the linker, rather than just the library
1775 name and type. This is feature gated and the exact behavior is
1776 implementation-defined (due to variety of linker invocation syntax).
1777 - `link` - indicate that a native library should be linked to for the
1778 declarations in this block to be linked correctly. See [external
1779 blocks](#external-blocks)
1781 On declarations inside an `extern` block, the following attributes are
1784 - `link_name` - the name of the symbol that this function or static should be
1786 - `linkage` - on a static, this specifies the [linkage
1787 type](http://llvm.org/docs/LangRef.html#linkage-types).
1789 ### Miscellaneous attributes
1791 - `link_section` - on statics and functions, this specifies the section of the
1792 object file that this item's contents will be placed into.
1793 - `macro_export` - export a macro for cross-crate usage.
1794 - `no_mangle` - on any item, do not apply the standard name mangling. Set the
1795 symbol for this item to its identifier.
1796 - `packed` - on structs or enums, eliminate any padding that would be used to
1798 - `repr` - on C-like enums, this sets the underlying type used for
1799 representation. Useful for FFI. Takes one argument, which is the primitive
1800 type this enum should be represented for, or `C`, which specifies that it
1801 should be the default `enum` size of the C ABI for that platform. Note that
1802 enum representation in C is undefined, and this may be incorrect when the C
1803 code is compiled with certain flags.
1804 - `simd` - on certain tuple structs, derive the arithmetic operators, which
1805 lower to the target's SIMD instructions, if any.
1806 - `static_assert` - on statics whose type is `bool`, terminates compilation
1807 with an error if it is not initialized to `true`.
1808 - `unsafe_destructor` - allow implementations of the "drop" language item
1809 where the type it is implemented for does not implement the "send" language
1811 - `unsafe_no_drop_flag` - on structs, remove the flag that prevents
1812 destructors from being run twice. Destructors might be run multiple times on
1813 the same object with this attribute.
1815 ### Conditional compilation
1817 Sometimes one wants to have different compiler outputs from the same code,
1818 depending on build target, such as targeted operating system, or to enable
1821 There are two kinds of configuration options, one that is either defined or not
1822 (`#[cfg(foo)]`), and the other that contains a string that can be checked
1823 against (`#[cfg(bar = "baz")]` (currently only compiler-defined configuration
1824 options can have the latter form).
1827 // The function is only included in the build when compiling for OSX
1828 #[cfg(target_os = "macos")]
1833 // This function is only included when either foo or bar is defined
1836 fn needs_foo_or_bar() {
1840 // This function is only included when compiling for a unixish OS with a 32-bit
1842 #[cfg(unix, target_word_size = "32")]
1843 fn on_32bit_unix() {
1848 This illustrates some conditional compilation can be achieved using the
1849 `#[cfg(...)]` attribute. Note that `#[cfg(foo, bar)]` is a condition that needs
1850 both `foo` and `bar` to be defined while `#[cfg(foo)] #[cfg(bar)]` only needs
1851 one of `foo` and `bar` to be defined (this resembles in the disjunctive normal
1852 form). Additionally, one can reverse a condition by enclosing it in a
1853 `not(...)`, like e. g. `#[cfg(not(target_os = "win32"))]`.
1855 The following configurations must be defined by the implementation:
1857 * `target_arch = "..."`. Target CPU architecture, such as `"x86"`, `"x86_64"`
1858 `"mips"`, or `"arm"`.
1859 * `target_endian = "..."`. Endianness of the target CPU, either `"little"` or
1861 * `target_family = "..."`. Operating system family of the target, e. g.
1862 `"unix"` or `"windows"`. The value of this configuration option is defined as
1863 a configuration itself, like `unix` or `windows`.
1864 * `target_os = "..."`. Operating system of the target, examples include
1865 `"win32"`, `"macos"`, `"linux"`, `"android"` or `"freebsd"`.
1866 * `target_word_size = "..."`. Target word size in bits. This is set to `"32"`
1867 for targets with 32-bit pointers, and likewise set to `"64"` for 64-bit
1869 * `unix`. See `target_family`.
1870 * `windows`. See `target_family`.
1872 ### Lint check attributes
1874 A lint check names a potentially undesirable coding pattern, such as
1875 unreachable code or omitted documentation, for the static entity to
1876 which the attribute applies.
1878 For any lint check `C`:
1880 * `warn(C)` warns about violations of `C` but continues compilation,
1881 * `deny(C)` signals an error after encountering a violation of `C`,
1882 * `allow(C)` overrides the check for `C` so that violations will go
1884 * `forbid(C)` is the same as `deny(C)`, but also forbids uses of
1885 `allow(C)` within the attribute.
1887 The lint checks supported by the compiler can be found via `rustc -W help`,
1888 along with their default settings.
1892 // Missing documentation is ignored here
1893 #[allow(missing_doc)]
1894 pub fn undocumented_one() -> int { 1 }
1896 // Missing documentation signals a warning here
1897 #[warn(missing_doc)]
1898 pub fn undocumented_too() -> int { 2 }
1900 // Missing documentation signals an error here
1901 #[deny(missing_doc)]
1902 pub fn undocumented_end() -> int { 3 }
1906 This example shows how one can use `allow` and `warn` to toggle
1907 a particular check on and off.
1910 #[warn(missing_doc)]
1912 #[allow(missing_doc)]
1914 // Missing documentation is ignored here
1915 pub fn undocumented_one() -> int { 1 }
1917 // Missing documentation signals a warning here,
1918 // despite the allow above.
1919 #[warn(missing_doc)]
1920 pub fn undocumented_two() -> int { 2 }
1923 // Missing documentation signals a warning here
1924 pub fn undocumented_too() -> int { 3 }
1928 This example shows how one can use `forbid` to disallow uses
1929 of `allow` for that lint check.
1932 #[forbid(missing_doc)]
1934 // Attempting to toggle warning signals an error here
1935 #[allow(missing_doc)]
1937 pub fn undocumented_too() -> int { 2 }
1943 Some primitive Rust operations are defined in Rust code, rather than being
1944 implemented directly in C or assembly language. The definitions of these
1945 operations have to be easy for the compiler to find. The `lang` attribute
1946 makes it possible to declare these operations. For example, the `str` module
1947 in the Rust standard library defines the string equality function:
1951 pub fn eq_slice(a: &str, b: &str) -> bool {
1956 The name `str_eq` has a special meaning to the Rust compiler,
1957 and the presence of this definition means that it will use this definition
1958 when generating calls to the string equality function.
1960 A complete list of the built-in language items follows:
1962 #### Built-in Traits
1965 : Able to be sent across task boundaries.
1967 : Has a size known at compile time.
1969 : Types that do not move ownership when used by-value.
1971 : Able to be safely shared between tasks when aliased.
1977 These language items are traits:
1980 : Elements can be added (for example, integers and floats).
1982 : Elements can be subtracted.
1984 : Elements can be multiplied.
1986 : Elements have a division operation.
1988 : Elements have a remainder operation.
1990 : Elements can be negated arithmetically.
1992 : Elements can be negated logically.
1994 : Elements have an exclusive-or operation.
1996 : Elements have a bitwise `and` operation.
1998 : Elements have a bitwise `or` operation.
2000 : Elements have a left shift operation.
2002 : Elements have a right shift operation.
2004 : Elements can be indexed.
2006 : Elements can be compared for equality.
2008 : Elements have a partial ordering.
2010 : `*` can be applied, yielding a reference to another type
2012 : `*` can be applied, yielding a mutable reference to another type
2015 These are functions:
2018 : Compare two strings (`&str`) for equality.
2020 : Compare two owned strings (`~str`) for equality.
2022 : Return a new unique string
2023 containing a copy of the contents of a unique string.
2028 : A type whose contents can be mutated through an immutable reference
2030 : The type returned by the `type_id` intrinsic.
2034 These types help drive the compiler's analysis
2037 : The type parameter should be considered covariant
2038 `contravariant_type`
2039 : The type parameter should be considered contravariant
2041 : The type parameter should be considered invariant
2042 `covariant_lifetime`
2043 : The lifetime parameter should be considered covariant
2044 `contravariant_lifetime`
2045 : The lifetime parameter should be considered contravariant
2046 `invariant_lifetime`
2047 : The lifetime parameter should be considered invariant
2049 : This type does not implement "send", even if eligible
2051 : This type does not implement "copy", even if eligible
2053 : This type does not implement "share", even if eligible
2055 : This type implements "managed"
2058 : Abort the program with an error.
2060 : Abort the program with a bounds check error.
2062 : Allocate memory on the exchange heap.
2064 : Free memory that was allocated on the exchange heap.
2066 : Allocate memory on the managed heap.
2068 : Free memory that was allocated on the managed heap.
2070 > **Note:** This list is likely to become out of date. We should auto-generate it
2071 > from `librustc/middle/lang_items.rs`.
2073 ### Inline attributes
2075 The inline attribute is used to suggest to the compiler to perform an inline
2076 expansion and place a copy of the function in the caller rather than generating
2077 code to call the function where it is defined.
2079 The compiler automatically inlines functions based on internal heuristics.
2080 Incorrectly inlining functions can actually making the program slower, so it
2081 should be used with care.
2083 `#[inline]` and `#[inline(always)]` always causes the function to be serialized
2084 into crate metadata to allow cross-crate inlining.
2086 There are three different types of inline attributes:
2088 * `#[inline]` hints the compiler to perform an inline expansion.
2089 * `#[inline(always)]` asks the compiler to always perform an inline expansion.
2090 * `#[inline(never)]` asks the compiler to never perform an inline expansion.
2094 The `deriving` attribute allows certain traits to be automatically
2095 implemented for data structures. For example, the following will
2096 create an `impl` for the `Eq` and `Clone` traits for `Foo`, the type
2097 parameter `T` will be given the `Eq` or `Clone` constraints for the
2101 #[deriving(Eq, Clone)]
2108 The generated `impl` for `Eq` is equivalent to
2111 # struct Foo<T> { a: int, b: T }
2112 impl<T: Eq> Eq for Foo<T> {
2113 fn eq(&self, other: &Foo<T>) -> bool {
2114 self.a == other.a && self.b == other.b
2117 fn ne(&self, other: &Foo<T>) -> bool {
2118 self.a != other.a || self.b != other.b
2123 Supported traits for `deriving` are:
2125 * Comparison traits: `Eq`, `TotalEq`, `Ord`, `TotalOrd`.
2126 * Serialization: `Encodable`, `Decodable`. These require `serialize`.
2127 * `Clone`, to create `T` from `&T` via a copy.
2128 * `Hash`, to iterate over the bytes in a data type.
2129 * `Rand`, to create a random instance of a data type.
2130 * `Default`, to create an empty instance of a data type.
2131 * `Zero`, to create an zero instance of a numeric data type.
2132 * `FromPrimitive`, to create an instance from a numeric primitive.
2133 * `Show`, to format a value using the `{}` formatter.
2137 One can indicate the stability of an API using the following attributes:
2139 * `deprecated`: This item should no longer be used, e.g. it has been
2140 replaced. No guarantee of backwards-compatibility.
2141 * `experimental`: This item was only recently introduced or is
2142 otherwise in a state of flux. It may change significantly, or even
2143 be removed. No guarantee of backwards-compatibility.
2144 * `unstable`: This item is still under development, but requires more
2145 testing to be considered stable. No guarantee of backwards-compatibility.
2146 * `stable`: This item is considered stable, and will not change
2147 significantly. Guarantee of backwards-compatibility.
2148 * `frozen`: This item is very stable, and is unlikely to
2149 change. Guarantee of backwards-compatibility.
2150 * `locked`: This item will never change unless a serious bug is
2151 found. Guarantee of backwards-compatibility.
2153 These levels are directly inspired by
2154 [Node.js' "stability index"](http://nodejs.org/api/documentation.html).
2156 There are lints for disallowing items marked with certain levels:
2157 `deprecated`, `experimental` and `unstable`; the first two will warn
2158 by default. Items with not marked with a stability are considered to
2159 be unstable for the purposes of the lint. One can give an optional
2160 string that will be displayed when the lint flags the use of an item.
2165 #[deprecated="replaced by `best`"]
2167 // delete everything
2171 // delete fewer things
2180 bad(); // "warning: use of deprecated item: replaced by `best`"
2182 better(); // "warning: use of unmarked item"
2184 best(); // no warning
2188 > **Note:** Currently these are only checked when applied to
2189 > individual functions, structs, methods and enum variants, *not* to
2190 > entire modules, traits, impls or enums themselves.
2192 ### Compiler Features
2194 Certain aspects of Rust may be implemented in the compiler, but they're not
2195 necessarily ready for every-day use. These features are often of "prototype
2196 quality" or "almost production ready", but may not be stable enough to be
2197 considered a full-fleged language feature.
2199 For this reason, Rust recognizes a special crate-level attribute of the form:
2202 #![feature(feature1, feature2, feature3)]
2205 This directive informs the compiler that the feature list: `feature1`,
2206 `feature2`, and `feature3` should all be enabled. This is only recognized at a
2207 crate-level, not at a module-level. Without this directive, all features are
2208 considered off, and using the features will result in a compiler error.
2210 The currently implemented features of the reference compiler are:
2212 * `macro_rules` - The definition of new macros. This does not encompass
2213 macro-invocation, that is always enabled by default, this only
2214 covers the definition of new macros. There are currently
2215 various problems with invoking macros, how they interact with
2216 their environment, and possibly how they are used outside of
2217 location in which they are defined. Macro definitions are
2218 likely to change slightly in the future, so they are currently
2219 hidden behind this feature.
2221 * `globs` - Importing everything in a module through `*`. This is currently a
2222 large source of bugs in name resolution for Rust, and it's not clear
2223 whether this will continue as a feature or not. For these reasons,
2224 the glob import statement has been hidden behind this feature flag.
2226 * `struct_variant` - Structural enum variants (those with named fields). It is
2227 currently unknown whether this style of enum variant is as
2228 fully supported as the tuple-forms, and it's not certain
2229 that this style of variant should remain in the language.
2230 For now this style of variant is hidden behind a feature
2233 * `once_fns` - Onceness guarantees a closure is only executed once. Defining a
2234 closure as `once` is unlikely to be supported going forward. So
2235 they are hidden behind this feature until they are to be removed.
2237 * `managed_boxes` - Usage of `@` pointers is gated due to many
2238 planned changes to this feature. In the past, this has meant
2239 "a GC pointer", but the current implementation uses
2240 reference counting and will likely change drastically over
2241 time. Additionally, the `@` syntax will no longer be used to
2244 * `asm` - The `asm!` macro provides a means for inline assembly. This is often
2245 useful, but the exact syntax for this feature along with its semantics
2246 are likely to change, so this macro usage must be opted into.
2248 * `non_ascii_idents` - The compiler supports the use of non-ascii identifiers,
2249 but the implementation is a little rough around the
2250 edges, so this can be seen as an experimental feature for
2251 now until the specification of identifiers is fully
2254 * `thread_local` - The usage of the `#[thread_local]` attribute is experimental
2255 and should be seen as unstable. This attribute is used to
2256 declare a `static` as being unique per-thread leveraging
2257 LLVM's implementation which works in concert with the kernel
2258 loader and dynamic linker. This is not necessarily available
2259 on all platforms, and usage of it is discouraged (rust
2260 focuses more on task-local data instead of thread-local
2263 * `link_args` - This attribute is used to specify custom flags to the linker,
2264 but usage is strongly discouraged. The compiler's usage of the
2265 system linker is not guaranteed to continue in the future, and
2266 if the system linker is not used then specifying custom flags
2267 doesn't have much meaning.
2269 If a feature is promoted to a language feature, then all existing programs will
2270 start to receive compilation warnings about #[feature] directives which enabled
2271 the new feature (because the directive is no longer necessary). However, if
2272 a feature is decided to be removed from the language, errors will be issued (if
2273 there isn't a parser error first). The directive in this case is no longer
2274 necessary, and it's likely that existing code will break if the feature isn't
2277 If a unknown feature is found in a directive, it results in a compiler error. An
2278 unknown feature is one which has never been recognized by the compiler.
2280 # Statements and expressions
2282 Rust is _primarily_ an expression language. This means that most forms of
2283 value-producing or effect-causing evaluation are directed by the uniform
2284 syntax category of _expressions_. Each kind of expression can typically _nest_
2285 within each other kind of expression, and rules for evaluation of expressions
2286 involve specifying both the value produced by the expression and the order in
2287 which its sub-expressions are themselves evaluated.
2289 In contrast, statements in Rust serve _mostly_ to contain and explicitly
2290 sequence expression evaluation.
2294 A _statement_ is a component of a block, which is in turn a component of an
2295 outer [expression](#expressions) or [function](#functions).
2297 Rust has two kinds of statement:
2298 [declaration statements](#declaration-statements) and
2299 [expression statements](#expression-statements).
2301 ### Declaration statements
2303 A _declaration statement_ is one that introduces one or more *names* into the enclosing statement block.
2304 The declared names may denote new slots or new items.
2306 #### Item declarations
2308 An _item declaration statement_ has a syntactic form identical to an
2309 [item](#items) declaration within a module. Declaring an item -- a function,
2310 enumeration, structure, type, static, trait, implementation or module -- locally
2311 within a statement block is simply a way of restricting its scope to a narrow
2312 region containing all of its uses; it is otherwise identical in meaning to
2313 declaring the item outside the statement block.
2315 Note: there is no implicit capture of the function's dynamic environment when
2316 declaring a function-local item.
2318 #### Slot declarations
2320 ~~~~ {.notrust .ebnf .gram}
2321 let_decl : "let" pat [':' type ] ? [ init ] ? ';' ;
2322 init : [ '=' ] expr ;
2325 A _slot declaration_ introduces a new set of slots, given by a pattern.
2326 The pattern may be followed by a type annotation, and/or an initializer expression.
2327 When no type annotation is given, the compiler will infer the type,
2328 or signal an error if insufficient type information is available for definite inference.
2329 Any slots introduced by a slot declaration are visible from the point of declaration until the end of the enclosing block scope.
2331 ### Expression statements
2333 An _expression statement_ is one that evaluates an [expression](#expressions)
2334 and ignores its result.
2335 The type of an expression statement `e;` is always `()`, regardless of the type of `e`.
2336 As a rule, an expression statement's purpose is to trigger the effects of evaluating its expression.
2340 An expression may have two roles: it always produces a *value*, and it may have *effects*
2341 (otherwise known as "side effects").
2342 An expression *evaluates to* a value, and has effects during *evaluation*.
2343 Many expressions contain sub-expressions (operands).
2344 The meaning of each kind of expression dictates several things:
2345 * Whether or not to evaluate the sub-expressions when evaluating the expression
2346 * The order in which to evaluate the sub-expressions
2347 * How to combine the sub-expressions' values to obtain the value of the expression.
2349 In this way, the structure of expressions dictates the structure of execution.
2350 Blocks are just another kind of expression,
2351 so blocks, statements, expressions, and blocks again can recursively nest inside each other
2352 to an arbitrary depth.
2354 #### Lvalues, rvalues and temporaries
2356 Expressions are divided into two main categories: _lvalues_ and _rvalues_.
2357 Likewise within each expression, sub-expressions may occur in _lvalue context_ or _rvalue context_.
2358 The evaluation of an expression depends both on its own category and the context it occurs within.
2360 An lvalue is an expression that represents a memory location. These
2361 expressions are [paths](#path-expressions) (which refer to local
2362 variables, function and method arguments, or static variables),
2363 dereferences (`*expr`), [indexing expressions](#index-expressions)
2364 (`expr[expr]`), and [field references](#field-expressions) (`expr.f`).
2365 All other expressions are rvalues.
2367 The left operand of an [assignment](#assignment-expressions) or
2368 [compound-assignment](#compound-assignment-expressions) expression is an lvalue context,
2369 as is the single operand of a unary [borrow](#unary-operator-expressions).
2370 All other expression contexts are rvalue contexts.
2372 When an lvalue is evaluated in an _lvalue context_, it denotes a memory location;
2373 when evaluated in an _rvalue context_, it denotes the value held _in_ that memory location.
2375 When an rvalue is used in lvalue context, a temporary un-named lvalue is created and used instead.
2376 A temporary's lifetime equals the largest lifetime of any reference that points to it.
2378 #### Moved and copied types
2380 When a [local variable](#memory-slots) is used
2381 as an [rvalue](#lvalues-rvalues-and-temporaries)
2382 the variable will either be moved or copied, depending on its type.
2383 For types that contain [owning pointers](#pointer-types)
2384 or values that implement the special trait `Drop`,
2385 the variable is moved.
2386 All other types are copied.
2388 ### Literal expressions
2390 A _literal expression_ consists of one of the [literal](#literals)
2391 forms described earlier. It directly describes a number, character,
2392 string, boolean value, or the unit value.
2396 "hello"; // string type
2397 '5'; // character type
2401 ### Path expressions
2403 A [path](#paths) used as an expression context denotes either a local variable or an item.
2404 Path expressions are [lvalues](#lvalues-rvalues-and-temporaries).
2406 ### Tuple expressions
2408 Tuples are written by enclosing one or more comma-separated
2409 expressions in parentheses. They are used to create [tuple-typed](#tuple-types)
2418 ### Structure expressions
2420 ~~~~ {.notrust .ebnf .gram}
2421 struct_expr : expr_path '{' ident ':' expr
2422 [ ',' ident ':' expr ] *
2425 [ ',' expr ] * ')' |
2429 There are several forms of structure expressions.
2430 A _structure expression_ consists of the [path](#paths) of a [structure item](#structures),
2431 followed by a brace-enclosed list of one or more comma-separated name-value pairs,
2432 providing the field values of a new instance of the structure.
2433 A field name can be any identifier, and is separated from its value expression by a colon.
2434 The location denoted by a structure field is mutable if and only if the enclosing structure is mutable.
2436 A _tuple structure expression_ consists of the [path](#paths) of a [structure item](#structures),
2437 followed by a parenthesized list of one or more comma-separated expressions
2438 (in other words, the path of a structure item followed by a tuple expression).
2439 The structure item must be a tuple structure item.
2441 A _unit-like structure expression_ consists only of the [path](#paths) of a [structure item](#structures).
2443 The following are examples of structure expressions:
2446 # struct Point { x: f64, y: f64 }
2447 # struct TuplePoint(f64, f64);
2448 # mod game { pub struct User<'a> { pub name: &'a str, pub age: uint, pub score: uint } }
2449 # struct Cookie; fn some_fn<T>(t: T) {}
2450 Point {x: 10.0, y: 20.0};
2451 TuplePoint(10.0, 20.0);
2452 let u = game::User {name: "Joe", age: 35, score: 100_000};
2453 some_fn::<Cookie>(Cookie);
2456 A structure expression forms a new value of the named structure type.
2457 Note that for a given *unit-like* structure type, this will always be the same value.
2459 A structure expression can terminate with the syntax `..` followed by an expression to denote a functional update.
2460 The expression following `..` (the base) must have the same structure type as the new structure type being formed.
2461 The entire expression denotes the result of allocating a new structure
2462 (with the same type as the base expression)
2463 with the given values for the fields that were explicitly specified
2464 and the values in the base record for all other fields.
2467 # struct Point3d { x: int, y: int, z: int }
2468 let base = Point3d {x: 1, y: 2, z: 3};
2469 Point3d {y: 0, z: 10, .. base};
2472 ### Block expressions
2474 ~~~~ {.notrust .ebnf .gram}
2475 block_expr : '{' [ view_item ] *
2476 [ stmt ';' | item ] *
2480 A _block expression_ is similar to a module in terms of the declarations that
2481 are possible. Each block conceptually introduces a new namespace scope. View
2482 items can bring new names into scopes and declared items are in scope for only
2485 A block will execute each statement sequentially, and then execute the
2486 expression (if given). If the final expression is omitted, the type and return
2487 value of the block are `()`, but if it is provided, the type and return value
2488 of the block are that of the expression itself.
2490 ### Method-call expressions
2492 ~~~~ {.notrust .ebnf .gram}
2493 method_call_expr : expr '.' ident paren_expr_list ;
2496 A _method call_ consists of an expression followed by a single dot, an identifier, and a parenthesized expression-list.
2497 Method calls are resolved to methods on specific traits,
2498 either statically dispatching to a method if the exact `self`-type of the left-hand-side is known,
2499 or dynamically dispatching if the left-hand-side expression is an indirect [object type](#object-types).
2501 ### Field expressions
2503 ~~~~ {.notrust .ebnf .gram}
2504 field_expr : expr '.' ident ;
2507 A _field expression_ consists of an expression followed by a single dot and an identifier,
2508 when not immediately followed by a parenthesized expression-list (the latter is a [method call expression](#method-call-expressions)).
2509 A field expression denotes a field of a [structure](#structure-types).
2511 ~~~~ {.ignore .field}
2514 (Struct {a: 10, b: 20}).a;
2517 A field access on a record is an [lvalue](#lvalues-rvalues-and-temporaries) referring to the value of that field.
2518 When the field is mutable, it can be [assigned](#assignment-expressions) to.
2520 When the type of the expression to the left of the dot is a pointer to a record or structure,
2521 it is automatically dereferenced to make the field access possible.
2523 ### Vector expressions
2525 ~~~~ {.notrust .ebnf .gram}
2526 vec_expr : '[' "mut" ? vec_elems? ']' ;
2528 vec_elems : [expr [',' expr]*] | [expr ',' ".." expr] ;
2531 A [_vector_](#vector-types) _expression_ is written by enclosing zero or
2532 more comma-separated expressions of uniform type in square brackets.
2534 In the `[expr ',' ".." expr]` form, the expression after the `".."`
2535 must be a constant expression that can be evaluated at compile time, such
2536 as a [literal](#literals) or a [static item](#static-items).
2540 ["a", "b", "c", "d"];
2541 [0, ..128]; // vector with 128 zeros
2542 [0u8, 0u8, 0u8, 0u8];
2545 ### Index expressions
2547 ~~~~ {.notrust .ebnf .gram}
2548 idx_expr : expr '[' expr ']' ;
2551 [Vector](#vector-types)-typed expressions can be indexed by writing a
2552 square-bracket-enclosed expression (the index) after them. When the
2553 vector is mutable, the resulting [lvalue](#lvalues-rvalues-and-temporaries) can be assigned to.
2555 Indices are zero-based, and may be of any integral type. Vector access
2556 is bounds-checked at run-time. When the check fails, it will put the
2557 task in a _failing state_.
2564 (["a", "b"])[10]; // fails
2569 ### Unary operator expressions
2571 Rust defines six symbolic unary operators.
2572 They are all written as prefix operators,
2573 before the expression they apply to.
2576 : Negation. May only be applied to numeric types.
2578 : Dereference. When applied to a [pointer](#pointer-types) it denotes the pointed-to location.
2579 For pointers to mutable locations, the resulting [lvalue](#lvalues-rvalues-and-temporaries) can be assigned to.
2580 On non-pointer types, it calls the `deref` method of the `std::ops::Deref` trait, or the
2581 `deref_mut` method of the `std::ops::DerefMut` trait (if implemented by the type and required
2582 for an outer expression that will or could mutate the dereference), and produces the
2583 result of dereferencing the `&` or `&mut` borrowed pointer returned from the overload method.
2586 : Logical negation. On the boolean type, this flips between `true` and
2587 `false`. On integer types, this inverts the individual bits in the
2588 two's complement representation of the value.
2590 : [Boxing](#pointer-types) operators. Allocate a box to hold the value they are applied to,
2591 and store the value in it. `~` creates an owned box.
2593 : Borrow operator. Returns a reference, pointing to its operand.
2594 The operand of a borrow is statically proven to outlive the resulting pointer.
2595 If the borrow-checker cannot prove this, it is a compilation error.
2597 ### Binary operator expressions
2599 ~~~~ {.notrust .ebnf .gram}
2600 binop_expr : expr binop expr ;
2603 Binary operators expressions are given in terms of
2604 [operator precedence](#operator-precedence).
2606 #### Arithmetic operators
2608 Binary arithmetic expressions are syntactic sugar for calls to built-in traits,
2609 defined in the `std::ops` module of the `std` library.
2610 This means that arithmetic operators can be overridden for user-defined types.
2611 The default meaning of the operators on standard types is given here.
2614 : Addition and vector/string concatenation.
2615 Calls the `add` method on the `std::ops::Add` trait.
2618 Calls the `sub` method on the `std::ops::Sub` trait.
2621 Calls the `mul` method on the `std::ops::Mul` trait.
2624 Calls the `div` method on the `std::ops::Div` trait.
2627 Calls the `rem` method on the `std::ops::Rem` trait.
2629 #### Bitwise operators
2631 Like the [arithmetic operators](#arithmetic-operators), bitwise operators
2632 are syntactic sugar for calls to methods of built-in traits.
2633 This means that bitwise operators can be overridden for user-defined types.
2634 The default meaning of the operators on standard types is given here.
2638 Calls the `bitand` method of the `std::ops::BitAnd` trait.
2641 Calls the `bitor` method of the `std::ops::BitOr` trait.
2644 Calls the `bitxor` method of the `std::ops::BitXor` trait.
2646 : Logical left shift.
2647 Calls the `shl` method of the `std::ops::Shl` trait.
2649 : Logical right shift.
2650 Calls the `shr` method of the `std::ops::Shr` trait.
2652 #### Lazy boolean operators
2654 The operators `||` and `&&` may be applied to operands of boolean type.
2655 The `||` operator denotes logical 'or', and the `&&` operator denotes logical 'and'.
2656 They differ from `|` and `&` in that the right-hand operand is only evaluated
2657 when the left-hand operand does not already determine the result of the expression.
2658 That is, `||` only evaluates its right-hand operand
2659 when the left-hand operand evaluates to `false`, and `&&` only when it evaluates to `true`.
2661 #### Comparison operators
2663 Comparison operators are, like the [arithmetic operators](#arithmetic-operators),
2664 and [bitwise operators](#bitwise-operators),
2665 syntactic sugar for calls to built-in traits.
2666 This means that comparison operators can be overridden for user-defined types.
2667 The default meaning of the operators on standard types is given here.
2671 Calls the `eq` method on the `std::cmp::Eq` trait.
2674 Calls the `ne` method on the `std::cmp::Eq` trait.
2677 Calls the `lt` method on the `std::cmp::Ord` trait.
2680 Calls the `gt` method on the `std::cmp::Ord` trait.
2682 : Less than or equal.
2683 Calls the `le` method on the `std::cmp::Ord` trait.
2685 : Greater than or equal.
2686 Calls the `ge` method on the `std::cmp::Ord` trait.
2688 #### Type cast expressions
2690 A type cast expression is denoted with the binary operator `as`.
2692 Executing an `as` expression casts the value on the left-hand side to the type
2693 on the right-hand side.
2695 A numeric value can be cast to any numeric type.
2696 A raw pointer value can be cast to or from any integral type or raw pointer type.
2697 Any other cast is unsupported and will fail to compile.
2699 An example of an `as` expression:
2702 # fn sum(v: &[f64]) -> f64 { 0.0 }
2703 # fn len(v: &[f64]) -> int { 0 }
2705 fn avg(v: &[f64]) -> f64 {
2706 let sum: f64 = sum(v);
2707 let sz: f64 = len(v) as f64;
2712 #### Assignment expressions
2714 An _assignment expression_ consists of an [lvalue](#lvalues-rvalues-and-temporaries) expression followed by an
2715 equals sign (`=`) and an [rvalue](#lvalues-rvalues-and-temporaries) expression.
2717 Evaluating an assignment expression [either copies or moves](#moved-and-copied-types) its right-hand operand to its left-hand operand.
2726 #### Compound assignment expressions
2728 The `+`, `-`, `*`, `/`, `%`, `&`, `|`, `^`, `<<`, and `>>`
2729 operators may be composed with the `=` operator. The expression `lval
2730 OP= val` is equivalent to `lval = lval OP val`. For example, `x = x +
2731 1` may be written as `x += 1`.
2733 Any such expression always has the [`unit`](#primitive-types) type.
2735 #### Operator precedence
2737 The precedence of Rust binary operators is ordered as follows, going
2738 from strong to weak:
2740 ~~~~ {.notrust .precedence}
2755 Operators at the same precedence level are evaluated left-to-right. [Unary operators](#unary-operator-expressions)
2756 have the same precedence level and it is stronger than any of the binary operators'.
2758 ### Grouped expressions
2760 An expression enclosed in parentheses evaluates to the result of the enclosed
2761 expression. Parentheses can be used to explicitly specify evaluation order
2762 within an expression.
2764 ~~~~ {.notrust .ebnf .gram}
2765 paren_expr : '(' expr ')' ;
2768 An example of a parenthesized expression:
2771 let x = (2 + 3) * 4;
2775 ### Call expressions
2777 ~~~~ {.notrust .ebnf .gram}
2778 expr_list : [ expr [ ',' expr ]* ] ? ;
2779 paren_expr_list : '(' expr_list ')' ;
2780 call_expr : expr paren_expr_list ;
2783 A _call expression_ invokes a function, providing zero or more input slots and
2784 an optional reference slot to serve as the function's output, bound to the
2785 `lval` on the right hand side of the call. If the function eventually returns,
2786 then the expression completes.
2788 Some examples of call expressions:
2791 # use std::from_str::FromStr;
2792 # fn add(x: int, y: int) -> int { 0 }
2794 let x: int = add(1, 2);
2795 let pi: Option<f32> = FromStr::from_str("3.14");
2798 ### Lambda expressions
2800 ~~~~ {.notrust .ebnf .gram}
2801 ident_list : [ ident [ ',' ident ]* ] ? ;
2802 lambda_expr : '|' ident_list '|' expr ;
2805 A _lambda expression_ (sometimes called an "anonymous function expression") defines a function and denotes it as a value,
2806 in a single expression.
2807 A lambda expression is a pipe-symbol-delimited (`|`) list of identifiers followed by an expression.
2809 A lambda expression denotes a function that maps a list of parameters (`ident_list`)
2810 onto the expression that follows the `ident_list`.
2811 The identifiers in the `ident_list` are the parameters to the function.
2812 These parameters' types need not be specified, as the compiler infers them from context.
2814 Lambda expressions are most useful when passing functions as arguments to other functions,
2815 as an abbreviation for defining and capturing a separate function.
2817 Significantly, lambda expressions _capture their environment_,
2818 which regular [function definitions](#functions) do not.
2819 The exact type of capture depends on the [function type](#function-types) inferred for the lambda expression.
2820 In the simplest and least-expensive form (analogous to a ```|| { }``` expression),
2821 the lambda expression captures its environment by reference,
2822 effectively borrowing pointers to all outer variables mentioned inside the function.
2823 Alternately, the compiler may infer that a lambda expression should copy or move values (depending on their type.)
2824 from the environment into the lambda expression's captured environment.
2826 In this example, we define a function `ten_times` that takes a higher-order function argument,
2827 and call it with a lambda expression as an argument.
2830 fn ten_times(f: |int|) {
2838 ten_times(|j| println!("hello, {}", j));
2843 ~~~~ {.notrust .ebnf .gram}
2844 while_expr : "while" expr '{' block '}' ;
2847 A `while` loop begins by evaluating the boolean loop conditional expression.
2848 If the loop conditional expression evaluates to `true`, the loop body block
2849 executes and control returns to the loop conditional expression. If the loop
2850 conditional expression evaluates to `false`, the `while` expression completes.
2865 The keyword `loop` in Rust appears both in _loop expressions_ and in _continue expressions_.
2866 A loop expression denotes an infinite loop;
2867 see [Continue expressions](#continue-expressions) for continue expressions.
2869 ~~~~ {.notrust .ebnf .gram}
2870 loop_expr : [ lifetime ':' ] "loop" '{' block '}';
2873 A `loop` expression may optionally have a _label_.
2874 If a label is present,
2875 then labeled `break` and `loop` expressions nested within this loop may exit out of this loop or return control to its head.
2876 See [Break expressions](#break-expressions).
2878 ### Break expressions
2880 ~~~~ {.notrust .ebnf .gram}
2881 break_expr : "break" [ lifetime ];
2884 A `break` expression has an optional `label`.
2885 If the label is absent, then executing a `break` expression immediately terminates the innermost loop enclosing it.
2886 It is only permitted in the body of a loop.
2887 If the label is present, then `break foo` terminates the loop with label `foo`,
2888 which need not be the innermost label enclosing the `break` expression,
2889 but must enclose it.
2891 ### Continue expressions
2893 ~~~~ {.notrust .ebnf .gram}
2894 continue_expr : "loop" [ lifetime ];
2897 A continue expression, written `loop`, also has an optional `label`.
2898 If the label is absent,
2899 then executing a `loop` expression immediately terminates the current iteration of the innermost loop enclosing it,
2900 returning control to the loop *head*.
2901 In the case of a `while` loop,
2902 the head is the conditional expression controlling the loop.
2903 In the case of a `for` loop, the head is the call-expression controlling the loop.
2904 If the label is present, then `loop foo` returns control to the head of the loop with label `foo`,
2905 which need not be the innermost label enclosing the `break` expression,
2906 but must enclose it.
2908 A `loop` expression is only permitted in the body of a loop.
2912 ~~~~ {.notrust .ebnf .gram}
2913 for_expr : "for" pat "in" expr '{' block '}' ;
2916 A `for` expression is a syntactic construct for looping over elements
2917 provided by an implementation of `std::iter::Iterator`.
2919 An example of a for loop over the contents of a vector:
2923 # fn bar(f: Foo) { }
2928 let v: &[Foo] = &[a, b, c];
2935 An example of a for loop over a series of integers:
2938 # fn bar(b:uint) { }
2939 for i in range(0u, 256) {
2946 ~~~~ {.notrust .ebnf .gram}
2947 if_expr : "if" expr '{' block '}'
2950 else_tail : "else" [ if_expr
2954 An `if` expression is a conditional branch in program control. The form of
2955 an `if` expression is a condition expression, followed by a consequent
2956 block, any number of `else if` conditions and blocks, and an optional
2957 trailing `else` block. The condition expressions must have type
2958 `bool`. If a condition expression evaluates to `true`, the
2959 consequent block is executed and any subsequent `else if` or `else`
2960 block is skipped. If a condition expression evaluates to `false`, the
2961 consequent block is skipped and any subsequent `else if` condition is
2962 evaluated. If all `if` and `else if` conditions evaluate to `false`
2963 then any `else` block is executed.
2965 ### Match expressions
2967 ~~~~ {.notrust .ebnf .gram}
2968 match_expr : "match" expr '{' match_arm [ '|' match_arm ] * '}' ;
2970 match_arm : match_pat "=>" [ expr "," | '{' block '}' ] ;
2972 match_pat : pat [ ".." pat ] ? [ "if" expr ] ;
2975 A `match` expression branches on a *pattern*. The exact form of matching that
2976 occurs depends on the pattern. Patterns consist of some combination of
2977 literals, destructured vectors or enum constructors, structures, records and
2978 tuples, variable binding specifications, wildcards (`..`), and placeholders
2979 (`_`). A `match` expression has a *head expression*, which is the value to
2980 compare to the patterns. The type of the patterns must equal the type of the
2983 In a pattern whose head expression has an `enum` type, a placeholder (`_`)
2984 stands for a *single* data field, whereas a wildcard `..` stands for *all* the
2985 fields of a particular variant. For example:
2988 enum List<X> { Nil, Cons(X, ~List<X>) }
2990 let x: List<int> = Cons(10, ~Cons(11, ~Nil));
2993 Cons(_, ~Nil) => fail!("singleton list"),
2995 Nil => fail!("empty list")
2999 The first pattern matches lists constructed by applying `Cons` to any head
3000 value, and a tail value of `~Nil`. The second pattern matches _any_ list
3001 constructed with `Cons`, ignoring the values of its arguments. The difference
3002 between `_` and `..` is that the pattern `C(_)` is only type-correct if `C` has
3003 exactly one argument, while the pattern `C(..)` is type-correct for any enum
3004 variant `C`, regardless of how many arguments `C` has.
3006 Used inside a vector pattern, `..` stands for any number of elements. This
3007 wildcard can be used at most once for a given vector, which implies that it
3008 cannot be used to specifically match elements that are at an unknown distance
3009 from both ends of a vector, like `[.., 42, ..]`. If followed by a variable name,
3010 it will bind the corresponding slice to the variable. Example:
3013 fn is_symmetric(list: &[uint]) -> bool {
3016 [x, ..inside, y] if x == y => is_symmetric(inside),
3022 let sym = &[0, 1, 4, 2, 4, 1, 0];
3023 let not_sym = &[0, 1, 7, 2, 4, 1, 0];
3024 assert!(is_symmetric(sym));
3025 assert!(!is_symmetric(not_sym));
3029 A `match` behaves differently depending on whether or not the head expression
3030 is an [lvalue or an rvalue](#lvalues-rvalues-and-temporaries).
3031 If the head expression is an rvalue, it is
3032 first evaluated into a temporary location, and the resulting value
3033 is sequentially compared to the patterns in the arms until a match
3034 is found. The first arm with a matching pattern is chosen as the branch target
3035 of the `match`, any variables bound by the pattern are assigned to local
3036 variables in the arm's block, and control enters the block.
3038 When the head expression is an lvalue, the match does not allocate a
3039 temporary location (however, a by-value binding may copy or move from
3040 the lvalue). When possible, it is preferable to match on lvalues, as the
3041 lifetime of these matches inherits the lifetime of the lvalue, rather
3042 than being restricted to the inside of the match.
3044 An example of a `match` expression:
3047 # fn process_pair(a: int, b: int) { }
3048 # fn process_ten() { }
3050 enum List<X> { Nil, Cons(X, ~List<X>) }
3052 let x: List<int> = Cons(10, ~Cons(11, ~Nil));
3055 Cons(a, ~Cons(b, _)) => {
3070 Patterns that bind variables
3071 default to binding to a copy or move of the matched value
3072 (depending on the matched value's type).
3073 This can be changed to bind to a reference by
3074 using the `ref` keyword,
3075 or to a mutable reference using `ref mut`.
3077 Subpatterns can also be bound to variables by the use of the syntax
3078 `variable @ pattern`.
3082 enum List { Nil, Cons(uint, ~List) }
3084 fn is_sorted(list: &List) -> bool {
3086 Nil | Cons(_, ~Nil) => true,
3087 Cons(x, ref r @ ~Cons(y, _)) => (x <= y) && is_sorted(*r)
3092 let a = Cons(6, ~Cons(7, ~Cons(42, ~Nil)));
3093 assert!(is_sorted(&a));
3098 Patterns can also dereference pointers by using the `&`,
3099 `~` or `@` symbols, as appropriate. For example, these two matches
3100 on `x: &int` are equivalent:
3104 let y = match *x { 0 => "zero", _ => "some" };
3105 let z = match x { &0 => "zero", _ => "some" };
3110 A pattern that's just an identifier, like `Nil` in the previous example,
3111 could either refer to an enum variant that's in scope, or bind a new variable.
3112 The compiler resolves this ambiguity by forbidding variable bindings that occur
3113 in `match` patterns from shadowing names of variants that are in scope.
3114 For example, wherever `List` is in scope,
3115 a `match` pattern would not be able to bind `Nil` as a new name.
3116 The compiler interprets a variable pattern `x` as a binding _only_ if there is
3117 no variant named `x` in scope.
3118 A convention you can use to avoid conflicts is simply to name variants with
3119 upper-case letters, and local variables with lower-case letters.
3121 Multiple match patterns may be joined with the `|` operator.
3122 A range of values may be specified with `..`.
3128 let message = match x {
3129 0 | 1 => "not many",
3135 Range patterns only work on scalar types
3136 (like integers and characters; not like vectors and structs, which have sub-components).
3137 A range pattern may not be a sub-range of another range pattern inside the same `match`.
3139 Finally, match patterns can accept *pattern guards* to further refine the
3140 criteria for matching a case. Pattern guards appear after the pattern and
3141 consist of a bool-typed expression following the `if` keyword. A pattern
3142 guard may refer to the variables bound within the pattern they follow.
3145 # let maybe_digit = Some(0);
3146 # fn process_digit(i: int) { }
3147 # fn process_other(i: int) { }
3149 let message = match maybe_digit {
3150 Some(x) if x < 10 => process_digit(x),
3151 Some(x) => process_other(x),
3156 ### Return expressions
3158 ~~~~ {.notrust .ebnf .gram}
3159 return_expr : "return" expr ? ;
3162 Return expressions are denoted with the keyword `return`. Evaluating a `return`
3163 expression moves its argument into the output slot of the current
3164 function, destroys the current function activation frame, and transfers
3165 control to the caller frame.
3167 An example of a `return` expression:
3170 fn max(a: int, b: int) -> int {
3182 Every slot, item and value in a Rust program has a type. The _type_ of a *value*
3183 defines the interpretation of the memory holding it.
3185 Built-in types and type-constructors are tightly integrated into the language,
3186 in nontrivial ways that are not possible to emulate in user-defined
3187 types. User-defined types have limited capabilities.
3191 The primitive types are the following:
3193 * The "unit" type `()`, having the single "unit" value `()` (occasionally called "nil").
3194 ^[The "unit" value `()` is *not* a sentinel "null pointer" value for reference slots; the "unit" type is the implicit return type from functions otherwise lacking a return type, and can be used in other contexts (such as message-sending or type-parametric code) as a zero-size type.]
3195 * The boolean type `bool` with values `true` and `false`.
3196 * The machine types.
3197 * The machine-dependent integer and floating-point types.
3201 The machine types are the following:
3203 * The unsigned word types `u8`, `u16`, `u32` and `u64`, with values drawn from
3204 the integer intervals $[0, 2^8 - 1]$, $[0, 2^{16} - 1]$, $[0, 2^{32} - 1]$ and
3205 $[0, 2^{64} - 1]$ respectively.
3207 * The signed two's complement word types `i8`, `i16`, `i32` and `i64`, with
3208 values drawn from the integer intervals $[-(2^7), 2^7 - 1]$,
3209 $[-(2^{15}), 2^{15} - 1]$, $[-(2^{31}), 2^{31} - 1]$, $[-(2^{63}), 2^{63} - 1]$
3212 * The IEEE 754-2008 `binary32` and `binary64` floating-point types: `f32` and
3213 `f64`, respectively.
3215 #### Machine-dependent integer types
3217 The Rust type `uint`^[A Rust `uint` is analogous to a C99 `uintptr_t`.] is an
3218 unsigned integer type with target-machine-dependent size. Its size, in
3219 bits, is equal to the number of bits required to hold any memory address on
3222 The Rust type `int`^[A Rust `int` is analogous to a C99 `intptr_t`.] is a
3223 two's complement signed integer type with target-machine-dependent size. Its
3224 size, in bits, is equal to the size of the rust type `uint` on the same target
3229 The types `char` and `str` hold textual data.
3231 A value of type `char` is a [Unicode scalar value](
3232 http://www.unicode.org/glossary/#unicode_scalar_value)
3233 (ie. a code point that is not a surrogate),
3234 represented as a 32-bit unsigned word in the 0x0000 to 0xD7FF
3235 or 0xE000 to 0x10FFFF range.
3236 A `[char]` vector is effectively an UCS-4 / UTF-32 string.
3238 A value of type `str` is a Unicode string,
3239 represented as a vector of 8-bit unsigned bytes holding a sequence of UTF-8 codepoints.
3240 Since `str` is of unknown size, it is not a _first class_ type,
3241 but can only be instantiated through a pointer type,
3242 such as `&str` or `~str`.
3246 The tuple type-constructor forms a new heterogeneous product of values similar
3247 to the record type-constructor. The differences are as follows:
3249 * tuple elements cannot be mutable, unlike record fields
3250 * tuple elements are not named and can be accessed only by pattern-matching
3252 Tuple types and values are denoted by listing the types or values of their
3253 elements, respectively, in a parenthesized, comma-separated
3256 The members of a tuple are laid out in memory contiguously, like a record, in
3257 order specified by the tuple type.
3259 An example of a tuple type and its use:
3262 type Pair<'a> = (int,&'a str);
3263 let p: Pair<'static> = (10,"hello");
3265 assert!(b != "world");
3270 The vector type constructor represents a homogeneous array of values of a given type.
3271 A vector has a fixed size.
3272 (Operations like `vec.push` operate solely on owned vectors.)
3273 A vector type can be annotated with a _definite_ size, such as `[int, ..10]`.
3274 Such a definite-sized vector type is a first-class type, since its size is known statically.
3275 A vector without such a size is said to be of _indefinite_ size,
3276 and is therefore not a _first-class_ type.
3277 An indefinite-size vector can only be instantiated through a pointer type,
3278 such as `&[T]` or `~[T]`.
3279 The kind of a vector type depends on the kind of its element type,
3280 as with other simple structural types.
3282 Expressions producing vectors of definite size cannot be evaluated in a
3283 context expecting a vector of indefinite size; one must copy the
3284 definite-sized vector contents into a distinct vector of indefinite size.
3286 An example of a vector type and its use:
3289 let v: &[int] = &[7, 5, 3];
3294 All in-bounds elements of a vector are always initialized,
3295 and access to a vector is always bounds-checked.
3299 A `struct` *type* is a heterogeneous product of other types, called the *fields* of the type.
3300 ^[`struct` types are analogous `struct` types in C,
3301 the *record* types of the ML family,
3302 or the *structure* types of the Lisp family.]
3304 New instances of a `struct` can be constructed with a [struct expression](#structure-expressions).
3306 The memory order of fields in a `struct` is given by the item defining it.
3307 Fields may be given in any order in a corresponding struct *expression*;
3308 the resulting `struct` value will always be laid out in memory in the order specified by the corresponding *item*.
3310 The fields of a `struct` may be qualified by [visibility modifiers](#re-exporting-and-visibility),
3311 to restrict access to implementation-private data in a structure.
3313 A _tuple struct_ type is just like a structure type, except that the fields are anonymous.
3315 A _unit-like struct_ type is like a structure type, except that it has no fields.
3316 The one value constructed by the associated [structure expression](#structure-expressions)
3317 is the only value that inhabits such a type.
3319 ### Enumerated types
3321 An *enumerated type* is a nominal, heterogeneous disjoint union type,
3322 denoted by the name of an [`enum` item](#enumerations).
3323 ^[The `enum` type is analogous to a `data` constructor declaration in ML,
3324 or a *pick ADT* in Limbo.]
3326 An [`enum` item](#enumerations) declares both the type and a number of *variant constructors*,
3327 each of which is independently named and takes an optional tuple of arguments.
3329 New instances of an `enum` can be constructed by calling one of the variant constructors,
3330 in a [call expression](#call-expressions).
3332 Any `enum` value consumes as much memory as the largest variant constructor for its corresponding `enum` type.
3334 Enum types cannot be denoted *structurally* as types,
3335 but must be denoted by named reference to an [`enum` item](#enumerations).
3339 Nominal types -- [enumerations](#enumerated-types) and [structures](#structure-types) -- may be recursive.
3340 That is, each `enum` constructor or `struct` field may refer, directly or indirectly, to the enclosing `enum` or `struct` type itself.
3341 Such recursion has restrictions:
3343 * Recursive types must include a nominal type in the recursion
3344 (not mere [type definitions](#type-definitions),
3345 or other structural types such as [vectors](#vector-types) or [tuples](#tuple-types)).
3346 * A recursive `enum` item must have at least one non-recursive constructor
3347 (in order to give the recursion a basis case).
3348 * The size of a recursive type must be finite;
3349 in other words the recursive fields of the type must be [pointer types](#pointer-types).
3350 * Recursive type definitions can cross module boundaries, but not module *visibility* boundaries,
3351 or crate boundaries (in order to simplify the module system and type checker).
3353 An example of a *recursive* type and its use:
3361 let a: List<int> = Cons(7, ~Cons(13, ~Nil));
3366 All pointers in Rust are explicit first-class values.
3367 They can be copied, stored into data structures, and returned from functions.
3368 There are four varieties of pointer in Rust:
3370 Owning pointers (`~`)
3371 : These point to owned heap allocations (or "boxes") in the shared, inter-task heap.
3372 Each owned box has a single owning pointer; pointer and pointee retain a 1:1 relationship at all times.
3373 Owning pointers are written `~content`,
3374 for example `~int` means an owning pointer to an owned box containing an integer.
3375 Copying an owned box is a "deep" operation:
3376 it involves allocating a new owned box and copying the contents of the old box into the new box.
3377 Releasing an owning pointer immediately releases its corresponding owned box.
3380 : These point to memory _owned by some other value_.
3381 References arise by (automatic) conversion from owning pointers, managed pointers,
3382 or by applying the borrowing operator `&` to some other value,
3383 including [lvalues, rvalues or temporaries](#lvalues-rvalues-and-temporaries).
3384 References are written `&content`, or in some cases `&'f content` for some lifetime-variable `f`,
3385 for example `&int` means a reference to an integer.
3386 Copying a reference is a "shallow" operation:
3387 it involves only copying the pointer itself.
3388 Releasing a reference typically has no effect on the value it points to,
3389 with the exception of temporary values,
3390 which are released when the last reference to them is released.
3393 : Raw pointers are pointers without safety or liveness guarantees.
3394 Raw pointers are written `*content`,
3395 for example `*int` means a raw pointer to an integer.
3396 Copying or dropping a raw pointer has no effect on the lifecycle of any other value.
3397 Dereferencing a raw pointer or converting it to any other pointer type is an [`unsafe` operation](#unsafe-functions).
3398 Raw pointers are generally discouraged in Rust code;
3399 they exist to support interoperability with foreign code,
3400 and writing performance-critical or low-level functions.
3404 The function type constructor `fn` forms new function types.
3405 A function type consists of a possibly-empty set of function-type modifiers
3406 (such as `unsafe` or `extern`), a sequence of input types and an output type.
3408 An example of a `fn` type:
3411 fn add(x: int, y: int) -> int {
3415 let mut x = add(5,7);
3417 type Binop<'a> = 'a |int,int| -> int;
3418 let bo: Binop = add;
3424 The type of a closure mapping an input of type `A` to an output of type `B` is `|A| -> B`. A closure with no arguments or return values has type `||`.
3427 An example of creating and calling a closure:
3430 let captured_var = 10;
3432 let closure_no_args = || println!("captured_var={}", captured_var);
3434 let closure_args = |arg: int| -> int {
3435 println!("captured_var={}, arg={}", captured_var, arg);
3436 arg // Note lack of semicolon after 'arg'
3439 fn call_closure(c1: ||, c2: |int| -> int) {
3444 call_closure(closure_no_args, closure_args);
3450 Every trait item (see [traits](#traits)) defines a type with the same name as the trait.
3451 This type is called the _object type_ of the trait.
3452 Object types permit "late binding" of methods, dispatched using _virtual method tables_ ("vtables").
3453 Whereas most calls to trait methods are "early bound" (statically resolved) to specific implementations at compile time,
3454 a call to a method on an object type is only resolved to a vtable entry at compile time.
3455 The actual implementation for each vtable entry can vary on an object-by-object basis.
3457 Given a pointer-typed expression `E` of type `&T` or `~T`, where `T` implements trait `R`,
3458 casting `E` to the corresponding pointer type `&R` or `~R` results in a value of the _object type_ `R`.
3459 This result is represented as a pair of pointers:
3460 the vtable pointer for the `T` implementation of `R`, and the pointer value of `E`.
3462 An example of an object type:
3466 fn to_string(&self) -> ~str;
3469 impl Printable for int {
3470 fn to_string(&self) -> ~str { self.to_str() }
3473 fn print(a: ~Printable) {
3474 println!("{}", a.to_string());
3478 print(~10 as ~Printable);
3482 In this example, the trait `Printable` occurs as an object type in both the type signature of `print`,
3483 and the cast expression in `main`.
3487 Within the body of an item that has type parameter declarations, the names of its type parameters are types:
3490 fn map<A: Clone, B: Clone>(f: |A| -> B, xs: &[A]) -> ~[B] {
3494 let first: B = f(xs[0].clone());
3495 let rest: ~[B] = map(f, xs.slice(1, xs.len()));
3496 return ~[first] + rest;
3500 Here, `first` has type `B`, referring to `map`'s `B` type parameter;
3501 and `rest` has type `~[B]`, a vector type with element type `B`.
3505 The special type `self` has a meaning within methods inside an
3506 impl item. It refers to the type of the implicit `self` argument. For
3511 fn make_string(&self) -> ~str;
3514 impl Printable for ~str {
3515 fn make_string(&self) -> ~str {
3521 `self` refers to the value of type `~str` that is the receiver for a
3522 call to the method `make_string`.
3526 Types in Rust are categorized into kinds, based on various properties of the components of the type.
3530 : Types of this kind can be safely sent between tasks.
3531 This kind includes scalars, owning pointers, owned closures, and
3532 structural types containing only other owned types.
3533 All `Send` types are `'static`.
3535 : Types of this kind consist of "Plain Old Data"
3536 which can be copied by simply moving bits.
3537 All values of this kind can be implicitly copied.
3538 This kind includes scalars and immutable references,
3539 as well as structural types containing other `Copy` types.
3541 : Types of this kind do not contain any references (except for
3542 references with the `static` lifetime, which are allowed).
3543 This can be a useful guarantee for code
3544 that breaks borrowing assumptions
3545 using [`unsafe` operations](#unsafe-functions).
3547 : This is not strictly a kind,
3548 but its presence interacts with kinds:
3549 the `Drop` trait provides a single method `drop`
3550 that takes no parameters,
3551 and is run when values of the type are dropped.
3552 Such a method is called a "destructor",
3553 and are always executed in "top-down" order:
3554 a value is completely destroyed
3555 before any of the values it owns run their destructors.
3556 Only `Send` types can implement `Drop`.
3559 : Types with destructors, closure environments,
3560 and various other _non-first-class_ types,
3561 are not copyable at all.
3562 Such types can usually only be accessed through pointers,
3563 or in some cases, moved between mutable locations.
3565 Kinds can be supplied as _bounds_ on type parameters, like traits,
3566 in which case the parameter is constrained to types satisfying that kind.
3568 By default, type parameters do not carry any assumed kind-bounds at all.
3569 When instantiating a type parameter,
3570 the kind bounds on the parameter are checked
3571 to be the same or narrower than the kind
3572 of the type that it is instantiated with.
3574 Sending operations are not part of the Rust language,
3575 but are implemented in the library.
3576 Generic functions that send values
3577 bound the kind of these values to sendable.
3579 # Memory and concurrency models
3581 Rust has a memory model centered around concurrently-executing _tasks_. Thus
3582 its memory model and its concurrency model are best discussed simultaneously,
3583 as parts of each only make sense when considered from the perspective of the
3586 When reading about the memory model, keep in mind that it is partitioned in
3587 order to support tasks; and when reading about tasks, keep in mind that their
3588 isolation and communication mechanisms are only possible due to the ownership
3589 and lifetime semantics of the memory model.
3593 A Rust program's memory consists of a static set of *items*, a set of
3594 [tasks](#tasks) each with its own *stack*, and a *heap*. Immutable portions of
3595 the heap may be shared between tasks, mutable portions may not.
3597 Allocations in the stack consist of *slots*, and allocations in the heap
3600 ### Memory allocation and lifetime
3602 The _items_ of a program are those functions, modules and types
3603 that have their value calculated at compile-time and stored uniquely in the
3604 memory image of the rust process. Items are neither dynamically allocated nor
3607 A task's _stack_ consists of activation frames automatically allocated on
3608 entry to each function as the task executes. A stack allocation is reclaimed
3609 when control leaves the frame containing it.
3611 The _heap_ is a general term that describes two separate sets of boxes:
3612 managed boxes -- which may be subject to garbage collection -- and owned
3613 boxes. The lifetime of an allocation in the heap depends on the lifetime of
3614 the box values pointing to it. Since box values may themselves be passed in
3615 and out of frames, or stored in the heap, heap allocations may outlive the
3616 frame they are allocated within.
3618 ### Memory ownership
3620 A task owns all memory it can *safely* reach through local variables,
3621 as well as managed, owned boxes and references.
3623 When a task sends a value that has the `Send` trait to another task,
3624 it loses ownership of the value sent and can no longer refer to it.
3625 This is statically guaranteed by the combined use of "move semantics",
3626 and the compiler-checked _meaning_ of the `Send` trait:
3627 it is only instantiated for (transitively) sendable kinds of data constructor and pointers,
3628 never including managed boxes or references.
3630 When a stack frame is exited, its local allocations are all released, and its
3631 references to boxes (both managed and owned) are dropped.
3633 A managed box may (in the case of a recursive, mutable managed type) be cyclic;
3634 in this case the release of memory inside the managed structure may be deferred
3635 until task-local garbage collection can reclaim it. Code can ensure no such
3636 delayed deallocation occurs by restricting itself to owned boxes and similar
3637 unmanaged kinds of data.
3639 When a task finishes, its stack is necessarily empty and it therefore has no
3640 references to any boxes; the remainder of its heap is immediately freed.
3644 A task's stack contains slots.
3646 A _slot_ is a component of a stack frame, either a function parameter,
3647 a [temporary](#lvalues-rvalues-and-temporaries), or a local variable.
3649 A _local variable_ (or *stack-local* allocation) holds a value directly,
3650 allocated within the stack's memory. The value is a part of the stack frame.
3652 Local variables are immutable unless declared otherwise like: `let mut x = ...`.
3654 Function parameters are immutable unless declared with `mut`. The
3655 `mut` keyword applies only to the following parameter (so `|mut x, y|`
3656 and `fn f(mut x: ~int, y: ~int)` declare one mutable variable `x` and
3657 one immutable variable `y`).
3659 Methods that take either `self` or `~self` can optionally place them in a
3660 mutable slot by prefixing them with `mut` (similar to regular arguments):
3664 fn change(mut self) -> Self;
3665 fn modify(mut ~self) -> ~Self;
3669 Local variables are not initialized when allocated; the entire frame worth of
3670 local variables are allocated at once, on frame-entry, in an uninitialized
3671 state. Subsequent statements within a function may or may not initialize the
3672 local variables. Local variables can be used only after they have been
3673 initialized; this is enforced by the compiler.
3677 An _owned box_ is a reference to a heap allocation holding another value, which is constructed
3678 by the prefix *tilde* sigil `~`
3680 An example of an owned box type and value:
3686 Owned box values exist in 1:1 correspondence with their heap allocation
3687 copying an owned box value makes a shallow copy of the pointer
3688 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.
3693 // attempting to use `x` will result in an error here
3700 An executing Rust program consists of a tree of tasks.
3701 A Rust _task_ consists of an entry function, a stack,
3702 a set of outgoing communication channels and incoming communication ports,
3703 and ownership of some portion of the heap of a single operating-system process.
3704 (We expect that many programs will not use channels and ports directly,
3705 but will instead use higher-level abstractions provided in standard libraries,
3708 Multiple Rust tasks may coexist in a single operating-system process.
3709 The runtime scheduler maps tasks to a certain number of operating-system threads.
3710 By default, the scheduler chooses the number of threads based on
3711 the number of concurrent physical CPUs detected at startup.
3712 It's also possible to override this choice at runtime.
3713 When the number of tasks exceeds the number of threads -- which is likely --
3714 the scheduler multiplexes the tasks onto threads.^[
3715 This is an M:N scheduler,
3716 which is known to give suboptimal results for CPU-bound concurrency problems.
3717 In such cases, running with the same number of threads and tasks can yield better results.
3718 Rust has M:N scheduling in order to support very large numbers of tasks
3719 in contexts where threads are too resource-intensive to use in large number.
3720 The cost of threads varies substantially per operating system, and is sometimes quite low,
3721 so this flexibility is not always worth exploiting.]
3723 ### Communication between tasks
3725 Rust tasks are isolated and generally unable to interfere with one another's memory directly,
3726 except through [`unsafe` code](#unsafe-functions).
3727 All contact between tasks is mediated by safe forms of ownership transfer,
3728 and data races on memory are prohibited by the type system.
3730 Inter-task communication and co-ordination facilities are provided in the standard library.
3733 - synchronous and asynchronous communication channels with various communication topologies
3734 - read-only and read-write shared variables with various safe mutual exclusion patterns
3735 - simple locks and semaphores
3737 When such facilities carry values, the values are restricted to the [`Send` type-kind](#type-kinds).
3738 Restricting communication interfaces to this kind ensures that no references or managed pointers move between tasks.
3739 Thus access to an entire data structure can be mediated through its owning "root" value;
3740 no further locking or copying is required to avoid data races within the substructure of such a value.
3744 The _lifecycle_ of a task consists of a finite set of states and events
3745 that cause transitions between the states. The lifecycle states of a task are:
3752 A task begins its lifecycle -- once it has been spawned -- in the *running*
3753 state. In this state it executes the statements of its entry function, and any
3754 functions called by the entry function.
3756 A task may transition from the *running* state to the *blocked*
3757 state any time it makes a blocking communication call. When the
3758 call can be completed -- when a message arrives at a sender, or a
3759 buffer opens to receive a message -- then the blocked task will
3760 unblock and transition back to *running*.
3762 A task may transition to the *failing* state at any time, due being
3763 killed by some external event or internally, from the evaluation of a
3764 `fail!()` macro. Once *failing*, a task unwinds its stack and
3765 transitions to the *dead* state. Unwinding the stack of a task is done by
3766 the task itself, on its own control stack. If a value with a destructor is
3767 freed during unwinding, the code for the destructor is run, also on the task's
3768 control stack. Running the destructor code causes a temporary transition to a
3769 *running* state, and allows the destructor code to cause any subsequent
3770 state transitions. The original task of unwinding and failing thereby may
3771 suspend temporarily, and may involve (recursive) unwinding of the stack of a
3772 failed destructor. Nonetheless, the outermost unwinding activity will continue
3773 until the stack is unwound and the task transitions to the *dead*
3774 state. There is no way to "recover" from task failure. Once a task has
3775 temporarily suspended its unwinding in the *failing* state, failure
3776 occurring from within this destructor results in *hard* failure.
3777 A hard failure currently results in the process aborting.
3779 A task in the *dead* state cannot transition to other states; it exists
3780 only to have its termination status inspected by other tasks, and/or to await
3781 reclamation when the last reference to it drops.
3785 The currently scheduled task is given a finite *time slice* in which to
3786 execute, after which it is *descheduled* at a loop-edge or similar
3787 preemption point, and another task within is scheduled, pseudo-randomly.
3789 An executing task can yield control at any time, by making a library call to
3790 `std::task::yield`, which deschedules it immediately. Entering any other
3791 non-executing state (blocked, dead) similarly deschedules the task.
3793 # Runtime services, linkage and debugging
3795 The Rust _runtime_ is a relatively compact collection of C++ and Rust code
3796 that provides fundamental services and datatypes to all Rust tasks at
3797 run-time. It is smaller and simpler than many modern language runtimes. It is
3798 tightly integrated into the language's execution model of memory, tasks,
3799 communication and logging.
3801 > **Note:** The runtime library will merge with the `std` library in future versions of Rust.
3803 ### Memory allocation
3805 The runtime memory-management system is based on a _service-provider interface_,
3806 through which the runtime requests blocks of memory from its environment
3807 and releases them back to its environment when they are no longer needed.
3808 The default implementation of the service-provider interface
3809 consists of the C runtime functions `malloc` and `free`.
3811 The runtime memory-management system, in turn, supplies Rust tasks with
3812 facilities for allocating releasing stacks, as well as allocating and freeing
3817 The runtime provides C and Rust code to assist with various built-in types,
3818 such as vectors, strings, and the low level communication system (ports,
3821 Support for other built-in types such as simple types, tuples, records, and
3822 enums is open-coded by the Rust compiler.
3824 ### Task scheduling and communication
3826 The runtime provides code to manage inter-task communication. This includes
3827 the system of task-lifecycle state transitions depending on the contents of
3828 queues, as well as code to copy values between queues and their recipients and
3829 to serialize values for transmission over operating-system inter-process
3830 communication facilities.
3834 The Rust compiler supports various methods to link crates together both
3835 statically and dynamically. This section will explore the various methods to
3836 link Rust crates together, and more information about native libraries can be
3837 found in the [ffi tutorial][ffi].
3839 In one session of compilation, the compiler can generate multiple artifacts
3840 through the usage of command line flags and the `crate_type` attribute.
3842 * `--crate-type=bin`, `#[crate_type = "bin"]` - A runnable executable will be
3843 produced. This requires that there is a `main` function in the crate which
3844 will be run when the program begins executing. This will link in all Rust and
3845 native dependencies, producing a distributable binary.
3847 * `--crate-type=lib`, `#[crate_type = "lib"]` - A Rust library will be produced.
3848 This is an ambiguous concept as to what exactly is produced because a library
3849 can manifest itself in several forms. The purpose of this generic `lib` option
3850 is to generate the "compiler recommended" style of library. The output library
3851 will always be usable by rustc, but the actual type of library may change from
3852 time-to-time. The remaining output types are all different flavors of
3853 libraries, and the `lib` type can be seen as an alias for one of them (but the
3854 actual one is compiler-defined).
3856 * `--crate-type=dylib`, `#[crate_type = "dylib"]` - A dynamic Rust library will
3857 be produced. This is different from the `lib` output type in that this forces
3858 dynamic library generation. The resulting dynamic library can be used as a
3859 dependency for other libraries and/or executables. This output type will
3860 create `*.so` files on linux, `*.dylib` files on osx, and `*.dll` files on
3863 * `--crate-type=staticlib`, `#[crate_type = "staticlib"]` - A static system
3864 library will be produced. This is different from other library outputs in that
3865 the Rust compiler will never attempt to link to `staticlib` outputs. The
3866 purpose of this output type is to create a static library containing all of
3867 the local crate's code along with all upstream dependencies. The static
3868 library is actually a `*.a` archive on linux and osx and a `*.lib` file on
3869 windows. This format is recommended for use in situtations such as linking
3870 Rust code into an existing non-Rust application because it will not have
3871 dynamic dependencies on other Rust code.
3873 * `--crate-type=rlib`, `#[crate_type = "rlib"]` - A "Rust library" file will be
3874 produced. This is used as an intermediate artifact and can be thought of as a
3875 "static Rust library". These `rlib` files, unlike `staticlib` files, are
3876 interpreted by the Rust compiler in future linkage. This essentially means
3877 that `rustc` will look for metadata in `rlib` files like it looks for metadata
3878 in dynamic libraries. This form of output is used to produce statically linked
3879 executables as well as `staticlib` outputs.
3881 Note that these outputs are stackable in the sense that if multiple are
3882 specified, then the compiler will produce each form of output at once without
3883 having to recompile.
3885 With all these different kinds of outputs, if crate A depends on crate B, then
3886 the compiler could find B in various different forms throughout the system. The
3887 only forms looked for by the compiler, however, are the `rlib` format and the
3888 dynamic library format. With these two options for a dependent library, the
3889 compiler must at some point make a choice between these two formats. With this
3890 in mind, the compiler follows these rules when determining what format of
3891 dependencies will be used:
3893 1. If a dynamic library is being produced, then it is required for all upstream
3894 Rust dependencies to also be dynamic. This is a limitation of the current
3895 implementation of the linkage model. The reason behind this limitation is to
3896 prevent multiple copies of the same upstream library from showing up, and in
3897 the future it is planned to support a mixture of dynamic and static linking.
3899 When producing a dynamic library, the compiler will generate an error if an
3900 upstream dependency could not be found, and also if an upstream dependency
3901 could only be found in an `rlib` format. Remember that `staticlib` formats
3902 are always ignored by `rustc` for crate-linking purposes.
3904 2. If a static library is being produced, all upstream dependecies are
3905 required to be available in `rlib` formats. This requirement stems from the
3906 same reasons that a dynamic library must have all dynamic dependencies.
3908 Note that it is impossible to link in native dynamic dependencies to a static
3909 library, and in this case warnings will be printed about all unlinked native
3910 dynamic dependencies.
3912 3. If an `rlib` file is being produced, then there are no restrictions on what
3913 format the upstream dependencies are available in. It is simply required that
3914 all upstream dependencies be available for reading metadata from.
3916 The reason for this is that `rlib` files do not contain any of their upstream
3917 dependencies. It wouldn't be very efficient for all `rlib` files to contain a
3918 copy of `libstd.rlib`!
3920 4. If an executable is being produced, then things get a little interesting. As
3921 with the above limitations in dynamic and static libraries, it is required
3922 for all upstream dependencies to be in the same format. The next question is
3923 whether to prefer a dynamic or a static format. The compiler currently favors
3924 static linking over dynamic linking, but this can be inverted with the `-C
3925 prefer-dynamic` flag to the compiler.
3927 What this means is that first the compiler will attempt to find all upstream
3928 dependencies as `rlib` files, and if successful, it will create a statically
3929 linked executable. If an upstream dependency is missing as an `rlib` file,
3930 then the compiler will force all dependencies to be dynamic and will generate
3931 errors if dynamic versions could not be found.
3933 In general, `--crate-type=bin` or `--crate-type=lib` should be sufficient for
3934 all compilation needs, and the other options are just available if more
3935 fine-grained control is desired over the output format of a Rust crate.
3939 The runtime contains a system for directing [logging
3940 expressions](#logging-expressions) to a logging console and/or internal logging
3941 buffers. Logging can be enabled per module.
3943 Logging output is enabled by setting the `RUST_LOG` environment
3944 variable. `RUST_LOG` accepts a logging specification made up of a
3945 comma-separated list of paths, with optional log levels. For each
3946 module containing log expressions, if `RUST_LOG` contains the path to
3947 that module or a parent of that module, then logs of the appropriate
3948 level will be output to the console.
3950 The path to a module consists of the crate name, any parent modules,
3951 then the module itself, all separated by double colons (`::`). The
3952 optional log level can be appended to the module path with an equals
3953 sign (`=`) followed by the log level, from 1 to 4, inclusive. Level 1
3954 is the error level, 2 is warning, 3 info, and 4 debug. You can also
3955 use the symbolic constants `error`, `warn`, `info`, and `debug`. Any
3956 logs less than or equal to the specified level will be output. If not
3957 specified then log level 4 is assumed. Debug messages can be omitted
3958 by passing `--cfg ndebug` to `rustc`.
3960 As an example, to see all the logs generated by the compiler, you would set
3961 `RUST_LOG` to `rustc`, which is the crate name (as specified in its `crate_id`
3962 [attribute](#attributes)). To narrow down the logs to just crate resolution,
3963 you would set it to `rustc::metadata::creader`. To see just error logging
3966 Note that when compiling source files that don't specify a
3967 crate name the crate is given a default name that matches the source file,
3968 with the extension removed. In that case, to turn on logging for a program
3969 compiled from, e.g. `helloworld.rs`, `RUST_LOG` should be set to `helloworld`.
3971 As a convenience, the logging spec can also be set to a special pseudo-crate,
3972 `::help`. In this case, when the application starts, the runtime will
3973 simply output a list of loaded modules containing log expressions, then exit.
3975 #### Logging Expressions
3977 Rust provides several macros to log information. Here's a simple Rust program
3978 that demonstrates all four of them:
3982 #[phase(syntax, link)] extern crate log;
3985 error!("This is an error log")
3986 warn!("This is a warn log")
3987 info!("this is an info log")
3988 debug!("This is a debug log")
3992 These four log levels correspond to levels 1-4, as controlled by `RUST_LOG`:
3994 ``` {.bash .notrust}
3995 $ RUST_LOG=rust=3 ./rust
3996 This is an error log
4001 # Appendix: Rationales and design tradeoffs
4005 # Appendix: Influences and further references
4009 > The essential problem that must be solved in making a fault-tolerant
4010 > software system is therefore that of fault-isolation. Different programmers
4011 > will write different modules, some modules will be correct, others will have
4012 > errors. We do not want the errors in one module to adversely affect the
4013 > behaviour of a module which does not have any errors.
4015 > — Joe Armstrong
4017 > In our approach, all data is private to some process, and processes can
4018 > only communicate through communications channels. *Security*, as used
4019 > in this paper, is the property which guarantees that processes in a system
4020 > cannot affect each other except by explicit communication.
4022 > When security is absent, nothing which can be proven about a single module
4023 > in isolation can be guaranteed to hold when that module is embedded in a
4026 > — Robert Strom and Shaula Yemini
4028 > Concurrent and applicative programming complement each other. The
4029 > ability to send messages on channels provides I/O without side effects,
4030 > while the avoidance of shared data helps keep concurrent processes from
4035 Rust is not a particularly original language. It may however appear unusual
4036 by contemporary standards, as its design elements are drawn from a number of
4037 "historical" languages that have, with a few exceptions, fallen out of
4038 favour. Five prominent lineages contribute the most, though their influences
4039 have come and gone during the course of Rust's development:
4041 * The NIL (1981) and Hermes (1990) family. These languages were developed by
4042 Robert Strom, Shaula Yemini, David Bacon and others in their group at IBM
4043 Watson Research Center (Yorktown Heights, NY, USA).
4045 * The Erlang (1987) language, developed by Joe Armstrong, Robert Virding, Claes
4046 Wikström, Mike Williams and others in their group at the Ericsson Computer
4047 Science Laboratory (Älvsjö, Stockholm, Sweden) .
4049 * The Sather (1990) language, developed by Stephen Omohundro, Chu-Cheow Lim,
4050 Heinz Schmidt and others in their group at The International Computer
4051 Science Institute of the University of California, Berkeley (Berkeley, CA,
4054 * The Newsqueak (1988), Alef (1995), and Limbo (1996) family. These
4055 languages were developed by Rob Pike, Phil Winterbottom, Sean Dorward and
4056 others in their group at Bell Labs Computing Sciences Research Center
4057 (Murray Hill, NJ, USA).
4059 * The Napier (1985) and Napier88 (1988) family. These languages were
4060 developed by Malcolm Atkinson, Ron Morrison and others in their group at
4061 the University of St. Andrews (St. Andrews, Fife, UK).
4063 Additional specific influences can be seen from the following languages:
4065 * The structural algebraic types and compilation manager of SML.
4066 * The attribute and assembly systems of C#.
4067 * The references and deterministic destructor system of C++.
4068 * The memory region systems of the ML Kit and Cyclone.
4069 * The typeclass system of Haskell.
4070 * The lexical identifier rule of Python.
4071 * The block syntax of Ruby.
4073 [ffi]: guide-ffi.html