1 % The Rust Reference Manual
5 This document is the reference manual for the Rust programming language. It
6 provides three kinds of material:
8 - Chapters that formally define the language grammar and, for each
9 construct, informally describe its semantics and give examples of its
11 - Chapters that informally describe the memory model, concurrency model,
12 runtime services, linkage model and debugging facilities.
13 - Appendix chapters providing rationale and references to languages that
14 influenced the design.
16 This document does not serve as a tutorial introduction to the
17 language. Background familiarity with the language is assumed. A separate
18 [tutorial] document is available to help acquire such background familiarity.
20 This document also does not serve as a reference to the [standard]
21 library included in the language distribution. Those libraries are
22 documented separately by extracting documentation attributes from their
25 [tutorial]: tutorial.html
26 [standard]: std/index.html
30 Rust is a work in progress. The language continues to evolve as the design
31 shifts and is fleshed out in working code. Certain parts work, certain parts
32 do not, certain parts will be removed or changed.
34 This manual is a snapshot written in the present tense. All features described
35 exist in working code unless otherwise noted, but some are quite primitive or
36 remain to be further modified by planned work. Some may be temporary. It is a
37 *draft*, and we ask that you not take anything you read here as final.
39 If you have suggestions to make, please try to focus them on *reductions* to
40 the language: possible features that can be combined or omitted. We aim to
41 keep the size and complexity of the language under control.
43 > **Note:** The grammar for Rust given in this document is rough and
44 > very incomplete; only a modest number of sections have accompanying grammar
45 > rules. Formalizing the grammar accepted by the Rust parser is ongoing work,
46 > but future versions of this document will contain a complete
47 > grammar. Moreover, we hope that this grammar will be extracted and verified
48 > as LL(1) by an automated grammar-analysis tool, and further tested against the
49 > Rust sources. Preliminary versions of this automation exist, but are not yet
54 Rust's grammar is defined over Unicode codepoints, each conventionally
55 denoted `U+XXXX`, for 4 or more hexadecimal digits `X`. _Most_ of Rust's
56 grammar is confined to the ASCII range of Unicode, and is described in this
57 document by a dialect of Extended Backus-Naur Form (EBNF), specifically a
58 dialect of EBNF supported by common automated LL(k) parsing tools such as
59 `llgen`, rather than the dialect given in ISO 14977. The dialect can be
60 defined self-referentially as follows:
62 ~~~~ {.ebnf .notation}
64 rule : nonterminal ':' productionrule ';' ;
65 productionrule : production [ '|' production ] * ;
67 term : element repeats ;
68 element : LITERAL | IDENTIFIER | '[' productionrule ']' ;
69 repeats : [ '*' | '+' ] NUMBER ? | NUMBER ? | '?' ;
74 - Whitespace in the grammar is ignored.
75 - Square brackets are used to group rules.
76 - `LITERAL` is a single printable ASCII character, or an escaped hexadecimal
77 ASCII code of the form `\xQQ`, in single quotes, denoting the corresponding
78 Unicode codepoint `U+00QQ`.
79 - `IDENTIFIER` is a nonempty string of ASCII letters and underscores.
80 - The `repeat` forms apply to the adjacent `element`, and are as follows:
81 - `?` means zero or one repetition
82 - `*` means zero or more repetitions
83 - `+` means one or more repetitions
84 - NUMBER trailing a repeat symbol gives a maximum repetition count
85 - NUMBER on its own gives an exact repetition count
87 This EBNF dialect should hopefully be familiar to many readers.
89 ## Unicode productions
91 A few productions in Rust's grammar permit Unicode codepoints outside the ASCII range.
92 We define these productions in terms of character properties specified in the Unicode standard,
93 rather than in terms of ASCII-range codepoints.
94 The section [Special Unicode Productions](#special-unicode-productions) lists these productions.
96 ## String table productions
98 Some rules in the grammar — notably [unary
99 operators](#unary-operator-expressions), [binary
100 operators](#binary-operator-expressions), and [keywords](#keywords) —
101 are given in a simplified form: as a listing of a table of unquoted,
102 printable whitespace-separated strings. These cases form a subset of
103 the rules regarding the [token](#tokens) rule, and are assumed to be
104 the result of a lexical-analysis phase feeding the parser, driven by a
105 DFA, operating over the disjunction of all such string table entries.
107 When such a string enclosed in double-quotes (`"`) occurs inside the
108 grammar, it is an implicit reference to a single member of such a string table
109 production. See [tokens](#tokens) for more information.
115 Rust input is interpreted as a sequence of Unicode codepoints encoded in UTF-8.
116 Most Rust grammar rules are defined in terms of printable ASCII-range codepoints,
117 but a small number are defined in terms of Unicode properties or explicit
118 codepoint lists. [^inputformat]
120 [^inputformat]: Substitute definitions for the special Unicode productions are
121 provided to the grammar verifier, restricted to ASCII range, when verifying
122 the grammar in this document.
124 ## Special Unicode Productions
126 The following productions in the Rust grammar are defined in terms of Unicode properties:
127 `ident`, `non_null`, `non_star`, `non_eol`, `non_slash_or_star`, `non_single_quote` and `non_double_quote`.
131 The `ident` production is any nonempty Unicode string of the following form:
133 - The first character has property `XID_start`
134 - The remaining characters have property `XID_continue`
136 that does _not_ occur in the set of [keywords](#keywords).
138 > **Note**: `XID_start` and `XID_continue` as character properties cover the
139 > character ranges used to form the more familiar C and Java language-family
142 ### Delimiter-restricted productions
144 Some productions are defined by exclusion of particular Unicode characters:
146 - `non_null` is any single Unicode character aside from `U+0000` (null)
147 - `non_eol` is `non_null` restricted to exclude `U+000A` (`'\n'`)
148 - `non_star` is `non_null` restricted to exclude `U+002A` (`*`)
149 - `non_slash_or_star` is `non_null` restricted to exclude `U+002F` (`/`) and `U+002A` (`*`)
150 - `non_single_quote` is `non_null` restricted to exclude `U+0027` (`'`)
151 - `non_double_quote` is `non_null` restricted to exclude `U+0022` (`"`)
156 comment : block_comment | line_comment ;
157 block_comment : "/*" block_comment_body * '*' + '/' ;
158 block_comment_body : [block_comment | character] * ;
159 line_comment : "//" non_eol * ;
162 Comments in Rust code follow the general C++ style of line and block-comment forms.
163 Nested block comments are supported.
165 Line comments beginning with exactly _three_ slashes (`///`), and block
166 comments beginning with exactly one repeated asterisk in the block-open
167 sequence (`/**`), are interpreted as a special syntax for `doc`
168 [attributes](#attributes). That is, they are equivalent to writing
169 `#[doc="..."]` around the body of the comment (this includes the comment
170 characters themselves, ie `/// Foo` turns into `#[doc="/// Foo"]`).
172 `//!` comments apply to the parent of the comment, rather than the item that
173 follows. `//!` comments are usually used to display information on the crate
176 Non-doc comments are interpreted as a form of whitespace.
181 whitespace_char : '\x20' | '\x09' | '\x0a' | '\x0d' ;
182 whitespace : [ whitespace_char | comment ] + ;
185 The `whitespace_char` production is any nonempty Unicode string consisting of any
186 of the following Unicode characters: `U+0020` (space, `' '`), `U+0009` (tab,
187 `'\t'`), `U+000A` (LF, `'\n'`), `U+000D` (CR, `'\r'`).
189 Rust is a "free-form" language, meaning that all forms of whitespace serve
190 only to separate _tokens_ in the grammar, and have no semantic significance.
192 A Rust program has identical meaning if each whitespace element is replaced
193 with any other legal whitespace element, such as a single space character.
198 simple_token : keyword | unop | binop ;
199 token : simple_token | ident | literal | symbol | whitespace token ;
202 Tokens are primitive productions in the grammar defined by regular
203 (non-recursive) languages. "Simple" tokens are given in [string table
204 production](#string-table-productions) form, and occur in the rest of the
205 grammar as double-quoted strings. Other tokens have exact rules given.
209 The keywords are the following strings:
211 ~~~~ {.text .keyword}
222 self static struct super
228 Each of these keywords has special meaning in its grammar,
229 and all of them are excluded from the `ident` rule.
233 A literal is an expression consisting of a single token, rather than a
234 sequence of tokens, that immediately and directly denotes the value it
235 evaluates to, rather than referring to it by name or some other evaluation
236 rule. A literal is a form of constant expression, so is evaluated (primarily)
240 literal : string_lit | char_lit | byte_string_lit | byte_lit | num_lit ;
243 #### Character and string literals
246 char_lit : '\x27' char_body '\x27' ;
247 string_lit : '"' string_body * '"' | 'r' raw_string ;
249 char_body : non_single_quote
250 | '\x5c' [ '\x27' | common_escape | unicode_escape ] ;
252 string_body : non_double_quote
253 | '\x5c' [ '\x22' | common_escape | unicode_escape ] ;
254 raw_string : '"' raw_string_body '"' | '#' raw_string '#' ;
256 common_escape : '\x5c'
257 | 'n' | 'r' | 't' | '0'
259 unicode_escape : 'u' hex_digit 4
262 hex_digit : 'a' | 'b' | 'c' | 'd' | 'e' | 'f'
263 | 'A' | 'B' | 'C' | 'D' | 'E' | 'F'
265 oct_digit : '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' ;
266 dec_digit : '0' | nonzero_dec ;
267 nonzero_dec: '1' | '2' | '3' | '4'
268 | '5' | '6' | '7' | '8' | '9' ;
271 A _character literal_ is a single Unicode character enclosed within two
272 `U+0027` (single-quote) characters, with the exception of `U+0027` itself,
273 which must be _escaped_ by a preceding U+005C character (`\`).
275 A _string literal_ is a sequence of any Unicode characters enclosed within
276 two `U+0022` (double-quote) characters, with the exception of `U+0022`
277 itself, which must be _escaped_ by a preceding `U+005C` character (`\`),
278 or a _raw string literal_.
280 Some additional _escapes_ are available in either character or non-raw string
281 literals. An escape starts with a `U+005C` (`\`) and continues with one of
284 * An _8-bit codepoint escape_ escape starts with `U+0078` (`x`) and is
285 followed by exactly two _hex digits_. It denotes the Unicode codepoint
286 equal to the provided hex value.
287 * A _16-bit codepoint escape_ starts with `U+0075` (`u`) and is followed
288 by exactly four _hex digits_. It denotes the Unicode codepoint equal to
289 the provided hex value.
290 * A _32-bit codepoint escape_ starts with `U+0055` (`U`) and is followed
291 by exactly eight _hex digits_. It denotes the Unicode codepoint equal to
292 the provided hex value.
293 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
294 (`r`), or `U+0074` (`t`), denoting the unicode values `U+000A` (LF),
295 `U+000D` (CR) or `U+0009` (HT) respectively.
296 * The _backslash escape_ is the character `U+005C` (`\`) which must be
297 escaped in order to denote *itself*.
299 Raw string literals do not process any escapes. They start with the character
300 `U+0072` (`r`), followed by zero or more of the character `U+0023` (`#`) and a
301 `U+0022` (double-quote) character. The _raw string body_ is not defined in the
302 EBNF grammar above: it can contain any sequence of Unicode characters and is
303 terminated only by another `U+0022` (double-quote) character, followed by the
304 same number of `U+0023` (`#`) characters that preceded the opening `U+0022`
305 (double-quote) character.
307 All Unicode characters contained in the raw string body represent themselves,
308 the characters `U+0022` (double-quote) (except when followed by at least as
309 many `U+0023` (`#`) characters as were used to start the raw string literal) or
310 `U+005C` (`\`) do not have any special meaning.
312 Examples for string literals:
315 "foo"; r"foo"; // foo
316 "\"foo\""; r#""foo""#; // "foo"
319 r##"foo #"# bar"##; // foo #"# bar
321 "\x52"; "R"; r"R"; // R
322 "\\x52"; r"\x52"; // \x52
325 #### Byte and byte string literals
328 byte_lit : 'b' '\x27' byte_body '\x27' ;
329 byte_string_lit : 'b' '"' string_body * '"' | 'b' 'r' raw_byte_string ;
331 byte_body : ascii_non_single_quote
332 | '\x5c' [ '\x27' | common_escape ] ;
334 byte_string_body : ascii_non_double_quote
335 | '\x5c' [ '\x22' | common_escape ] ;
336 raw_byte_string : '"' raw_byte_string_body '"' | '#' raw_byte_string '#' ;
340 A _byte literal_ is a single ASCII character (in the `U+0000` to `U+007F` range)
341 enclosed within two `U+0027` (single-quote) characters,
342 with the exception of `U+0027` itself,
343 which must be _escaped_ by a preceding U+005C character (`\`),
344 or a single _escape_.
345 It is equivalent to a `u8` unsigned 8-bit integer _number literal_.
347 A _byte string literal_ is a sequence of ASCII characters and _escapes_
348 enclosed within two `U+0022` (double-quote) characters,
349 with the exception of `U+0022` itself,
350 which must be _escaped_ by a preceding `U+005C` character (`\`),
351 or a _raw byte string literal_.
352 It is equivalent to a `&'static [u8]` borrowed vector of unsigned 8-bit integers.
354 Some additional _escapes_ are available in either byte or non-raw byte string
355 literals. An escape starts with a `U+005C` (`\`) and continues with one of
358 * An _byte escape_ escape starts with `U+0078` (`x`) and is
359 followed by exactly two _hex digits_. It denotes the byte
360 equal to the provided hex value.
361 * A _whitespace escape_ is one of the characters `U+006E` (`n`), `U+0072`
362 (`r`), or `U+0074` (`t`), denoting the bytes values `0x0A` (ASCII LF),
363 `0x0D` (ASCII CR) or `0x09` (ASCII HT) respectively.
364 * The _backslash escape_ is the character `U+005C` (`\`) which must be
365 escaped in order to denote its ASCII encoding `0x5C`.
367 Raw byte string literals do not process any escapes.
368 They start with the character `U+0072` (`r`),
369 followed by `U+0062` (`b`),
370 followed by zero or more of the character `U+0023` (`#`),
371 and a `U+0022` (double-quote) character.
372 The _raw string body_ is not defined in the EBNF grammar above:
373 it can contain any sequence of ASCII characters and is
374 terminated only by another `U+0022` (double-quote) character, followed by the
375 same number of `U+0023` (`#`) characters that preceded the opening `U+0022`
376 (double-quote) character.
377 A raw byte string literal can not contain any non-ASCII byte.
379 All characters contained in the raw string body represent their ASCII encoding,
380 the characters `U+0022` (double-quote) (except when followed by at least as
381 many `U+0023` (`#`) characters as were used to start the raw string literal) or
382 `U+005C` (`\`) do not have any special meaning.
384 Examples for byte string literals:
387 b"foo"; br"foo"; // foo
388 b"\"foo\""; br#""foo""#; // "foo"
391 br##"foo #"# bar"##; // foo #"# bar
393 b"\x52"; b"R"; br"R"; // R
394 b"\\x52"; br"\x52"; // \x52
400 num_lit : nonzero_dec [ dec_digit | '_' ] * num_suffix ?
401 | '0' [ [ dec_digit | '_' ] * num_suffix ?
402 | 'b' [ '1' | '0' | '_' ] + int_suffix ?
403 | 'o' [ oct_digit | '_' ] + int_suffix ?
404 | 'x' [ hex_digit | '_' ] + int_suffix ? ] ;
406 num_suffix : int_suffix | float_suffix ;
408 int_suffix : 'u' int_suffix_size ?
409 | 'i' int_suffix_size ? ;
410 int_suffix_size : [ '8' | '1' '6' | '3' '2' | '6' '4' ] ;
412 float_suffix : [ exponent | '.' dec_lit exponent ? ] ? float_suffix_ty ? ;
413 float_suffix_ty : 'f' [ '3' '2' | '6' '4' ] ;
414 exponent : ['E' | 'e'] ['-' | '+' ] ? dec_lit ;
415 dec_lit : [ dec_digit | '_' ] + ;
418 A _number literal_ is either an _integer literal_ or a _floating-point
419 literal_. The grammar for recognizing the two kinds of literals is mixed,
420 as they are differentiated by suffixes.
422 ##### Integer literals
424 An _integer literal_ has one of four forms:
426 * A _decimal literal_ starts with a *decimal digit* and continues with any
427 mixture of *decimal digits* and _underscores_.
428 * A _hex literal_ starts with the character sequence `U+0030` `U+0078`
429 (`0x`) and continues as any mixture hex digits and underscores.
430 * An _octal literal_ starts with the character sequence `U+0030` `U+006F`
431 (`0o`) and continues as any mixture octal digits and underscores.
432 * A _binary literal_ starts with the character sequence `U+0030` `U+0062`
433 (`0b`) and continues as any mixture binary digits and underscores.
435 An integer literal may be followed (immediately, without any spaces) by an
436 _integer suffix_, which changes the type of the literal. There are two kinds
437 of integer literal suffix:
439 * The `i` and `u` suffixes give the literal type `int` or `uint`,
441 * Each of the signed and unsigned machine types `u8`, `i8`,
442 `u16`, `i16`, `u32`, `i32`, `u64` and `i64`
443 give the literal the corresponding machine type.
445 The type of an _unsuffixed_ integer literal is determined by type inference.
446 If an integer type can be _uniquely_ determined from the surrounding program
447 context, the unsuffixed integer literal has that type. If the program context
448 underconstrains the type, it is considered a static type error;
449 if the program context overconstrains the type,
450 it is also considered a static type error.
452 Examples of integer literals of various forms:
459 0o70_i16; // type i16
460 0b1111_1111_1001_0000_i32; // type i32
463 ##### Floating-point literals
465 A _floating-point literal_ has one of two forms:
467 * Two _decimal literals_ separated by a period
468 character `U+002E` (`.`), with an optional _exponent_ trailing after the
469 second decimal literal.
470 * A single _decimal literal_ followed by an _exponent_.
472 By default, a floating-point literal has a generic type,
473 and, like integer literals, the type must be uniquely determined
475 A floating-point literal may be followed (immediately, without any
476 spaces) by a _floating-point suffix_, which changes the type of the literal.
477 There are two floating-point suffixes: `f32`, and `f64` (the 32-bit and 64-bit
478 floating point types).
480 Examples of floating-point literals of various forms:
483 123.0f64; // type f64
486 12E+99_f64; // type f64
489 ##### Unit and boolean literals
491 The _unit value_, the only value of the type that has the same name, is written as `()`.
492 The two values of the boolean type are written `true` and `false`.
498 | '#' | '[' | ']' | '(' | ')' | '{' | '}'
502 Symbols are a general class of printable [token](#tokens) that play structural
503 roles in a variety of grammar productions. They are catalogued here for
504 completeness as the set of remaining miscellaneous printable tokens that do not
505 otherwise appear as [unary operators](#unary-operator-expressions), [binary
506 operators](#binary-operator-expressions), or [keywords](#keywords).
512 expr_path : [ "::" ] ident [ "::" expr_path_tail ] + ;
513 expr_path_tail : '<' type_expr [ ',' type_expr ] + '>'
516 type_path : ident [ type_path_tail ] + ;
517 type_path_tail : '<' type_expr [ ',' type_expr ] + '>'
521 A _path_ is a sequence of one or more path components _logically_ separated by
522 a namespace qualifier (`::`). If a path consists of only one component, it may
523 refer to either an [item](#items) or a [slot](#memory-slots) in a local
524 control scope. If a path has multiple components, it refers to an item.
526 Every item has a _canonical path_ within its crate, but the path naming an
527 item is only meaningful within a given crate. There is no global namespace
528 across crates; an item's canonical path merely identifies it within the crate.
530 Two examples of simple paths consisting of only identifier components:
537 Path components are usually [identifiers](#identifiers), but the trailing
538 component of a path may be an angle-bracket-enclosed list of type
539 arguments. In [expression](#expressions) context, the type argument list is
540 given after a final (`::`) namespace qualifier in order to disambiguate it
541 from a relational expression involving the less-than symbol (`<`). In type
542 expression context, the final namespace qualifier is omitted.
544 Two examples of paths with type arguments:
547 # struct HashMap<K, V>;
549 # fn id<T>(t: T) -> T { t }
550 type T = HashMap<int,String>; // Type arguments used in a type expression
551 let x = id::<int>(10); // Type arguments used in a call expression
555 Paths can be denoted with various leading qualifiers to change the meaning of
558 * Paths starting with `::` are considered to be global paths where the
559 components of the path start being resolved from the crate root. Each
560 identifier in the path must resolve to an item.
568 ::a::foo(); // call a's foo function
574 * Paths starting with the keyword `super` begin resolution relative to the
575 parent module. Each further identifier must resolve to an item
583 super::a::foo(); // call a's foo function
589 * Paths starting with the keyword `self` begin resolution relative to the
590 current module. Each further identifier must resolve to an item.
602 A number of minor features of Rust are not central enough to have their own
603 syntax, and yet are not implementable as functions. Instead, they are given
604 names, and invoked through a consistent syntax: `name!(...)`. Examples
607 * `format!` : format data into a string
608 * `env!` : look up an environment variable's value at compile time
609 * `file!`: return the path to the file being compiled
610 * `stringify!` : pretty-print the Rust expression given as an argument
611 * `include!` : include the Rust expression in the given file
612 * `include_str!` : include the contents of the given file as a string
613 * `include_bin!` : include the contents of the given file as a binary blob
614 * `error!`, `warn!`, `info!`, `debug!` : provide diagnostic information.
616 All of the above extensions are expressions with values.
621 expr_macro_rules : "macro_rules" '!' ident '(' macro_rule * ')' ;
622 macro_rule : '(' matcher * ')' "=>" '(' transcriber * ')' ';' ;
623 matcher : '(' matcher * ')' | '[' matcher * ']'
624 | '{' matcher * '}' | '$' ident ':' ident
625 | '$' '(' matcher * ')' sep_token? [ '*' | '+' ]
626 | non_special_token ;
627 transcriber : '(' transcriber * ')' | '[' transcriber * ']'
628 | '{' transcriber * '}' | '$' ident
629 | '$' '(' transcriber * ')' sep_token? [ '*' | '+' ]
630 | non_special_token ;
633 User-defined syntax extensions are called "macros",
634 and the `macro_rules` syntax extension defines them.
635 Currently, user-defined macros can expand to expressions, statements, or items.
637 (A `sep_token` is any token other than `*` and `+`.
638 A `non_special_token` is any token other than a delimiter or `$`.)
640 The macro expander looks up macro invocations by name,
641 and tries each macro rule in turn.
642 It transcribes the first successful match.
643 Matching and transcription are closely related to each other,
644 and we will describe them together.
648 The macro expander matches and transcribes every token that does not begin with a `$` literally, including delimiters.
649 For parsing reasons, delimiters must be balanced, but they are otherwise not special.
651 In the matcher, `$` _name_ `:` _designator_ matches the nonterminal in the
652 Rust syntax named by _designator_. Valid designators are `item`, `block`,
653 `stmt`, `pat`, `expr`, `ty` (type), `ident`, `path`, `matchers` (lhs of the `=>` in macro rules),
654 `tt` (rhs of the `=>` in macro rules). In the transcriber, the designator is already known, and so only
655 the name of a matched nonterminal comes after the dollar sign.
657 In both the matcher and transcriber, the Kleene star-like operator indicates repetition.
658 The Kleene star operator consists of `$` and parens, optionally followed by a separator token, followed by `*` or `+`.
659 `*` means zero or more repetitions, `+` means at least one repetition.
660 The parens are not matched or transcribed.
661 On the matcher side, a name is bound to _all_ of the names it
662 matches, in a structure that mimics the structure of the repetition
663 encountered on a successful match. The job of the transcriber is to sort that
666 The rules for transcription of these repetitions are called "Macro By Example".
667 Essentially, one "layer" of repetition is discharged at a time, and all of
668 them must be discharged by the time a name is transcribed. Therefore,
669 `( $( $i:ident ),* ) => ( $i )` is an invalid macro, but
670 `( $( $i:ident ),* ) => ( $( $i:ident ),* )` is acceptable (if trivial).
672 When Macro By Example encounters a repetition, it examines all of the `$`
673 _name_ s that occur in its body. At the "current layer", they all must repeat
674 the same number of times, so
675 ` ( $( $i:ident ),* ; $( $j:ident ),* ) => ( $( ($i,$j) ),* )` is valid if
676 given the argument `(a,b,c ; d,e,f)`, but not `(a,b,c ; d,e)`. The repetition
677 walks through the choices at that layer in lockstep, so the former input
678 transcribes to `( (a,d), (b,e), (c,f) )`.
680 Nested repetitions are allowed.
682 ### Parsing limitations
684 The parser used by the macro system is reasonably powerful, but the parsing of
685 Rust syntax is restricted in two ways:
687 1. The parser will always parse as much as possible. If it attempts to match
688 `$i:expr [ , ]` against `8 [ , ]`, it will attempt to parse `i` as an array
689 index operation and fail. Adding a separator can solve this problem.
690 2. The parser must have eliminated all ambiguity by the time it reaches a `$` _name_ `:` _designator_.
691 This requirement most often affects name-designator pairs when they occur at the beginning of, or immediately after, a `$(...)*`; requiring a distinctive token in front can solve the problem.
693 ## Syntax extensions useful for the macro author
695 * `log_syntax!` : print out the arguments at compile time
696 * `trace_macros!` : supply `true` or `false` to enable or disable macro expansion logging
697 * `stringify!` : turn the identifier argument into a string literal
698 * `concat!` : concatenates a comma-separated list of literals
699 * `concat_idents!` : create a new identifier by concatenating the arguments
701 # Crates and source files
703 Rust is a *compiled* language.
704 Its semantics obey a *phase distinction* between compile-time and run-time.
705 Those semantic rules that have a *static interpretation* govern the success or failure of compilation.
706 We refer to these rules as "static semantics".
707 Semantic rules called "dynamic semantics" govern the behavior of programs at run-time.
708 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.
710 The compilation model centres on artifacts called _crates_.
711 Each compilation processes a single crate in source form, and if successful,
712 produces a single crate in binary form: either an executable or a
713 library.[^cratesourcefile]
715 [^cratesourcefile]: A crate is somewhat analogous to an *assembly* in the
716 ECMA-335 CLI model, a *library* in the SML/NJ Compilation Manager, a *unit*
717 in the Owens and Flatt module system, or a *configuration* in Mesa.
719 A _crate_ is a unit of compilation and linking, as well as versioning, distribution and runtime loading.
720 A crate contains a _tree_ of nested [module](#modules) scopes.
721 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.
723 The Rust compiler is always invoked with a single source file as input, and always produces a single output crate.
724 The processing of that source file may result in other source files being loaded as modules.
725 Source files have the extension `.rs`.
727 A Rust source file describes a module, the name and
728 location of which — in the module tree of the current crate — are defined
729 from outside the source file: either by an explicit `mod_item` in
730 a referencing source file, or by the name of the crate itself.
732 Each source file contains a sequence of zero or more `item` definitions,
733 and may optionally begin with any number of `attributes` that apply to the containing module.
734 Attributes on the anonymous crate module define important metadata that influences
735 the behavior of the compiler.
738 # #![allow(unused_attribute)]
740 #![crate_id = "projx#2.5"]
742 // Additional metadata attributes
743 #![desc = "Project X"]
745 #![comment = "This is a comment on Project X."]
747 // Specify the output type
748 #![crate_type = "lib"]
751 #![warn(non_camel_case_types)]
754 A crate that contains a `main` function can be compiled to an executable.
755 If a `main` function is present, its return type must be [`unit`](#primitive-types) and it must take no arguments.
757 # Items and attributes
759 Crates contain [items](#items),
760 each of which may have some number of [attributes](#attributes) attached to it.
765 item : mod_item | fn_item | type_item | struct_item | enum_item
766 | static_item | trait_item | impl_item | extern_block ;
769 An _item_ is a component of a crate; some module items can be defined in crate
770 files, but most are defined in source files. Items are organized within a
771 crate by a nested set of [modules](#modules). Every crate has a single
772 "outermost" anonymous module; all further items within the crate have
773 [paths](#paths) within the module tree of the crate.
775 Items are entirely determined at compile-time, generally remain fixed during
776 execution, and may reside in read-only memory.
778 There are several kinds of item:
780 * [modules](#modules)
781 * [functions](#functions)
782 * [type definitions](#type-definitions)
783 * [structures](#structures)
784 * [enumerations](#enumerations)
785 * [static items](#static-items)
787 * [implementations](#implementations)
789 Some items form an implicit scope for the declaration of sub-items. In other
790 words, within a function or module, declarations of items can (in many cases)
791 be mixed with the statements, control blocks, and similar artifacts that
792 otherwise compose the item body. The meaning of these scoped items is the same
793 as if the item was declared outside the scope — it is still a static item —
794 except that the item's *path name* within the module namespace is qualified by
795 the name of the enclosing item, or is private to the enclosing item (in the
797 The grammar specifies the exact locations in which sub-item declarations may appear.
801 All items except modules may be *parameterized* by type. Type parameters are
802 given as a comma-separated list of identifiers enclosed in angle brackets
803 (`<...>`), after the name of the item and before its definition.
804 The type parameters of an item are considered "part of the name", not part of the type of the item.
805 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.
806 In practice, the type-inference system can usually infer such argument types from context.
807 There are no general type-parametric types, only type-parametric items.
808 That is, Rust has no notion of type abstraction: there are no first-class "forall" types.
813 mod_item : "mod" ident ( ';' | '{' mod '}' );
814 mod : [ view_item | item ] * ;
817 A module is a container for zero or more [view items](#view-items) and zero or
818 more [items](#items). The view items manage the visibility of the items
819 defined within the module, as well as the visibility of names from outside the
820 module when referenced from inside the module.
822 A _module item_ is a module, surrounded in braces, named, and prefixed with
823 the keyword `mod`. A module item introduces a new, named module into the tree
824 of modules making up a crate. Modules can nest arbitrarily.
826 An example of a module:
830 type Complex = (f64, f64);
831 fn sin(f: f64) -> f64 {
835 fn cos(f: f64) -> f64 {
839 fn tan(f: f64) -> f64 {
846 Modules and types share the same namespace.
847 Declaring a named type that has the same name as a module in scope is forbidden:
848 that is, a type definition, trait, struct, enumeration, or type parameter
849 can't shadow the name of a module in scope, or vice versa.
851 A module without a body is loaded from an external file, by default with the same
852 name as the module, plus the `.rs` extension.
853 When a nested submodule is loaded from an external file,
854 it is loaded from a subdirectory path that mirrors the module hierarchy.
857 // Load the `vec` module from `vec.rs`
861 // Load the `local_data` module from `task/local_data.rs`
866 The directories and files used for loading external file modules can be influenced
867 with the `path` attribute.
870 #[path = "task_files"]
872 // Load the `local_data` module from `task_files/tls.rs`
881 view_item : extern_crate_decl | use_decl ;
884 A view item manages the namespace of a module.
885 View items do not define new items, but rather, simply change other items' visibility.
886 There are several kinds of view item:
888 * [`extern crate` declarations](#extern-crate-declarations)
889 * [`use` declarations](#use-declarations)
891 ##### Extern crate declarations
894 extern_crate_decl : "extern" "crate" crate_name
895 crate_name: ident | ( string_lit as ident )
898 An _`extern crate` declaration_ specifies a dependency on an external crate.
899 The external crate is then bound into the declaring scope as the `ident` provided
900 in the `extern_crate_decl`.
902 The external crate is resolved to a specific `soname` at compile time, and a
903 runtime linkage requirement to that `soname` is passed to the linker for
904 loading at runtime. The `soname` is resolved at compile time by scanning the
905 compiler's library path and matching the optional `crateid` provided as a string literal
906 against the `crateid` attributes that were declared on the external crate when
907 it was compiled. If no `crateid` is provided, a default `name` attribute is
908 assumed, equal to the `ident` given in the `extern_crate_decl`.
910 Four examples of `extern crate` declarations:
915 extern crate std; // equivalent to: extern crate std as std;
917 extern crate "std" as ruststd; // linking to 'std' under another name
920 ##### Use declarations
923 use_decl : "pub" ? "use" [ path "as" ident
926 path_glob : ident [ "::" [ path_glob
928 | '{' path_item [ ',' path_item ] * '}' ;
930 path_item : ident | "mod" ;
933 A _use declaration_ creates one or more local name bindings synonymous
934 with some other [path](#paths).
935 Usually a `use` declaration is used to shorten the path required to refer to a
936 module item. These declarations may appear at the top of [modules](#modules) and
939 > **Note**: Unlike in many languages,
940 > `use` declarations in Rust do *not* declare linkage dependency with external crates.
941 > Rather, [`extern crate` declarations](#extern-crate-declarations) declare linkage dependencies.
943 Use declarations support a number of convenient shortcuts:
945 * Rebinding the target name as a new local name, using the syntax `use p::q::r as x;`.
946 * Simultaneously binding a list of paths differing only in their final element,
947 using the glob-like brace syntax `use a::b::{c,d,e,f};`
948 * Binding all paths matching a given prefix, using the asterisk wildcard syntax `use a::b::*;`
949 * Simultaneously binding a list of paths differing only in their final element
950 and their immediate parent module, using the `mod` keyword, such as `use a::b::{mod, c, d};`
952 An example of `use` declarations:
955 use std::iter::range_step;
956 use std::option::{Some, None};
957 use std::collections::hashmap::{mod, HashMap};
960 # fn bar(map: HashMap<String, uint>, set: hashmap::HashSet<String>){}
963 // Equivalent to 'std::iter::range_step(0u, 10u, 2u);'
964 range_step(0u, 10u, 2u);
966 // Equivalent to 'foo(vec![std::option::Some(1.0f64),
967 // std::option::None]);'
968 foo(vec![Some(1.0f64), None]);
970 // Both `hash` and `HashMap` are in scope.
971 let map = HashMap::new();
972 let set = hashmap::HashSet::new();
977 Like items, `use` declarations are private to the containing module, by default.
978 Also like items, a `use` declaration can be public, if qualified by the `pub` keyword.
979 Such a `use` declaration serves to _re-export_ a name.
980 A public `use` declaration can therefore _redirect_ some public name to a different target definition:
981 even a definition with a private canonical path, inside a different module.
982 If a sequence of such redirections form a cycle or cannot be resolved unambiguously,
983 they represent a compile-time error.
985 An example of re-exporting:
990 pub use quux::foo::{bar, baz};
999 In this example, the module `quux` re-exports two public names defined in `foo`.
1001 Also note that the paths contained in `use` items are relative to the crate root.
1002 So, in the previous example, the `use` refers to `quux::foo::{bar, baz}`, and not simply to `foo::{bar, baz}`.
1003 This also means that top-level module declarations should be at the crate root if direct usage
1004 of the declared modules within `use` items is desired. It is also possible to use `self` and `super`
1005 at the beginning of a `use` item to refer to the current and direct parent modules respectively.
1006 All rules regarding accessing declared modules in `use` declarations applies to both module declarations
1007 and `extern crate` declarations.
1009 An example of what will and will not work for `use` items:
1012 # #![allow(unused_imports)]
1013 use foo::native::start; // good: foo is at the root of the crate
1014 use foo::baz::foobaz; // good: foo is at the root of the crate
1017 extern crate native;
1019 use foo::native::start; // good: foo is at crate root
1020 // use native::start; // bad: native is not at the crate root
1021 use self::baz::foobaz; // good: self refers to module 'foo'
1022 use foo::bar::foobar; // good: foo is at crate root
1029 use super::bar::foobar; // good: super refers to module 'foo'
1039 A _function item_ defines a sequence of [statements](#statements) and an optional final [expression](#expressions), along with a name and a set of parameters.
1040 Functions are declared with the keyword `fn`.
1041 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.
1043 A function may also be copied into a first class *value*, in which case the
1044 value has the corresponding [*function type*](#function-types), and can be
1045 used otherwise exactly as a function item (with a minor additional cost of
1046 calling the function indirectly).
1048 Every control path in a function logically ends with a `return` expression or a
1049 diverging expression. If the outermost block of a function has a
1050 value-producing expression in its final-expression position, that expression
1051 is interpreted as an implicit `return` expression applied to the
1054 An example of a function:
1057 fn add(x: int, y: int) -> int {
1062 As with `let` bindings, function arguments are irrefutable patterns,
1063 so any pattern that is valid in a let binding is also valid as an argument.
1066 fn first((value, _): (int, int)) -> int { value }
1070 #### Generic functions
1072 A _generic function_ allows one or more _parameterized types_ to
1073 appear in its signature. Each type parameter must be explicitly
1074 declared, in an angle-bracket-enclosed, comma-separated list following
1078 fn iter<T>(seq: &[T], f: |T|) {
1079 for elt in seq.iter() { f(elt); }
1081 fn map<T, U>(seq: &[T], f: |T| -> U) -> Vec<U> {
1082 let mut acc = vec![];
1083 for elt in seq.iter() { acc.push(f(elt)); }
1088 Inside the function signature and body, the name of the type parameter
1089 can be used as a type name.
1091 When a generic function is referenced, its type is instantiated based
1092 on the context of the reference. For example, calling the `iter`
1093 function defined above on `[1, 2]` will instantiate type parameter `T`
1094 with `int`, and require the closure parameter to have type
1097 The type parameters can also be explicitly supplied in a trailing
1098 [path](#paths) component after the function name. This might be necessary
1099 if there is not sufficient context to determine the type parameters. For
1100 example, `mem::size_of::<u32>() == 4`.
1102 Since a parameter type is opaque to the generic function, the set of
1103 operations that can be performed on it is limited. Values of parameter
1104 type can only be moved, not copied.
1107 fn id<T>(x: T) -> T { x }
1110 Similarly, [trait](#traits) bounds can be specified for type
1111 parameters to allow methods with that trait to be called on values
1117 Unsafe operations are those that potentially violate the memory-safety guarantees of Rust's static semantics.
1119 The following language level features cannot be used in the safe subset of Rust:
1121 - Dereferencing a [raw pointer](#pointer-types).
1122 - Reading or writing a [mutable static variable](#mutable-statics).
1123 - Calling an unsafe function (including an intrinsic or foreign function).
1125 ##### Unsafe functions
1127 Unsafe functions are functions that are not safe in all contexts and/or for all possible inputs.
1128 Such a function must be prefixed with the keyword `unsafe`.
1132 A block of code can also be prefixed with the `unsafe` keyword, to permit calling `unsafe` functions
1133 or dereferencing raw pointers within a safe function.
1135 When a programmer has sufficient conviction that a sequence of potentially unsafe operations is
1136 actually safe, they can encapsulate that sequence (taken as a whole) within an `unsafe` block. The
1137 compiler will consider uses of such code safe, in the surrounding context.
1139 Unsafe blocks are used to wrap foreign libraries, make direct use of hardware or implement features
1140 not directly present in the language. For example, Rust provides the language features necessary to
1141 implement memory-safe concurrency in the language but the implementation of tasks and message
1142 passing is in the standard library.
1144 Rust's type system is a conservative approximation of the dynamic safety requirements, so in some
1145 cases there is a performance cost to using safe code. For example, a doubly-linked list is not a
1146 tree structure and can only be represented with managed or reference-counted pointers in safe code.
1147 By using `unsafe` blocks to represent the reverse links as raw pointers, it can be implemented with
1148 only owned pointers.
1150 ##### Behavior considered unsafe
1152 This is a list of behavior which is forbidden in all Rust code. Type checking provides the guarantee
1153 that these issues are never caused by safe code. An `unsafe` block or function is responsible for
1154 never invoking this behaviour or exposing an API making it possible for it to occur in safe code.
1157 * Dereferencing a null/dangling raw pointer
1158 * Mutating an immutable value/reference
1159 * Reads of [undef](http://llvm.org/docs/LangRef.html#undefined-values) (uninitialized) memory
1160 * Breaking the [pointer aliasing rules](http://llvm.org/docs/LangRef.html#pointer-aliasing-rules)
1161 with raw pointers (a subset of the rules used by C)
1162 * Invoking undefined behavior via compiler intrinsics:
1163 * Indexing outside of the bounds of an object with `std::ptr::offset` (`offset` intrinsic), with
1164 the exception of one byte past the end which is permitted.
1165 * Using `std::ptr::copy_nonoverlapping_memory` (`memcpy32`/`memcpy64` instrinsics) on
1167 * Invalid values in primitive types, even in private fields/locals:
1168 * Dangling/null pointers in non-raw pointers, or slices
1169 * A value other than `false` (0) or `true` (1) in a `bool`
1170 * A discriminant in an `enum` not included in the type definition
1171 * A value in a `char` which is a surrogate or above `char::MAX`
1172 * non-UTF-8 byte sequences in a `str`
1174 ##### Behaviour not considered unsafe
1176 This is a list of behaviour not considered *unsafe* in Rust terms, but that may be undesired.
1179 * Reading data from private fields (`std::repr`, `format!("{:?}", x)`)
1180 * Leaks due to reference count cycles, even in the global heap
1181 * Exiting without calling destructors
1183 * Accessing/modifying the file system
1184 * Unsigned integer overflow (well-defined as wrapping)
1185 * Signed integer overflow (well-defined as two's complement representation wrapping)
1187 #### Diverging functions
1189 A special kind of function can be declared with a `!` character where the
1190 output slot type would normally be. For example:
1193 fn my_err(s: &str) -> ! {
1199 We call such functions "diverging" because they never return a value to the
1200 caller. Every control path in a diverging function must end with a
1201 `fail!()` or a call to another diverging function on every
1202 control path. The `!` annotation does *not* denote a type. Rather, the result
1203 type of a diverging function is a special type called $\bot$ ("bottom") that
1204 unifies with any type. Rust has no syntax for $\bot$.
1206 It might be necessary to declare a diverging function because as mentioned
1207 previously, the typechecker checks that every control path in a function ends
1208 with a [`return`](#return-expressions) or diverging expression. So, if `my_err`
1209 were declared without the `!` annotation, the following code would not
1213 # fn my_err(s: &str) -> ! { fail!() }
1215 fn f(i: int) -> int {
1220 my_err("Bad number!");
1225 This will not compile without the `!` annotation on `my_err`,
1226 since the `else` branch of the conditional in `f` does not return an `int`,
1227 as required by the signature of `f`.
1228 Adding the `!` annotation to `my_err` informs the typechecker that,
1229 should control ever enter `my_err`, no further type judgments about `f` need to hold,
1230 since control will never resume in any context that relies on those judgments.
1231 Thus the return type on `f` only needs to reflect the `if` branch of the conditional.
1234 #### Extern functions
1236 Extern functions are part of Rust's foreign function interface,
1237 providing the opposite functionality to [external blocks](#external-blocks).
1238 Whereas external blocks allow Rust code to call foreign code,
1239 extern functions with bodies defined in Rust code _can be called by foreign
1240 code_. They are defined in the same way as any other Rust function,
1241 except that they have the `extern` modifier.
1244 // Declares an extern fn, the ABI defaults to "C"
1245 extern fn new_int() -> int { 0 }
1247 // Declares an extern fn with "stdcall" ABI
1248 extern "stdcall" fn new_int_stdcall() -> int { 0 }
1251 Unlike normal functions, extern fns have an `extern "ABI" fn()`.
1252 This is the same type as the functions declared in an extern
1256 # extern fn new_int() -> int { 0 }
1257 let fptr: extern "C" fn() -> int = new_int;
1260 Extern functions may be called directly from Rust code as Rust uses large,
1261 contiguous stack segments like C.
1263 ### Type definitions
1265 A _type definition_ defines a new name for an existing [type](#types). Type
1266 definitions are declared with the keyword `type`. Every value has a single,
1267 specific type; the type-specified aspects of a value include:
1269 * Whether the value is composed of sub-values or is indivisible.
1270 * Whether the value represents textual or numerical information.
1271 * Whether the value represents integral or floating-point information.
1272 * The sequence of memory operations required to access the value.
1273 * The [kind](#type-kinds) of the type.
1275 For example, the type `(u8, u8)` defines the set of immutable values that are composite pairs,
1276 each containing two unsigned 8-bit integers accessed by pattern-matching and laid out in memory with the `x` component preceding the `y` component.
1280 A _structure_ is a nominal [structure type](#structure-types) defined with the keyword `struct`.
1282 An example of a `struct` item and its use:
1285 struct Point {x: int, y: int}
1286 let p = Point {x: 10, y: 11};
1290 A _tuple structure_ is a nominal [tuple type](#tuple-types), also defined with the keyword `struct`.
1294 struct Point(int, int);
1295 let p = Point(10, 11);
1296 let px: int = match p { Point(x, _) => x };
1299 A _unit-like struct_ is a structure without any fields, defined by leaving off the list of fields entirely.
1300 Such types will have a single value, just like the [unit value `()`](#unit-and-boolean-literals) of the unit type.
1305 let c = [Cookie, Cookie, Cookie, Cookie];
1308 The precise memory layout of a structure is not specified. One can specify a
1309 particular layout using the [`repr` attribute](#ffi-attributes).
1311 By using the `struct_inherit` feature gate, structures may use single inheritance. A Structure may only
1312 inherit from a single other structure, called the _super-struct_. The inheriting structure (sub-struct)
1313 acts as if all fields in the super-struct were present in the sub-struct. Fields declared in a sub-struct
1314 must not have the same name as any field in any (transitive) super-struct. All fields (both declared
1315 and inherited) must be specified in any initializers. Inheritance between structures does not give
1316 subtyping or coercion. The super-struct and sub-struct must be defined in the same crate. The super-struct
1317 must be declared using the `virtual` keyword.
1321 virtual struct Sup { x: int }
1322 struct Sub : Sup { y: int }
1323 let s = Sub {x: 10, y: 11};
1329 An _enumeration_ is a simultaneous definition of a nominal [enumerated type](#enumerated-types) as well as a set of *constructors*,
1330 that can be used to create or pattern-match values of the corresponding enumerated type.
1332 Enumerations are declared with the keyword `enum`.
1334 An example of an `enum` item and its use:
1342 let mut a: Animal = Dog;
1346 Enumeration constructors can have either named or unnamed fields:
1349 # #![feature(struct_variant)]
1353 Cat { name: String, weight: f64 }
1356 let mut a: Animal = Dog("Cocoa".to_string(), 37.2);
1357 a = Cat { name: "Spotty".to_string(), weight: 2.7 };
1361 In this example, `Cat` is a _struct-like enum variant_,
1362 whereas `Dog` is simply called an enum variant.
1367 static_item : "static" ident ':' type '=' expr ';' ;
1370 A *static item* is a named _constant value_ stored in the global data section of a crate.
1371 Immutable static items are stored in the read-only data section.
1372 The constant value bound to a static item is, like all constant values, evaluated at compile time.
1373 Static items have the `static` lifetime, which outlives all other lifetimes in a Rust program.
1374 Only values stored in the global data section (such as string constants
1375 and static items) can have the `static` lifetime;
1376 dynamically constructed values cannot safely be assigned the `static` lifetime.
1377 Static items are declared with the `static` keyword.
1378 A static item must have a _constant expression_ giving its definition.
1380 Static items must be explicitly typed.
1381 The type may be ```bool```, ```char```, a number, or a type derived from those primitive types.
1382 The derived types are references with the `static` lifetime,
1383 fixed-size arrays, tuples, and structs.
1386 static BIT1: uint = 1 << 0;
1387 static BIT2: uint = 1 << 1;
1389 static BITS: [uint, ..2] = [BIT1, BIT2];
1390 static STRING: &'static str = "bitstring";
1392 struct BitsNStrings<'a> {
1393 mybits: [uint, ..2],
1397 static bits_n_strings: BitsNStrings<'static> = BitsNStrings {
1403 #### Mutable statics
1405 If a static item is declared with the ```mut``` keyword, then it is allowed to
1406 be modified by the program. One of Rust's goals is to make concurrency bugs hard
1407 to run into, and this is obviously a very large source of race conditions or
1408 other bugs. For this reason, an ```unsafe``` block is required when either
1409 reading or writing a mutable static variable. Care should be taken to ensure
1410 that modifications to a mutable static are safe with respect to other tasks
1411 running in the same process.
1413 Mutable statics are still very useful, however. They can be used with C
1414 libraries and can also be bound from C libraries (in an ```extern``` block).
1417 # fn atomic_add(_: &mut uint, _: uint) -> uint { 2 }
1419 static mut LEVELS: uint = 0;
1421 // This violates the idea of no shared state, and this doesn't internally
1422 // protect against races, so this function is `unsafe`
1423 unsafe fn bump_levels_unsafe1() -> uint {
1429 // Assuming that we have an atomic_add function which returns the old value,
1430 // this function is "safe" but the meaning of the return value may not be what
1431 // callers expect, so it's still marked as `unsafe`
1432 unsafe fn bump_levels_unsafe2() -> uint {
1433 return atomic_add(&mut LEVELS, 1);
1439 A _trait_ describes a set of method types.
1441 Traits can include default implementations of methods,
1442 written in terms of some unknown [`self` type](#self-types);
1443 the `self` type may either be completely unspecified,
1444 or constrained by some other trait.
1446 Traits are implemented for specific types through separate [implementations](#implementations).
1449 # type Surface = int;
1450 # type BoundingBox = int;
1452 fn draw(&self, Surface);
1453 fn bounding_box(&self) -> BoundingBox;
1457 This defines a trait with two methods.
1458 All values that have [implementations](#implementations) of this trait in scope can have their `draw` and `bounding_box` methods called,
1459 using `value.bounding_box()` [syntax](#method-call-expressions).
1461 Type parameters can be specified for a trait to make it generic.
1462 These appear after the trait name, using the same syntax used in [generic functions](#generic-functions).
1466 fn len(&self) -> uint;
1467 fn elt_at(&self, n: uint) -> T;
1468 fn iter(&self, |T|);
1472 Generic functions may use traits as _bounds_ on their type parameters.
1473 This will have two effects: only types that have the trait may instantiate the parameter,
1474 and within the generic function,
1475 the methods of the trait can be called on values that have the parameter's type.
1479 # type Surface = int;
1480 # trait Shape { fn draw(&self, Surface); }
1481 fn draw_twice<T: Shape>(surface: Surface, sh: T) {
1487 Traits also define an [object type](#object-types) with the same name as the trait.
1488 Values of this type are created by [casting](#type-cast-expressions) pointer values
1489 (pointing to a type for which an implementation of the given trait is in scope)
1490 to pointers to the trait name, used as a type.
1494 # impl Shape for int { }
1495 # let mycircle = 0i;
1496 let myshape: Box<Shape> = box mycircle as Box<Shape>;
1499 The resulting value is a box containing the value that was cast,
1500 along with information that identifies the methods of the implementation that was used.
1501 Values with a trait type can have [methods called](#method-call-expressions) on them,
1502 for any method in the trait,
1503 and can be used to instantiate type parameters that are bounded by the trait.
1505 Trait methods may be static,
1506 which means that they lack a `self` argument.
1507 This means that they can only be called with function call syntax (`f(x)`)
1508 and not method call syntax (`obj.f()`).
1509 The way to refer to the name of a static method is to qualify it with the trait name,
1510 treating the trait name like a module.
1515 fn from_int(n: int) -> Self;
1518 fn from_int(n: int) -> f64 { n as f64 }
1520 let x: f64 = Num::from_int(42);
1523 Traits may inherit from other traits. For example, in
1526 trait Shape { fn area() -> f64; }
1527 trait Circle : Shape { fn radius() -> f64; }
1530 the syntax `Circle : Shape` means that types that implement `Circle` must also have an implementation for `Shape`.
1531 Multiple supertraits are separated by `+`, `trait Circle : Shape + PartialEq { }`.
1532 In an implementation of `Circle` for a given type `T`, methods can refer to `Shape` methods,
1533 since the typechecker checks that any type with an implementation of `Circle` also has an implementation of `Shape`.
1535 In type-parameterized functions,
1536 methods of the supertrait may be called on values of subtrait-bound type parameters.
1537 Referring to the previous example of `trait Circle : Shape`:
1540 # trait Shape { fn area(&self) -> f64; }
1541 # trait Circle : Shape { fn radius(&self) -> f64; }
1542 fn radius_times_area<T: Circle>(c: T) -> f64 {
1543 // `c` is both a Circle and a Shape
1544 c.radius() * c.area()
1548 Likewise, supertrait methods may also be called on trait objects.
1551 # trait Shape { fn area(&self) -> f64; }
1552 # trait Circle : Shape { fn radius(&self) -> f64; }
1553 # impl Shape for int { fn area(&self) -> f64 { 0.0 } }
1554 # impl Circle for int { fn radius(&self) -> f64 { 0.0 } }
1556 let mycircle = box mycircle as Box<Circle>;
1557 let nonsense = mycircle.radius() * mycircle.area();
1562 An _implementation_ is an item that implements a [trait](#traits) for a specific type.
1564 Implementations are defined with the keyword `impl`.
1567 # struct Point {x: f64, y: f64};
1568 # type Surface = int;
1569 # struct BoundingBox {x: f64, y: f64, width: f64, height: f64};
1570 # trait Shape { fn draw(&self, Surface); fn bounding_box(&self) -> BoundingBox; }
1571 # fn do_draw_circle(s: Surface, c: Circle) { }
1577 impl Shape for Circle {
1578 fn draw(&self, s: Surface) { do_draw_circle(s, *self); }
1579 fn bounding_box(&self) -> BoundingBox {
1580 let r = self.radius;
1581 BoundingBox{x: self.center.x - r, y: self.center.y - r,
1582 width: 2.0 * r, height: 2.0 * r}
1587 It is possible to define an implementation without referring to a trait.
1588 The methods in such an implementation can only be used
1589 as direct calls on the values of the type that the implementation targets.
1590 In such an implementation, the trait type and `for` after `impl` are omitted.
1591 Such implementations are limited to nominal types (enums, structs),
1592 and the implementation must appear in the same module or a sub-module as the `self` type.
1594 When a trait _is_ specified in an `impl`,
1595 all methods declared as part of the trait must be implemented,
1596 with matching types and type parameter counts.
1598 An implementation can take type parameters,
1599 which can be different from the type parameters taken by the trait it implements.
1600 Implementation parameters are written after the `impl` keyword.
1604 impl<T> Seq<T> for Vec<T> {
1607 impl Seq<bool> for u32 {
1608 /* Treat the integer as a sequence of bits */
1615 extern_block_item : "extern" '{' extern_block '}' ;
1616 extern_block : [ foreign_fn ] * ;
1619 External blocks form the basis for Rust's foreign function interface.
1620 Declarations in an external block describe symbols
1621 in external, non-Rust libraries.
1623 Functions within external blocks
1624 are declared in the same way as other Rust functions,
1625 with the exception that they may not have a body
1626 and are instead terminated by a semicolon.
1630 use libc::{c_char, FILE};
1633 fn fopen(filename: *const c_char, mode: *const c_char) -> *mut FILE;
1638 Functions within external blocks may be called by Rust code,
1639 just like functions defined in Rust.
1640 The Rust compiler automatically translates
1641 between the Rust ABI and the foreign ABI.
1643 A number of [attributes](#attributes) control the behavior of external
1646 By default external blocks assume that the library they are calling
1647 uses the standard C "cdecl" ABI. Other ABIs may be specified using
1648 an `abi` string, as shown here:
1651 // Interface to the Windows API
1652 extern "stdcall" { }
1655 The `link` attribute allows the name of the library to be specified. When
1656 specified the compiler will attempt to link against the native library of the
1660 #[link(name = "crypto")]
1664 The type of a function declared in an extern block is `extern "abi" fn(A1,
1665 ..., An) -> R`, where `A1...An` are the declared types of its arguments and
1666 `R` is the declared return type.
1668 ## Visibility and Privacy
1670 These two terms are often used interchangeably, and what they are attempting to
1671 convey is the answer to the question "Can this item be used at this location?"
1673 Rust's name resolution operates on a global hierarchy of namespaces. Each level
1674 in the hierarchy can be thought of as some item. The items are one of those
1675 mentioned above, but also include external crates. Declaring or defining a new
1676 module can be thought of as inserting a new tree into the hierarchy at the
1677 location of the definition.
1679 To control whether interfaces can be used across modules, Rust checks each use
1680 of an item to see whether it should be allowed or not. This is where privacy
1681 warnings are generated, or otherwise "you used a private item of another module
1682 and weren't allowed to."
1684 By default, everything in rust is *private*, with one exception. Enum variants
1685 in a `pub` enum are also public by default. You are allowed to alter this
1686 default visibility with the `priv` keyword. When an item is declared as `pub`,
1687 it can be thought of as being accessible to the outside world. For example:
1691 // Declare a private struct
1694 // Declare a public struct with a private field
1699 // Declare a public enum with two public variants
1701 PubliclyAccessibleState,
1702 PubliclyAccessibleState2,
1706 With the notion of an item being either public or private, Rust allows item
1707 accesses in two cases:
1709 1. If an item is public, then it can be used externally through any of its
1711 2. If an item is private, it may be accessed by the current module and its
1714 These two cases are surprisingly powerful for creating module hierarchies
1715 exposing public APIs while hiding internal implementation details. To help
1716 explain, here's a few use cases and what they would entail.
1718 * A library developer needs to expose functionality to crates which link against
1719 their library. As a consequence of the first case, this means that anything
1720 which is usable externally must be `pub` from the root down to the destination
1721 item. Any private item in the chain will disallow external accesses.
1723 * A crate needs a global available "helper module" to itself, but it doesn't
1724 want to expose the helper module as a public API. To accomplish this, the root
1725 of the crate's hierarchy would have a private module which then internally has
1726 a "public api". Because the entire crate is a descendant of the root, then the
1727 entire local crate can access this private module through the second case.
1729 * When writing unit tests for a module, it's often a common idiom to have an
1730 immediate child of the module to-be-tested named `mod test`. This module could
1731 access any items of the parent module through the second case, meaning that
1732 internal implementation details could also be seamlessly tested from the child
1735 In the second case, it mentions that a private item "can be accessed" by the
1736 current module and its descendants, but the exact meaning of accessing an item
1737 depends on what the item is. Accessing a module, for example, would mean looking
1738 inside of it (to import more items). On the other hand, accessing a function
1739 would mean that it is invoked. Additionally, path expressions and import
1740 statements are considered to access an item in the sense that the
1741 import/expression is only valid if the destination is in the current visibility
1744 Here's an example of a program which exemplifies the three cases outlined above.
1747 // This module is private, meaning that no external crate can access this
1748 // module. Because it is private at the root of this current crate, however, any
1749 // module in the crate may access any publicly visible item in this module.
1750 mod crate_helper_module {
1752 // This function can be used by anything in the current crate
1753 pub fn crate_helper() {}
1755 // This function *cannot* be used by anything else in the crate. It is not
1756 // publicly visible outside of the `crate_helper_module`, so only this
1757 // current module and its descendants may access it.
1758 fn implementation_detail() {}
1761 // This function is "public to the root" meaning that it's available to external
1762 // crates linking against this one.
1763 pub fn public_api() {}
1765 // Similarly to 'public_api', this module is public so external crates may look
1768 use crate_helper_module;
1770 pub fn my_method() {
1771 // Any item in the local crate may invoke the helper module's public
1772 // interface through a combination of the two rules above.
1773 crate_helper_module::crate_helper();
1776 // This function is hidden to any module which is not a descendant of
1778 fn my_implementation() {}
1784 fn test_my_implementation() {
1785 // Because this module is a descendant of `submodule`, it's allowed
1786 // to access private items inside of `submodule` without a privacy
1788 super::my_implementation();
1796 For a rust program to pass the privacy checking pass, all paths must be valid
1797 accesses given the two rules above. This includes all use statements,
1798 expressions, types, etc.
1800 ### Re-exporting and Visibility
1802 Rust allows publicly re-exporting items through a `pub use` directive. Because
1803 this is a public directive, this allows the item to be used in the current
1804 module through the rules above. It essentially allows public access into the
1805 re-exported item. For example, this program is valid:
1808 pub use self::implementation as api;
1810 mod implementation {
1817 This means that any external crate referencing `implementation::f` would receive
1818 a privacy violation, while the path `api::f` would be allowed.
1820 When re-exporting a private item, it can be thought of as allowing the "privacy
1821 chain" being short-circuited through the reexport instead of passing through the
1822 namespace hierarchy as it normally would.
1824 ### Glob imports and Visibility
1826 Currently glob imports are considered an "experimental" language feature. For
1827 sanity purpose along with helping the implementation, glob imports will only
1828 import public items from their destination, not private items.
1830 > **Note:** This is subject to change, glob exports may be removed entirely or
1831 > they could possibly import private items for a privacy error to later be
1832 > issued if the item is used.
1837 attribute : '#' '!' ? '[' meta_item ']' ;
1838 meta_item : ident [ '=' literal
1839 | '(' meta_seq ')' ] ? ;
1840 meta_seq : meta_item [ ',' meta_seq ] ? ;
1843 Any item declaration may have an _attribute_ applied to it. Attributes in Rust
1844 are modeled on Attributes in ECMA-335, with the syntax coming from ECMA-334
1845 (C#). An attribute is a general, free-form metadatum that is interpreted
1846 according to name, convention, and language and compiler version. Attributes
1847 may appear as any of:
1849 * A single identifier, the attribute name
1850 * An identifier followed by the equals sign '=' and a literal, providing a
1852 * An identifier followed by a parenthesized list of sub-attribute arguments
1854 Attributes with a bang ("!") after the hash ("#") apply to the item that the
1855 attribute is declared within. Attributes that do not have a bang after the
1856 hash apply to the item that follows the attribute.
1858 An example of attributes:
1861 // General metadata applied to the enclosing module or crate.
1864 // A function marked as a unit test
1870 // A conditionally-compiled module
1871 #[cfg(target_os="linux")]
1876 // A lint attribute used to suppress a warning/error
1877 #[allow(non_camel_case_types)]
1881 > **Note:** At some point in the future, the compiler will distinguish between
1882 > language-reserved and user-available attributes. Until then, there is
1883 > effectively no difference between an attribute handled by a loadable syntax
1884 > extension and the compiler.
1886 ### Crate-only attributes
1888 - `crate_id` - specify the this crate's crate ID.
1889 - `crate_type` - see [linkage](#linkage).
1890 - `feature` - see [compiler features](#compiler-features).
1891 - `no_builtins` - disable optimizing certain code patterns to invocations of
1892 library functions that are assumed to exist
1893 - `no_main` - disable emitting the `main` symbol. Useful when some other
1894 object being linked to defines `main`.
1895 - `no_start` - disable linking to the `native` crate, which specifies the
1896 "start" language item.
1897 - `no_std` - disable linking to the `std` crate.
1899 ### Module-only attributes
1901 - `macro_escape` - macros defined in this module will be visible in the
1902 module's parent, after this module has been included.
1903 - `no_implicit_prelude` - disable injecting `use std::prelude::*` in this
1905 - `path` - specifies the file to load the module from. `#[path="foo.rs"] mod
1906 bar;` is equivalent to `mod bar { /* contents of foo.rs */ }`. The path is
1907 taken relative to the directory that the current module is in.
1909 ### Function-only attributes
1911 - `main` - indicates that this function should be passed to the entry point,
1912 rather than the function in the crate root named `main`.
1913 - `plugin_registrar` - mark this function as the registration point for
1914 compiler plugins, such as loadable syntax extensions.
1915 - `start` - indicates that this function should be used as the entry point,
1916 overriding the "start" language item. See the "start" [language
1917 item](#language-items) for more details.
1919 ### Static-only attributes
1921 - `thread_local` - on a `static mut`, this signals that the value of this
1922 static may change depending on the current thread. The exact consequences of
1923 this are implementation-defined.
1927 On an `extern` block, the following attributes are interpreted:
1929 - `link_args` - specify arguments to the linker, rather than just the library
1930 name and type. This is feature gated and the exact behavior is
1931 implementation-defined (due to variety of linker invocation syntax).
1932 - `link` - indicate that a native library should be linked to for the
1933 declarations in this block to be linked correctly. See [external
1934 blocks](#external-blocks)
1936 On declarations inside an `extern` block, the following attributes are
1939 - `link_name` - the name of the symbol that this function or static should be
1941 - `linkage` - on a static, this specifies the [linkage
1942 type](http://llvm.org/docs/LangRef.html#linkage-types).
1946 - `repr` - on C-like enums, this sets the underlying type used for
1947 representation. Takes one argument, which is the primitive
1948 type this enum should be represented for, or `C`, which specifies that it
1949 should be the default `enum` size of the C ABI for that platform. Note that
1950 enum representation in C is undefined, and this may be incorrect when the C
1951 code is compiled with certain flags.
1955 - `repr` - specifies the representation to use for this struct. Takes a list
1956 of options. The currently accepted ones are `C` and `packed`, which may be
1957 combined. `C` will use a C ABI compatible struct layout, and `packed` will
1958 remove any padding between fields (note that this is very fragile and may
1959 break platforms which require aligned access).
1961 ### Miscellaneous attributes
1963 - `export_name` - on statics and functions, this determines the name of the
1965 - `link_section` - on statics and functions, this specifies the section of the
1966 object file that this item's contents will be placed into.
1967 - `macro_export` - export a macro for cross-crate usage.
1968 - `no_mangle` - on any item, do not apply the standard name mangling. Set the
1969 symbol for this item to its identifier.
1970 - `packed` - on structs or enums, eliminate any padding that would be used to
1972 - `phase` - on `extern crate` statements, allows specifying which "phase" of
1973 compilation the crate should be loaded for. Currently, there are two
1974 choices: `link` and `plugin`. `link` is the default. `plugin` will load the
1975 crate at compile-time and use any syntax extensions or lints that the crate
1976 defines. They can both be specified, `#[phase(link, plugin)]` to use a crate
1977 both at runtime and compiletime.
1978 - `simd` - on certain tuple structs, derive the arithmetic operators, which
1979 lower to the target's SIMD instructions, if any; the `simd` feature gate
1980 is necessary to use this attribute.
1981 - `static_assert` - on statics whose type is `bool`, terminates compilation
1982 with an error if it is not initialized to `true`.
1983 - `unsafe_destructor` - allow implementations of the "drop" language item
1984 where the type it is implemented for does not implement the "send" language
1985 item; the `unsafe_destructor` feature gate is needed to use this attribute
1986 - `unsafe_no_drop_flag` - on structs, remove the flag that prevents
1987 destructors from being run twice. Destructors might be run multiple times on
1988 the same object with this attribute.
1990 ### Conditional compilation
1992 Sometimes one wants to have different compiler outputs from the same code,
1993 depending on build target, such as targeted operating system, or to enable
1996 There are two kinds of configuration options, one that is either defined or not
1997 (`#[cfg(foo)]`), and the other that contains a string that can be checked
1998 against (`#[cfg(bar = "baz")]` (currently only compiler-defined configuration
1999 options can have the latter form).
2002 // The function is only included in the build when compiling for OSX
2003 #[cfg(target_os = "macos")]
2008 // This function is only included when either foo or bar is defined
2011 fn needs_foo_or_bar() {
2015 // This function is only included when compiling for a unixish OS with a 32-bit
2017 #[cfg(unix, target_word_size = "32")]
2018 fn on_32bit_unix() {
2023 This illustrates some conditional compilation can be achieved using the
2024 `#[cfg(...)]` attribute. Note that `#[cfg(foo, bar)]` is a condition that needs
2025 both `foo` and `bar` to be defined while `#[cfg(foo)] #[cfg(bar)]` only needs
2026 one of `foo` and `bar` to be defined (this resembles in the disjunctive normal
2027 form). Additionally, one can reverse a condition by enclosing it in a
2028 `not(...)`, like e. g. `#[cfg(not(target_os = "win32"))]`.
2030 The following configurations must be defined by the implementation:
2032 * `target_arch = "..."`. Target CPU architecture, such as `"x86"`, `"x86_64"`
2033 `"mips"`, or `"arm"`.
2034 * `target_endian = "..."`. Endianness of the target CPU, either `"little"` or
2036 * `target_family = "..."`. Operating system family of the target, e. g.
2037 `"unix"` or `"windows"`. The value of this configuration option is defined as
2038 a configuration itself, like `unix` or `windows`.
2039 * `target_os = "..."`. Operating system of the target, examples include
2040 `"win32"`, `"macos"`, `"linux"`, `"android"`, `"freebsd"` or `"dragonfly"`.
2041 * `target_word_size = "..."`. Target word size in bits. This is set to `"32"`
2042 for targets with 32-bit pointers, and likewise set to `"64"` for 64-bit
2044 * `unix`. See `target_family`.
2045 * `windows`. See `target_family`.
2047 ### Lint check attributes
2049 A lint check names a potentially undesirable coding pattern, such as
2050 unreachable code or omitted documentation, for the static entity to
2051 which the attribute applies.
2053 For any lint check `C`:
2055 * `allow(C)` overrides the check for `C` so that violations will go
2057 * `deny(C)` signals an error after encountering a violation of `C`,
2058 * `forbid(C)` is the same as `deny(C)`, but also forbids changing the lint
2060 * `warn(C)` warns about violations of `C` but continues compilation.
2062 The lint checks supported by the compiler can be found via `rustc -W help`,
2063 along with their default settings.
2067 // Missing documentation is ignored here
2068 #[allow(missing_doc)]
2069 pub fn undocumented_one() -> int { 1 }
2071 // Missing documentation signals a warning here
2072 #[warn(missing_doc)]
2073 pub fn undocumented_too() -> int { 2 }
2075 // Missing documentation signals an error here
2076 #[deny(missing_doc)]
2077 pub fn undocumented_end() -> int { 3 }
2081 This example shows how one can use `allow` and `warn` to toggle
2082 a particular check on and off.
2085 #[warn(missing_doc)]
2087 #[allow(missing_doc)]
2089 // Missing documentation is ignored here
2090 pub fn undocumented_one() -> int { 1 }
2092 // Missing documentation signals a warning here,
2093 // despite the allow above.
2094 #[warn(missing_doc)]
2095 pub fn undocumented_two() -> int { 2 }
2098 // Missing documentation signals a warning here
2099 pub fn undocumented_too() -> int { 3 }
2103 This example shows how one can use `forbid` to disallow uses
2104 of `allow` for that lint check.
2107 #[forbid(missing_doc)]
2109 // Attempting to toggle warning signals an error here
2110 #[allow(missing_doc)]
2112 pub fn undocumented_too() -> int { 2 }
2118 Some primitive Rust operations are defined in Rust code, rather than being
2119 implemented directly in C or assembly language. The definitions of these
2120 operations have to be easy for the compiler to find. The `lang` attribute
2121 makes it possible to declare these operations. For example, the `str` module
2122 in the Rust standard library defines the string equality function:
2126 pub fn eq_slice(a: &str, b: &str) -> bool {
2131 The name `str_eq` has a special meaning to the Rust compiler,
2132 and the presence of this definition means that it will use this definition
2133 when generating calls to the string equality function.
2135 A complete list of the built-in language items follows:
2137 #### Built-in Traits
2140 : Types that do not move ownership when used by-value.
2144 : Able to be sent across task boundaries.
2146 : Has a size known at compile time.
2148 : Able to be safely shared between tasks when aliased.
2152 These language items are traits:
2155 : Elements can be added (for example, integers and floats).
2157 : Elements can be subtracted.
2159 : Elements can be multiplied.
2161 : Elements have a division operation.
2163 : Elements have a remainder operation.
2165 : Elements can be negated arithmetically.
2167 : Elements can be negated logically.
2169 : Elements have an exclusive-or operation.
2171 : Elements have a bitwise `and` operation.
2173 : Elements have a bitwise `or` operation.
2175 : Elements have a left shift operation.
2177 : Elements have a right shift operation.
2179 : Elements can be indexed.
2181 : ___Needs filling in___
2183 : Elements can be compared for equality.
2185 : Elements have a partial ordering.
2187 : `*` can be applied, yielding a reference to another type
2189 : `*` can be applied, yielding a mutable reference to another type
2191 These are functions:
2194 : ___Needs filling in___
2196 : ___Needs filling in___
2198 : ___Needs filling in___
2200 : Compare two strings (`&str`) for equality.
2202 : Return a new unique string
2203 containing a copy of the contents of a unique string.
2208 : The type returned by the `type_id` intrinsic.
2210 : A type whose contents can be mutated through an immutable reference
2214 These types help drive the compiler's analysis
2217 : ___Needs filling in___
2219 : This type implements "managed"
2221 : This type does not implement "copy", even if eligible
2223 : This type does not implement "send", even if eligible
2225 : This type does not implement "sync", even if eligible
2227 : ___Needs filling in___
2229 : Free memory that was allocated on the exchange heap.
2231 : Allocate memory on the exchange heap.
2232 * `closure_exchange_malloc`
2233 : ___Needs filling in___
2235 : Abort the program with an error.
2236 * `fail_bounds_check`
2237 : Abort the program with a bounds check error.
2239 : Free memory that was allocated on the managed heap.
2241 : ___Needs filling in___
2243 : ___Needs filling in___
2245 : ___Needs filling in___
2247 : ___Needs filling in___
2248 * `contravariant_lifetime`
2249 : The lifetime parameter should be considered contravariant
2250 * `covariant_lifetime`
2251 : The lifetime parameter should be considered covariant
2252 * `invariant_lifetime`
2253 : The lifetime parameter should be considered invariant
2255 : Allocate memory on the managed heap.
2257 : ___Needs filling in___
2259 : ___Needs filling in___
2261 : ___Needs filling in___
2263 : ___Needs filling in___
2264 * `contravariant_type`
2265 : The type parameter should be considered contravariant
2267 : The type parameter should be considered covariant
2269 : The type parameter should be considered invariant
2271 : ___Needs filling in___
2273 : ___Needs filling in___
2275 > **Note:** This list is likely to become out of date. We should auto-generate it
2276 > from `librustc/middle/lang_items.rs`.
2278 ### Inline attributes
2280 The inline attribute is used to suggest to the compiler to perform an inline
2281 expansion and place a copy of the function or static in the caller rather than
2282 generating code to call the function or access the static where it is defined.
2284 The compiler automatically inlines functions based on internal heuristics.
2285 Incorrectly inlining functions can actually making the program slower, so it
2286 should be used with care.
2288 Immutable statics are always considered inlineable
2289 unless marked with `#[inline(never)]`.
2291 whether two different inlineable statics
2292 have the same memory address.
2294 the compiler is free
2295 to collapse duplicate inlineable statics together.
2297 `#[inline]` and `#[inline(always)]` always causes the function to be serialized
2298 into crate metadata to allow cross-crate inlining.
2300 There are three different types of inline attributes:
2302 * `#[inline]` hints the compiler to perform an inline expansion.
2303 * `#[inline(always)]` asks the compiler to always perform an inline expansion.
2304 * `#[inline(never)]` asks the compiler to never perform an inline expansion.
2308 The `deriving` attribute allows certain traits to be automatically
2309 implemented for data structures. For example, the following will
2310 create an `impl` for the `PartialEq` and `Clone` traits for `Foo`, the type
2311 parameter `T` will be given the `PartialEq` or `Clone` constraints for the
2315 #[deriving(PartialEq, Clone)]
2322 The generated `impl` for `PartialEq` is equivalent to
2325 # struct Foo<T> { a: int, b: T }
2326 impl<T: PartialEq> PartialEq for Foo<T> {
2327 fn eq(&self, other: &Foo<T>) -> bool {
2328 self.a == other.a && self.b == other.b
2331 fn ne(&self, other: &Foo<T>) -> bool {
2332 self.a != other.a || self.b != other.b
2337 Supported traits for `deriving` are:
2339 * Comparison traits: `PartialEq`, `Eq`, `PartialOrd`, `Ord`.
2340 * Serialization: `Encodable`, `Decodable`. These require `serialize`.
2341 * `Clone`, to create `T` from `&T` via a copy.
2342 * `Default`, to create an empty instance of a data type.
2343 * `FromPrimitive`, to create an instance from a numeric primitive.
2344 * `Hash`, to iterate over the bytes in a data type.
2345 * `Rand`, to create a random instance of a data type.
2346 * `Show`, to format a value using the `{}` formatter.
2347 * `Zero`, to create a zero instance of a numeric data type.
2351 One can indicate the stability of an API using the following attributes:
2353 * `deprecated`: This item should no longer be used, e.g. it has been
2354 replaced. No guarantee of backwards-compatibility.
2355 * `experimental`: This item was only recently introduced or is
2356 otherwise in a state of flux. It may change significantly, or even
2357 be removed. No guarantee of backwards-compatibility.
2358 * `unstable`: This item is still under development, but requires more
2359 testing to be considered stable. No guarantee of backwards-compatibility.
2360 * `stable`: This item is considered stable, and will not change
2361 significantly. Guarantee of backwards-compatibility.
2362 * `frozen`: This item is very stable, and is unlikely to
2363 change. Guarantee of backwards-compatibility.
2364 * `locked`: This item will never change unless a serious bug is
2365 found. Guarantee of backwards-compatibility.
2367 These levels are directly inspired by
2368 [Node.js' "stability index"](http://nodejs.org/api/documentation.html).
2370 Stability levels are inherited, so an item's stability attribute is the
2371 default stability for everything nested underneath it.
2373 There are lints for disallowing items marked with certain levels: `deprecated`,
2374 `experimental` and `unstable`. For now, only `deprecated` warns by default, but
2375 this will change once the standard library has been stabilized.
2376 Stability levels are meant to be promises at the crate
2377 level, so these lints only apply when referencing
2378 items from an _external_ crate, not to items defined within the
2379 current crate. Items with no stability level are considered
2380 to be unstable for the purposes of the lint. One can give an optional
2381 string that will be displayed when the lint flags the use of an item.
2383 For example, if we define one crate called `stability_levels`:
2386 #[deprecated="replaced by `best`"]
2388 // delete everything
2392 // delete fewer things
2401 then the lints will work as follows for a client crate:
2405 extern crate stability_levels;
2406 use stability_levels::{bad, better, best};
2409 bad(); // "warning: use of deprecated item: replaced by `best`"
2411 better(); // "warning: use of unmarked item"
2413 best(); // no warning
2417 > **Note:** Currently these are only checked when applied to
2418 > individual functions, structs, methods and enum variants, *not* to
2419 > entire modules, traits, impls or enums themselves.
2421 ### Compiler Features
2423 Certain aspects of Rust may be implemented in the compiler, but they're not
2424 necessarily ready for every-day use. These features are often of "prototype
2425 quality" or "almost production ready", but may not be stable enough to be
2426 considered a full-fledged language feature.
2428 For this reason, Rust recognizes a special crate-level attribute of the form:
2431 #![feature(feature1, feature2, feature3)]
2434 This directive informs the compiler that the feature list: `feature1`,
2435 `feature2`, and `feature3` should all be enabled. This is only recognized at a
2436 crate-level, not at a module-level. Without this directive, all features are
2437 considered off, and using the features will result in a compiler error.
2439 The currently implemented features of the reference compiler are:
2441 * `asm` - The `asm!` macro provides a means for inline assembly. This is often
2442 useful, but the exact syntax for this feature along with its semantics
2443 are likely to change, so this macro usage must be opted into.
2445 * `concat_idents` - Allows use of the `concat_idents` macro, which is in many
2446 ways insufficient for concatenating identifiers, and may
2447 be removed entirely for something more wholesome.
2449 * `default_type_params` - Allows use of default type parameters. The future of
2450 this feature is uncertain.
2452 * `globs` - Importing everything in a module through `*`. This is currently a
2453 large source of bugs in name resolution for Rust, and it's not clear
2454 whether this will continue as a feature or not. For these reasons,
2455 the glob import statement has been hidden behind this feature flag.
2457 * `intrinsics` - Allows use of the "rust-intrinsics" ABI. Compiler intrinsics
2458 are inherently unstable and no promise about them is made.
2460 * `lang_items` - Allows use of the `#[lang]` attribute. Like `intrinsics`,
2461 lang items are inherently unstable and no promise about
2464 * `link_args` - This attribute is used to specify custom flags to the linker,
2465 but usage is strongly discouraged. The compiler's usage of the
2466 system linker is not guaranteed to continue in the future, and
2467 if the system linker is not used then specifying custom flags
2468 doesn't have much meaning.
2470 * `linkage` - Allows use of the `linkage` attribute, which is not portable.
2472 * `log_syntax` - Allows use of the `log_syntax` macro attribute, which is a
2473 nasty hack that will certainly be removed.
2475 * `macro_rules` - The definition of new macros. This does not encompass
2476 macro-invocation, that is always enabled by default, this only
2477 covers the definition of new macros. There are currently
2478 various problems with invoking macros, how they interact with
2479 their environment, and possibly how they are used outside of
2480 location in which they are defined. Macro definitions are
2481 likely to change slightly in the future, so they are currently
2482 hidden behind this feature.
2484 * `managed_boxes` - Usage of `@` is gated due to many
2485 planned changes to this feature. In the past, this has meant
2486 "a GC pointer", but the current implementation uses
2487 reference counting and will likely change drastically over
2488 time. Additionally, the `@` syntax will no longer be used to
2491 * `non_ascii_idents` - The compiler supports the use of non-ascii identifiers,
2492 but the implementation is a little rough around the
2493 edges, so this can be seen as an experimental feature for
2494 now until the specification of identifiers is fully
2497 * `once_fns` - Onceness guarantees a closure is only executed once. Defining a
2498 closure as `once` is unlikely to be supported going forward. So
2499 they are hidden behind this feature until they are to be removed.
2501 * `overloaded_calls` - Allow implementing the `Fn*` family of traits on user
2502 types, allowing overloading the call operator (`()`).
2503 This feature may still undergo changes before being
2506 * `phase` - Usage of the `#[phase]` attribute allows loading compiler plugins
2507 for custom lints or syntax extensions. The implementation is considered
2508 unwholesome and in need of overhaul, and it is not clear what they
2509 will look like moving forward.
2511 * `plugin_registrar` - Indicates that a crate has compiler plugins that it
2512 wants to load. As with `phase`, the implementation is
2513 in need of a overhaul, and it is not clear that plugins
2514 defined using this will continue to work.
2516 * `quote` - Allows use of the `quote_*!` family of macros, which are
2517 implemented very poorly and will likely change significantly
2518 with a proper implementation.
2520 * `rustc_diagnostic_macros`- A mysterious feature, used in the implementation
2521 of rustc, not meant for mortals.
2523 * `simd` - Allows use of the `#[simd]` attribute, which is overly simple and
2524 not the SIMD interface we want to expose in the long term.
2526 * `struct_inherit` - Allows using struct inheritance, which is barely
2527 implemented and will probably be removed. Don't use this.
2529 * `struct_variant` - Structural enum variants (those with named fields). It is
2530 currently unknown whether this style of enum variant is as
2531 fully supported as the tuple-forms, and it's not certain
2532 that this style of variant should remain in the language.
2533 For now this style of variant is hidden behind a feature
2536 * `thread_local` - The usage of the `#[thread_local]` attribute is experimental
2537 and should be seen as unstable. This attribute is used to
2538 declare a `static` as being unique per-thread leveraging
2539 LLVM's implementation which works in concert with the kernel
2540 loader and dynamic linker. This is not necessarily available
2541 on all platforms, and usage of it is discouraged (rust
2542 focuses more on task-local data instead of thread-local
2545 * `trace_macros` - Allows use of the `trace_macros` macro, which is a nasty
2546 hack that will certainly be removed.
2548 * `unboxed_closure_sugar` - Allows using `|Foo| -> Bar` as a trait bound
2549 meaning one of the `Fn` traits. Still
2552 * `unboxed_closures` - A work in progress feature with many known bugs.
2554 * `unsafe_destructor` - Allows use of the `#[unsafe_destructor]` attribute,
2555 which is considered wildly unsafe and will be
2556 obsoleted by language improvements.
2558 If a feature is promoted to a language feature, then all existing programs will
2559 start to receive compilation warnings about #[feature] directives which enabled
2560 the new feature (because the directive is no longer necessary). However, if
2561 a feature is decided to be removed from the language, errors will be issued (if
2562 there isn't a parser error first). The directive in this case is no longer
2563 necessary, and it's likely that existing code will break if the feature isn't
2566 If a unknown feature is found in a directive, it results in a compiler error. An
2567 unknown feature is one which has never been recognized by the compiler.
2569 # Statements and expressions
2571 Rust is _primarily_ an expression language. This means that most forms of
2572 value-producing or effect-causing evaluation are directed by the uniform
2573 syntax category of _expressions_. Each kind of expression can typically _nest_
2574 within each other kind of expression, and rules for evaluation of expressions
2575 involve specifying both the value produced by the expression and the order in
2576 which its sub-expressions are themselves evaluated.
2578 In contrast, statements in Rust serve _mostly_ to contain and explicitly
2579 sequence expression evaluation.
2583 A _statement_ is a component of a block, which is in turn a component of an
2584 outer [expression](#expressions) or [function](#functions).
2586 Rust has two kinds of statement:
2587 [declaration statements](#declaration-statements) and
2588 [expression statements](#expression-statements).
2590 ### Declaration statements
2592 A _declaration statement_ is one that introduces one or more *names* into the enclosing statement block.
2593 The declared names may denote new slots or new items.
2595 #### Item declarations
2597 An _item declaration statement_ has a syntactic form identical to an
2598 [item](#items) declaration within a module. Declaring an item — a function,
2599 enumeration, structure, type, static, trait, implementation or module — locally
2600 within a statement block is simply a way of restricting its scope to a narrow
2601 region containing all of its uses; it is otherwise identical in meaning to
2602 declaring the item outside the statement block.
2604 > **Note**: there is no implicit capture of the function's dynamic environment when
2605 > declaring a function-local item.
2607 #### Slot declarations
2610 let_decl : "let" pat [':' type ] ? [ init ] ? ';' ;
2611 init : [ '=' ] expr ;
2614 A _slot declaration_ introduces a new set of slots, given by a pattern.
2615 The pattern may be followed by a type annotation, and/or an initializer expression.
2616 When no type annotation is given, the compiler will infer the type,
2617 or signal an error if insufficient type information is available for definite inference.
2618 Any slots introduced by a slot declaration are visible from the point of declaration until the end of the enclosing block scope.
2620 ### Expression statements
2622 An _expression statement_ is one that evaluates an [expression](#expressions)
2623 and ignores its result.
2624 The type of an expression statement `e;` is always `()`, regardless of the type of `e`.
2625 As a rule, an expression statement's purpose is to trigger the effects of evaluating its expression.
2629 An expression may have two roles: it always produces a *value*, and it may have *effects*
2630 (otherwise known as "side effects").
2631 An expression *evaluates to* a value, and has effects during *evaluation*.
2632 Many expressions contain sub-expressions (operands).
2633 The meaning of each kind of expression dictates several things:
2634 * Whether or not to evaluate the sub-expressions when evaluating the expression
2635 * The order in which to evaluate the sub-expressions
2636 * How to combine the sub-expressions' values to obtain the value of the expression.
2638 In this way, the structure of expressions dictates the structure of execution.
2639 Blocks are just another kind of expression,
2640 so blocks, statements, expressions, and blocks again can recursively nest inside each other
2641 to an arbitrary depth.
2643 #### Lvalues, rvalues and temporaries
2645 Expressions are divided into two main categories: _lvalues_ and _rvalues_.
2646 Likewise within each expression, sub-expressions may occur in _lvalue context_ or _rvalue context_.
2647 The evaluation of an expression depends both on its own category and the context it occurs within.
2649 An lvalue is an expression that represents a memory location. These
2650 expressions are [paths](#path-expressions) (which refer to local
2651 variables, function and method arguments, or static variables),
2652 dereferences (`*expr`), [indexing expressions](#index-expressions)
2653 (`expr[expr]`), and [field references](#field-expressions) (`expr.f`).
2654 All other expressions are rvalues.
2656 The left operand of an [assignment](#assignment-expressions) or
2657 [compound-assignment](#compound-assignment-expressions) expression is an lvalue context,
2658 as is the single operand of a unary [borrow](#unary-operator-expressions).
2659 All other expression contexts are rvalue contexts.
2661 When an lvalue is evaluated in an _lvalue context_, it denotes a memory location;
2662 when evaluated in an _rvalue context_, it denotes the value held _in_ that memory location.
2664 When an rvalue is used in lvalue context, a temporary un-named lvalue is created and used instead.
2665 A temporary's lifetime equals the largest lifetime of any reference that points to it.
2667 #### Moved and copied types
2669 When a [local variable](#memory-slots) is used
2670 as an [rvalue](#lvalues,-rvalues-and-temporaries)
2671 the variable will either be moved or copied, depending on its type.
2672 For types that contain [owning pointers](#pointer-types)
2673 or values that implement the special trait `Drop`,
2674 the variable is moved.
2675 All other types are copied.
2677 ### Literal expressions
2679 A _literal expression_ consists of one of the [literal](#literals)
2680 forms described earlier. It directly describes a number, character,
2681 string, boolean value, or the unit value.
2685 "hello"; // string type
2686 '5'; // character type
2690 ### Path expressions
2692 A [path](#paths) used as an expression context denotes either a local variable or an item.
2693 Path expressions are [lvalues](#lvalues,-rvalues-and-temporaries).
2695 ### Tuple expressions
2697 Tuples are written by enclosing one or more comma-separated
2698 expressions in parentheses. They are used to create [tuple-typed](#tuple-types)
2707 ### Structure expressions
2710 struct_expr : expr_path '{' ident ':' expr
2711 [ ',' ident ':' expr ] *
2714 [ ',' expr ] * ')' |
2718 There are several forms of structure expressions.
2719 A _structure expression_ consists of the [path](#paths) of a [structure item](#structures),
2720 followed by a brace-enclosed list of one or more comma-separated name-value pairs,
2721 providing the field values of a new instance of the structure.
2722 A field name can be any identifier, and is separated from its value expression by a colon.
2723 The location denoted by a structure field is mutable if and only if the enclosing structure is mutable.
2725 A _tuple structure expression_ consists of the [path](#paths) of a [structure item](#structures),
2726 followed by a parenthesized list of one or more comma-separated expressions
2727 (in other words, the path of a structure item followed by a tuple expression).
2728 The structure item must be a tuple structure item.
2730 A _unit-like structure expression_ consists only of the [path](#paths) of a [structure item](#structures).
2732 The following are examples of structure expressions:
2735 # struct Point { x: f64, y: f64 }
2736 # struct TuplePoint(f64, f64);
2737 # mod game { pub struct User<'a> { pub name: &'a str, pub age: uint, pub score: uint } }
2738 # struct Cookie; fn some_fn<T>(t: T) {}
2739 Point {x: 10.0, y: 20.0};
2740 TuplePoint(10.0, 20.0);
2741 let u = game::User {name: "Joe", age: 35, score: 100_000};
2742 some_fn::<Cookie>(Cookie);
2745 A structure expression forms a new value of the named structure type.
2746 Note that for a given *unit-like* structure type, this will always be the same value.
2748 A structure expression can terminate with the syntax `..` followed by an expression to denote a functional update.
2749 The expression following `..` (the base) must have the same structure type as the new structure type being formed.
2750 The entire expression denotes the result of constructing a new structure
2751 (with the same type as the base expression)
2752 with the given values for the fields that were explicitly specified
2753 and the values in the base expression for all other fields.
2756 # struct Point3d { x: int, y: int, z: int }
2757 let base = Point3d {x: 1, y: 2, z: 3};
2758 Point3d {y: 0, z: 10, .. base};
2761 ### Block expressions
2764 block_expr : '{' [ view_item ] *
2765 [ stmt ';' | item ] *
2769 A _block expression_ is similar to a module in terms of the declarations that
2770 are possible. Each block conceptually introduces a new namespace scope. View
2771 items can bring new names into scopes and declared items are in scope for only
2774 A block will execute each statement sequentially, and then execute the
2775 expression (if given). If the final expression is omitted, the type and return
2776 value of the block are `()`, but if it is provided, the type and return value
2777 of the block are that of the expression itself.
2779 ### Method-call expressions
2782 method_call_expr : expr '.' ident paren_expr_list ;
2785 A _method call_ consists of an expression followed by a single dot, an identifier, and a parenthesized expression-list.
2786 Method calls are resolved to methods on specific traits,
2787 either statically dispatching to a method if the exact `self`-type of the left-hand-side is known,
2788 or dynamically dispatching if the left-hand-side expression is an indirect [object type](#object-types).
2790 ### Field expressions
2793 field_expr : expr '.' ident ;
2796 A _field expression_ consists of an expression followed by a single dot and an identifier,
2797 when not immediately followed by a parenthesized expression-list (the latter is a [method call expression](#method-call-expressions)).
2798 A field expression denotes a field of a [structure](#structure-types).
2800 ~~~~ {.ignore .field}
2803 (Struct {a: 10, b: 20}).a;
2806 A field access is an [lvalue](#lvalues,-rvalues-and-temporaries) referring to the value of that field.
2807 When the type providing the field inherits mutabilty, it can be [assigned](#assignment-expressions) to.
2809 Also, if the type of the expression to the left of the dot is a pointer,
2810 it is automatically dereferenced to make the field access possible.
2812 ### Vector expressions
2815 vec_expr : '[' "mut" ? vec_elems? ']' ;
2817 vec_elems : [expr [',' expr]*] | [expr ',' ".." expr] ;
2820 A [_vector_](#vector-types) _expression_ is written by enclosing zero or
2821 more comma-separated expressions of uniform type in square brackets.
2823 In the `[expr ',' ".." expr]` form, the expression after the `".."`
2824 must be a constant expression that can be evaluated at compile time, such
2825 as a [literal](#literals) or a [static item](#static-items).
2829 ["a", "b", "c", "d"];
2830 [0i, ..128]; // vector with 128 zeros
2831 [0u8, 0u8, 0u8, 0u8];
2834 ### Index expressions
2837 idx_expr : expr '[' expr ']' ;
2840 [Vector](#vector-types)-typed expressions can be indexed by writing a
2841 square-bracket-enclosed expression (the index) after them. When the
2842 vector is mutable, the resulting [lvalue](#lvalues,-rvalues-and-temporaries) can be assigned to.
2844 Indices are zero-based, and may be of any integral type. Vector access
2845 is bounds-checked at run-time. When the check fails, it will put the
2846 task in a _failing state_.
2850 # task::spawn(proc() {
2853 (["a", "b"])[10]; // fails
2858 ### Unary operator expressions
2860 Rust defines six symbolic unary operators.
2861 They are all written as prefix operators,
2862 before the expression they apply to.
2865 : Negation. May only be applied to numeric types.
2867 : Dereference. When applied to a [pointer](#pointer-types) it denotes the pointed-to location.
2868 For pointers to mutable locations, the resulting [lvalue](#lvalues,-rvalues-and-temporaries) can be assigned to.
2869 On non-pointer types, it calls the `deref` method of the `std::ops::Deref` trait, or the
2870 `deref_mut` method of the `std::ops::DerefMut` trait (if implemented by the type and required
2871 for an outer expression that will or could mutate the dereference), and produces the
2872 result of dereferencing the `&` or `&mut` borrowed pointer returned from the overload method.
2875 : Logical negation. On the boolean type, this flips between `true` and
2876 `false`. On integer types, this inverts the individual bits in the
2877 two's complement representation of the value.
2879 : [Boxing](#pointer-types) operators. Allocate a box to hold the value they are applied to,
2880 and store the value in it. `box` creates an owned box.
2882 : Borrow operator. Returns a reference, pointing to its operand.
2883 The operand of a borrow is statically proven to outlive the resulting pointer.
2884 If the borrow-checker cannot prove this, it is a compilation error.
2886 ### Binary operator expressions
2889 binop_expr : expr binop expr ;
2892 Binary operators expressions are given in terms of
2893 [operator precedence](#operator-precedence).
2895 #### Arithmetic operators
2897 Binary arithmetic expressions are syntactic sugar for calls to built-in traits,
2898 defined in the `std::ops` module of the `std` library.
2899 This means that arithmetic operators can be overridden for user-defined types.
2900 The default meaning of the operators on standard types is given here.
2903 : Addition and vector/string concatenation.
2904 Calls the `add` method on the `std::ops::Add` trait.
2907 Calls the `sub` method on the `std::ops::Sub` trait.
2910 Calls the `mul` method on the `std::ops::Mul` trait.
2913 Calls the `div` method on the `std::ops::Div` trait.
2916 Calls the `rem` method on the `std::ops::Rem` trait.
2918 #### Bitwise operators
2920 Like the [arithmetic operators](#arithmetic-operators), bitwise operators
2921 are syntactic sugar for calls to methods of built-in traits.
2922 This means that bitwise operators can be overridden for user-defined types.
2923 The default meaning of the operators on standard types is given here.
2927 Calls the `bitand` method of the `std::ops::BitAnd` trait.
2930 Calls the `bitor` method of the `std::ops::BitOr` trait.
2933 Calls the `bitxor` method of the `std::ops::BitXor` trait.
2935 : Logical left shift.
2936 Calls the `shl` method of the `std::ops::Shl` trait.
2938 : Logical right shift.
2939 Calls the `shr` method of the `std::ops::Shr` trait.
2941 #### Lazy boolean operators
2943 The operators `||` and `&&` may be applied to operands of boolean type.
2944 The `||` operator denotes logical 'or', and the `&&` operator denotes logical 'and'.
2945 They differ from `|` and `&` in that the right-hand operand is only evaluated
2946 when the left-hand operand does not already determine the result of the expression.
2947 That is, `||` only evaluates its right-hand operand
2948 when the left-hand operand evaluates to `false`, and `&&` only when it evaluates to `true`.
2950 #### Comparison operators
2952 Comparison operators are, like the [arithmetic operators](#arithmetic-operators),
2953 and [bitwise operators](#bitwise-operators),
2954 syntactic sugar for calls to built-in traits.
2955 This means that comparison operators can be overridden for user-defined types.
2956 The default meaning of the operators on standard types is given here.
2960 Calls the `eq` method on the `std::cmp::PartialEq` trait.
2963 Calls the `ne` method on the `std::cmp::PartialEq` trait.
2966 Calls the `lt` method on the `std::cmp::PartialOrd` trait.
2969 Calls the `gt` method on the `std::cmp::PartialOrd` trait.
2971 : Less than or equal.
2972 Calls the `le` method on the `std::cmp::PartialOrd` trait.
2974 : Greater than or equal.
2975 Calls the `ge` method on the `std::cmp::PartialOrd` trait.
2977 #### Type cast expressions
2979 A type cast expression is denoted with the binary operator `as`.
2981 Executing an `as` expression casts the value on the left-hand side to the type
2982 on the right-hand side.
2984 A numeric value can be cast to any numeric type.
2985 A raw pointer value can be cast to or from any integral type or raw pointer type.
2986 Any other cast is unsupported and will fail to compile.
2988 An example of an `as` expression:
2991 # fn sum(v: &[f64]) -> f64 { 0.0 }
2992 # fn len(v: &[f64]) -> int { 0 }
2994 fn avg(v: &[f64]) -> f64 {
2995 let sum: f64 = sum(v);
2996 let sz: f64 = len(v) as f64;
3001 #### Assignment expressions
3003 An _assignment expression_ consists of an [lvalue](#lvalues,-rvalues-and-temporaries) expression followed by an
3004 equals sign (`=`) and an [rvalue](#lvalues,-rvalues-and-temporaries) expression.
3006 Evaluating an assignment expression [either copies or moves](#moved-and-copied-types) its right-hand operand to its left-hand operand.
3015 #### Compound assignment expressions
3017 The `+`, `-`, `*`, `/`, `%`, `&`, `|`, `^`, `<<`, and `>>`
3018 operators may be composed with the `=` operator. The expression `lval
3019 OP= val` is equivalent to `lval = lval OP val`. For example, `x = x +
3020 1` may be written as `x += 1`.
3022 Any such expression always has the [`unit`](#primitive-types) type.
3024 #### Operator precedence
3026 The precedence of Rust binary operators is ordered as follows, going
3027 from strong to weak:
3029 ~~~~ {.text .precedence}
3044 Operators at the same precedence level are evaluated left-to-right. [Unary operators](#unary-operator-expressions)
3045 have the same precedence level and it is stronger than any of the binary operators'.
3047 ### Grouped expressions
3049 An expression enclosed in parentheses evaluates to the result of the enclosed
3050 expression. Parentheses can be used to explicitly specify evaluation order
3051 within an expression.
3054 paren_expr : '(' expr ')' ;
3057 An example of a parenthesized expression:
3060 let x: int = (2 + 3) * 4;
3064 ### Call expressions
3067 expr_list : [ expr [ ',' expr ]* ] ? ;
3068 paren_expr_list : '(' expr_list ')' ;
3069 call_expr : expr paren_expr_list ;
3072 A _call expression_ invokes a function, providing zero or more input slots and
3073 an optional reference slot to serve as the function's output, bound to the
3074 `lval` on the right hand side of the call. If the function eventually returns,
3075 then the expression completes.
3077 Some examples of call expressions:
3080 # use std::from_str::FromStr;
3081 # fn add(x: int, y: int) -> int { 0 }
3083 let x: int = add(1, 2);
3084 let pi: Option<f32> = FromStr::from_str("3.14");
3087 ### Lambda expressions
3090 ident_list : [ ident [ ',' ident ]* ] ? ;
3091 lambda_expr : '|' ident_list '|' expr ;
3094 A _lambda expression_ (sometimes called an "anonymous function expression") defines a function and denotes it as a value,
3095 in a single expression.
3096 A lambda expression is a pipe-symbol-delimited (`|`) list of identifiers followed by an expression.
3098 A lambda expression denotes a function that maps a list of parameters (`ident_list`)
3099 onto the expression that follows the `ident_list`.
3100 The identifiers in the `ident_list` are the parameters to the function.
3101 These parameters' types need not be specified, as the compiler infers them from context.
3103 Lambda expressions are most useful when passing functions as arguments to other functions,
3104 as an abbreviation for defining and capturing a separate function.
3106 Significantly, lambda expressions _capture their environment_,
3107 which regular [function definitions](#functions) do not.
3108 The exact type of capture depends on the [function type](#function-types) inferred for the lambda expression.
3109 In the simplest and least-expensive form (analogous to a ```|| { }``` expression),
3110 the lambda expression captures its environment by reference,
3111 effectively borrowing pointers to all outer variables mentioned inside the function.
3112 Alternately, the compiler may infer that a lambda expression should copy or move values (depending on their type.)
3113 from the environment into the lambda expression's captured environment.
3115 In this example, we define a function `ten_times` that takes a higher-order function argument,
3116 and call it with a lambda expression as an argument.
3119 fn ten_times(f: |int|) {
3127 ten_times(|j| println!("hello, {}", j));
3133 while_expr : "while" no_struct_literal_expr '{' block '}' ;
3136 A `while` loop begins by evaluating the boolean loop conditional expression.
3137 If the loop conditional expression evaluates to `true`, the loop body block
3138 executes and control returns to the loop conditional expression. If the loop
3139 conditional expression evaluates to `false`, the `while` expression completes.
3154 A `loop` expression denotes an infinite loop.
3157 loop_expr : [ lifetime ':' ] "loop" '{' block '}';
3160 A `loop` expression may optionally have a _label_.
3161 If a label is present,
3162 then labeled `break` and `continue` expressions nested within this loop may exit out of this loop or return control to its head.
3163 See [Break expressions](#break-expressions) and [Continue expressions](#continue-expressions).
3165 ### Break expressions
3168 break_expr : "break" [ lifetime ];
3171 A `break` expression has an optional _label_.
3172 If the label is absent, then executing a `break` expression immediately terminates the innermost loop enclosing it.
3173 It is only permitted in the body of a loop.
3174 If the label is present, then `break foo` terminates the loop with label `foo`,
3175 which need not be the innermost label enclosing the `break` expression,
3176 but must enclose it.
3178 ### Continue expressions
3181 continue_expr : "continue" [ lifetime ];
3184 A `continue` expression has an optional _label_.
3185 If the label is absent,
3186 then executing a `continue` expression immediately terminates the current iteration of the innermost loop enclosing it,
3187 returning control to the loop *head*.
3188 In the case of a `while` loop,
3189 the head is the conditional expression controlling the loop.
3190 In the case of a `for` loop, the head is the call-expression controlling the loop.
3191 If the label is present, then `continue foo` returns control to the head of the loop with label `foo`,
3192 which need not be the innermost label enclosing the `break` expression,
3193 but must enclose it.
3195 A `continue` expression is only permitted in the body of a loop.
3200 for_expr : "for" pat "in" no_struct_literal_expr '{' block '}' ;
3203 A `for` expression is a syntactic construct for looping over elements
3204 provided by an implementation of `std::iter::Iterator`.
3206 An example of a for loop over the contents of a vector:
3210 # fn bar(f: Foo) { }
3215 let v: &[Foo] = &[a, b, c];
3222 An example of a for loop over a series of integers:
3225 # fn bar(b:uint) { }
3226 for i in range(0u, 256) {
3234 if_expr : "if" no_struct_literal_expr '{' block '}'
3237 else_tail : "else" [ if_expr
3241 An `if` expression is a conditional branch in program control. The form of
3242 an `if` expression is a condition expression, followed by a consequent
3243 block, any number of `else if` conditions and blocks, and an optional
3244 trailing `else` block. The condition expressions must have type
3245 `bool`. If a condition expression evaluates to `true`, the
3246 consequent block is executed and any subsequent `else if` or `else`
3247 block is skipped. If a condition expression evaluates to `false`, the
3248 consequent block is skipped and any subsequent `else if` condition is
3249 evaluated. If all `if` and `else if` conditions evaluate to `false`
3250 then any `else` block is executed.
3252 ### Match expressions
3255 match_expr : "match" no_struct_literal_expr '{' match_arm * '}' ;
3257 match_arm : attribute * match_pat "=>" [ expr "," | '{' block '}' ] ;
3259 match_pat : pat [ '|' pat ] * [ "if" expr ] ? ;
3262 A `match` expression branches on a *pattern*. The exact form of matching that
3263 occurs depends on the pattern. Patterns consist of some combination of
3264 literals, destructured vectors or enum constructors, structures and
3265 tuples, variable binding specifications, wildcards (`..`), and placeholders
3266 (`_`). A `match` expression has a *head expression*, which is the value to
3267 compare to the patterns. The type of the patterns must equal the type of the
3270 In a pattern whose head expression has an `enum` type, a placeholder (`_`)
3271 stands for a *single* data field, whereas a wildcard `..` stands for *all* the
3272 fields of a particular variant. For example:
3275 enum List<X> { Nil, Cons(X, Box<List<X>>) }
3277 let x: List<int> = Cons(10, box Cons(11, box Nil));
3280 Cons(_, box Nil) => fail!("singleton list"),
3282 Nil => fail!("empty list")
3286 The first pattern matches lists constructed by applying `Cons` to any head
3287 value, and a tail value of `box Nil`. The second pattern matches _any_ list
3288 constructed with `Cons`, ignoring the values of its arguments. The difference
3289 between `_` and `..` is that the pattern `C(_)` is only type-correct if `C` has
3290 exactly one argument, while the pattern `C(..)` is type-correct for any enum
3291 variant `C`, regardless of how many arguments `C` has.
3293 Used inside a vector pattern, `..` stands for any number of elements. This
3294 wildcard can be used at most once for a given vector, which implies that it
3295 cannot be used to specifically match elements that are at an unknown distance
3296 from both ends of a vector, like `[.., 42, ..]`. If followed by a variable name,
3297 it will bind the corresponding slice to the variable. Example:
3300 fn is_symmetric(list: &[uint]) -> bool {
3303 [x, ..inside, y] if x == y => is_symmetric(inside),
3309 let sym = &[0, 1, 4, 2, 4, 1, 0];
3310 let not_sym = &[0, 1, 7, 2, 4, 1, 0];
3311 assert!(is_symmetric(sym));
3312 assert!(!is_symmetric(not_sym));
3316 A `match` behaves differently depending on whether or not the head expression
3317 is an [lvalue or an rvalue](#lvalues,-rvalues-and-temporaries).
3318 If the head expression is an rvalue, it is
3319 first evaluated into a temporary location, and the resulting value
3320 is sequentially compared to the patterns in the arms until a match
3321 is found. The first arm with a matching pattern is chosen as the branch target
3322 of the `match`, any variables bound by the pattern are assigned to local
3323 variables in the arm's block, and control enters the block.
3325 When the head expression is an lvalue, the match does not allocate a
3326 temporary location (however, a by-value binding may copy or move from
3327 the lvalue). When possible, it is preferable to match on lvalues, as the
3328 lifetime of these matches inherits the lifetime of the lvalue, rather
3329 than being restricted to the inside of the match.
3331 An example of a `match` expression:
3334 # fn process_pair(a: int, b: int) { }
3335 # fn process_ten() { }
3337 enum List<X> { Nil, Cons(X, Box<List<X>>) }
3339 let x: List<int> = Cons(10, box Cons(11, box Nil));
3342 Cons(a, box Cons(b, _)) => {
3357 Patterns that bind variables
3358 default to binding to a copy or move of the matched value
3359 (depending on the matched value's type).
3360 This can be changed to bind to a reference by
3361 using the `ref` keyword,
3362 or to a mutable reference using `ref mut`.
3364 Subpatterns can also be bound to variables by the use of the syntax
3365 `variable @ subpattern`.
3369 enum List { Nil, Cons(uint, Box<List>) }
3371 fn is_sorted(list: &List) -> bool {
3373 Nil | Cons(_, box Nil) => true,
3374 Cons(x, ref r @ box Cons(_, _)) => {
3376 box Cons(y, _) => (x <= y) && is_sorted(&**r),
3384 let a = Cons(6, box Cons(7, box Cons(42, box Nil)));
3385 assert!(is_sorted(&a));
3390 Patterns can also dereference pointers by using the `&`,
3391 `box` or `@` symbols, as appropriate. For example, these two matches
3392 on `x: &int` are equivalent:
3396 let y = match *x { 0 => "zero", _ => "some" };
3397 let z = match x { &0 => "zero", _ => "some" };
3402 A pattern that's just an identifier, like `Nil` in the previous example,
3403 could either refer to an enum variant that's in scope, or bind a new variable.
3404 The compiler resolves this ambiguity by forbidding variable bindings that occur
3405 in `match` patterns from shadowing names of variants that are in scope.
3406 For example, wherever `List` is in scope,
3407 a `match` pattern would not be able to bind `Nil` as a new name.
3408 The compiler interprets a variable pattern `x` as a binding _only_ if there is
3409 no variant named `x` in scope.
3410 A convention you can use to avoid conflicts is simply to name variants with
3411 upper-case letters, and local variables with lower-case letters.
3413 Multiple match patterns may be joined with the `|` operator.
3414 A range of values may be specified with `..`.
3420 let message = match x {
3421 0 | 1 => "not many",
3427 Range patterns only work on scalar types
3428 (like integers and characters; not like vectors and structs, which have sub-components).
3429 A range pattern may not be a sub-range of another range pattern inside the same `match`.
3431 Finally, match patterns can accept *pattern guards* to further refine the
3432 criteria for matching a case. Pattern guards appear after the pattern and
3433 consist of a bool-typed expression following the `if` keyword. A pattern
3434 guard may refer to the variables bound within the pattern they follow.
3437 # let maybe_digit = Some(0);
3438 # fn process_digit(i: int) { }
3439 # fn process_other(i: int) { }
3441 let message = match maybe_digit {
3442 Some(x) if x < 10 => process_digit(x),
3443 Some(x) => process_other(x),
3448 ### Return expressions
3451 return_expr : "return" expr ? ;
3454 Return expressions are denoted with the keyword `return`. Evaluating a `return`
3455 expression moves its argument into the output slot of the current
3456 function, destroys the current function activation frame, and transfers
3457 control to the caller frame.
3459 An example of a `return` expression:
3462 fn max(a: int, b: int) -> int {
3474 Every slot, item and value in a Rust program has a type. The _type_ of a *value*
3475 defines the interpretation of the memory holding it.
3477 Built-in types and type-constructors are tightly integrated into the language,
3478 in nontrivial ways that are not possible to emulate in user-defined
3479 types. User-defined types have limited capabilities.
3483 The primitive types are the following:
3485 * The "unit" type `()`, having the single "unit" value `()` (occasionally called
3487 * The boolean type `bool` with values `true` and `false`.
3488 * The machine types.
3489 * The machine-dependent integer and floating-point types.
3491 [^unittype]: The "unit" value `()` is *not* a sentinel "null pointer" value for
3492 reference slots; the "unit" type is the implicit return type from functions
3493 otherwise lacking a return type, and can be used in other contexts (such as
3494 message-sending or type-parametric code) as a zero-size type.]
3498 The machine types are the following:
3500 * The unsigned word types `u8`, `u16`, `u32` and `u64`, with values drawn from
3501 the integer intervals [0, 2^8 - 1], [0, 2^16 - 1], [0, 2^32 - 1] and
3502 [0, 2^64 - 1] respectively.
3504 * The signed two's complement word types `i8`, `i16`, `i32` and `i64`, with
3505 values drawn from the integer intervals [-(2^(7)), 2^7 - 1],
3506 [-(2^(15)), 2^15 - 1], [-(2^(31)), 2^31 - 1], [-(2^(63)), 2^63 - 1]
3509 * The IEEE 754-2008 `binary32` and `binary64` floating-point types: `f32` and
3510 `f64`, respectively.
3512 #### Machine-dependent integer types
3514 The Rust type `uint` [^rustuint] is an
3515 unsigned integer type with target-machine-dependent size. Its size, in
3516 bits, is equal to the number of bits required to hold any memory address on
3519 The Rust type `int` [^rustint] is a
3520 two's complement signed integer type with target-machine-dependent size. Its
3521 size, in bits, is equal to the size of the rust type `uint` on the same target
3524 [^rustuint]: A Rust `uint` is analogous to a C99 `uintptr_t`.
3525 [^rustint]: A Rust `int` is analogous to a C99 `intptr_t`.
3529 The types `char` and `str` hold textual data.
3531 A value of type `char` is a [Unicode scalar value](
3532 http://www.unicode.org/glossary/#unicode_scalar_value)
3533 (ie. a code point that is not a surrogate),
3534 represented as a 32-bit unsigned word in the 0x0000 to 0xD7FF
3535 or 0xE000 to 0x10FFFF range.
3536 A `[char]` vector is effectively an UCS-4 / UTF-32 string.
3538 A value of type `str` is a Unicode string,
3539 represented as a vector of 8-bit unsigned bytes holding a sequence of UTF-8 codepoints.
3540 Since `str` is of unknown size, it is not a _first class_ type,
3541 but can only be instantiated through a pointer type,
3542 such as `&str` or `String`.
3546 A tuple *type* is a heterogeneous product of other types, called the *elements*
3547 of the tuple. It has no nominal name and is instead structurally typed.
3549 Tuple types and values are denoted by listing the types or values of their
3550 elements, respectively, in a parenthesized, comma-separated
3553 Because tuple elements don't have a name, they can only be accessed by pattern-matching.
3555 The members of a tuple are laid out in memory contiguously, in
3556 order specified by the tuple type.
3558 An example of a tuple type and its use:
3561 type Pair<'a> = (int, &'a str);
3562 let p: Pair<'static> = (10, "hello");
3564 assert!(b != "world");
3569 The vector type constructor represents a homogeneous array of values of a given type.
3570 A vector has a fixed size.
3571 (Operations like `vec.push` operate solely on owned vectors.)
3572 A vector type can be annotated with a _definite_ size, such as `[int, ..10]`.
3573 Such a definite-sized vector type is a first-class type, since its size is known statically.
3574 A vector without such a size is said to be of _indefinite_ size,
3575 and is therefore not a _first-class_ type.
3576 An indefinite-size vector can only be instantiated through a pointer type,
3577 such as `&[T]` or `Vec<T>`.
3578 The kind of a vector type depends on the kind of its element type,
3579 as with other simple structural types.
3581 Expressions producing vectors of definite size cannot be evaluated in a
3582 context expecting a vector of indefinite size; one must copy the
3583 definite-sized vector contents into a distinct vector of indefinite size.
3585 An example of a vector type and its use:
3588 let v: &[int] = &[7, 5, 3];
3593 All in-bounds elements of a vector are always initialized,
3594 and access to a vector is always bounds-checked.
3598 A `struct` *type* is a heterogeneous product of other types, called the *fields*
3599 of the type.[^structtype]
3601 [^structtype]: `struct` types are analogous `struct` types in C,
3602 the *record* types of the ML family,
3603 or the *structure* types of the Lisp family.
3605 New instances of a `struct` can be constructed with a [struct expression](#structure-expressions).
3607 The memory layout of a `struct` is undefined by default to allow for compiler optimizations like
3608 field reordering, but it can be fixed with the `#[repr(...)]` attribute.
3609 In either case, fields may be given in any order in a corresponding struct *expression*;
3610 the resulting `struct` value will always have the same memory layout.
3612 The fields of a `struct` may be qualified by [visibility modifiers](#re-exporting-and-visibility),
3613 to allow access to data in a structure outside a module.
3615 A _tuple struct_ type is just like a structure type, except that the fields are anonymous.
3617 A _unit-like struct_ type is like a structure type, except that it has no fields.
3618 The one value constructed by the associated [structure expression](#structure-expressions)
3619 is the only value that inhabits such a type.
3621 ### Enumerated types
3623 An *enumerated type* is a nominal, heterogeneous disjoint union type,
3624 denoted by the name of an [`enum` item](#enumerations). [^enumtype]
3626 [^enumtype]: The `enum` type is analogous to a `data` constructor declaration in
3627 ML, or a *pick ADT* in Limbo.
3629 An [`enum` item](#enumerations) declares both the type and a number of *variant constructors*,
3630 each of which is independently named and takes an optional tuple of arguments.
3632 New instances of an `enum` can be constructed by calling one of the variant constructors,
3633 in a [call expression](#call-expressions).
3635 Any `enum` value consumes as much memory as the largest variant constructor for its corresponding `enum` type.
3637 Enum types cannot be denoted *structurally* as types,
3638 but must be denoted by named reference to an [`enum` item](#enumerations).
3642 Nominal types — [enumerations](#enumerated-types) and [structures](#structure-types) — may be recursive.
3643 That is, each `enum` constructor or `struct` field may refer, directly or indirectly, to the enclosing `enum` or `struct` type itself.
3644 Such recursion has restrictions:
3646 * Recursive types must include a nominal type in the recursion
3647 (not mere [type definitions](#type-definitions),
3648 or other structural types such as [vectors](#vector-types) or [tuples](#tuple-types)).
3649 * A recursive `enum` item must have at least one non-recursive constructor
3650 (in order to give the recursion a basis case).
3651 * The size of a recursive type must be finite;
3652 in other words the recursive fields of the type must be [pointer types](#pointer-types).
3653 * Recursive type definitions can cross module boundaries, but not module *visibility* boundaries,
3654 or crate boundaries (in order to simplify the module system and type checker).
3656 An example of a *recursive* type and its use:
3661 Cons(T, Box<List<T>>)
3664 let a: List<int> = Cons(7, box Cons(13, box Nil));
3669 All pointers in Rust are explicit first-class values.
3670 They can be copied, stored into data structures, and returned from functions.
3671 There are two varieties of pointer in Rust:
3674 : These point to memory _owned by some other value_.
3675 A reference type is written `&type` for some lifetime-variable `f`,
3676 or just `&'a type` when you need an explicit lifetime.
3677 Copying a reference is a "shallow" operation:
3678 it involves only copying the pointer itself.
3679 Releasing a reference typically has no effect on the value it points to,
3680 with the exception of temporary values, which are released when the last
3681 reference to them is released.
3683 * Raw pointers (`*`)
3684 : Raw pointers are pointers without safety or liveness guarantees.
3685 Raw pointers are written as `*const T` or `*mut T`,
3686 for example `*const int` means a raw pointer to an integer.
3687 Copying or dropping a raw pointer has no effect on the lifecycle of any
3688 other value. Dereferencing a raw pointer or converting it to any other
3689 pointer type is an [`unsafe` operation](#unsafe-functions).
3690 Raw pointers are generally discouraged in Rust code;
3691 they exist to support interoperability with foreign code,
3692 and writing performance-critical or low-level functions.
3694 The standard library contains addtional 'smart pointer' types beyond references
3699 The function type constructor `fn` forms new function types.
3700 A function type consists of a possibly-empty set of function-type modifiers
3701 (such as `unsafe` or `extern`), a sequence of input types and an output type.
3703 An example of a `fn` type:
3706 fn add(x: int, y: int) -> int {
3710 let mut x = add(5,7);
3712 type Binop<'a> = |int,int|: 'a -> int;
3713 let bo: Binop = add;
3719 ~~~~ {.ebnf .notation}
3720 closure_type := [ 'unsafe' ] [ '<' lifetime-list '>' ] '|' arg-list '|'
3721 [ ':' bound-list ] [ '->' type ]
3722 procedure_type := 'proc' [ '<' lifetime-list '>' ] '(' arg-list ')'
3723 [ ':' bound-list ] [ '->' type ]
3724 lifetime-list := lifetime | lifetime ',' lifetime-list
3725 arg-list := ident ':' type | ident ':' type ',' arg-list
3726 bound-list := bound | bound '+' bound-list
3727 bound := path | lifetime
3730 The type of a closure mapping an input of type `A` to an output of type `B` is
3731 `|A| -> B`. A closure with no arguments or return values has type `||`.
3732 Similarly, a procedure mapping `A` to `B` is `proc(A) -> B` and a no-argument
3733 and no-return value closure has type `proc()`.
3735 An example of creating and calling a closure:
3738 let captured_var = 10i;
3740 let closure_no_args = || println!("captured_var={}", captured_var);
3742 let closure_args = |arg: int| -> int {
3743 println!("captured_var={}, arg={}", captured_var, arg);
3744 arg // Note lack of semicolon after 'arg'
3747 fn call_closure(c1: ||, c2: |int| -> int) {
3752 call_closure(closure_no_args, closure_args);
3756 Unlike closures, procedures may only be invoked once, but own their
3757 environment, and are allowed to move out of their environment. Procedures are
3758 allocated on the heap (unlike closures). An example of creating and calling a
3762 let string = "Hello".to_string();
3764 // Creates a new procedure, passing it to the `spawn` function.
3766 println!("{} world!", string);
3769 // the variable `string` has been moved into the previous procedure, so it is
3770 // no longer usable.
3773 // Create an invoke a procedure. Note that the procedure is *moved* when
3774 // invoked, so it cannot be invoked again.
3775 let f = proc(n: int) { n + 22 };
3776 println!("answer: {}", f(20));
3782 Every trait item (see [traits](#traits)) defines a type with the same name as the trait.
3783 This type is called the _object type_ of the trait.
3784 Object types permit "late binding" of methods, dispatched using _virtual method tables_ ("vtables").
3785 Whereas most calls to trait methods are "early bound" (statically resolved) to specific implementations at compile time,
3786 a call to a method on an object type is only resolved to a vtable entry at compile time.
3787 The actual implementation for each vtable entry can vary on an object-by-object basis.
3789 Given a pointer-typed expression `E` of type `&T` or `Box<T>`, where `T` implements trait `R`,
3790 casting `E` to the corresponding pointer type `&R` or `Box<R>` results in a value of the _object type_ `R`.
3791 This result is represented as a pair of pointers:
3792 the vtable pointer for the `T` implementation of `R`, and the pointer value of `E`.
3794 An example of an object type:
3798 fn stringify(&self) -> String;
3801 impl Printable for int {
3802 fn stringify(&self) -> String { self.to_string() }
3805 fn print(a: Box<Printable>) {
3806 println!("{}", a.stringify());
3810 print(box 10i as Box<Printable>);
3814 In this example, the trait `Printable` occurs as an object type in both the type signature of `print`,
3815 and the cast expression in `main`.
3819 Within the body of an item that has type parameter declarations, the names of its type parameters are types:
3822 fn map<A: Clone, B: Clone>(f: |A| -> B, xs: &[A]) -> Vec<B> {
3826 let first: B = f(xs[0].clone());
3827 let rest: Vec<B> = map(f, xs.slice(1, xs.len()));
3828 return vec![first].append(rest.as_slice());
3832 Here, `first` has type `B`, referring to `map`'s `B` type parameter;
3833 and `rest` has type `Vec<B>`, a vector type with element type `B`.
3837 The special type `self` has a meaning within methods inside an
3838 impl item. It refers to the type of the implicit `self` argument. For
3843 fn make_string(&self) -> String;
3846 impl Printable for String {
3847 fn make_string(&self) -> String {
3853 `self` refers to the value of type `String` that is the receiver for a
3854 call to the method `make_string`.
3858 Types in Rust are categorized into kinds, based on various properties of the components of the type.
3862 : Types of this kind can be safely sent between tasks.
3863 This kind includes scalars, owning pointers, owned closures, and
3864 structural types containing only other owned types.
3865 All `Send` types are `'static`.
3867 : Types of this kind consist of "Plain Old Data"
3868 which can be copied by simply moving bits.
3869 All values of this kind can be implicitly copied.
3870 This kind includes scalars and immutable references,
3871 as well as structural types containing other `Copy` types.
3873 : Types of this kind do not contain any references (except for
3874 references with the `static` lifetime, which are allowed).
3875 This can be a useful guarantee for code
3876 that breaks borrowing assumptions
3877 using [`unsafe` operations](#unsafe-functions).
3879 : This is not strictly a kind,
3880 but its presence interacts with kinds:
3881 the `Drop` trait provides a single method `drop`
3882 that takes no parameters,
3883 and is run when values of the type are dropped.
3884 Such a method is called a "destructor",
3885 and are always executed in "top-down" order:
3886 a value is completely destroyed
3887 before any of the values it owns run their destructors.
3888 Only `Send` types can implement `Drop`.
3891 : Types with destructors, closure environments,
3892 and various other _non-first-class_ types,
3893 are not copyable at all.
3894 Such types can usually only be accessed through pointers,
3895 or in some cases, moved between mutable locations.
3897 Kinds can be supplied as _bounds_ on type parameters, like traits,
3898 in which case the parameter is constrained to types satisfying that kind.
3900 By default, type parameters do not carry any assumed kind-bounds at all.
3901 When instantiating a type parameter,
3902 the kind bounds on the parameter are checked
3903 to be the same or narrower than the kind
3904 of the type that it is instantiated with.
3906 Sending operations are not part of the Rust language,
3907 but are implemented in the library.
3908 Generic functions that send values
3909 bound the kind of these values to sendable.
3911 # Memory and concurrency models
3913 Rust has a memory model centered around concurrently-executing _tasks_. Thus
3914 its memory model and its concurrency model are best discussed simultaneously,
3915 as parts of each only make sense when considered from the perspective of the
3918 When reading about the memory model, keep in mind that it is partitioned in
3919 order to support tasks; and when reading about tasks, keep in mind that their
3920 isolation and communication mechanisms are only possible due to the ownership
3921 and lifetime semantics of the memory model.
3925 A Rust program's memory consists of a static set of *items*, a set of
3926 [tasks](#tasks) each with its own *stack*, and a *heap*. Immutable portions of
3927 the heap may be shared between tasks, mutable portions may not.
3929 Allocations in the stack consist of *slots*, and allocations in the heap
3932 ### Memory allocation and lifetime
3934 The _items_ of a program are those functions, modules and types
3935 that have their value calculated at compile-time and stored uniquely in the
3936 memory image of the rust process. Items are neither dynamically allocated nor
3939 A task's _stack_ consists of activation frames automatically allocated on
3940 entry to each function as the task executes. A stack allocation is reclaimed
3941 when control leaves the frame containing it.
3943 The _heap_ is a general term that describes two separate sets of boxes:
3944 managed boxes — which may be subject to garbage collection — and owned
3945 boxes. The lifetime of an allocation in the heap depends on the lifetime of
3946 the box values pointing to it. Since box values may themselves be passed in
3947 and out of frames, or stored in the heap, heap allocations may outlive the
3948 frame they are allocated within.
3950 ### Memory ownership
3952 A task owns all memory it can *safely* reach through local variables,
3953 as well as managed, owned boxes and references.
3955 When a task sends a value that has the `Send` trait to another task,
3956 it loses ownership of the value sent and can no longer refer to it.
3957 This is statically guaranteed by the combined use of "move semantics",
3958 and the compiler-checked _meaning_ of the `Send` trait:
3959 it is only instantiated for (transitively) sendable kinds of data constructor and pointers,
3960 never including managed boxes or references.
3962 When a stack frame is exited, its local allocations are all released, and its
3963 references to boxes (both managed and owned) are dropped.
3965 A managed box may (in the case of a recursive, mutable managed type) be cyclic;
3966 in this case the release of memory inside the managed structure may be deferred
3967 until task-local garbage collection can reclaim it. Code can ensure no such
3968 delayed deallocation occurs by restricting itself to owned boxes and similar
3969 unmanaged kinds of data.
3971 When a task finishes, its stack is necessarily empty and it therefore has no
3972 references to any boxes; the remainder of its heap is immediately freed.
3976 A task's stack contains slots.
3978 A _slot_ is a component of a stack frame, either a function parameter,
3979 a [temporary](#lvalues,-rvalues-and-temporaries), or a local variable.
3981 A _local variable_ (or *stack-local* allocation) holds a value directly,
3982 allocated within the stack's memory. The value is a part of the stack frame.
3984 Local variables are immutable unless declared otherwise like: `let mut x = ...`.
3986 Function parameters are immutable unless declared with `mut`. The
3987 `mut` keyword applies only to the following parameter (so `|mut x, y|`
3988 and `fn f(mut x: Box<int>, y: Box<int>)` declare one mutable variable `x` and
3989 one immutable variable `y`).
3991 Methods that take either `self` or `Box<Self>` can optionally place them in a
3992 mutable slot by prefixing them with `mut` (similar to regular arguments):
3996 fn change(mut self) -> Self;
3997 fn modify(mut self: Box<Self>) -> Box<Self>;
4001 Local variables are not initialized when allocated; the entire frame worth of
4002 local variables are allocated at once, on frame-entry, in an uninitialized
4003 state. Subsequent statements within a function may or may not initialize the
4004 local variables. Local variables can be used only after they have been
4005 initialized; this is enforced by the compiler.
4009 An _owned box_ is a reference to a heap allocation holding another value, which is constructed
4010 by the prefix operator `box`. When the standard library is in use, the type of an owned box is
4011 `std::owned::Box<T>`.
4013 An example of an owned box type and value:
4016 let x: Box<int> = box 10;
4019 Owned box values exist in 1:1 correspondence with their heap allocation,
4020 copying an owned box value makes a shallow copy of the pointer.
4021 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.
4024 let x: Box<int> = box 10;
4026 // attempting to use `x` will result in an error here
4033 An executing Rust program consists of a tree of tasks.
4034 A Rust _task_ consists of an entry function, a stack,
4035 a set of outgoing communication channels and incoming communication ports,
4036 and ownership of some portion of the heap of a single operating-system process.
4037 (We expect that many programs will not use channels and ports directly,
4038 but will instead use higher-level abstractions provided in standard libraries,
4041 Multiple Rust tasks may coexist in a single operating-system process.
4042 The runtime scheduler maps tasks to a certain number of operating-system threads.
4043 By default, the scheduler chooses the number of threads based on
4044 the number of concurrent physical CPUs detected at startup.
4045 It's also possible to override this choice at runtime.
4046 When the number of tasks exceeds the number of threads — which is likely —
4047 the scheduler multiplexes the tasks onto threads.[^mnscheduler]
4049 [^mnscheduler]: This is an M:N scheduler, which is known to give suboptimal
4050 results for CPU-bound concurrency problems. In such cases, running with the
4051 same number of threads and tasks can yield better results. Rust has M:N
4052 scheduling in order to support very large numbers of tasks in contexts where
4053 threads are too resource-intensive to use in large number. The cost of
4054 threads varies substantially per operating system, and is sometimes quite
4055 low, so this flexibility is not always worth exploiting.
4057 ### Communication between tasks
4059 Rust tasks are isolated and generally unable to interfere with one another's memory directly,
4060 except through [`unsafe` code](#unsafe-functions).
4061 All contact between tasks is mediated by safe forms of ownership transfer,
4062 and data races on memory are prohibited by the type system.
4064 Inter-task communication and co-ordination facilities are provided in the standard library.
4067 - synchronous and asynchronous communication channels with various communication topologies
4068 - read-only and read-write shared variables with various safe mutual exclusion patterns
4069 - simple locks and semaphores
4071 When such facilities carry values, the values are restricted to the [`Send` type-kind](#type-kinds).
4072 Restricting communication interfaces to this kind ensures that no references or managed pointers move between tasks.
4073 Thus access to an entire data structure can be mediated through its owning "root" value;
4074 no further locking or copying is required to avoid data races within the substructure of such a value.
4078 The _lifecycle_ of a task consists of a finite set of states and events
4079 that cause transitions between the states. The lifecycle states of a task are:
4086 A task begins its lifecycle — once it has been spawned — in the *running*
4087 state. In this state it executes the statements of its entry function, and any
4088 functions called by the entry function.
4090 A task may transition from the *running* state to the *blocked*
4091 state any time it makes a blocking communication call. When the
4092 call can be completed — when a message arrives at a sender, or a
4093 buffer opens to receive a message — then the blocked task will
4094 unblock and transition back to *running*.
4096 A task may transition to the *failing* state at any time, due being
4097 killed by some external event or internally, from the evaluation of a
4098 `fail!()` macro. Once *failing*, a task unwinds its stack and
4099 transitions to the *dead* state. Unwinding the stack of a task is done by
4100 the task itself, on its own control stack. If a value with a destructor is
4101 freed during unwinding, the code for the destructor is run, also on the task's
4102 control stack. Running the destructor code causes a temporary transition to a
4103 *running* state, and allows the destructor code to cause any subsequent
4104 state transitions. The original task of unwinding and failing thereby may
4105 suspend temporarily, and may involve (recursive) unwinding of the stack of a
4106 failed destructor. Nonetheless, the outermost unwinding activity will continue
4107 until the stack is unwound and the task transitions to the *dead*
4108 state. There is no way to "recover" from task failure. Once a task has
4109 temporarily suspended its unwinding in the *failing* state, failure
4110 occurring from within this destructor results in *hard* failure.
4111 A hard failure currently results in the process aborting.
4113 A task in the *dead* state cannot transition to other states; it exists
4114 only to have its termination status inspected by other tasks, and/or to await
4115 reclamation when the last reference to it drops.
4119 The currently scheduled task is given a finite *time slice* in which to
4120 execute, after which it is *descheduled* at a loop-edge or similar
4121 preemption point, and another task within is scheduled, pseudo-randomly.
4123 An executing task can yield control at any time, by making a library call to
4124 `std::task::yield`, which deschedules it immediately. Entering any other
4125 non-executing state (blocked, dead) similarly deschedules the task.
4127 # Runtime services, linkage and debugging
4129 The Rust _runtime_ is a relatively compact collection of C++ and Rust code
4130 that provides fundamental services and datatypes to all Rust tasks at
4131 run-time. It is smaller and simpler than many modern language runtimes. It is
4132 tightly integrated into the language's execution model of memory, tasks,
4133 communication and logging.
4135 > **Note:** The runtime library will merge with the `std` library in future versions of Rust.
4137 ### Memory allocation
4139 The runtime memory-management system is based on a _service-provider interface_,
4140 through which the runtime requests blocks of memory from its environment
4141 and releases them back to its environment when they are no longer needed.
4142 The default implementation of the service-provider interface
4143 consists of the C runtime functions `malloc` and `free`.
4145 The runtime memory-management system, in turn, supplies Rust tasks with
4146 facilities for allocating releasing stacks, as well as allocating and freeing
4151 The runtime provides C and Rust code to assist with various built-in types,
4152 such as vectors, strings, and the low level communication system (ports,
4155 Support for other built-in types such as simple types, tuples and
4156 enums is open-coded by the Rust compiler.
4158 ### Task scheduling and communication
4160 The runtime provides code to manage inter-task communication. This includes
4161 the system of task-lifecycle state transitions depending on the contents of
4162 queues, as well as code to copy values between queues and their recipients and
4163 to serialize values for transmission over operating-system inter-process
4164 communication facilities.
4168 The Rust compiler supports various methods to link crates together both
4169 statically and dynamically. This section will explore the various methods to
4170 link Rust crates together, and more information about native libraries can be
4171 found in the [ffi tutorial][ffi].
4173 In one session of compilation, the compiler can generate multiple artifacts
4174 through the usage of either command line flags or the `crate_type` attribute.
4175 If one or more command line flag is specified, all `crate_type` attributes will
4176 be ignored in favor of only building the artifacts specified by command line.
4178 * `--crate-type=bin`, `#[crate_type = "bin"]` - A runnable executable will be
4179 produced. This requires that there is a `main` function in the crate which
4180 will be run when the program begins executing. This will link in all Rust and
4181 native dependencies, producing a distributable binary.
4183 * `--crate-type=lib`, `#[crate_type = "lib"]` - A Rust library will be produced.
4184 This is an ambiguous concept as to what exactly is produced because a library
4185 can manifest itself in several forms. The purpose of this generic `lib` option
4186 is to generate the "compiler recommended" style of library. The output library
4187 will always be usable by rustc, but the actual type of library may change from
4188 time-to-time. The remaining output types are all different flavors of
4189 libraries, and the `lib` type can be seen as an alias for one of them (but the
4190 actual one is compiler-defined).
4192 * `--crate-type=dylib`, `#[crate_type = "dylib"]` - A dynamic Rust library will
4193 be produced. This is different from the `lib` output type in that this forces
4194 dynamic library generation. The resulting dynamic library can be used as a
4195 dependency for other libraries and/or executables. This output type will
4196 create `*.so` files on linux, `*.dylib` files on osx, and `*.dll` files on
4199 * `--crate-type=staticlib`, `#[crate_type = "staticlib"]` - A static system
4200 library will be produced. This is different from other library outputs in that
4201 the Rust compiler will never attempt to link to `staticlib` outputs. The
4202 purpose of this output type is to create a static library containing all of
4203 the local crate's code along with all upstream dependencies. The static
4204 library is actually a `*.a` archive on linux and osx and a `*.lib` file on
4205 windows. This format is recommended for use in situations such as linking
4206 Rust code into an existing non-Rust application because it will not have
4207 dynamic dependencies on other Rust code.
4209 * `--crate-type=rlib`, `#[crate_type = "rlib"]` - A "Rust library" file will be
4210 produced. This is used as an intermediate artifact and can be thought of as a
4211 "static Rust library". These `rlib` files, unlike `staticlib` files, are
4212 interpreted by the Rust compiler in future linkage. This essentially means
4213 that `rustc` will look for metadata in `rlib` files like it looks for metadata
4214 in dynamic libraries. This form of output is used to produce statically linked
4215 executables as well as `staticlib` outputs.
4217 Note that these outputs are stackable in the sense that if multiple are
4218 specified, then the compiler will produce each form of output at once without
4219 having to recompile. However, this only applies for outputs specified by the same
4220 method. If only `crate_type` attributes are specified, then they will all be
4221 built, but if one or more `--crate-type` command line flag is specified,
4222 then only those outputs will be built.
4224 With all these different kinds of outputs, if crate A depends on crate B, then
4225 the compiler could find B in various different forms throughout the system. The
4226 only forms looked for by the compiler, however, are the `rlib` format and the
4227 dynamic library format. With these two options for a dependent library, the
4228 compiler must at some point make a choice between these two formats. With this
4229 in mind, the compiler follows these rules when determining what format of
4230 dependencies will be used:
4232 1. If a static library is being produced, all upstream dependencies are
4233 required to be available in `rlib` formats. This requirement stems from the
4234 reason that a dynamic library cannot be converted into a static format.
4236 Note that it is impossible to link in native dynamic dependencies to a static
4237 library, and in this case warnings will be printed about all unlinked native
4238 dynamic dependencies.
4240 2. If an `rlib` file is being produced, then there are no restrictions on what
4241 format the upstream dependencies are available in. It is simply required that
4242 all upstream dependencies be available for reading metadata from.
4244 The reason for this is that `rlib` files do not contain any of their upstream
4245 dependencies. It wouldn't be very efficient for all `rlib` files to contain a
4246 copy of `libstd.rlib`!
4248 3. If an executable is being produced and the `-C prefer-dynamic` flag is not
4249 specified, then dependencies are first attempted to be found in the `rlib`
4250 format. If some dependencies are not available in an rlib format, then
4251 dynamic linking is attempted (see below).
4253 4. If a dynamic library or an executable that is being dynamically linked is
4254 being produced, then the compiler will attempt to reconcile the available
4255 dependencies in either the rlib or dylib format to create a final product.
4257 A major goal of the compiler is to ensure that a library never appears more
4258 than once in any artifact. For example, if dynamic libraries B and C were
4259 each statically linked to library A, then a crate could not link to B and C
4260 together because there would be two copies of A. The compiler allows mixing
4261 the rlib and dylib formats, but this restriction must be satisfied.
4263 The compiler currently implements no method of hinting what format a library
4264 should be linked with. When dynamically linking, the compiler will attempt to
4265 maximize dynamic dependencies while still allowing some dependencies to be
4266 linked in via an rlib.
4268 For most situations, having all libraries available as a dylib is recommended
4269 if dynamically linking. For other situations, the compiler will emit a
4270 warning if it is unable to determine which formats to link each library with.
4272 In general, `--crate-type=bin` or `--crate-type=lib` should be sufficient for
4273 all compilation needs, and the other options are just available if more
4274 fine-grained control is desired over the output format of a Rust crate.
4278 The runtime contains a system for directing [logging
4279 expressions](#logging-expressions) to a logging console and/or internal logging
4280 buffers. Logging can be enabled per module.
4282 Logging output is enabled by setting the `RUST_LOG` environment
4283 variable. `RUST_LOG` accepts a logging specification made up of a
4284 comma-separated list of paths, with optional log levels. For each
4285 module containing log expressions, if `RUST_LOG` contains the path to
4286 that module or a parent of that module, then logs of the appropriate
4287 level will be output to the console.
4289 The path to a module consists of the crate name, any parent modules,
4290 then the module itself, all separated by double colons (`::`). The
4291 optional log level can be appended to the module path with an equals
4292 sign (`=`) followed by the log level, from 1 to 4, inclusive. Level 1
4293 is the error level, 2 is warning, 3 info, and 4 debug. You can also
4294 use the symbolic constants `error`, `warn`, `info`, and `debug`. Any
4295 logs less than or equal to the specified level will be output. If not
4296 specified then log level 4 is assumed. Debug messages can be omitted
4297 by passing `--cfg ndebug` to `rustc`.
4299 As an example, to see all the logs generated by the compiler, you would set
4300 `RUST_LOG` to `rustc`, which is the crate name (as specified in its `crate_id`
4301 [attribute](#attributes)). To narrow down the logs to just crate resolution,
4302 you would set it to `rustc::metadata::creader`. To see just error logging
4305 Note that when compiling source files that don't specify a
4306 crate name the crate is given a default name that matches the source file,
4307 with the extension removed. In that case, to turn on logging for a program
4308 compiled from, e.g. `helloworld.rs`, `RUST_LOG` should be set to `helloworld`.
4310 #### Logging Expressions
4312 Rust provides several macros to log information. Here's a simple Rust program
4313 that demonstrates all four of them:
4317 #[phase(plugin, link)] extern crate log;
4320 error!("This is an error log")
4321 warn!("This is a warn log")
4322 info!("this is an info log")
4323 debug!("This is a debug log")
4327 These four log levels correspond to levels 1-4, as controlled by `RUST_LOG`:
4330 $ RUST_LOG=rust=3 ./rust
4331 This is an error log
4336 # Appendix: Rationales and design tradeoffs
4340 # Appendix: Influences and further references
4344 > The essential problem that must be solved in making a fault-tolerant
4345 > software system is therefore that of fault-isolation. Different programmers
4346 > will write different modules, some modules will be correct, others will have
4347 > errors. We do not want the errors in one module to adversely affect the
4348 > behaviour of a module which does not have any errors.
4350 > — Joe Armstrong
4352 > In our approach, all data is private to some process, and processes can
4353 > only communicate through communications channels. *Security*, as used
4354 > in this paper, is the property which guarantees that processes in a system
4355 > cannot affect each other except by explicit communication.
4357 > When security is absent, nothing which can be proven about a single module
4358 > in isolation can be guaranteed to hold when that module is embedded in a
4361 > — Robert Strom and Shaula Yemini
4363 > Concurrent and applicative programming complement each other. The
4364 > ability to send messages on channels provides I/O without side effects,
4365 > while the avoidance of shared data helps keep concurrent processes from
4370 Rust is not a particularly original language. It may however appear unusual
4371 by contemporary standards, as its design elements are drawn from a number of
4372 "historical" languages that have, with a few exceptions, fallen out of
4373 favour. Five prominent lineages contribute the most, though their influences
4374 have come and gone during the course of Rust's development:
4376 * The NIL (1981) and Hermes (1990) family. These languages were developed by
4377 Robert Strom, Shaula Yemini, David Bacon and others in their group at IBM
4378 Watson Research Center (Yorktown Heights, NY, USA).
4380 * The Erlang (1987) language, developed by Joe Armstrong, Robert Virding, Claes
4381 Wikström, Mike Williams and others in their group at the Ericsson Computer
4382 Science Laboratory (Älvsjö, Stockholm, Sweden) .
4384 * The Sather (1990) language, developed by Stephen Omohundro, Chu-Cheow Lim,
4385 Heinz Schmidt and others in their group at The International Computer
4386 Science Institute of the University of California, Berkeley (Berkeley, CA,
4389 * The Newsqueak (1988), Alef (1995), and Limbo (1996) family. These
4390 languages were developed by Rob Pike, Phil Winterbottom, Sean Dorward and
4391 others in their group at Bell Labs Computing Sciences Research Center
4392 (Murray Hill, NJ, USA).
4394 * The Napier (1985) and Napier88 (1988) family. These languages were
4395 developed by Malcolm Atkinson, Ron Morrison and others in their group at
4396 the University of St. Andrews (St. Andrews, Fife, UK).
4398 Additional specific influences can be seen from the following languages:
4400 * The structural algebraic types and compilation manager of SML.
4401 * The attribute and assembly systems of C#.
4402 * The references and deterministic destructor system of C++.
4403 * The memory region systems of the ML Kit and Cyclone.
4404 * The typeclass system of Haskell.
4405 * The lexical identifier rule of Python.
4406 * The block syntax of Ruby.
4408 [ffi]: guide-ffi.html