<!DOCTYPE HTML> <html lang="en"> <head> <meta charset="UTF-8"> <title>The Rust Reference</title> <meta content="text/html; charset=utf-8" http-equiv="Content-Type"> <meta name="description" content=""> <meta name="viewport" content="width=device-width, initial-scale=1"> <meta name="theme-color" content="#ffffff" /> <base href=""> <link rel="stylesheet" href="book.css"> <link href="https://fonts.googleapis.com/css?family=Open+Sans:300italic,400italic,600italic,700italic,800italic,400,300,600,700,800" rel="stylesheet" type="text/css"> <link href="https://fonts.googleapis.com/css?family=Source+Code+Pro:500" rel="stylesheet" type="text/css"> <link rel="shortcut icon" href="favicon.png"> <!-- Font Awesome --> <link rel="stylesheet" href="https://maxcdn.bootstrapcdn.com/font-awesome/4.3.0/css/font-awesome.min.css"> <link rel="stylesheet" href="highlight.css"> <link rel="stylesheet" href="tomorrow-night.css"> <link rel="stylesheet" href="ayu-highlight.css"> <!-- Custom theme --> <link rel="stylesheet" href="src/theme/reference.css"> <!-- Fetch Clipboard.js from CDN but have a local fallback --> <script src="https://cdn.jsdelivr.net/clipboard.js/1.6.1/clipboard.min.js"></script> <script> if (typeof Clipboard == 'undefined') { document.write(unescape("%3Cscript src='clipboard.min.js'%3E%3C/script%3E")); } </script> </head> <body class="light"> <!-- Work around some values being stored in localStorage wrapped in quotes --> <script type="text/javascript"> try { var theme = localStorage.getItem('mdbook-theme'); var sidebar = localStorage.getItem('mdbook-sidebar'); if (theme.startsWith('"') && theme.endsWith('"')) { localStorage.setItem('mdbook-theme', theme.slice(1, theme.length - 1)); } if (sidebar.startsWith('"') && sidebar.endsWith('"')) { localStorage.setItem('mdbook-sidebar', sidebar.slice(1, sidebar.length - 1)); } } catch (e) { } </script> <!-- Set the theme before any content is loaded, prevents flash --> <script type="text/javascript"> var theme; try { theme = localStorage.getItem('mdbook-theme'); } catch(e) { } if (theme === null || theme === undefined) { theme = 'light'; } document.body.className = theme; document.querySelector('html').className = theme; </script> <!-- Hide / unhide sidebar before it is displayed --> <script type="text/javascript"> var sidebar = 'hidden'; if (document.body.clientWidth >= 1080) { try { sidebar = localStorage.getItem('mdbook-sidebar'); } catch(e) { } sidebar = sidebar || 'visible'; } document.querySelector('html').classList.add("sidebar-" + sidebar); </script> <nav id="sidebar" class="sidebar" aria-label="Table of contents"> <ol class="chapter"><li class="affix"><a href="introduction.html">Introduction</a></li><li><a href="notation.html"><strong aria-hidden="true">1.</strong> Notation</a></li><li><a href="lexical-structure.html"><strong aria-hidden="true">2.</strong> Lexical structure</a></li><li><ol class="section"><li><a href="input-format.html"><strong aria-hidden="true">2.1.</strong> Input format</a></li><li><a href="keywords.html"><strong aria-hidden="true">2.2.</strong> Keywords</a></li><li><a href="identifiers.html"><strong aria-hidden="true">2.3.</strong> Identifiers</a></li><li><a href="comments.html"><strong aria-hidden="true">2.4.</strong> Comments</a></li><li><a href="whitespace.html"><strong aria-hidden="true">2.5.</strong> Whitespace</a></li><li><a href="tokens.html"><strong aria-hidden="true">2.6.</strong> Tokens</a></li><li><a href="paths.html"><strong aria-hidden="true">2.7.</strong> Paths</a></li></ol></li><li><a href="macros.html"><strong aria-hidden="true">3.</strong> Macros</a></li><li><ol class="section"><li><a href="macros-by-example.html"><strong aria-hidden="true">3.1.</strong> Macros By Example</a></li><li><a href="procedural-macros.html"><strong aria-hidden="true">3.2.</strong> Procedural Macros</a></li></ol></li><li><a href="crates-and-source-files.html"><strong aria-hidden="true">4.</strong> Crates and source files</a></li><li><a href="items-and-attributes.html"><strong aria-hidden="true">5.</strong> Items and attributes</a></li><li><ol class="section"><li><a href="items.html"><strong aria-hidden="true">5.1.</strong> Items</a></li><li><ol class="section"><li><a href="items/modules.html"><strong aria-hidden="true">5.1.1.</strong> Modules</a></li><li><a href="items/extern-crates.html"><strong aria-hidden="true">5.1.2.</strong> Extern crates</a></li><li><a href="items/use-declarations.html"><strong aria-hidden="true">5.1.3.</strong> Use declarations</a></li><li><a href="items/functions.html"><strong aria-hidden="true">5.1.4.</strong> Functions</a></li><li><a href="items/type-aliases.html"><strong aria-hidden="true">5.1.5.</strong> Type aliases</a></li><li><a href="items/structs.html"><strong aria-hidden="true">5.1.6.</strong> Structs</a></li><li><a href="items/enumerations.html"><strong aria-hidden="true">5.1.7.</strong> Enumerations</a></li><li><a href="items/unions.html"><strong aria-hidden="true">5.1.8.</strong> Unions</a></li><li><a href="items/constant-items.html"><strong aria-hidden="true">5.1.9.</strong> Constant items</a></li><li><a href="items/static-items.html"><strong aria-hidden="true">5.1.10.</strong> Static items</a></li><li><a href="items/traits.html"><strong aria-hidden="true">5.1.11.</strong> Traits</a></li><li><a href="items/implementations.html"><strong aria-hidden="true">5.1.12.</strong> Implementations</a></li><li><a href="items/external-blocks.html"><strong aria-hidden="true">5.1.13.</strong> External blocks</a></li></ol></li><li><a href="visibility-and-privacy.html"><strong aria-hidden="true">5.2.</strong> Visibility and Privacy</a></li><li><a href="attributes.html"><strong aria-hidden="true">5.3.</strong> Attributes</a></li></ol></li><li><a href="statements-and-expressions.html"><strong aria-hidden="true">6.</strong> Statements and expressions</a></li><li><ol class="section"><li><a href="statements.html"><strong aria-hidden="true">6.1.</strong> Statements</a></li><li><a href="expressions.html"><strong aria-hidden="true">6.2.</strong> Expressions</a></li><li><ol class="section"><li><a href="expressions/literal-expr.html"><strong aria-hidden="true">6.2.1.</strong> Literal expressions</a></li><li><a href="expressions/path-expr.html"><strong aria-hidden="true">6.2.2.</strong> Path expressions</a></li><li><a href="expressions/block-expr.html"><strong aria-hidden="true">6.2.3.</strong> Block expressions</a></li><li><a href="expressions/operator-expr.html"><strong aria-hidden="true">6.2.4.</strong> Operator expressions</a></li><li><a href="expressions/array-expr.html"><strong aria-hidden="true">6.2.5.</strong> Array and index expressions</a></li><li><a href="expressions/tuple-expr.html"><strong aria-hidden="true">6.2.6.</strong> Tuple and index expressions</a></li><li><a href="expressions/struct-expr.html"><strong aria-hidden="true">6.2.7.</strong> Struct expressions</a></li><li><a href="expressions/enum-variant-expr.html"><strong aria-hidden="true">6.2.8.</strong> Enum variant expressions</a></li><li><a href="expressions/call-expr.html"><strong aria-hidden="true">6.2.9.</strong> Call expressions</a></li><li><a href="expressions/method-call-expr.html"><strong aria-hidden="true">6.2.10.</strong> Method call expressions</a></li><li><a href="expressions/field-expr.html"><strong aria-hidden="true">6.2.11.</strong> Field access expressions</a></li><li><a href="expressions/closure-expr.html"><strong aria-hidden="true">6.2.12.</strong> Closure expressions</a></li><li><a href="expressions/loop-expr.html"><strong aria-hidden="true">6.2.13.</strong> Loop expressions</a></li><li><a href="expressions/range-expr.html"><strong aria-hidden="true">6.2.14.</strong> Range expressions</a></li><li><a href="expressions/if-expr.html"><strong aria-hidden="true">6.2.15.</strong> If and if let expressions</a></li><li><a href="expressions/match-expr.html"><strong aria-hidden="true">6.2.16.</strong> Match expressions</a></li><li><a href="expressions/return-expr.html"><strong aria-hidden="true">6.2.17.</strong> Return expressions</a></li></ol></li></ol></li><li><a href="type-system.html"><strong aria-hidden="true">7.</strong> Type system</a></li><li><ol class="section"><li><a href="types.html"><strong aria-hidden="true">7.1.</strong> Types</a></li><li><a href="dynamically-sized-types.html"><strong aria-hidden="true">7.2.</strong> Dynamically Sized Types</a></li><li><a href="type-layout.html"><strong aria-hidden="true">7.3.</strong> Type layout</a></li><li><a href="interior-mutability.html"><strong aria-hidden="true">7.4.</strong> Interior mutability</a></li><li><a href="subtyping.html"><strong aria-hidden="true">7.5.</strong> Subtyping</a></li><li><a href="type-coercions.html"><strong aria-hidden="true">7.6.</strong> Type coercions</a></li><li><a href="destructors.html"><strong aria-hidden="true">7.7.</strong> Destructors</a></li></ol></li><li><a href="special-types-and-traits.html"><strong aria-hidden="true">8.</strong> Special types and traits</a></li><li><a href="memory-model.html"><strong aria-hidden="true">9.</strong> Memory model</a></li><li><ol class="section"><li><a href="memory-allocation-and-lifetime.html"><strong aria-hidden="true">9.1.</strong> Memory allocation and lifetime</a></li><li><a href="memory-ownership.html"><strong aria-hidden="true">9.2.</strong> Memory ownership</a></li><li><a href="variables.html"><strong aria-hidden="true">9.3.</strong> Variables</a></li></ol></li><li><a href="linkage.html"><strong aria-hidden="true">10.</strong> Linkage</a></li><li><a href="unsafety.html"><strong aria-hidden="true">11.</strong> Unsafety</a></li><li><ol class="section"><li><a href="unsafe-functions.html"><strong aria-hidden="true">11.1.</strong> Unsafe functions</a></li><li><a href="unsafe-blocks.html"><strong aria-hidden="true">11.2.</strong> Unsafe blocks</a></li><li><a href="behavior-considered-undefined.html"><strong aria-hidden="true">11.3.</strong> Behavior considered undefined</a></li><li><a href="behavior-not-considered-unsafe.html"><strong aria-hidden="true">11.4.</strong> Behavior not considered unsafe</a></li></ol></li><li><a href="influences.html">Appendix: Influences</a></li><li class="affix"><a href="undocumented.html">Appendix: As-yet-undocumented Features</a></li><li class="affix"><a href="glossary.html">Appendix: Glossary</a></li></ol> </nav> <div id="page-wrapper" class="page-wrapper"> <div class="page"> <header><p class="warning"> For now, this reference is a best-effort document. We strive for validity and completeness, but are not yet there. In the future, the docs and lang teams will work together to figure out how best to do this. Until then, this is a best-effort attempt. If you find something wrong or missing, file an <a href="https://github.com/rust-lang-nursery/reference/issues">issue</a> or send in a pull request. </p></header> <div id="menu-bar" class="menu-bar"> <div id="menu-bar-sticky-container"> <div class="left-buttons"> <button id="sidebar-toggle" class="icon-button" type="button" title="Toggle Table of Contents" aria-label="Toggle Table of Contents" aria-controls="sidebar"> <i class="fa fa-bars"></i> </button> <button id="theme-toggle" class="icon-button" type="button" title="Change theme" aria-label="Change theme" aria-haspopup="true" aria-expanded="false" aria-controls="theme-list"> <i class="fa fa-paint-brush"></i> </button> <ul id="theme-list" class="theme-popup" aria-label="submenu"> <li><button class="theme" id="light">Light <span class="default">(default)</span></button></li> <li><button class="theme" id="rust">Rust</button></li> <li><button class="theme" id="coal">Coal</button></li> <li><button class="theme" id="navy">Navy</button></li> <li><button class="theme" id="ayu">Ayu</button></li> </ul> </div> <h1 class="menu-title">The Rust Reference</h1> <div class="right-buttons"> <a href="print.html" title="Print this book" aria-label="Print this book"> <i id="print-button" class="fa fa-print"></i> </a> </div> </div> </div> <!-- Apply ARIA attributes after the sidebar and the sidebar toggle button are added to the DOM --> <script type="text/javascript"> document.getElementById('sidebar-toggle').setAttribute('aria-expanded', sidebar === 'visible'); document.getElementById('sidebar').setAttribute('aria-hidden', sidebar !== 'visible'); Array.from(document.querySelectorAll('#sidebar a')).forEach(function(link) { link.setAttribute('tabIndex', sidebar === 'visible' ? 0 : -1); }); </script> <div id="content" class="content"> <main> <a class="header" href="print.html#introduction" id="introduction"><h1>Introduction</h1></a> <p>This document is the primary reference for the Rust programming language. It provides three kinds of material:</p> <ul> <li>Chapters that informally describe each language construct and their use.</li> <li>Chapters that informally describe the memory model, concurrency model, runtime services, linkage model and debugging facilities.</li> <li>Appendix chapters providing rationale and references to languages that influenced the design.</li> </ul> <p>This document does not serve as an introduction to the language. Background familiarity with the language is assumed. A separate <a href="../book/index.html">book</a> is available to help acquire such background familiarity.</p> <p>This document also does not serve as a reference to the <a href="../std/index.html">standard</a> library included in the language distribution. Those libraries are documented separately by extracting documentation attributes from their source code. Many of the features that one might expect to be language features are library features in Rust, so what you're looking for may be there, not here.</p> <p>This document also only serves as a reference to what is available in stable Rust. For unstable features being worked on, see the <a href="https://doc.rust-lang.org/nightly/unstable-book/">Unstable Book</a>. This was a recent change in scope, so unstable features are still documented, but are in the process of being removed.</p> <p>Finally, this document is not normative. It may include details that are specific to <code>rustc</code> itself, and should not be taken as a specification for the Rust language. We intend to produce such a document someday, but this is what we have for now.</p> <p>You may also be interested in the <a href="../grammar.html">grammar</a>.</p> <p>You can contribute to this document by opening an issue or sending a pull request to <a href="https://github.com/rust-lang-nursery/reference/">the Rust Reference repository</a>.</p> <p>N. B. This document may be incomplete. Documenting everything might take a while. We have a <a href="https://github.com/rust-lang-nursery/reference/issues/9">big issue</a> to track documentation for every Rust feature, so check that out if you can't find something here.</p> <a class="header" href="print.html#notation" id="notation"><h1>Notation</h1></a> <a class="header" href="print.html#unicode-productions" id="unicode-productions"><h2>Unicode productions</h2></a> <p>A few productions in Rust's grammar permit Unicode code points outside the ASCII range. We define these productions in terms of character properties specified in the Unicode standard, rather than in terms of ASCII-range code points. The grammar has a <a href="../grammar.html#special-unicode-productions">Special Unicode Productions</a> section that lists these productions.</p> <a class="header" href="print.html#string-table-productions" id="string-table-productions"><h2>String table productions</h2></a> <p>Some rules in the grammar — notably <a href="expressions/operator-expr.html#borrow-operators">unary operators</a>, <a href="expressions/operator-expr.html#arithmetic-and-logical-binary-operators">binary operators</a>, and <a href="keywords.html">keywords</a> — are given in a simplified form: as a listing of a table of unquoted, printable whitespace-separated strings. These cases form a subset of the rules regarding the <a href="tokens.html">token</a> rule, and are assumed to be the result of a lexical-analysis phase feeding the parser, driven by a <abbr title="Deterministic Finite Automaton">DFA</abbr>, operating over the disjunction of all such string table entries.</p> <p>When such a string enclosed in double-quotes (<code>"</code>) occurs inside the grammar, it is an implicit reference to a single member of such a string table production. See <a href="tokens.html">tokens</a> for more information.</p> <a class="header" href="print.html#lexical-structure" id="lexical-structure"><h1>Lexical structure</h1></a> <a class="header" href="print.html#input-format" id="input-format"><h1>Input format</h1></a> <p>Rust input is interpreted as a sequence of Unicode code points encoded in UTF-8. Most Rust grammar rules are defined in terms of printable ASCII-range code points, but a small number are defined in terms of Unicode properties or explicit code point lists. <sup class="footnote-reference"><a href="print.html#inputformat">1</a></sup></p> <div class="footnote-definition" id="inputformat"><sup class="footnote-definition-label">1</sup> <p>Substitute definitions for the special Unicode productions are provided to the grammar verifier, restricted to ASCII range, when verifying the grammar in this document.</p> </div> <a class="header" href="print.html#keywords" id="keywords"><h1>Keywords</h1></a> <p>Rust divides keywords into three categories:</p> <ul> <li><a href="print.html#strict-keywords">strict</a></li> <li><a href="print.html#weak-keywords">weak</a></li> <li><a href="print.html#reserved-keywords">reserved</a></li> </ul> <a class="header" href="print.html#strict-keywords" id="strict-keywords"><h2>Strict keywords</h2></a> <p>These keywords can only be used in their correct contexts. For example, it is not allowed to declare a variable with name <code>struct</code>.</p> <blockquote> <p><strong><sup>Lexer:<sup></strong><br /> KW_AS : <code>as</code><br /> KW_BOX : <code>box</code><br /> KW_BREAK : <code>break</code><br /> KW_CONST : <code>const</code><br /> KW_CONTINUE : <code>continue</code><br /> KW_CRATE : <code>crate</code><br /> KW_ELSE : <code>else</code><br /> KW_ENUM : <code>enum</code><br /> KW_EXTERN : <code>extern</code><br /> KW_FALSE : <code>false</code><br /> KW_FN : <code>fn</code><br /> KW_FOR : <code>for</code><br /> KW_IF : <code>if</code><br /> KW_IMPL : <code>impl</code><br /> KW_IN : <code>in</code><br /> KW_LET : <code>let</code><br /> KW_LOOP : <code>loop</code><br /> KW_MATCH : <code>match</code><br /> KW_MOD : <code>mod</code><br /> KW_MOVE : <code>move</code><br /> KW_MUT : <code>mut</code><br /> KW_PUB : <code>pub</code><br /> KW_REF : <code>ref</code><br /> KW_RETURN : <code>return</code><br /> KW_SELFVALUE : <code>self</code><br /> KW_SELFTYPE : <code>Self</code><br /> KW_STATIC : <code>static</code><br /> KW_STRUCT : <code>struct</code><br /> KW_SUPER : <code>super</code><br /> KW_TRAIT : <code>trait</code><br /> KW_TRUE : <code>true</code><br /> KW_TYPE : <code>type</code><br /> KW_UNSAFE : <code>unsafe</code><br /> KW_USE : <code>use</code><br /> KW_WHERE : <code>where</code><br /> KW_WHILE : <code>while</code></p> </blockquote> <a class="header" href="print.html#weak-keywords" id="weak-keywords"><h2>Weak keywords</h2></a> <p>These keywords have special meaning only in certain contexts. For example, it is possible to declare a variable or method with the name <code>union</code>.</p> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> KW_CATCH : <code>catch</code><br /> KW_DEFAULT : <code>default</code><br /> KW_UNION : <code>union</code><br /> KW_STATICLIFETIME : <code>'static</code></p> </blockquote> <a class="header" href="print.html#reserved-keywords" id="reserved-keywords"><h2>Reserved keywords</h2></a> <p>These keywords aren't used yet, but they are reserved for future use. The reasoning behind this is to make current programs forward compatible with future versions of Rust by forbidding them to use these keywords.</p> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> KW_ABSTRACT : <code>abstract</code><br /> KW_ALIGNOF : <code>alignof</code><br /> KW_BECOME : <code>become</code><br /> KW_DO : <code>do</code><br /> KW_FINAL : <code>final</code><br /> KW_MACRO : <code>macro</code><br /> KW_OFFSETOF : <code>offsetof</code><br /> KW_OVERRIDE : <code>override</code><br /> KW_PRIV : <code>priv</code><br /> KW_PROC : <code>proc</code><br /> KW_PURE : <code>pure</code><br /> KW_SIZEOF : <code>sizeof</code><br /> KW_TYPEOF : <code>typeof</code><br /> KW_UNSIZED : <code>unsized</code><br /> KW_VIRTUAL : <code>virtual</code><br /> KW_YIELD : <code>yield</code></p> </blockquote> <a class="header" href="print.html#identifiers" id="identifiers"><h1>Identifiers</h1></a> <blockquote> <p><strong><sup>Lexer:<sup></strong><br /> IDENTIFIER :<br /> [<code>a</code>-<code>z</code> <code>A</code>-<code>Z</code>] [<code>a</code>-<code>z</code> <code>A</code>-<code>Z</code> <code>0</code>-<code>9</code> <code>_</code>]<sup>*</sup><br /> | <code>_</code> [<code>a</code>-<code>z</code> <code>A</code>-<code>Z</code> <code>0</code>-<code>9</code> <code>_</code>]<sup>+</sup></p> </blockquote> <p>An identifier is any nonempty ASCII string of the following form:</p> <p>Either</p> <ul> <li>The first character is a letter</li> <li>The remaining characters are alphanumeric or <code>_</code></li> </ul> <p>Or</p> <ul> <li>The first character is <code>_</code></li> <li>The identifier is more than one character, <code>_</code> alone is not an identifier</li> <li>The remaining characters are alphanumeric or <code>_</code></li> </ul> <a class="header" href="print.html#comments" id="comments"><h1>Comments</h1></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> LINE_COMMENT :<br /> <code>//</code> (~[<code>/</code> <code>!</code>] | <code>//</code>) ~<code>\n</code><sup>*</sup><br /> | <code>//</code></p> <p>BLOCK_COMMENT :<br /> <code>/*</code> (~[<code>*</code> <code>!</code>] | <code>**</code> | <em>BlockCommentOrDoc</em>) (<em>BlockCommentOrDoc</em> | ~<code>*/</code>)<sup>*</sup> <code>*/</code><br /> | <code>/**/</code><br /> | <code>/***/</code></p> <p>INNER_LINE_DOC :<br /> <code>//!</code> ~[<code>\n</code> <em>IsolatedCR</em>]<sup>*</sup></p> <p>INNER_BLOCK_DOC :<br /> <code>/*!</code> ( <em>BlockCommentOrDoc</em> | ~[<code>*/</code> <em>IsolatedCR</em>] )<sup>*</sup> <code>*/</code></p> <p>OUTER_LINE_DOC :<br /> <code>///</code> (~<code>/</code> ~[<code>\n</code> <em>IsolatedCR</em>]<sup>*</sup>)<sup>?</sup></p> <p>OUTER_BLOCK_DOC :<br /> <code>/**</code> (~<code>*</code> | <em>BlockCommentOrDoc</em> ) (<em>BlockCommentOrDoc</em> | ~[<code>*/</code> <em>IsolatedCR</em>])<sup>*</sup> <code>*/</code></p> <p><em>BlockCommentOrDoc</em> :<br /> BLOCK_COMMENT<br /> | OUTER_BLOCK_DOC<br /> | INNER_BLOCK_DOC</p> <p><em>IsolatedCR</em> :<br /> <em>A <code>\r</code> not followed by a <code>\n</code></em></p> </blockquote> <a class="header" href="print.html#non-doc-comments" id="non-doc-comments"><h2>Non-doc comments</h2></a> <p>Comments in Rust code follow the general C++ style of line (<code>//</code>) and block (<code>/* ... */</code>) comment forms. Nested block comments are supported.</p> <p>Non-doc comments are interpreted as a form of whitespace.</p> <a class="header" href="print.html#doc-comments" id="doc-comments"><h2>Doc comments</h2></a> <p>Line doc comments beginning with exactly <em>three</em> slashes (<code>///</code>), and block doc comments (<code>/** ... */</code>), both inner doc comments, are interpreted as a special syntax for <code>doc</code> <a href="attributes.html">attributes</a>. That is, they are equivalent to writing <code>#[doc="..."]</code> around the body of the comment, i.e., <code>/// Foo</code> turns into <code>#[doc="Foo"]</code> and <code>/** Bar */</code> turns into <code>#[doc="Bar"]</code>.</p> <p>Line comments beginning with <code>//!</code> and block comments <code>/*! ... */</code> are doc comments that apply to the parent of the comment, rather than the item that follows. That is, they are equivalent to writing <code>#![doc="..."]</code> around the body of the comment. <code>//!</code> comments are usually used to document modules that occupy a source file.</p> <p>Isolated CRs (<code>\r</code>), i.e. not followed by LF (<code>\n</code>), are not allowed in doc comments.</p> <a class="header" href="print.html#examples" id="examples"><h2>Examples</h2></a> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { //! A doc comment that applies to the implicit anonymous module of this crate pub mod outer_module { //! - Inner line doc //!! - Still an inner line doc (but with a bang at the beginning) /*! - Inner block doc */ /*!! - Still an inner block doc (but with a bang at the beginning) */ // - Only a comment /// - Outer line doc (exactly 3 slashes) //// - Only a comment /* - Only a comment */ /** - Outer block doc (exactly) 2 asterisks */ /*** - Only a comment */ pub mod inner_module {} pub mod nested_comments { /* In Rust /* we can /* nest comments */ */ */ // All three types of block comments can contain or be nested inside // any other type: /* /* */ /** */ /*! */ */ /*! /* */ /** */ /*! */ */ /** /* */ /** */ /*! */ */ pub mod dummy_item {} } pub mod degenerate_cases { // empty inner line doc //! // empty inner block doc /*!*/ // empty line comment // // empty outer line doc /// // empty block comment /**/ pub mod dummy_item {} // empty 2-asterisk block isn't a doc block, it is a block comment /***/ } /* The next one isn't allowed because outer doc comments require an item that will receive the doc */ /// Where is my item? # mod boo {} } #}</code></pre></pre> <a class="header" href="print.html#whitespace" id="whitespace"><h1>Whitespace</h1></a> <p>Whitespace is any non-empty string containing only characters that have the <code>Pattern_White_Space</code> Unicode property, namely:</p> <ul> <li><code>U+0009</code> (horizontal tab, <code>'\t'</code>)</li> <li><code>U+000A</code> (line feed, <code>'\n'</code>)</li> <li><code>U+000B</code> (vertical tab)</li> <li><code>U+000C</code> (form feed)</li> <li><code>U+000D</code> (carriage return, <code>'\r'</code>)</li> <li><code>U+0020</code> (space, <code>' '</code>)</li> <li><code>U+0085</code> (next line)</li> <li><code>U+200E</code> (left-to-right mark)</li> <li><code>U+200F</code> (right-to-left mark)</li> <li><code>U+2028</code> (line separator)</li> <li><code>U+2029</code> (paragraph separator)</li> </ul> <p>Rust is a "free-form" language, meaning that all forms of whitespace serve only to separate <em>tokens</em> in the grammar, and have no semantic significance.</p> <p>A Rust program has identical meaning if each whitespace element is replaced with any other legal whitespace element, such as a single space character.</p> <a class="header" href="print.html#tokens" id="tokens"><h1>Tokens</h1></a> <p>Tokens are primitive productions in the grammar defined by regular (non-recursive) languages. "Simple" tokens are given in <a href="notation.html#string-table-productions">string table production</a> form, and occur in the rest of the grammar as double-quoted strings. Other tokens have exact rules given.</p> <a class="header" href="print.html#literals" id="literals"><h2>Literals</h2></a> <p>A literal is an expression consisting of a single token, rather than a sequence of tokens, that immediately and directly denotes the value it evaluates to, rather than referring to it by name or some other evaluation rule. A literal is a form of <a href="expressions.html#constant-expressions">constant expression</a>, so is evaluated (primarily) at compile time.</p> <a class="header" href="print.html#examples-1" id="examples-1"><h3>Examples</h3></a> <a class="header" href="print.html#characters-and-strings" id="characters-and-strings"><h4>Characters and strings</h4></a> <table><thead><tr><th> </th><th> Example </th><th> <code>#</code> sets </th><th> Characters </th><th> Escapes </th></tr></thead><tbody> <tr><td> <a href="print.html#character-literals">Character</a> </td><td> <code>'H'</code> </td><td> <code>N/A</code> </td><td> All Unicode </td><td> <a href="print.html#quote-escapes">Quote</a> & <a href="print.html#ascii-escapes">ASCII</a> & <a href="print.html#unicode-escapes">Unicode</a> </td></tr> <tr><td> <a href="print.html#string-literals">String</a> </td><td> <code>"hello"</code> </td><td> <code>N/A</code> </td><td> All Unicode </td><td> <a href="print.html#quote-escapes">Quote</a> & <a href="print.html#ascii-escapes">ASCII</a> & <a href="print.html#unicode-escapes">Unicode</a> </td></tr> <tr><td> <a href="print.html#raw-string-literals">Raw</a> </td><td> <code>r#"hello"#</code> </td><td> <code>0...</code> </td><td> All Unicode </td><td> <code>N/A</code> </td></tr> <tr><td> <a href="print.html#byte-literals">Byte</a> </td><td> <code>b'H'</code> </td><td> <code>N/A</code> </td><td> All ASCII </td><td> <a href="print.html#quote-escapes">Quote</a> & <a href="print.html#byte-escapes">Byte</a> </td></tr> <tr><td> <a href="print.html#byte-string-literals">Byte string</a> </td><td> <code>b"hello"</code> </td><td> <code>N/A</code> </td><td> All ASCII </td><td> <a href="print.html#quote-escapes">Quote</a> & <a href="print.html#byte-escapes">Byte</a> </td></tr> <tr><td> <a href="print.html#raw-byte-string-literals">Raw byte string</a> </td><td> <code>br#"hello"#</code> </td><td> <code>0...</code> </td><td> All ASCII </td><td> <code>N/A</code> </td></tr> </tbody></table> <a class="header" href="print.html#ascii-escapes" id="ascii-escapes"><h4>ASCII escapes</h4></a> <table><thead><tr><th> </th><th> Name </th></tr></thead><tbody> <tr><td> <code>\x41</code> </td><td> 7-bit character code (exactly 2 digits, up to 0x7F) </td></tr> <tr><td> <code>\n</code> </td><td> Newline </td></tr> <tr><td> <code>\r</code> </td><td> Carriage return </td></tr> <tr><td> <code>\t</code> </td><td> Tab </td></tr> <tr><td> <code>\\</code> </td><td> Backslash </td></tr> <tr><td> <code>\0</code> </td><td> Null </td></tr> </tbody></table> <a class="header" href="print.html#byte-escapes" id="byte-escapes"><h4>Byte escapes</h4></a> <table><thead><tr><th> </th><th> Name </th></tr></thead><tbody> <tr><td> <code>\x7F</code> </td><td> 8-bit character code (exactly 2 digits) </td></tr> <tr><td> <code>\n</code> </td><td> Newline </td></tr> <tr><td> <code>\r</code> </td><td> Carriage return </td></tr> <tr><td> <code>\t</code> </td><td> Tab </td></tr> <tr><td> <code>\\</code> </td><td> Backslash </td></tr> <tr><td> <code>\0</code> </td><td> Null </td></tr> </tbody></table> <a class="header" href="print.html#unicode-escapes" id="unicode-escapes"><h4>Unicode escapes</h4></a> <table><thead><tr><th> </th><th> Name </th></tr></thead><tbody> <tr><td> <code>\u{7FFF}</code> </td><td> 24-bit Unicode character code (up to 6 digits) </td></tr> </tbody></table> <a class="header" href="print.html#quote-escapes" id="quote-escapes"><h4>Quote escapes</h4></a> <table><thead><tr><th> </th><th> Name </th></tr></thead><tbody> <tr><td> <code>\'</code> </td><td> Single quote </td></tr> <tr><td> <code>\"</code> </td><td> Double quote </td></tr> </tbody></table> <a class="header" href="print.html#numbers" id="numbers"><h4>Numbers</h4></a> <table><thead><tr><th> <a href="print.html#number-literals">Number literals</a><code>*</code> </th><th> Example </th><th> Exponentiation </th><th> Suffixes </th></tr></thead><tbody> <tr><td> Decimal integer </td><td> <code>98_222</code> </td><td> <code>N/A</code> </td><td> Integer suffixes </td></tr> <tr><td> Hex integer </td><td> <code>0xff</code> </td><td> <code>N/A</code> </td><td> Integer suffixes </td></tr> <tr><td> Octal integer </td><td> <code>0o77</code> </td><td> <code>N/A</code> </td><td> Integer suffixes </td></tr> <tr><td> Binary integer </td><td> <code>0b1111_0000</code> </td><td> <code>N/A</code> </td><td> Integer suffixes </td></tr> <tr><td> Floating-point </td><td> <code>123.0E+77</code> </td><td> <code>Optional</code> </td><td> Floating-point suffixes </td></tr> </tbody></table> <p><code>*</code> All number literals allow <code>_</code> as a visual separator: <code>1_234.0E+18f64</code></p> <a class="header" href="print.html#suffixes" id="suffixes"><h4>Suffixes</h4></a> <table><thead><tr><th> Integer </th><th> Floating-point </th></tr></thead><tbody> <tr><td> <code>u8</code>, <code>i8</code>, <code>u16</code>, <code>i16</code>, <code>u32</code>, <code>i32</code>, <code>u64</code>, <code>i64</code>, <code>isize</code>, <code>usize</code> </td><td> <code>f32</code>, <code>f64</code> </td></tr> </tbody></table> <a class="header" href="print.html#character-and-string-literals" id="character-and-string-literals"><h3>Character and string literals</h3></a> <a class="header" href="print.html#character-literals" id="character-literals"><h4>Character literals</h4></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> CHAR_LITERAL :<br /> <code>'</code> ( ~[<code>'</code> <code>\</code> \n \r \t] | QUOTE_ESCAPE | ASCII_ESCAPE | UNICODE_ESCAPE ) <code>'</code></p> <p>QUOTE_ESCAPE :<br /> <code>\'</code> | <code>\"</code></p> <p>ASCII_ESCAPE :<br /> <code>\x</code> OCT_DIGIT HEX_DIGIT<br /> | <code>\n</code> | <code>\r</code> | <code>\t</code> | <code>\\</code> | <code>\0</code></p> <p>UNICODE_ESCAPE :<br /> <code>\u{</code> ( HEX_DIGIT <code>_</code><sup>*</sup> )<sup>1..6</sup> <code>}</code></p> </blockquote> <p>A <em>character literal</em> is a single Unicode character enclosed within two <code>U+0027</code> (single-quote) characters, with the exception of <code>U+0027</code> itself, which must be <em>escaped</em> by a preceding <code>U+005C</code> character (<code>\</code>).</p> <a class="header" href="print.html#string-literals" id="string-literals"><h4>String literals</h4></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> STRING_LITERAL :<br /> <code>"</code> (<br /> ~[<code>"</code> <code>\</code> <em>IsolatedCR</em>]<br /> | QUOTE_ESCAPE<br /> | ASCII_ESCAPE<br /> | UNICODE_ESCAPE<br /> | STRING_CONTINUE<br /> )<sup>*</sup> <code>"</code></p> <p>STRING_CONTINUE :<br /> <code>\</code> <em>followed by</em> \n</p> </blockquote> <p>A <em>string literal</em> is a sequence of any Unicode characters enclosed within two <code>U+0022</code> (double-quote) characters, with the exception of <code>U+0022</code> itself, which must be <em>escaped</em> by a preceding <code>U+005C</code> character (<code>\</code>).</p> <p>Line-break characters are allowed in string literals. Normally they represent themselves (i.e. no translation), but as a special exception, when an unescaped <code>U+005C</code> character (<code>\</code>) occurs immediately before the newline (<code>U+000A</code>), the <code>U+005C</code> character, the newline, and all whitespace at the beginning of the next line are ignored. Thus <code>a</code> and <code>b</code> are equal:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let a = "foobar"; let b = "foo\ bar"; assert_eq!(a,b); #}</code></pre></pre> <a class="header" href="print.html#character-escapes" id="character-escapes"><h4>Character escapes</h4></a> <p>Some additional <em>escapes</em> are available in either character or non-raw string literals. An escape starts with a <code>U+005C</code> (<code>\</code>) and continues with one of the following forms:</p> <ul> <li>An <em>8-bit code point escape</em> starts with <code>U+0078</code> (<code>x</code>) and is followed by exactly two <em>hex digits</em>. It denotes the Unicode code point equal to the provided hex value.</li> <li>A <em>24-bit code point escape</em> starts with <code>U+0075</code> (<code>u</code>) and is followed by up to six <em>hex digits</em> surrounded by braces <code>U+007B</code> (<code>{</code>) and <code>U+007D</code> (<code>}</code>). It denotes the Unicode code point equal to the provided hex value.</li> <li>A <em>whitespace escape</em> is one of the characters <code>U+006E</code> (<code>n</code>), <code>U+0072</code> (<code>r</code>), or <code>U+0074</code> (<code>t</code>), denoting the Unicode values <code>U+000A</code> (LF), <code>U+000D</code> (CR) or <code>U+0009</code> (HT) respectively.</li> <li>The <em>null escape</em> is the character <code>U+0030</code> (<code>0</code>) and denotes the Unicode value <code>U+0000</code> (NUL).</li> <li>The <em>backslash escape</em> is the character <code>U+005C</code> (<code>\</code>) which must be escaped in order to denote <em>itself</em>.</li> </ul> <a class="header" href="print.html#raw-string-literals" id="raw-string-literals"><h4>Raw string literals</h4></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> RAW_STRING_LITERAL :<br /> <code>r</code> RAW_STRING_CONTENT</p> <p>RAW_STRING_CONTENT :<br /> <code>"</code> ( ~ <em>IsolatedCR</em> )<sup>* (non-greedy)</sup> <code>"</code><br /> | <code>#</code> RAW_STRING_CONTENT <code>#</code></p> </blockquote> <p>Raw string literals do not process any escapes. They start with the character <code>U+0072</code> (<code>r</code>), followed by zero or more of the character <code>U+0023</code> (<code>#</code>) and a <code>U+0022</code> (double-quote) character. The <em>raw string body</em> can contain any sequence of Unicode characters and is terminated only by another <code>U+0022</code> (double-quote) character, followed by the same number of <code>U+0023</code> (<code>#</code>) characters that preceded the opening <code>U+0022</code> (double-quote) character.</p> <p>All Unicode characters contained in the raw string body represent themselves, the characters <code>U+0022</code> (double-quote) (except when followed by at least as many <code>U+0023</code> (<code>#</code>) characters as were used to start the raw string literal) or <code>U+005C</code> (<code>\</code>) do not have any special meaning.</p> <p>Examples for string literals:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { "foo"; r"foo"; // foo "\"foo\""; r#""foo""#; // "foo" "foo #\"# bar"; r##"foo #"# bar"##; // foo #"# bar "\x52"; "R"; r"R"; // R "\\x52"; r"\x52"; // \x52 #}</code></pre></pre> <a class="header" href="print.html#byte-and-byte-string-literals" id="byte-and-byte-string-literals"><h3>Byte and byte string literals</h3></a> <a class="header" href="print.html#byte-literals" id="byte-literals"><h4>Byte literals</h4></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> BYTE_LITERAL :<br /> <code>b'</code> ( ASCII_FOR_CHAR | BYTE_ESCAPE ) <code>'</code></p> <p>ASCII_FOR_CHAR :<br /> <em>any ASCII (i.e. 0x00 to 0x7F), except</em> <code>'</code>, <code>/</code>, \n, \r or \t</p> <p>BYTE_ESCAPE :<br /> <code>\x</code> HEX_DIGIT HEX_DIGIT<br /> | <code>\n</code> | <code>\r</code> | <code>\t</code> | <code>\\</code> | <code>\0</code></p> </blockquote> <p>A <em>byte literal</em> is a single ASCII character (in the <code>U+0000</code> to <code>U+007F</code> range) or a single <em>escape</em> preceded by the characters <code>U+0062</code> (<code>b</code>) and <code>U+0027</code> (single-quote), and followed by the character <code>U+0027</code>. If the character <code>U+0027</code> is present within the literal, it must be <em>escaped</em> by a preceding <code>U+005C</code> (<code>\</code>) character. It is equivalent to a <code>u8</code> unsigned 8-bit integer <em>number literal</em>.</p> <a class="header" href="print.html#byte-string-literals" id="byte-string-literals"><h4>Byte string literals</h4></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> BYTE_STRING_LITERAL :<br /> <code>b"</code> ( ASCII_FOR_STRING | BYTE_ESCAPE | STRING_CONTINUE )<sup>*</sup> <code>"</code></p> <p>ASCII_FOR_STRING :<br /> <em>any ASCII (i.e 0x00 to 0x7F), except</em> <code>"</code>, <code>/</code> <em>and IsolatedCR</em></p> </blockquote> <p>A non-raw <em>byte string literal</em> is a sequence of ASCII characters and <em>escapes</em>, preceded by the characters <code>U+0062</code> (<code>b</code>) and <code>U+0022</code> (double-quote), and followed by the character <code>U+0022</code>. If the character <code>U+0022</code> is present within the literal, it must be <em>escaped</em> by a preceding <code>U+005C</code> (<code>\</code>) character. Alternatively, a byte string literal can be a <em>raw byte string literal</em>, defined below. A byte string literal of length <code>n</code> is equivalent to a <code>&'static [u8; n]</code> borrowed fixed-sized array of unsigned 8-bit integers.</p> <p>Some additional <em>escapes</em> are available in either byte or non-raw byte string literals. An escape starts with a <code>U+005C</code> (<code>\</code>) and continues with one of the following forms:</p> <ul> <li>A <em>byte escape</em> escape starts with <code>U+0078</code> (<code>x</code>) and is followed by exactly two <em>hex digits</em>. It denotes the byte equal to the provided hex value.</li> <li>A <em>whitespace escape</em> is one of the characters <code>U+006E</code> (<code>n</code>), <code>U+0072</code> (<code>r</code>), or <code>U+0074</code> (<code>t</code>), denoting the bytes values <code>0x0A</code> (ASCII LF), <code>0x0D</code> (ASCII CR) or <code>0x09</code> (ASCII HT) respectively.</li> <li>The <em>null escape</em> is the character <code>U+0030</code> (<code>0</code>) and denotes the byte value <code>0x00</code> (ASCII NUL).</li> <li>The <em>backslash escape</em> is the character <code>U+005C</code> (<code>\</code>) which must be escaped in order to denote its ASCII encoding <code>0x5C</code>.</li> </ul> <a class="header" href="print.html#raw-byte-string-literals" id="raw-byte-string-literals"><h4>Raw byte string literals</h4></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> RAW_BYTE_STRING_LITERAL :<br /> <code>br</code> RAW_BYTE_STRING_CONTENT</p> <p>RAW_BYTE_STRING_CONTENT :<br /> <code>"</code> ASCII<sup>* (non-greedy)</sup> <code>"</code><br /> | <code>#</code> RAW_STRING_CONTENT <code>#</code></p> <p>ASCII :<br /> <em>any ASCII (i.e. 0x00 to 0x7F)</em></p> </blockquote> <p>Raw byte string literals do not process any escapes. They start with the character <code>U+0062</code> (<code>b</code>), followed by <code>U+0072</code> (<code>r</code>), followed by zero or more of the character <code>U+0023</code> (<code>#</code>), and a <code>U+0022</code> (double-quote) character. The <em>raw string body</em> can contain any sequence of ASCII characters and is terminated only by another <code>U+0022</code> (double-quote) character, followed by the same number of <code>U+0023</code> (<code>#</code>) characters that preceded the opening <code>U+0022</code> (double-quote) character. A raw byte string literal can not contain any non-ASCII byte.</p> <p>All characters contained in the raw string body represent their ASCII encoding, the characters <code>U+0022</code> (double-quote) (except when followed by at least as many <code>U+0023</code> (<code>#</code>) characters as were used to start the raw string literal) or <code>U+005C</code> (<code>\</code>) do not have any special meaning.</p> <p>Examples for byte string literals:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { b"foo"; br"foo"; // foo b"\"foo\""; br#""foo""#; // "foo" b"foo #\"# bar"; br##"foo #"# bar"##; // foo #"# bar b"\x52"; b"R"; br"R"; // R b"\\x52"; br"\x52"; // \x52 #}</code></pre></pre> <a class="header" href="print.html#number-literals" id="number-literals"><h3>Number literals</h3></a> <p>A <em>number literal</em> is either an <em>integer literal</em> or a <em>floating-point literal</em>. The grammar for recognizing the two kinds of literals is mixed.</p> <a class="header" href="print.html#integer-literals" id="integer-literals"><h4>Integer literals</h4></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> INTEGER_LITERAL :<br /> ( DEC_LITERAL | BIN_LITERAL | OCT_LITERAL | HEX_LITERAL ) INTEGER_SUFFIX<sup>?</sup></p> <p>DEC_LITERAL :<br /> DEC_DIGIT (DEC_DIGIT|<code>_</code>)<sup>*</sup></p> <p>BIN_LITERAL :<br /> <code>0b</code> (BIN_DIGIT|<code>_</code>)<sup>*</sup> BIN_DIGIT (BIN_DIGIT|<code>_</code>)<sup>*</sup></p> <p>OCT_LITERAL :<br /> <code>0o</code> (OCT_DIGIT|<code>_</code>)<sup>*</sup> OCT_DIGIT (OCT_DIGIT|<code>_</code>)<sup>*</sup></p> <p>HEX_LITERAL :<br /> <code>0x</code> (HEX_DIGIT|<code>_</code>)<sup>*</sup> HEX_DIGIT (HEX_DIGIT|<code>_</code>)<sup>*</sup></p> <p>BIN_DIGIT : [<code>0</code>-<code>1</code>]</p> <p>OCT_DIGIT : [<code>0</code>-<code>7</code>]</p> <p>DEC_DIGIT : [<code>0</code>-<code>9</code>]</p> <p>HEX_DIGIT : [<code>0</code>-<code>9</code> <code>a</code>-<code>f</code> <code>A</code>-<code>F</code>]</p> <p>INTEGER_SUFFIX :<br /> <code>u8</code> | <code>u16</code> | <code>u32</code> | <code>u64</code> | <code>usize</code><br /> | <code>i8</code> | <code>i16</code> | <code>i32</code> | <code>i64</code> | <code>isize</code></p> </blockquote> <!-- FIXME: separate the DECIMAL_LITERAL with no prefix or suffix (used on tuple indexing and float_literal --> <!-- FIXME: u128 and i128 --> <p>An <em>integer literal</em> has one of four forms:</p> <ul> <li>A <em>decimal literal</em> starts with a <em>decimal digit</em> and continues with any mixture of <em>decimal digits</em> and <em>underscores</em>.</li> <li>A <em>hex literal</em> starts with the character sequence <code>U+0030</code> <code>U+0078</code> (<code>0x</code>) and continues as any mixture (with at least one digit) of hex digits and underscores.</li> <li>An <em>octal literal</em> starts with the character sequence <code>U+0030</code> <code>U+006F</code> (<code>0o</code>) and continues as any mixture (with at least one digit) of octal digits and underscores.</li> <li>A <em>binary literal</em> starts with the character sequence <code>U+0030</code> <code>U+0062</code> (<code>0b</code>) and continues as any mixture (with at least one digit) of binary digits and underscores.</li> </ul> <p>Like any literal, an integer literal may be followed (immediately, without any spaces) by an <em>integer suffix</em>, which forcibly sets the type of the literal. The integer suffix must be the name of one of the integral types: <code>u8</code>, <code>i8</code>, <code>u16</code>, <code>i16</code>, <code>u32</code>, <code>i32</code>, <code>u64</code>, <code>i64</code>, <code>isize</code>, or <code>usize</code>.</p> <p>The type of an <em>unsuffixed</em> integer literal is determined by type inference:</p> <ul> <li> <p>If an integer type can be <em>uniquely</em> determined from the surrounding program context, the unsuffixed integer literal has that type.</p> </li> <li> <p>If the program context under-constrains the type, it defaults to the signed 32-bit integer <code>i32</code>.</p> </li> <li> <p>If the program context over-constrains the type, it is considered a static type error.</p> </li> </ul> <p>Examples of integer literals of various forms:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { 123; // type i32 123i32; // type i32 123u32; // type u32 123_u32; // type u32 let a: u64 = 123; // type u64 0xff; // type i32 0xff_u8; // type u8 0o70; // type i32 0o70_i16; // type i16 0b1111_1111_1001_0000; // type i32 0b1111_1111_1001_0000i64; // type i64 0b________1; // type i32 0usize; // type usize #}</code></pre></pre> <p>Examples of invalid integer literals:</p> <pre><code class="language-rust ignore">// invalid suffixes 0invalidSuffix; // uses numbers of the wrong base 123AFB43; 0b0102; 0o0581; // integers too big for their type (they overflow) 128_i8; 256_u8; // bin, hex and octal literals must have at least one digit 0b_; 0b____; </code></pre> <p>Note that the Rust syntax considers <code>-1i8</code> as an application of the <a href="expressions/operator-expr.html#negation-operators">unary minus operator</a> to an integer literal <code>1i8</code>, rather than a single integer literal.</p> <a class="header" href="print.html#floating-point-literals" id="floating-point-literals"><h4>Floating-point literals</h4></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> FLOAT_LITERAL :<br /> DEC_LITERAL <code>.</code> <em>(not immediately followed by <code>.</code>, <code>_</code> or an identifier</em>)<br /> | DEC_LITERAL FLOAT_EXPONENT<br /> | DEC_LITERAL <code>.</code> DEC_LITERAL FLOAT_EXPONENT<sup>?</sup><br /> | DEC_LITERAL (<code>.</code> DEC_LITERAL)<sup>?</sup> FLOAT_EXPONENT<sup>?</sup> FLOAT_SUFFIX</p> <p>FLOAT_EXPONENT :<br /> (<code>e</code>|<code>E</code>) (<code>+</code>|<code>-</code>)? (DEC_DIGIT|<code>_</code>)<sup>*</sup> DEC_DIGIT (DEC_DIGIT|<code>_</code>)<sup>*</sup></p> <p>FLOAT_SUFFIX :<br /> <code>f32</code> | <code>f64</code></p> </blockquote> <p>A <em>floating-point literal</em> has one of two forms:</p> <ul> <li>A <em>decimal literal</em> followed by a period character <code>U+002E</code> (<code>.</code>). This is optionally followed by another decimal literal, with an optional <em>exponent</em>.</li> <li>A single <em>decimal literal</em> followed by an <em>exponent</em>.</li> </ul> <p>Like integer literals, a floating-point literal may be followed by a suffix, so long as the pre-suffix part does not end with <code>U+002E</code> (<code>.</code>). The suffix forcibly sets the type of the literal. There are two valid <em>floating-point suffixes</em>, <code>f32</code> and <code>f64</code> (the 32-bit and 64-bit floating point types), which explicitly determine the type of the literal.</p> <p>The type of an <em>unsuffixed</em> floating-point literal is determined by type inference:</p> <ul> <li> <p>If a floating-point type can be <em>uniquely</em> determined from the surrounding program context, the unsuffixed floating-point literal has that type.</p> </li> <li> <p>If the program context under-constrains the type, it defaults to <code>f64</code>.</p> </li> <li> <p>If the program context over-constrains the type, it is considered a static type error.</p> </li> </ul> <p>Examples of floating-point literals of various forms:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { 123.0f64; // type f64 0.1f64; // type f64 0.1f32; // type f32 12E+99_f64; // type f64 let x: f64 = 2.; // type f64 #}</code></pre></pre> <p>This last example is different because it is not possible to use the suffix syntax with a floating point literal ending in a period. <code>2.f64</code> would attempt to call a method named <code>f64</code> on <code>2</code>.</p> <p>The representation semantics of floating-point numbers are described in <a href="types.html#machine-types">"Machine Types"</a>.</p> <a class="header" href="print.html#boolean-literals" id="boolean-literals"><h3>Boolean literals</h3></a> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> BOOLEAN_LITERAL :<br /> <code>true</code><br /> | <code>false</code></p> </blockquote> <p>The two values of the boolean type are written <code>true</code> and <code>false</code>.</p> <a class="header" href="print.html#symbols" id="symbols"><h2>Symbols</h2></a> <p>Symbols are a general class of printable <a href="print.html#tokens">tokens</a> that play structural roles in a variety of grammar productions. They are a set of remaining miscellaneous printable tokens that do not otherwise appear as <a href="expressions/operator-expr.html#borrow-operators">unary operators</a>, <a href="expressions/operator-expr.html#arithmetic-and-logical-binary-operators">binary operators</a>, or <a href="keywords.html">keywords</a>. They are catalogued in <a href="../grammar.html#symbols">the Symbols section</a> of the Grammar document.</p> <a class="header" href="print.html#paths" id="paths"><h1>Paths</h1></a> <p>A <em>path</em> is a sequence of one or more path components <em>logically</em> separated by a namespace qualifier (<code>::</code>). If a path consists of only one component, it refers to either an <a href="items.html">item</a> or a <a href="variables.html">variable</a> in a local control scope. If a path has multiple components, it always refers to an item.</p> <p>Two examples of simple paths consisting of only identifier components:</p> <pre><code class="language-rust ignore">x; x::y::z; </code></pre> <p>Path components are usually <a href="identifiers.html">identifiers</a>, but they may also include angle-bracket-enclosed lists of type arguments. In <a href="expressions.html">expression</a> context, the type argument list is given after a <code>::</code> namespace qualifier in order to disambiguate it from a relational expression involving the less-than symbol (<code><</code>). In type expression context, the final namespace qualifier is omitted.</p> <p>Two examples of paths with type arguments:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # struct HashMap<K, V>(K,V); # fn f() { # fn id<T>(t: T) -> T { t } type T = HashMap<i32,String>; // Type arguments used in a type expression let x = id::<i32>(10); // Type arguments used in a call expression # } #}</code></pre></pre> <p>Paths can be denoted with various leading qualifiers to change the meaning of how it is resolved:</p> <ul> <li>Paths starting with <code>::</code> are considered to be global paths where the components of the path start being resolved from the crate root. Each identifier in the path must resolve to an item.</li> </ul> <pre><pre class="playpen"><code class="language-rust">mod a { pub fn foo() {} } mod b { pub fn foo() { ::a::foo(); // call a's foo function } } # fn main() {} </code></pre></pre> <ul> <li>Paths starting with the keyword <code>super</code> begin resolution relative to the parent module. Each further identifier must resolve to an item.</li> </ul> <pre><pre class="playpen"><code class="language-rust">mod a { pub fn foo() {} } mod b { pub fn foo() { super::a::foo(); // call a's foo function } } # fn main() {} </code></pre></pre> <ul> <li>Paths starting with the keyword <code>self</code> begin resolution relative to the current module. Each further identifier must resolve to an item.</li> </ul> <pre><pre class="playpen"><code class="language-rust">fn foo() {} fn bar() { self::foo(); } # fn main() {} </code></pre></pre> <p>Additionally keyword <code>super</code> may be repeated several times after the first <code>super</code> or <code>self</code> to refer to ancestor modules.</p> <pre><pre class="playpen"><code class="language-rust">mod a { fn foo() {} mod b { mod c { fn foo() { super::super::foo(); // call a's foo function self::super::super::foo(); // call a's foo function } } } } # fn main() {} </code></pre></pre> <a class="header" href="print.html#canonical-paths" id="canonical-paths"><h2>Canonical paths</h2></a> <p>Items defined in a module or implementation have a <em>canonical path</em> that corresponds to where within its crate it is defined. All other paths to these items are aliases. The canonical path is defined as a <em>path prefix</em> appended by the path component the item itself defines.</p> <p><a href="items/implementations.html">Implementations</a> and <a href="items/use-declarations.html">use declarations</a> do not have canonical paths, although the items that implementations define do have them. Items defined in block expressions do not have canonical paths. Items defined in a module that does not have a canonical path do not have a canonical path. Associated items defined in an implementation that refers to an item without a canonical path, e.g. as the implementing type, the trait being implemented, a type parameter or bound on a type parameter, do not have canonical paths.</p> <p>The path prefix for modules is the canonical path to that module. For bare implementations, it is the canonical path of the item being implemented surrounded by angle (<code><></code>) brackets. For trait implementations, it is the canonical path of the item being implemented followed by <code>as</code> followed by the canonical path to the trait all surrounded in angle (<code><></code>) brackets.</p> <p>The canonical path is only meaningful within a given crate. There is no global namespace across crates; an item's canonical path merely identifies it within the crate.</p> <pre><pre class="playpen"><code class="language-rust">// Comments show the canonical path of the item. mod a { // ::a pub struct Struct; // ::a::Struct pub trait Trait { // ::a::Trait fn f(&self); // a::Trait::f } impl Trait for Struct { fn f(&self) {} // <::a::Struct as ::a::Trait>::f } impl Struct { fn g(&self) {} // <::a::Struct>::g } } mod without { // ::without fn canonicals() { // ::without::canonicals struct OtherStruct; // None trait OtherTrait { // None fn g(&self); // None } impl OtherTrait for OtherStruct { fn g(&self) {} // None } impl OtherTrait for ::a::Struct { fn g(&self) {} // None } impl ::a::Trait for OtherStruct { fn f(&self) {} // None } } } # fn main() {} </code></pre></pre> <a class="header" href="print.html#macros" id="macros"><h1>Macros</h1></a> <p>A number of minor features of Rust are not central enough to have their own syntax, and yet are not implementable as functions. Instead, they are given names, and invoked through a consistent syntax: <code>some_extension!(...)</code>.</p> <p>Thre are two ways to define new macros:</p> <ul> <li><a href="macros-by-example.html">Macros by Example</a> define new syntax in a higher-level, declarative way.</li> <li><a href="procedural-macros.html">Procedural Macros</a> can be used to implement custom derive.</li> </ul> <a class="header" href="print.html#macros-by-example" id="macros-by-example"><h1>Macros By Example</h1></a> <p><code>macro_rules</code> allows users to define syntax extension in a declarative way. We call such extensions "macros by example" or simply "macros".</p> <p>Currently, macros can expand to expressions, statements, items, or patterns.</p> <p>(A <code>sep_token</code> is any token other than <code>*</code> and <code>+</code>. A <code>non_special_token</code> is any token other than a delimiter or <code>$</code>.)</p> <p>The macro expander looks up macro invocations by name, and tries each macro rule in turn. It transcribes the first successful match. Matching and transcription are closely related to each other, and we will describe them together.</p> <p>The macro expander matches and transcribes every token that does not begin with a <code>$</code> literally, including delimiters. For parsing reasons, delimiters must be balanced, but they are otherwise not special.</p> <p>In the matcher, <code>$</code> <em>name</em> <code>:</code> <em>designator</em> matches the nonterminal in the Rust syntax named by <em>designator</em>. Valid designators are:</p> <ul> <li><code>item</code>: an <a href="items.html">item</a></li> <li><code>block</code>: a <a href="expressions/block-expr.html">block</a></li> <li><code>stmt</code>: a <a href="statements.html">statement</a></li> <li><code>pat</code>: a <a href="expressions/match-expr.html">pattern</a></li> <li><code>expr</code>: an <a href="expressions.html">expression</a></li> <li><code>ty</code>: a <a href="types.html">type</a></li> <li><code>ident</code>: an <a href="identifiers.html">identifier</a></li> <li><code>path</code>: a <a href="paths.html">path</a></li> <li><code>tt</code>: a token tree (a single <a href="tokens.html">token</a> by matching <code>()</code>, <code>[]</code>, or <code>{}</code>)</li> <li><code>meta</code>: the contents of an <a href="attributes.html">attribute</a></li> </ul> <p>In the transcriber, the designator is already known, and so only the name of a matched nonterminal comes after the dollar sign.</p> <p>In both the matcher and transcriber, the Kleene star-like operator indicates repetition. The Kleene star operator consists of <code>$</code> and parentheses, optionally followed by a separator token, followed by <code>*</code> or <code>+</code>. <code>*</code> means zero or more repetitions, <code>+</code> means at least one repetition. The parentheses are not matched or transcribed. On the matcher side, a name is bound to <em>all</em> of the names it matches, in a structure that mimics the structure of the repetition encountered on a successful match. The job of the transcriber is to sort that structure out.</p> <p>The rules for transcription of these repetitions are called "Macro By Example". Essentially, one "layer" of repetition is discharged at a time, and all of them must be discharged by the time a name is transcribed. Therefore, <code>( $( $i:ident ),* ) => ( $i )</code> is an invalid macro, but <code>( $( $i:ident ),* ) => ( $( $i:ident ),* )</code> is acceptable (if trivial).</p> <p>When Macro By Example encounters a repetition, it examines all of the <code>$</code> <em>name</em> s that occur in its body. At the "current layer", they all must repeat the same number of times, so <code>( $( $i:ident ),* ; $( $j:ident ),* ) => ( $( ($i,$j) ),* )</code> is valid if given the argument <code>(a,b,c ; d,e,f)</code>, but not <code>(a,b,c ; d,e)</code>. The repetition walks through the choices at that layer in lockstep, so the former input transcribes to <code>(a,d), (b,e), (c,f)</code>.</p> <p>Nested repetitions are allowed.</p> <a class="header" href="print.html#parsing-limitations" id="parsing-limitations"><h3>Parsing limitations</h3></a> <p>The parser used by the macro system is reasonably powerful, but the parsing of Rust syntax is restricted in two ways:</p> <ol> <li>Macro definitions are required to include suitable separators after parsing expressions and other bits of the Rust grammar. This implies that a macro definition like <code>$i:expr [ , ]</code> is not legal, because <code>[</code> could be part of an expression. A macro definition like <code>$i:expr,</code> or <code>$i:expr;</code> would be legal, however, because <code>,</code> and <code>;</code> are legal separators. See <a href="https://github.com/rust-lang/rfcs/blob/master/text/0550-macro-future-proofing.md">RFC 550</a> for more information.</li> <li>The parser must have eliminated all ambiguity by the time it reaches a <code>$</code> <em>name</em> <code>:</code> <em>designator</em>. This requirement most often affects name-designator pairs when they occur at the beginning of, or immediately after, a <code>$(...)*</code>; requiring a distinctive token in front can solve the problem.</li> </ol> <a class="header" href="print.html#procedural-macros" id="procedural-macros"><h2>Procedural Macros</h2></a> <p><em>Procedural macros</em> allow creating syntax extensions as execution of a function. Procedural macros can be used to implement custom <a href="attributes.html#derive">derive</a> on your own types. See <a href="../book/first-edition/procedural-macros.html">the book</a> for a tutorial.</p> <p>Procedural macros involve a few different parts of the language and its standard libraries. First is the <code>proc_macro</code> crate, included with Rust, that defines an interface for building a procedural macro. The <code>#[proc_macro_derive(Foo)]</code> attribute is used to mark the deriving function. This function must have the type signature:</p> <pre><code class="language-rust ignore">use proc_macro::TokenStream; #[proc_macro_derive(Hello)] pub fn hello_world(input: TokenStream) -> TokenStream </code></pre> <p>Finally, procedural macros must be in their own crate, with the <code>proc-macro</code> crate type.</p> <a class="header" href="print.html#crates-and-source-files" id="crates-and-source-files"><h1>Crates and source files</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>Crate</em> :<br /> UTF8BOM<sup>?</sup><br /> SHEBANG<sup>?</sup><br /> <a href="attributes.html"><em>InnerAttribute</em></a><sup>*</sup><br /> <a href="items.html"><em>Item</em></a><sup>*</sup></p> </blockquote> <blockquote> <p><strong><sup>Lexer</sup></strong><br /> UTF8BOM : <code>\uFEFF</code><br /> SHEBANG : <code>#!</code> ~[<code>[</code> <code>\n</code>] ~<code>\n</code><sup>*</sup></p> </blockquote> <p>Although Rust, like any other language, can be implemented by an interpreter as well as a compiler, the only existing implementation is a compiler, and the language has always been designed to be compiled. For these reasons, this section assumes a compiler.</p> <p>Rust's semantics obey a <em>phase distinction</em> between compile-time and run-time.<sup class="footnote-reference"><a href="print.html#phase-distinction">1</a></sup> Semantic rules that have a <em>static interpretation</em> govern the success or failure of compilation, while semantic rules that have a <em>dynamic interpretation</em> govern the behavior of the program at run-time.</p> <p>The compilation model centers on artifacts called <em>crates</em>. Each compilation processes a single crate in source form, and if successful, produces a single crate in binary form: either an executable or some sort of library.<sup class="footnote-reference"><a href="print.html#cratesourcefile">2</a></sup></p> <p>A <em>crate</em> is a unit of compilation and linking, as well as versioning, distribution and runtime loading. A crate contains a <em>tree</em> of nested <a href="items/modules.html">module</a> scopes. 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 <a href="paths.html">module path</a> denoting its location within the crate's module tree.</p> <p>The Rust compiler is always invoked with a single source file as input, and always produces a single output crate. The processing of that source file may result in other source files being loaded as modules. Source files have the extension <code>.rs</code>.</p> <p>A Rust source file describes a module, the name and location of which — in the module tree of the current crate — are defined from outside the source file: either by an explicit <code>mod_item</code> in a referencing source file, or by the name of the crate itself. Every source file is a module, but not every module needs its own source file: <a href="items/modules.html">module definitions</a> can be nested within one file.</p> <p>Each source file contains a sequence of zero or more <code>item</code> definitions, and may optionally begin with any number of <a href="items-and-attributes.html">attributes</a> that apply to the containing module, most of which influence the behavior of the compiler. The anonymous crate module can have additional attributes that apply to the crate as a whole.</p> <pre><pre class="playpen"><code class="language-rust no_run"> # #![allow(unused_variables)] #fn main() { // Specify the crate name. #![crate_name = "projx"] // Specify the type of output artifact. #![crate_type = "lib"] // Turn on a warning. // This can be done in any module, not just the anonymous crate module. #![warn(non_camel_case_types)] #}</code></pre></pre> <p>A crate that contains a <code>main</code> function can be compiled to an executable. If a <code>main</code> function is present, its return type must be <code>()</code> ("<a href="types.html#tuple-types">unit</a>") and it must take no arguments.</p> <p>The optional <a href="https://en.wikipedia.org/wiki/Byte_order_mark#UTF-8"><em>UTF8 byte order mark</em></a> (UTF8BOM production) indicates that the file is encoded in UTF8. It can only occur at the beginning of the file and is ignored by the compiler.</p> <p>A source file can have a <a href="https://en.wikipedia.org/wiki/Shebang_(Unix)"><em>shebang</em></a> (SHEBANG production), which indicates to the operating system what program to use to execute this file. It serves essentially to treat the source file as an executable script. The shebang can only occur at the beginning of the file (but after the optional <em>UTF8BOM</em>). It is ignored by the compiler. For example:</p> <pre><code class="language-text ignore">#!/usr/bin/env rustx fn main() { println!("Hello!"); } </code></pre> <div class="footnote-definition" id="phase-distinction"><sup class="footnote-definition-label">1</sup> <p>This distinction would also exist in an interpreter. Static checks like syntactic analysis, type checking, and lints should happen before the program is executed regardless of when it is executed.</p> </div> <div class="footnote-definition" id="cratesourcefile"><sup class="footnote-definition-label">2</sup> <p>A crate is somewhat analogous to an <em>assembly</em> in the ECMA-335 CLI model, a <em>library</em> in the SML/NJ Compilation Manager, a <em>unit</em> in the Owens and Flatt module system, or a <em>configuration</em> in Mesa.</p> </div> <a class="header" href="print.html#items-and-attributes" id="items-and-attributes"><h1>Items and attributes</h1></a> <p>Crates contain <a href="items.html">items</a>, each of which may have some number of <a href="attributes.html">attributes</a> attached to it.</p> <a class="header" href="print.html#items" id="items"><h1>Items</h1></a> <p>An <em>item</em> is a component of a crate. Items are organized within a crate by a nested set of <a href="items/modules.html">modules</a>. Every crate has a single "outermost" anonymous module; all further items within the crate have <a href="paths.html">paths</a> within the module tree of the crate.</p> <p>Items are entirely determined at compile-time, generally remain fixed during execution, and may reside in read-only memory.</p> <p>There are several kinds of items:</p> <ul> <li><a href="items/modules.html">modules</a></li> <li><a href="items/extern-crates.html"><code>extern crate</code> declarations</a></li> <li><a href="items/use-declarations.html"><code>use</code> declarations</a></li> <li><a href="items/functions.html">function definitions</a></li> <li><a href="items/type-aliases.html">type definitions</a></li> <li><a href="items/structs.html">struct definitions</a></li> <li><a href="items/enumerations.html">enumeration definitions</a></li> <li><a href="items/unions.html">union definitions</a></li> <li><a href="items/constant-items.html">constant items</a></li> <li><a href="items/static-items.html">static items</a></li> <li><a href="items/traits.html">trait definitions</a></li> <li><a href="items/implementations.html">implementations</a></li> <li><a href="items/external-blocks.html"><code>extern</code> blocks</a></li> </ul> <p>Some items form an implicit scope for the declaration of sub-items. In other words, within a function or module, declarations of items can (in many cases) be mixed with the statements, control blocks, and similar artifacts that otherwise compose the item body. The meaning of these scoped items is the same as if the item was declared outside the scope — it is still a static item — except that the item's <em>path name</em> within the module namespace is qualified by the name of the enclosing item, or is private to the enclosing item (in the case of functions). The grammar specifies the exact locations in which sub-item declarations may appear.</p> <a class="header" href="print.html#type-parameters" id="type-parameters"><h2>Type Parameters</h2></a> <p>Functions, type aliases, structs, enumerations, unions, traits and implementations may be <em>parameterized</em> by type. Type parameters are given as a comma-separated list of identifiers enclosed in angle brackets (<code><...></code>), after the name of the item (except for implementations, where they come directly after <code>impl</code>) and before its definition.</p> <p>The type parameters of an item are considered "part of the name", not part of the type of the item. A referencing <a href="paths.html">path</a> 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. In practice, the type-inference system can usually infer such argument types from context. There are no general type-parametric types, only type-parametric items. That is, Rust has no notion of type abstraction: there are no higher-ranked (or "forall") types abstracted over other types, though higher-ranked types do exist for lifetimes.</p> <a class="header" href="print.html#modules" id="modules"><h1>Modules</h1></a> <blockquote> <p><strong><sup>Syntax:<sup></strong><br /> <em>Module</em> :<br /> <code>mod</code> <a href="identifiers.html">IDENTIFIER</a> <code>;</code><br /> | <code>mod</code> <a href="identifiers.html">IDENTIFIER</a> <code>{</code><br /> <a href="attributes.html"><em>InnerAttribute</em></a><sup>*</sup><br /> <a href="items.html"><em>Item</em></a><sup>*</sup><br /> <code>}</code></p> </blockquote> <p>A module is a container for zero or more <a href="items.html">items</a>.</p> <p>A <em>module item</em> is a module, surrounded in braces, named, and prefixed with the keyword <code>mod</code>. A module item introduces a new, named module into the tree of modules making up a crate. Modules can nest arbitrarily.</p> <p>An example of a module:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { mod math { type Complex = (f64, f64); fn sin(f: f64) -> f64 { /* ... */ # panic!(); } fn cos(f: f64) -> f64 { /* ... */ # panic!(); } fn tan(f: f64) -> f64 { /* ... */ # panic!(); } } #}</code></pre></pre> <p>Modules and types share the same namespace. Declaring a named type with the same name as a module in scope is forbidden: that is, a type definition, trait, struct, enumeration, union, type parameter or crate can't shadow the name of a module in scope, or vice versa. Items brought into scope with <code>use</code> also have this restriction.</p> <p>A module without a body is loaded from an external file, by default with the same name as the module, plus the <code>.rs</code> extension. When a nested submodule is loaded from an external file, it is loaded from a subdirectory path that mirrors the module hierarchy.</p> <pre><code class="language-rust ignore">// Load the `vec` module from `vec.rs` mod vec; mod thread { // Load the `local_data` module from `thread/local_data.rs` // or `thread/local_data/mod.rs`. mod local_data; } </code></pre> <p>The directories and files used for loading external file modules can be influenced with the <code>path</code> attribute.</p> <pre><code class="language-rust ignore">#[path = "thread_files"] mod thread { // Load the `local_data` module from `thread_files/tls.rs` #[path = "tls.rs"] mod local_data; } </code></pre> <a class="header" href="print.html#extern-crate-declarations" id="extern-crate-declarations"><h1>Extern crate declarations</h1></a> <blockquote> <p><strong><sup>Syntax:<sup></strong><br /> <em>ExternCrate</em> :<br /> <code>extern</code> <code>crate</code> <a href="identifiers.html">IDENTIFIER</a> (<code>as</code> <a href="identifiers.html">IDENTIFIER</a>)<sup>?</sup> <code>;</code></p> </blockquote> <p>An <em><code>extern crate</code> declaration</em> specifies a dependency on an external crate. The external crate is then bound into the declaring scope as the <code>ident</code> provided in the <code>extern_crate_decl</code>.</p> <p>The external crate is resolved to a specific <code>soname</code> at compile time, and a runtime linkage requirement to that <code>soname</code> is passed to the linker for loading at runtime. The <code>soname</code> is resolved at compile time by scanning the compiler's library path and matching the optional <code>crateid</code> provided against the <code>crateid</code> attributes that were declared on the external crate when it was compiled. If no <code>crateid</code> is provided, a default <code>name</code> attribute is assumed, equal to the <code>ident</code> given in the <code>extern_crate_decl</code>.</p> <p>Three examples of <code>extern crate</code> declarations:</p> <pre><code class="language-rust ignore">extern crate pcre; extern crate std; // equivalent to: extern crate std as std; extern crate std as ruststd; // linking to 'std' under another name </code></pre> <p>When naming Rust crates, hyphens are disallowed. However, Cargo packages may make use of them. In such case, when <code>Cargo.toml</code> doesn't specify a crate name, Cargo will transparently replace <code>-</code> with <code>_</code> (Refer to <a href="https://github.com/rust-lang/rfcs/blob/master/text/0940-hyphens-considered-harmful.md">RFC 940</a> for more details).</p> <p>Here is an example:</p> <pre><code class="language-rust ignore">// Importing the Cargo package hello-world extern crate hello_world; // hyphen replaced with an underscore </code></pre> <a class="header" href="print.html#use-declarations" id="use-declarations"><h1>Use declarations</h1></a> <blockquote> <p><strong><sup>Syntax:</sup></strong><br /> <em>UseDeclaration</em> :<br /> (<a href="visibility-and-privacy.html"><em>Visibility</em></a>)<sup>?</sup> <code>use</code> <em>UseTree</em> <code>;</code></p> <p><em>UseTree</em> :<br /> (<a href="paths.html"><em>SimplePath</em></a><sup>?</sup> <code>::</code>)<sup>?</sup> <code>*</code><br /> | (<a href="paths.html"><em>SimplePath</em></a><sup>?</sup> <code>::</code>)<sup>?</sup> <code>{</code> (<em>UseTree</em> ( <code>,</code> <em>UseTree</em> )<sup>*</sup> <code>,</code><sup>?</sup>)<sup>?</sup> <code>}</code><br /> | <a href="paths.html"><em>SimplePath</em></a> <code>as</code> <a href="identifiers.html">IDENTIFIER</a></p> </blockquote> <p>A <em>use declaration</em> creates one or more local name bindings synonymous with some other <a href="paths.html">path</a>. Usually a <code>use</code> declaration is used to shorten the path required to refer to a module item. These declarations may appear in <a href="items/modules.html">modules</a> and <a href="expressions/block-expr.html">blocks</a>, usually at the top.</p> <blockquote> <p><strong>Note</strong>: Unlike in many languages, <code>use</code> declarations in Rust do <em>not</em> declare linkage dependency with external crates. Rather, <a href="items/extern-crates.html"><code>extern crate</code> declarations</a> declare linkage dependencies.</p> </blockquote> <p>Use declarations support a number of convenient shortcuts:</p> <ul> <li>Simultaneously binding a list of paths with a common prefix, using the glob-like brace syntax <code>use a::b::{c, d, e::f, g::h::i};</code></li> <li>Simultaneously binding a list of paths with a common prefix and their common parent module, using the <code>self</code> keyword, such as <code>use a::b::{self, c, d::e};</code></li> <li>Rebinding the target name as a new local name, using the syntax <code>use p::q::r as x;</code>. This can also be used with the last two features: <code>use a::b::{self as ab, c as abc}</code>.</li> <li>Binding all paths matching a given prefix, using the asterisk wildcard syntax <code>use a::b::*;</code>.</li> <li>Nesting groups of the previous features multiple times, such as <code>use a::b::{self as ab, c, d::{*, e::f}};</code></li> </ul> <p>An example of <code>use</code> declarations:</p> <pre><pre class="playpen"><code class="language-rust">use std::option::Option::{Some, None}; use std::collections::hash_map::{self, HashMap}; fn foo<T>(_: T){} fn bar(map1: HashMap<String, usize>, map2: hash_map::HashMap<String, usize>){} fn main() { // Equivalent to 'foo(vec![std::option::Option::Some(1.0f64), // std::option::Option::None]);' foo(vec![Some(1.0f64), None]); // Both `hash_map` and `HashMap` are in scope. let map1 = HashMap::new(); let map2 = hash_map::HashMap::new(); bar(map1, map2); } </code></pre></pre> <p>Like items, <code>use</code> declarations are private to the containing module, by default. Also like items, a <code>use</code> declaration can be public, if qualified by the <code>pub</code> keyword. Such a <code>use</code> declaration serves to <em>re-export</em> a name. A public <code>use</code> declaration can therefore <em>redirect</em> some public name to a different target definition: even a definition with a private canonical path, inside a different module. If a sequence of such redirections form a cycle or cannot be resolved unambiguously, they represent a compile-time error.</p> <p>An example of re-exporting:</p> <pre><pre class="playpen"><code class="language-rust"># fn main() { } mod quux { pub use quux::foo::{bar, baz}; pub mod foo { pub fn bar() { } pub fn baz() { } } } </code></pre></pre> <p>In this example, the module <code>quux</code> re-exports two public names defined in <code>foo</code>.</p> <p>Also note that the paths contained in <code>use</code> items are relative to the crate root. So, in the previous example, the <code>use</code> refers to <code>quux::foo::{bar, baz}</code>, and not simply to <code>foo::{bar, baz}</code>. This also means that top-level module declarations should be at the crate root if direct usage of the declared modules within <code>use</code> items is desired. It is also possible to use <code>self</code> and <code>super</code> at the beginning of a <code>use</code> item to refer to the current and direct parent modules respectively. All rules regarding accessing declared modules in <code>use</code> declarations apply to both module declarations and <code>extern crate</code> declarations.</p> <p>An example of what will and will not work for <code>use</code> items:</p> <pre><pre class="playpen"><code class="language-rust"># #![allow(unused_imports)] use foo::baz::foobaz; // good: foo is at the root of the crate mod foo { mod example { pub mod iter {} } use foo::example::iter; // good: foo is at crate root // use example::iter; // bad: example is not at the crate root use self::baz::foobaz; // good: self refers to module 'foo' use foo::bar::foobar; // good: foo is at crate root pub mod bar { pub fn foobar() { } } pub mod baz { use super::bar::foobar; // good: super refers to module 'foo' pub fn foobaz() { } } } fn main() {} </code></pre></pre> <a class="header" href="print.html#functions" id="functions"><h1>Functions</h1></a> <p>A <em>function</em> consists of a <a href="expressions/block-expr.html">block</a>, along with a name and a set of parameters. Other than a name, all these are optional. Functions are declared with the keyword <code>fn</code>. Functions may declare a set of <em>input</em> <a href="variables.html"><em>variables</em></a> as parameters, through which the caller passes arguments into the function, and the <em>output</em> <a href="types.html"><em>type</em></a> of the value the function will return to its caller on completion.</p> <p>When referred to, a <em>function</em> yields a first-class <em>value</em> of the corresponding zero-sized <a href="types.html#function-item-types"><em>function item type</em></a>, which when called evaluates to a direct call to the function.</p> <p>For example, this is a simple function:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn answer_to_life_the_universe_and_everything() -> i32 { return 42; } #}</code></pre></pre> <p>As with <code>let</code> bindings, function arguments are irrefutable patterns, so any pattern that is valid in a let binding is also valid as an argument:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn first((value, _): (i32, i32)) -> i32 { value } #}</code></pre></pre> <p>The block of a function is conceptually wrapped in a block that binds the argument patterns and then <code>return</code>s the value of the function's block. This means that the tail expression of the block, if evaluated, ends up being returned to the caller. As usual, an explicit return expression within the body of the function will short-cut that implicit return, if reached.</p> <p>For example, the function above behaves as if it was written as:</p> <pre><code class="language-rust ignore">// argument_0 is the actual first argument passed from the caller let (value, _) = argument_0; return { value }; </code></pre> <a class="header" href="print.html#generic-functions" id="generic-functions"><h2>Generic functions</h2></a> <p>A <em>generic function</em> allows one or more <em>parameterized types</em> to appear in its signature. Each type parameter must be explicitly declared in an angle-bracket-enclosed and comma-separated list, following the function name.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { // foo is generic over A and B fn foo<A, B>(x: A, y: B) { # } #}</code></pre></pre> <p>Inside the function signature and body, the name of the type parameter can be used as a type name. <a href="items/traits.html">Trait</a> bounds can be specified for type parameters to allow methods with that trait to be called on values of that type. This is specified using the <code>where</code> syntax:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # use std::fmt::Debug; fn foo<T>(x: T) where T: Debug { # } #}</code></pre></pre> <p>When a generic function is referenced, its type is instantiated based on the context of the reference. For example, calling the <code>foo</code> function here:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { use std::fmt::Debug; fn foo<T>(x: &[T]) where T: Debug { // details elided } foo(&[1, 2]); #}</code></pre></pre> <p>will instantiate type parameter <code>T</code> with <code>i32</code>.</p> <p>The type parameters can also be explicitly supplied in a trailing <a href="paths.html">path</a> component after the function name. This might be necessary if there is not sufficient context to determine the type parameters. For example, <code>mem::size_of::<u32>() == 4</code>.</p> <a class="header" href="print.html#diverging-functions" id="diverging-functions"><h2>Diverging functions</h2></a> <p>A special kind of function can be declared with a <code>!</code> character where the output type would normally be. For example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn my_err(s: &str) -> ! { println!("{}", s); panic!(); } #}</code></pre></pre> <p>We call such functions "diverging" because they never return a value to the caller. Every control path in a diverging function must end with a <code>panic!()</code>, a loop expression without an associated break expression, or a call to another diverging function on every control path. The <code>!</code> annotation does <em>not</em> denote a type.</p> <p>It might be necessary to declare a diverging function because as mentioned previously, the typechecker checks that every control path in a function ends with a <a href="expressions/return-expr.html"><code>return</code></a> or diverging expression. So, if <code>my_err</code> were declared without the <code>!</code> annotation, the following code would not typecheck:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # fn my_err(s: &str) -> ! { panic!() } fn f(i: i32) -> i32 { if i == 42 { return 42; } else { my_err("Bad number!"); } } #}</code></pre></pre> <p>This will not compile without the <code>!</code> annotation on <code>my_err</code>, since the <code>else</code> branch of the conditional in <code>f</code> does not return an <code>i32</code>, as required by the signature of <code>f</code>. Adding the <code>!</code> annotation to <code>my_err</code> informs the typechecker that, should control ever enter <code>my_err</code>, no further type judgments about <code>f</code> need to hold, since control will never resume in any context that relies on those judgments. Thus the return type on <code>f</code> only needs to reflect the <code>if</code> branch of the conditional.</p> <a class="header" href="print.html#extern-functions" id="extern-functions"><h2>Extern functions</h2></a> <p>Extern functions are part of Rust's foreign function interface, providing the opposite functionality to <a href="items/external-blocks.html">external blocks</a>. Whereas external blocks allow Rust code to call foreign code, extern functions with bodies defined in Rust code <em>can be called by foreign code</em>. They are defined in the same way as any other Rust function, except that they have the <code>extern</code> modifier.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { // Declares an extern fn, the ABI defaults to "C" extern fn new_i32() -> i32 { 0 } // Declares an extern fn with "stdcall" ABI # #[cfg(target_arch = "x86_64")] extern "stdcall" fn new_i32_stdcall() -> i32 { 0 } #}</code></pre></pre> <p>Unlike normal functions, extern fns have type <code>extern "ABI" fn()</code>. This is the same type as the functions declared in an extern block.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # extern fn new_i32() -> i32 { 0 } let fptr: extern "C" fn() -> i32 = new_i32; #}</code></pre></pre> <p>As non-Rust calling conventions do not support unwinding, unwinding past the end of an extern function will cause the process to abort. In LLVM, this is implemented by executing an illegal instruction.</p> <a class="header" href="print.html#type-aliases" id="type-aliases"><h1>Type aliases</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>TypeAlias</em> :<br /> <code>type</code> <a href="identifiers.html">IDENTIFIER</a> <a href="items.html#type-parameters"><em>Generics</em></a><sup>?</sup> <a href="items.html#type-parameters"><em>WhereClause</em></a><sup>?</sup> <code>=</code> <a href="types.html"><em>Type</em></a> <code>;</code></p> </blockquote> <p>A <em>type alias</em> defines a new name for an existing <a href="types.html">type</a>. Type aliases are declared with the keyword <code>type</code>. Every value has a single, specific type, but may implement several different traits, or be compatible with several different type constraints.</p> <p>For example, the following defines the type <code>Point</code> as a synonym for the type <code>(u8, u8)</code>, the type of pairs of unsigned 8 bit integers:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { type Point = (u8, u8); let p: Point = (41, 68); #}</code></pre></pre> <p>A type alias to an enum type cannot be used to qualify the constructors:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { enum E { A } type F = E; let _: F = E::A; // OK // let _: F = F::A; // Doesn't work #}</code></pre></pre> <a class="header" href="print.html#structs" id="structs"><h1>Structs</h1></a> <p>A <em>struct</em> is a nominal <a href="types.html#struct-types">struct type</a> defined with the keyword <code>struct</code>.</p> <p>An example of a <code>struct</code> item and its use:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct Point {x: i32, y: i32} let p = Point {x: 10, y: 11}; let px: i32 = p.x; #}</code></pre></pre> <p>A <em>tuple struct</em> is a nominal <a href="types.html#tuple-types">tuple type</a>, also defined with the keyword <code>struct</code>. For example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct Point(i32, i32); let p = Point(10, 11); let px: i32 = match p { Point(x, _) => x }; #}</code></pre></pre> <p>A <em>unit-like struct</em> is a struct without any fields, defined by leaving off the list of fields entirely. Such a struct implicitly defines a constant of its type with the same name. For example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct Cookie; let c = [Cookie, Cookie {}, Cookie, Cookie {}]; #}</code></pre></pre> <p>is equivalent to</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct Cookie {} const Cookie: Cookie = Cookie {}; let c = [Cookie, Cookie {}, Cookie, Cookie {}]; #}</code></pre></pre> <p>The precise memory layout of a struct is not specified. One can specify a particular layout using the <a href="attributes.html#ffi-attributes"><code>repr</code> attribute</a>.</p> <a class="header" href="print.html#enumerations" id="enumerations"><h1>Enumerations</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>Enumeration</em> :<br /> <code>enum</code> <a href="identifiers.html">IDENTIFIER</a> <a href="items.html#type-parameters"><em>Generics</em></a><sup>?</sup> <a href="items.html#type-parameters"><em>WhereClause</em></a><sup>?</sup> <code>{</code> <em>EnumItems</em><sup>?</sup> <code>}</code></p> <p><em>EnumItems</em> :<br /> <em>EnumItem</em> ( <code>,</code> <em>EnumItem</em> )<sup>*</sup> <code>,</code><sup>?</sup></p> <p><em>EnumItem</em> :<br /> <em>OuterAttribute</em><sup>*</sup><br /> <a href="identifiers.html">IDENTIFIER</a> ( <em>EnumItemTuple</em> | <em>EnumItemStruct</em> | <em>EnumItemDiscriminant</em> )<sup>?</sup></p> <p><em>EnumItemTuple</em> :<br /> <code>(</code> <a href="items/structs.html"><em>TupleFields</em></a><sup>?</sup> <code>)</code></p> <p><em>EnumItemStruct</em> :<br /> <code>{</code> <a href="items/structs.html"><em>StructFields</em></a><sup>?</sup> <code>}</code></p> <p><em>EnumItemDiscriminant</em> :<br /> <code>=</code> <a href="expressions.html"><em>Expression</em></a></p> </blockquote> <p>An <em>enumeration</em>, also referred to as <em>enum</em> is a simultaneous definition of a nominal <a href="types.html#enumerated-types">enumerated type</a> as well as a set of <em>constructors</em>, that can be used to create or pattern-match values of the corresponding enumerated type.</p> <p>Enumerations are declared with the keyword <code>enum</code>.</p> <p>An example of an <code>enum</code> item and its use:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { enum Animal { Dog, Cat, } let mut a: Animal = Animal::Dog; a = Animal::Cat; #}</code></pre></pre> <p>Enum constructors can have either named or unnamed fields:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { enum Animal { Dog(String, f64), Cat { name: String, weight: f64 }, } let mut a: Animal = Animal::Dog("Cocoa".to_string(), 37.2); a = Animal::Cat { name: "Spotty".to_string(), weight: 2.7 }; #}</code></pre></pre> <p>In this example, <code>Cat</code> is a <em>struct-like enum variant</em>, whereas <code>Dog</code> is simply called an enum variant. Each enum instance has a <em>discriminant</em> which is an integer associated to it that is used to determine which variant it holds. An opaque reference to this discriminant can be obtained with the <a href="../std/mem/fn.discriminant.html"><code>mem::discriminant</code></a> function.</p> <a class="header" href="print.html#custom-discriminant-values-for-field-less-enumerations" id="custom-discriminant-values-for-field-less-enumerations"><h2>Custom Discriminant Values for Field-Less Enumerations</h2></a> <p>If there is no data attached to <em>any</em> of the variants of an enumeration, then the discriminant can be directly chosen and accessed.</p> <p>These enumerations can be cast to integer types with the <code>as</code> operator by a <a href="expressions/operator-expr.html#semantics">numeric cast</a>. The enumeration can optionally specify which integer each discriminant gets by following the variant name with <code>=</code> and then an integer literal. If the first variant in the declaration is unspecified, then it is set to zero. For every unspecified discriminant, it is set to one higher than the previous variant in the declaration.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { enum Foo { Bar, // 0 Baz = 123, // 123 Quux, // 124 } let baz_discriminant = Foo::Baz as u32; assert_eq!(baz_discriminant, 123); #}</code></pre></pre> <p>Under the [default representation], the specified discriminant is interpreted as an <code>isize</code> value although the compiler is allowed to use a smaller type in the actual memory layout. The size and thus acceptable values can be changed by using a [primitive representation] or the [<code>C</code> representation].</p> <p>It is an error when two variants share the same discriminant.</p> <pre><code class="language-rust ignore">enum SharedDiscriminantError { SharedA = 1, SharedB = 1 } enum SharedDiscriminantError2 { Zero, // 0 One, // 1 OneToo = 1 // 1 (collision with previous!) } </code></pre> <p>It is also an error to have an unspecified discriminant where the previous discriminant is the maximum value for the size of the discriminant.</p> <pre><code class="language-rust ignore">#[repr(u8)] enum OverflowingDiscriminantError { Max = 255, MaxPlusOne // Would be 256, but that overflows the enum. } #[repr(u8)] enum OverflowingDiscriminantError2 { MaxMinusOne = 254, // 254 Max, // 255 MaxPlusOne // Would be 256, but that overflows the enum. } </code></pre> <a class="header" href="print.html#zero-variant-enums" id="zero-variant-enums"><h2>Zero-variant Enums</h2></a> <p>Enums with zero variants are known as <em>zero-variant enums</em>. As they have no valid values, they cannot be instantiated.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { enum ZeroVariants {} #}</code></pre></pre> <a class="header" href="print.html#unions" id="unions"><h1>Unions</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>Union</em> :<br /> <code>union</code> <a href="identifiers.html">IDENTIFIER</a> <a href="items.html#type-parameters"><em>Generics</em></a><sup>?</sup> <a href="items.html#type-parameters"><em>WhereClause</em></a><sup>?</sup> <code>{</code><a href="items/structs.html"><em>StructFields</em></a> <code>}</code></p> </blockquote> <p>A union declaration uses the same syntax as a struct declaration, except with <code>union</code> in place of <code>struct</code>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { #[repr(C)] union MyUnion { f1: u32, f2: f32, } #}</code></pre></pre> <p>The key property of unions is that all fields of a union share common storage. As a result writes to one field of a union can overwrite its other fields, and size of a union is determined by the size of its largest field.</p> <p>A value of a union type can be created using the same syntax that is used for struct types, except that it must specify exactly one field:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # union MyUnion { f1: u32, f2: f32 } # let u = MyUnion { f1: 1 }; #}</code></pre></pre> <p>The expression above creates a value of type <code>MyUnion</code> with active field <code>f1</code>. Active field of a union can be accessed using the same syntax as struct fields:</p> <pre><code class="language-rust ignore">let f = u.f1; </code></pre> <p>Inactive fields can be accessed as well (using the same syntax) if they are sufficiently layout compatible with the current value kept by the union. Reading incompatible fields results in undefined behavior. However, the active field is not generally known statically, so all reads of union fields have to be placed in <code>unsafe</code> blocks.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # union MyUnion { f1: u32, f2: f32 } # let u = MyUnion { f1: 1 }; # unsafe { let f = u.f1; } #}</code></pre></pre> <p>Writes to <code>Copy</code> union fields do not require reads for running destructors, so these writes don't have to be placed in <code>unsafe</code> blocks</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # union MyUnion { f1: u32, f2: f32 } # let mut u = MyUnion { f1: 1 }; # u.f1 = 2; #}</code></pre></pre> <p>Commonly, code using unions will provide safe wrappers around unsafe union field accesses.</p> <p>Another way to access union fields is to use pattern matching. Pattern matching on union fields uses the same syntax as struct patterns, except that the pattern must specify exactly one field. Since pattern matching accesses potentially inactive fields it has to be placed in <code>unsafe</code> blocks as well.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # union MyUnion { f1: u32, f2: f32 } # fn f(u: MyUnion) { unsafe { match u { MyUnion { f1: 10 } => { println!("ten"); } MyUnion { f2 } => { println!("{}", f2); } } } } #}</code></pre></pre> <p>Pattern matching may match a union as a field of a larger structure. In particular, when using a Rust union to implement a C tagged union via FFI, this allows matching on the tag and the corresponding field simultaneously:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { #[repr(u32)] enum Tag { I, F } #[repr(C)] union U { i: i32, f: f32, } #[repr(C)] struct Value { tag: Tag, u: U, } fn is_zero(v: Value) -> bool { unsafe { match v { Value { tag: I, u: U { i: 0 } } => true, Value { tag: F, u: U { f: 0.0 } } => true, _ => false, } } } #}</code></pre></pre> <p>Since union fields share common storage, gaining write access to one field of a union can give write access to all its remaining fields. Borrow checking rules have to be adjusted to account for this fact. As a result, if one field of a union is borrowed, all its remaining fields are borrowed as well for the same lifetime.</p> <pre><code class="language-rust ignore">// ERROR: cannot borrow `u` (via `u.f2`) as mutable more than once at a time fn test() { let mut u = MyUnion { f1: 1 }; unsafe { let b1 = &mut u.f1; ---- first mutable borrow occurs here (via `u.f1`) let b2 = &mut u.f2; ^^^^ second mutable borrow occurs here (via `u.f2`) *b1 = 5; } - first borrow ends here assert_eq!(unsafe { u.f1 }, 5); } </code></pre> <p>As you could see, in many aspects (except for layouts, safety and ownership) unions behave exactly like structs, largely as a consequence of inheriting their syntactic shape from structs. This is also true for many unmentioned aspects of Rust language (such as privacy, name resolution, type inference, generics, trait implementations, inherent implementations, coherence, pattern checking, etc etc etc).</p> <p>More detailed specification for unions, including unstable bits, can be found in <a href="https://github.com/rust-lang/rfcs/pull/1897">RFC 1897 "Unions v1.2"</a>.</p> <a class="header" href="print.html#constant-items" id="constant-items"><h1>Constant items</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong> <em>ConstantItem</em> : <code>const</code> <a href="identifiers.html">IDENTIFIER</a> <code>:</code> <a href="types.html"><em>Type</em></a> <code>=</code> <a href="expressions.html"><em>Expression</em></a> <code>;</code></p> </blockquote> <p>A <em>constant item</em> is a named <em><a href="expressions.html#constant-expressions">constant value</a></em> which is not associated with a specific memory location in the program. Constants are essentially inlined wherever they are used, meaning that they are copied directly into the relevant context when used. References to the same constant are not necessarily guaranteed to refer to the same memory address.</p> <p>Constants must be explicitly typed. The type must have a <code>'static</code> lifetime: any references it contains must have <code>'static</code> lifetimes.</p> <p>Constants may refer to the address of other constants, in which case the address will have elided lifetimes where applicable, otherwise – in most cases – defaulting to the <code>static</code> lifetime. (See <a href="items/static-items.html#static-lifetime-elision">static lifetime elision</a>.) The compiler is, however, still at liberty to translate the constant many times, so the address referred to may not be stable.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { const BIT1: u32 = 1 << 0; const BIT2: u32 = 1 << 1; const BITS: [u32; 2] = [BIT1, BIT2]; const STRING: &'static str = "bitstring"; struct BitsNStrings<'a> { mybits: [u32; 2], mystring: &'a str, } const BITS_N_STRINGS: BitsNStrings<'static> = BitsNStrings { mybits: BITS, mystring: STRING, }; #}</code></pre></pre> <a class="header" href="print.html#constants-with-destructors" id="constants-with-destructors"><h2>Constants with Destructors</h2></a> <p>Constants can contain destructors. Destructors are ran when the value goes out of scope.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct TypeWithDestructor(i32); impl Drop for TypeWithDestructor { fn drop(&mut self) { println!("Dropped. Held {}.", self.0); } } const ZERO_WITH_DESTRUCTOR: TypeWithDestructor = TypeWithDestructor(0); fn create_and_drop_zero_with_destructor() { let x = ZERO_WITH_DESTRUCTOR; // x gets dropped at end of function, calling drop. // prints "Dropped. Held 0.". } #}</code></pre></pre> <a class="header" href="print.html#static-items" id="static-items"><h1>Static items</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>StaticItem</em> :<br /> <code>static</code> <code>mut</code><sup>?</sup> <a href="identifiers.html">IDENTIFIER</a> <code>:</code> <a href="types.html"><em>Type</em></a> <code>=</code> <a href="expressions.html"><em>Expression</em></a> <code>;</code></p> </blockquote> <p>A <em>static item</em> is similar to a <a href="items/constant-items.html">constant</a>, except that it represents a precise memory location in the program. A static is never "inlined" at the usage site, and all references to it refer to the same memory location. Static items have the <code>static</code> lifetime, which outlives all other lifetimes in a Rust program. Static items may be placed in read-only memory if the type is not <a href="interior-mutability.html">interior mutable</a>. Static items do not call <code>drop</code> at the end of the program.</p> <p>All access to a static is safe, but there are a number of restrictions on statics:</p> <ul> <li>The type must have the <code>Sync</code> trait bound to allow thread-safe access.</li> <li>Statics allow using paths to statics in the <a href="expressions.html#constant-expressions">constant-expression</a> used to initialize them, but statics may not refer to other statics by value, only through a reference.</li> <li>Constants cannot refer to statics.</li> </ul> <a class="header" href="print.html#mutable-statics" id="mutable-statics"><h2>Mutable statics</h2></a> <p>If a static item is declared with the <code>mut</code> keyword, then it is allowed to be modified by the program. One of Rust's goals is to make concurrency bugs hard to run into, and this is obviously a very large source of race conditions or other bugs. For this reason, an <code>unsafe</code> block is required when either reading or writing a mutable static variable. Care should be taken to ensure that modifications to a mutable static are safe with respect to other threads running in the same process.</p> <p>Mutable statics are still very useful, however. They can be used with C libraries and can also be bound from C libraries (in an <code>extern</code> block).</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # fn atomic_add(_: &mut u32, _: u32) -> u32 { 2 } static mut LEVELS: u32 = 0; // This violates the idea of no shared state, and this doesn't internally // protect against races, so this function is `unsafe` unsafe fn bump_levels_unsafe1() -> u32 { let ret = LEVELS; LEVELS += 1; return ret; } // Assuming that we have an atomic_add function which returns the old value, // this function is "safe" but the meaning of the return value may not be what // callers expect, so it's still marked as `unsafe` unsafe fn bump_levels_unsafe2() -> u32 { return atomic_add(&mut LEVELS, 1); } #}</code></pre></pre> <p>Mutable statics have the same restrictions as normal statics, except that the type does not have to implement the <code>Sync</code> trait.</p> <a class="header" href="print.html#static-lifetime-elision" id="static-lifetime-elision"><h2><code>'static</code> lifetime elision</h2></a> <p>Both constant and static declarations of reference types have <em>implicit</em> <code>'static</code> lifetimes unless an explicit lifetime is specified. As such, the constant declarations involving <code>'static</code> above may be written without the lifetimes. Returning to our previous example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { const BIT1: u32 = 1 << 0; const BIT2: u32 = 1 << 1; const BITS: [u32; 2] = [BIT1, BIT2]; const STRING: &str = "bitstring"; struct BitsNStrings<'a> { mybits: [u32; 2], mystring: &'a str, } const BITS_N_STRINGS: BitsNStrings = BitsNStrings { mybits: BITS, mystring: STRING, }; #}</code></pre></pre> <p>Note that if the <code>static</code> or <code>const</code> items include function or closure references, which themselves include references, the compiler will first try the standard elision rules (<a href="../nomicon/lifetime-elision.html">see discussion in the nomicon</a>). If it is unable to resolve the lifetimes by its usual rules, it will default to using the <code>'static</code> lifetime. By way of example:</p> <pre><code class="language-rust ignore">// Resolved as `fn<'a>(&'a str) -> &'a str`. const RESOLVED_SINGLE: fn(&str) -> &str = .. // Resolved as `Fn<'a, 'b, 'c>(&'a Foo, &'b Bar, &'c Baz) -> usize`. const RESOLVED_MULTIPLE: Fn(&Foo, &Bar, &Baz) -> usize = .. // There is insufficient information to bound the return reference lifetime // relative to the argument lifetimes, so the signature is resolved as // `Fn(&'static Foo, &'static Bar) -> &'static Baz`. const RESOLVED_STATIC: Fn(&Foo, &Bar) -> &Baz = .. </code></pre> <a class="header" href="print.html#using-statics-or-consts" id="using-statics-or-consts"><h2>Using Statics or Consts</h2></a> <p>In can be confusing whether or not you should use a constant item or a static item. Constants should, in general, be preferred over statics unless one of the following are true:</p> <ul> <li>Large amounts of data are being stored</li> <li>The single-address or non-inlining property of statics is required.</li> <li>Interior mutability is required.</li> </ul> <a class="header" href="print.html#traits" id="traits"><h1>Traits</h1></a> <p>A <em>trait</em> describes an abstract interface that types can implement. This interface consists of associated items, which come in three varieties:</p> <ul> <li><a href="print.html#associated-functions-and-methods">functions</a></li> <li><a href="print.html#associated-types">types</a></li> <li><a href="print.html#associated-constants">constants</a></li> </ul> <p>All traits define an implicit type parameter <code>Self</code> that refers to "the type that is implementing this interface". Traits may also contain additional type parameters. These type parameters (including <code>Self</code>) may be constrained by other traits and so forth as usual.</p> <p>Traits are implemented for specific types through separate <a href="items/implementations.html">implementations</a>.</p> <a class="header" href="print.html#associated-functions-and-methods" id="associated-functions-and-methods"><h2>Associated functions and methods</h2></a> <p>Associated functions whose first parameter is named <code>self</code> are called methods and may be invoked using <code>.</code> notation (e.g., <code>x.foo()</code>) as well as the usual function call notation (<code>foo(x)</code>).</p> <p>Consider the following trait:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # type Surface = i32; # type BoundingBox = i32; trait Shape { fn draw(&self, Surface); fn bounding_box(&self) -> BoundingBox; } #}</code></pre></pre> <p>This defines a trait with two methods. All values that have <a href="items/implementations.html">implementations</a> of this trait in scope can have their <code>draw</code> and <code>bounding_box</code> methods called, using <code>value.bounding_box()</code> <a href="expressions/method-call-expr.html">syntax</a>. Note that <code>&self</code> is short for <code>self: &Self</code>, and similarly, <code>self</code> is short for <code>self: Self</code> and <code>&mut self</code> is short for <code>self: &mut Self</code>.</p> <p>Traits can include default implementations of methods, as in:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { trait Foo { fn bar(&self); fn baz(&self) { println!("We called baz."); } } #}</code></pre></pre> <p>Here the <code>baz</code> method has a default implementation, so types that implement <code>Foo</code> need only implement <code>bar</code>. It is also possible for implementing types to override a method that has a default implementation.</p> <p>Type parameters can be specified for a trait to make it generic. These appear after the trait name, using the same syntax used in <a href="items/functions.html#generic-functions">generic functions</a>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { trait Seq<T> { fn len(&self) -> u32; fn elt_at(&self, n: u32) -> T; fn iter<F>(&self, F) where F: Fn(T); } #}</code></pre></pre> <p>Associated functions may lack a <code>self</code> argument, sometimes called 'static methods'. This means that they can only be called with function call syntax (<code>f(x)</code>) and not method call syntax (<code>obj.f()</code>). The way to refer to the name of a static method is to qualify it with the trait name or type name, treating the trait name like a module. For example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { trait Num { fn from_i32(n: i32) -> Self; } impl Num for f64 { fn from_i32(n: i32) -> f64 { n as f64 } } let x: f64 = Num::from_i32(42); let x: f64 = f64::from_i32(42); #}</code></pre></pre> <a class="header" href="print.html#associated-types" id="associated-types"><h2>Associated Types</h2></a> <p>It is also possible to define associated types for a trait. Consider the following example of a <code>Container</code> trait. Notice how the type is available for use in the method signatures:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { trait Container { type E; fn empty() -> Self; fn insert(&mut self, Self::E); } #}</code></pre></pre> <p>In order for a type to implement this trait, it must not only provide implementations for every method, but it must specify the type <code>E</code>. Here's an implementation of <code>Container</code> for the standard library type <code>Vec</code>:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # trait Container { # type E; # fn empty() -> Self; # fn insert(&mut self, Self::E); # } impl<T> Container for Vec<T> { type E = T; fn empty() -> Vec<T> { Vec::new() } fn insert(&mut self, x: T) { self.push(x); } } #}</code></pre></pre> <a class="header" href="print.html#associated-constants" id="associated-constants"><h2>Associated Constants</h2></a> <p>A trait can define constants like this:</p> <pre><pre class="playpen"><code class="language-rust">trait Foo { const ID: i32; } impl Foo for i32 { const ID: i32 = 1; } fn main() { assert_eq!(1, i32::ID); } </code></pre></pre> <p>Any implementor of <code>Foo</code> will have to define <code>ID</code>. Without the definition:</p> <pre><pre class="playpen"><code class="language-rust compile_fail E0046"> # #![allow(unused_variables)] #fn main() { trait Foo { const ID: i32; } impl Foo for i32 { } #}</code></pre></pre> <p>gives</p> <pre><code class="language-text">error: not all trait items implemented, missing: `ID` [E0046] impl Foo for i32 { } </code></pre> <p>A default value can be implemented as well:</p> <pre><pre class="playpen"><code class="language-rust">trait Foo { const ID: i32 = 1; } impl Foo for i32 { } impl Foo for i64 { const ID: i32 = 5; } fn main() { assert_eq!(1, i32::ID); assert_eq!(5, i64::ID); } </code></pre></pre> <p>As you can see, when implementing <code>Foo</code>, you can leave it unimplemented, as with <code>i32</code>. It will then use the default value. But, as in <code>i64</code>, we can also add our own definition.</p> <p>Associated constants don’t have to be associated with a trait. An <code>impl</code> block for a <code>struct</code> or an <code>enum</code> works fine too:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct Foo; impl Foo { const FOO: u32 = 3; } #}</code></pre></pre> <a class="header" href="print.html#trait-bounds" id="trait-bounds"><h2>Trait bounds</h2></a> <p>Generic functions may use traits as <em>bounds</em> on their type parameters. This will have three effects:</p> <ul> <li>Only types that have the trait may instantiate the parameter.</li> <li>Within the generic function, the methods of the trait can be called on values that have the parameter's type. Associated types can be used in the function's signature, and associated constants can be used in expressions within the function body.</li> <li>Generic functions and types with the same or weaker bounds can use the generic type in the function body or signature.</li> </ul> <p>For example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # type Surface = i32; # trait Shape { fn draw(&self, Surface); } struct Figure<S: Shape>(S, S); fn draw_twice<T: Shape>(surface: Surface, sh: T) { sh.draw(surface); sh.draw(surface); } fn draw_figure<U: Shape>(surface: Surface, Figure(sh1, sh2): Figure<U>) { sh1.draw(surface); draw_twice(surface, sh2); // Can call this since U: Shape } #}</code></pre></pre> <a class="header" href="print.html#object-safety" id="object-safety"><h2>Object Safety</h2></a> <p>Object safe traits can be the base trait of a <a href="types.html#trait-objects">trait object</a>. A trait is <em>object safe</em> if it has the following qualities (defined in <a href="https://github.com/rust-lang/rfcs/blob/master/text/0255-object-safety.md">RFC 255</a>):</p> <ul> <li>It must not require <code>Self: Sized</code></li> <li>All associated functions must either have a <code>where Self: Sized</code> bound or <ul> <li>Not have any type parameters (although lifetime parameters are allowed)</li> <li>Must be a method: its first parameter must be called self, with type <code>Self</code>, <code>&Self</code>, <code>&mut Self</code>, <code>Box<Self></code>.</li> <li><code>Self</code> may only be used in the type of the receiver.</li> </ul> </li> <li>It must not have any associated constants.</li> </ul> <a class="header" href="print.html#supertraits" id="supertraits"><h2>Supertraits</h2></a> <p>Trait bounds on <code>Self</code> are considered "supertraits". These are required to be acyclic. Supertraits are somewhat different from other constraints in that they affect what methods are available in the vtable when the trait is used as a <a href="types.html#trait-objects">trait object</a>. Consider the following example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { trait Shape { fn area(&self) -> f64; } trait Circle : Shape { fn radius(&self) -> f64; } #}</code></pre></pre> <p>The syntax <code>Circle : Shape</code> means that types that implement <code>Circle</code> must also have an implementation for <code>Shape</code>. Multiple supertraits are separated by <code>+</code>, <code>trait Circle : Shape + PartialEq { }</code>. In an implementation of <code>Circle</code> for a given type <code>T</code>, methods can refer to <code>Shape</code> methods, since the typechecker checks that any type with an implementation of <code>Circle</code> also has an implementation of <code>Shape</code>:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct Foo; trait Shape { fn area(&self) -> f64; } trait Circle : Shape { fn radius(&self) -> f64; } impl Shape for Foo { fn area(&self) -> f64 { 0.0 } } impl Circle for Foo { fn radius(&self) -> f64 { println!("calling area: {}", self.area()); 0.0 } } let c = Foo; c.radius(); #}</code></pre></pre> <p>In type-parameterized functions, methods of the supertrait may be called on values of subtrait-bound type parameters. Referring to the previous example of <code>trait Circle : Shape</code>:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # trait Shape { fn area(&self) -> f64; } # trait Circle : Shape { fn radius(&self) -> f64; } fn radius_times_area<T: Circle>(c: T) -> f64 { // `c` is both a Circle and a Shape c.radius() * c.area() } #}</code></pre></pre> <p>Likewise, supertrait methods may also be called on trait objects.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # trait Shape { fn area(&self) -> f64; } # trait Circle : Shape { fn radius(&self) -> f64; } # impl Shape for i32 { fn area(&self) -> f64 { 0.0 } } # impl Circle for i32 { fn radius(&self) -> f64 { 0.0 } } # let mycircle = 0i32; let mycircle = Box::new(mycircle) as Box<Circle>; let nonsense = mycircle.radius() * mycircle.area(); #}</code></pre></pre> <a class="header" href="print.html#implementations" id="implementations"><h1>Implementations</h1></a> <p>An <em>implementation</em> is an item that associates items with an <em>implementing type</em>.</p> <p>There are two types of implementations: inherent implementations and <a href="items/traits.html">trait</a> implementations.</p> <p>Implementations are defined with the keyword <code>impl</code>.</p> <a class="header" href="print.html#inherent-implementations" id="inherent-implementations"><h2>Inherent Implementations</h2></a> <p>An inherent implementation is defined as the sequence of the <code>impl</code> keyword, generic type declarations, a path to a nomial type, a where clause, and a bracketed set of associable items.</p> <p>The nominal type is called the <em>implementing type</em> and the associable items are the <em>associated items</em> to the implementing type.</p> <p>Inherent implementations associate the associated items to the implementing type.</p> <p>The associated item has a path of a path to the implementing type followed by the associate item's path component.</p> <p>Inherent implementations cannot contain associated type aliases.</p> <p>A type can have multiple inherent implementations.</p> <p>The implementing type must be defined within the same crate.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct Point {x: i32, y: i32} impl Point { fn log(&self) { println!("Point is at ({}, {})", self.x, self.y); } } let my_point = Point {x: 10, y:11}; my_point.log(); #}</code></pre></pre> <a class="header" href="print.html#trait-implementations" id="trait-implementations"><h2>Trait Implementations</h2></a> <p>A <em>trait implementation</em> is defined like an inherent implementation except that the optional generic type declarations is followed by a <a href="items/traits.html">trait</a> followed by the keyword <code>for</code>. <!-- To understand this, you have to back-reference to the previous section. :( --></p> <p>The trait is known as the <em>implemented trait</em>.</p> <p>The implementing type implements the implemented trait.</p> <p>A trait implementation must define all non-default associated items declared by the implemented trait, may redefine default associated items defined by the implemented trait trait, and cannot define any other items.</p> <p>The path to the associated items is <code><</code> followed by a path to the implementing type followed by <code>as</code> followed by a path to the trait followed by <code>></code> as a path component followed by the associated item's path component.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # #[derive(Copy, Clone)] # struct Point {x: f64, y: f64}; # type Surface = i32; # struct BoundingBox {x: f64, y: f64, width: f64, height: f64}; # trait Shape { fn draw(&self, Surface); fn bounding_box(&self) -> BoundingBox; } # fn do_draw_circle(s: Surface, c: Circle) { } struct Circle { radius: f64, center: Point, } impl Copy for Circle {} impl Clone for Circle { fn clone(&self) -> Circle { *self } } impl Shape for Circle { fn draw(&self, s: Surface) { do_draw_circle(s, *self); } fn bounding_box(&self) -> BoundingBox { let r = self.radius; BoundingBox { x: self.center.x - r, y: self.center.y - r, width: 2.0 * r, height: 2.0 * r, } } } #}</code></pre></pre> <a class="header" href="print.html#trait-implementation-coherence" id="trait-implementation-coherence"><h3>Trait Implementation Coherence</h3></a> <p>A trait implementation is consider incoherent if either the orphan check fails or there are overlapping implementation instaces.</p> <p>Two trait implementations overlap when there is a non-empty intersection of the traits the implementation is for, the implementations can be instantiated with the same type. <!-- This is probably wrong? Source: No two implementations can be instantiable with the same set of types for the input type parameters. --></p> <p>The <code>Orphan Check</code> states that every trait implementation must meet either of the following conditions:</p> <ol> <li> <p>The trait being implemented is defined in the same crate.</p> </li> <li> <p>At least one of either <code>Self</code> or a generic type parameter of the trait must meet the following grammar, where <code>C</code> is a nominal type defined within the containing crate:</p> <pre><code class="language-ignore"> T = C | &T | &mut T | Box<T> </code></pre> </li> </ol> <a class="header" href="print.html#generic-implementations" id="generic-implementations"><h2>Generic Implementations</h2></a> <p>An implementation can take type and lifetime parameters, which can be used in the rest of the implementation. Type parameters declared for an implementation must be used at least once in either the trait or the implementing type of an implementation. Implementation parameters are written directly after the <code>impl</code> keyword.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # trait Seq<T> { fn dummy(&self, _: T) { } } impl<T> Seq<T> for Vec<T> { /* ... */ } impl Seq<bool> for u32 { /* Treat the integer as a sequence of bits */ } #}</code></pre></pre> <a class="header" href="print.html#external-blocks" id="external-blocks"><h1>External blocks</h1></a> <p>External blocks form the basis for Rust's foreign function interface. Declarations in an external block describe symbols in external, non-Rust libraries.</p> <p>Functions within external blocks are declared in the same way as other Rust functions, with the exception that they may not have a body and are instead terminated by a semicolon.</p> <p>Functions within external blocks may be called by Rust code, just like functions defined in Rust. The Rust compiler automatically translates between the Rust ABI and the foreign ABI.</p> <p>Functions within external blocks may be variadic by specifying <code>...</code> after one or more named arguments in the argument list:</p> <pre><code class="language-rust ignore">extern { fn foo(x: i32, ...); } </code></pre> <p>A number of <a href="attributes.html#ffi-attributes">attributes</a> control the behavior of external blocks.</p> <p>By default external blocks assume that the library they are calling uses the standard C ABI on the specific platform. Other ABIs may be specified using an <code>abi</code> string, as shown here:</p> <pre><code class="language-rust ignore">// Interface to the Windows API extern "stdcall" { } </code></pre> <p>There are three ABI strings which are cross-platform, and which all compilers are guaranteed to support:</p> <ul> <li><code>extern "Rust"</code> -- The default ABI when you write a normal <code>fn foo()</code> in any Rust code.</li> <li><code>extern "C"</code> -- This is the same as <code>extern fn foo()</code>; whatever the default your C compiler supports.</li> <li><code>extern "system"</code> -- Usually the same as <code>extern "C"</code>, except on Win32, in which case it's <code>"stdcall"</code>, or what you should use to link to the Windows API itself</li> </ul> <p>There are also some platform-specific ABI strings:</p> <ul> <li><code>extern "cdecl"</code> -- The default for x86_32 C code.</li> <li><code>extern "stdcall"</code> -- The default for the Win32 API on x86_32.</li> <li><code>extern "win64"</code> -- The default for C code on x86_64 Windows.</li> <li><code>extern "sysv64"</code> -- The default for C code on non-Windows x86_64.</li> <li><code>extern "aapcs"</code> -- The default for ARM.</li> <li><code>extern "fastcall"</code> -- The <code>fastcall</code> ABI -- corresponds to MSVC's <code>__fastcall</code> and GCC and clang's <code>__attribute__((fastcall))</code></li> <li><code>extern "vectorcall"</code> -- The <code>vectorcall</code> ABI -- corresponds to MSVC's <code>__vectorcall</code> and clang's <code>__attribute__((vectorcall))</code></li> </ul> <p>Finally, there are some rustc-specific ABI strings:</p> <ul> <li><code>extern "rust-intrinsic"</code> -- The ABI of rustc intrinsics.</li> <li><code>extern "rust-call"</code> -- The ABI of the Fn::call trait functions.</li> <li><code>extern "platform-intrinsic"</code> -- Specific platform intrinsics -- like, for example, <code>sqrt</code> -- have this ABI. You should never have to deal with it.</li> </ul> <p>The <code>link</code> attribute allows the name of the library to be specified. When specified the compiler will attempt to link against the native library of the specified name.</p> <pre><code class="language-rust ignore">#[link(name = "crypto")] extern { } </code></pre> <p>The type of a function declared in an extern block is <code>extern "abi" fn(A1, ..., An) -> R</code>, where <code>A1...An</code> are the declared types of its arguments and <code>R</code> is the declared return type.</p> <p>It is valid to add the <code>link</code> attribute on an empty extern block. You can use this to satisfy the linking requirements of extern blocks elsewhere in your code (including upstream crates) instead of adding the attribute to each extern block.</p> <a class="header" href="print.html#visibility-and-privacy" id="visibility-and-privacy"><h1>Visibility and Privacy</h1></a> <blockquote> <p><strong><sup>Syntax<sup></strong><br /> <em>Visibility</em> :<br /> EMPTY<br /> | <code>pub</code><br /> | <code>pub</code> <code>(</code> <code>crate</code> <code>)</code><br /> | <code>pub</code> <code>(</code> <code>in</code> <em>ModulePath</em> <code>)</code><br /> | <code>pub</code> <code>(</code> <code>in</code><sup>?</sup> <code>self</code> <code>)</code><br /> | <code>pub</code> <code>(</code> <code>in</code><sup>?</sup> <code>super</code> <code>)</code></p> </blockquote> <p>These two terms are often used interchangeably, and what they are attempting to convey is the answer to the question "Can this item be used at this location?"</p> <p>Rust's name resolution operates on a global hierarchy of namespaces. Each level in the hierarchy can be thought of as some item. The items are one of those mentioned above, but also include external crates. Declaring or defining a new module can be thought of as inserting a new tree into the hierarchy at the location of the definition.</p> <p>To control whether interfaces can be used across modules, Rust checks each use of an item to see whether it should be allowed or not. This is where privacy warnings are generated, or otherwise "you used a private item of another module and weren't allowed to."</p> <p>By default, everything in Rust is <em>private</em>, with two exceptions: Associated items in a <code>pub</code> Trait are public by default; Enum variants in a <code>pub</code> enum are also public by default. When an item is declared as <code>pub</code>, it can be thought of as being accessible to the outside world. For example:</p> <pre><pre class="playpen"><code class="language-rust"># fn main() {} // Declare a private struct struct Foo; // Declare a public struct with a private field pub struct Bar { field: i32, } // Declare a public enum with two public variants pub enum State { PubliclyAccessibleState, PubliclyAccessibleState2, } </code></pre></pre> <p>With the notion of an item being either public or private, Rust allows item accesses in two cases:</p> <ol> <li>If an item is public, then it can be accessed externally from some module <code>m</code> if you can access all the item's parent modules from <code>m</code>. You can also potentially be able to name the item through re-exports. See below.</li> <li>If an item is private, it may be accessed by the current module and its descendants.</li> </ol> <p>These two cases are surprisingly powerful for creating module hierarchies exposing public APIs while hiding internal implementation details. To help explain, here's a few use cases and what they would entail:</p> <ul> <li> <p>A library developer needs to expose functionality to crates which link against their library. As a consequence of the first case, this means that anything which is usable externally must be <code>pub</code> from the root down to the destination item. Any private item in the chain will disallow external accesses.</p> </li> <li> <p>A crate needs a global available "helper module" to itself, but it doesn't want to expose the helper module as a public API. To accomplish this, the root of the crate's hierarchy would have a private module which then internally has a "public API". Because the entire crate is a descendant of the root, then the entire local crate can access this private module through the second case.</p> </li> <li> <p>When writing unit tests for a module, it's often a common idiom to have an immediate child of the module to-be-tested named <code>mod test</code>. This module could access any items of the parent module through the second case, meaning that internal implementation details could also be seamlessly tested from the child module.</p> </li> </ul> <p>In the second case, it mentions that a private item "can be accessed" by the current module and its descendants, but the exact meaning of accessing an item depends on what the item is. Accessing a module, for example, would mean looking inside of it (to import more items). On the other hand, accessing a function would mean that it is invoked. Additionally, path expressions and import statements are considered to access an item in the sense that the import/expression is only valid if the destination is in the current visibility scope.</p> <p>Here's an example of a program which exemplifies the three cases outlined above:</p> <pre><pre class="playpen"><code class="language-rust">// This module is private, meaning that no external crate can access this // module. Because it is private at the root of this current crate, however, any // module in the crate may access any publicly visible item in this module. mod crate_helper_module { // This function can be used by anything in the current crate pub fn crate_helper() {} // This function *cannot* be used by anything else in the crate. It is not // publicly visible outside of the `crate_helper_module`, so only this // current module and its descendants may access it. fn implementation_detail() {} } // This function is "public to the root" meaning that it's available to external // crates linking against this one. pub fn public_api() {} // Similarly to 'public_api', this module is public so external crates may look // inside of it. pub mod submodule { use crate_helper_module; pub fn my_method() { // Any item in the local crate may invoke the helper module's public // interface through a combination of the two rules above. crate_helper_module::crate_helper(); } // This function is hidden to any module which is not a descendant of // `submodule` fn my_implementation() {} #[cfg(test)] mod test { #[test] fn test_my_implementation() { // Because this module is a descendant of `submodule`, it's allowed // to access private items inside of `submodule` without a privacy // violation. super::my_implementation(); } } } # fn main() {} </code></pre></pre> <p>For a Rust program to pass the privacy checking pass, all paths must be valid accesses given the two rules above. This includes all use statements, expressions, types, etc.</p> <a class="header" href="print.html#pubin-path-pubcrate-pubsuper-and-pubself" id="pubin-path-pubcrate-pubsuper-and-pubself"><h2><code>pub(in path)</code>, <code>pub(crate)</code>, <code>pub(super)</code>, and <code>pub(self)</code></h2></a> <p>In addition to public and private, Rust allows users to declare an item as visible within a given scope. The rules for <code>pub</code> restrictions are as follows:</p> <ul> <li><code>pub(in path)</code> makes an item visible within the provided <code>path</code>. <code>path</code> must be a parent module of the item whose visibility is being declared.</li> <li><code>pub(crate)</code> makes an item visible within the current crate.</li> <li><code>pub(super)</code> makes an item visible to the parent module. This equivalent to <code>pub(in super)</code>.</li> <li><code>pub(self)</code> makes an item visible to the current module. This is equivalent to <code>pub(in self)</code>.</li> </ul> <p>Here's an example:</p> <pre><pre class="playpen"><code class="language-rust">pub mod outer_mod { pub mod inner_mod { // This function is visible within `outer_mod` pub(in outer_mod) fn outer_mod_visible_fn() {} // This function is visible to the entire crate pub(crate) fn crate_visible_fn() {} // This function is visible within `outer_mod` pub(super) fn super_mod_visible_fn() { // This function is visible since we're in the same `mod` inner_mod_visible_fn(); } // This function is visible pub(self) fn inner_mod_visible_fn() {} } pub fn foo() { inner_mod::outer_mod_visible_fn(); inner_mod::crate_visible_fn(); inner_mod::super_mod_visible_fn(); // This function is no longer visible since we're outside of `inner_mod` // Error! `inner_mod_visible_fn` is private //inner_mod::inner_mod_visible_fn(); } } fn bar() { // This function is still visible since we're in the same crate outer_mod::inner_mod::crate_visible_fn(); // This function is no longer visible since we're outside of `outer_mod` // Error! `super_mod_visible_fn` is private //outer_mod::inner_mod::super_mod_visible_fn(); // This function is no longer visible since we're outside of `outer_mod` // Error! `outer_mod_visible_fn` is private //outer_mod::inner_mod::outer_mod_visible_fn(); outer_mod::foo(); } fn main() { bar() } </code></pre></pre> <a class="header" href="print.html#re-exporting-and-visibility" id="re-exporting-and-visibility"><h2>Re-exporting and Visibility</h2></a> <p>Rust allows publicly re-exporting items through a <code>pub use</code> directive. Because this is a public directive, this allows the item to be used in the current module through the rules above. It essentially allows public access into the re-exported item. For example, this program is valid:</p> <pre><pre class="playpen"><code class="language-rust">pub use self::implementation::api; mod implementation { pub mod api { pub fn f() {} } } # fn main() {} </code></pre></pre> <p>This means that any external crate referencing <code>implementation::api::f</code> would receive a privacy violation, while the path <code>api::f</code> would be allowed.</p> <p>When re-exporting a private item, it can be thought of as allowing the "privacy chain" being short-circuited through the reexport instead of passing through the namespace hierarchy as it normally would.</p> <a class="header" href="print.html#attributes" id="attributes"><h1>Attributes</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>Attribute</em> :<br /> <em>InnerAttribute</em> | <em>OuterAttribute</em></p> <p><em>InnerAttribute</em> :<br /> <code>#![</code> MetaItem <code>]</code></p> <p><em>OuterAttribute</em> :<br /> <code>#[</code> MetaItem <code>]</code></p> <p><em>MetaItem</em> :<br /> IDENTIFIER<br /> | IDENTIFIER <code>=</code> LITERAL<br /> | IDENTIFIER <code>(</code> LITERAL <code>)</code><br /> | IDENTIFIER <code>(</code> <em>MetaSeq</em> <code>)</code><br /> | IDENTIFIER <code>(</code> <em>MetaSeq</em> <code>,</code> <code>)</code></p> <p><em>MetaSeq</em> :<br /> EMPTY<br /> | <em>MetaItem</em><br /> | <em>MetaSeq</em> <code>,</code> <em>MetaItem</em></p> </blockquote> <p>Any item declaration may have an <em>attribute</em> applied to it. Attributes in Rust are modeled on Attributes in ECMA-335, with the syntax coming from ECMA-334 (C#). An attribute is a general, free-form metadatum that is interpreted according to name, convention, and language and compiler version. Attributes may appear as any of:</p> <ul> <li>A single identifier, the attribute name</li> <li>An identifier followed by the equals sign '=' and a literal, providing a key/value pair</li> <li>An identifier followed by a parenthesized literal, providing a key/value pair</li> <li>An identifier followed by a parenthesized list of sub-attribute arguments</li> </ul> <p>Attributes with a bang ("!") after the hash ("#") apply to the item that the attribute is declared within. Attributes that do not have a bang after the hash apply to the item that follows the attribute.</p> <p>An example of attributes:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { // General metadata applied to the enclosing module or crate. #![crate_type = "lib"] // A function marked as a unit test #[test] fn test_foo() { /* ... */ } // A conditionally-compiled module #[cfg(target_os = "linux")] mod bar { /* ... */ } // A lint attribute used to suppress a warning/error #[allow(non_camel_case_types)] type int8_t = i8; #}</code></pre></pre> <a class="header" href="print.html#crate-only-attributes" id="crate-only-attributes"><h2>Crate-only attributes</h2></a> <ul> <li><code>crate_name</code> - specify the crate's crate name.</li> <li><code>crate_type</code> - see <a href="linkage.html">linkage</a>.</li> <li><code>no_builtins</code> - disable optimizing certain code patterns to invocations of library functions that are assumed to exist</li> <li><code>no_main</code> - disable emitting the <code>main</code> symbol. Useful when some other object being linked to defines <code>main</code>.</li> <li><code>no_start</code> - disable linking to the <code>native</code> crate, which specifies the "start" language item.</li> <li><code>no_std</code> - disable linking to the <code>std</code> crate.</li> <li><code>recursion_limit</code> - Sets the maximum depth for potentially infinitely-recursive compile-time operations like auto-dereference or macro expansion. The default is <code>#![recursion_limit="64"]</code>.</li> <li><code>windows_subsystem</code> - Indicates that when this crate is linked for a Windows target it will configure the resulting binary's <a href="https://msdn.microsoft.com/en-us/library/fcc1zstk.aspx">subsystem</a> via the linker. Valid values for this attribute are <code>console</code> and <code>windows</code>, corresponding to those two respective subsystems. More subsystems may be allowed in the future, and this attribute is ignored on non-Windows targets.</li> </ul> <a class="header" href="print.html#module-only-attributes" id="module-only-attributes"><h2>Module-only attributes</h2></a> <ul> <li><code>no_implicit_prelude</code> - disable injecting <code>use std::prelude::*</code> in this module.</li> <li><code>path</code> - specifies the file to load the module from. <code>#[path="foo.rs"] mod bar;</code> is equivalent to <code>mod bar { /* contents of foo.rs */ }</code>. The path is taken relative to the directory that the current module is in.</li> </ul> <a class="header" href="print.html#function-only-attributes" id="function-only-attributes"><h2>Function-only attributes</h2></a> <ul> <li><code>main</code> - indicates that this function should be passed to the entry point, rather than the function in the crate root named <code>main</code>.</li> <li><code>test</code> - indicates that this function is a test function, to only be compiled in case of <code>--test</code>. <ul> <li><code>ignore</code> - indicates that this test function is disabled.</li> </ul> </li> <li><code>should_panic</code> - indicates that this test function should panic, inverting the success condition.</li> <li><code>cold</code> - The function is unlikely to be executed, so optimize it (and calls to it) differently.</li> </ul> <a class="header" href="print.html#ffi-attributes" id="ffi-attributes"><h2>FFI attributes</h2></a> <p>On an <code>extern</code> block, the following attributes are interpreted:</p> <ul> <li><code>link_args</code> - specify arguments to the linker, rather than just the library name and type. This is feature gated and the exact behavior is implementation-defined (due to variety of linker invocation syntax).</li> <li><code>link</code> - indicate that a native library should be linked to for the declarations in this block to be linked correctly. <code>link</code> supports an optional <code>kind</code> key with three possible values: <code>dylib</code>, <code>static</code>, and <code>framework</code>. See <a href="items/external-blocks.html">external blocks</a> for more about external blocks. Two examples: <code>#[link(name = "readline")]</code> and <code>#[link(name = "CoreFoundation", kind = "framework")]</code>.</li> <li><code>linked_from</code> - indicates what native library this block of FFI items is coming from. This attribute is of the form <code>#[linked_from = "foo"]</code> where <code>foo</code> is the name of a library in either <code>#[link]</code> or a <code>-l</code> flag. This attribute is currently required to export symbols from a Rust dynamic library on Windows, and it is feature gated behind the <code>linked_from</code> feature.</li> </ul> <p>On declarations inside an <code>extern</code> block, the following attributes are interpreted:</p> <ul> <li><code>link_name</code> - the name of the symbol that this function or static should be imported as.</li> <li><code>linkage</code> - on a static, this specifies the <a href="http://llvm.org/docs/LangRef.html#linkage-types">linkage type</a>.</li> </ul> <p>See <a href="type-layout.html">type layout</a> for documentation on the <code>repr</code> attribute which can be used to control type layout.</p> <a class="header" href="print.html#macro-related-attributes" id="macro-related-attributes"><h2>Macro-related attributes</h2></a> <ul> <li> <p><code>macro_use</code> on a <code>mod</code> — macros defined in this module will be visible in the module's parent, after this module has been included.</p> </li> <li> <p><code>macro_use</code> on an <code>extern crate</code> — load macros from this crate. An optional list of names <code>#[macro_use(foo, bar)]</code> restricts the import to just those macros named. The <code>extern crate</code> must appear at the crate root, not inside <code>mod</code>, which ensures proper function of the <a href="../book/first-edition/macros.html#the-variable-crate"><code>$crate</code> macro variable</a>.</p> </li> <li> <p><code>macro_reexport</code> on an <code>extern crate</code> — re-export the named macros.</p> </li> <li> <p><code>macro_export</code> - export a macro for cross-crate usage.</p> </li> <li> <p><code>no_link</code> on an <code>extern crate</code> — even if we load this crate for macros, don't link it into the output.</p> </li> </ul> <p>See the <a href="../book/first-edition/macros.html#scoping-and-macro-importexport">macros section of the book</a> for more information on macro scope.</p> <a class="header" href="print.html#miscellaneous-attributes" id="miscellaneous-attributes"><h2>Miscellaneous attributes</h2></a> <ul> <li><code>export_name</code> - on statics and functions, this determines the name of the exported symbol.</li> <li><code>link_section</code> - on statics and functions, this specifies the section of the object file that this item's contents will be placed into.</li> <li><code>no_mangle</code> - on any item, do not apply the standard name mangling. Set the symbol for this item to its identifier.</li> <li><code>must_use</code> - on structs and enums, will warn if a value of this type isn't used or assigned to a variable. You may also include an optional message by using <code>#[must_use = "message"]</code> which will be given alongside the warning.</li> </ul> <a class="header" href="print.html#deprecation" id="deprecation"><h3>Deprecation</h3></a> <p>The <code>deprecated</code> attribute marks an item as deprecated. It has two optional fields, <code>since</code> and <code>note</code>.</p> <ul> <li><code>since</code> expects a version number, as in <code>#[deprecated(since = "1.4.1")]</code> <ul> <li><code>rustc</code> doesn't know anything about versions, but external tools like <code>clippy</code> may check the validity of this field.</li> </ul> </li> <li><code>note</code> is a free text field, allowing you to provide an explanation about the deprecation and preferred alternatives.</li> </ul> <p>Only <a href="visibility-and-privacy.html">public items</a> can be given the <code>#[deprecated]</code> attribute. <code>#[deprecated]</code> on a module is inherited by all child items of that module.</p> <p><code>rustc</code> will issue warnings on usage of <code>#[deprecated]</code> items. <code>rustdoc</code> will show item deprecation, including the <code>since</code> version and <code>note</code>, if available.</p> <p>Here's an example.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { #[deprecated(since = "5.2", note = "foo was rarely used. Users should instead use bar")] pub fn foo() {} pub fn bar() {} #}</code></pre></pre> <p>The <a href="https://github.com/rust-lang/rfcs/blob/master/text/1270-deprecation.md">RFC</a> contains motivations and more details.</p> <a class="header" href="print.html#documentation" id="documentation"><h3>Documentation</h3></a> <p>The <code>doc</code> attribute is used to document items and fields. <a href="comments.html#doc-comments">Doc comments</a> are transformed into <code>doc</code> attributes.</p> <p>See <a href="../rustdoc/the-doc-attribute.html">The Rustdoc Book</a> for reference material on this attribute.</p> <a class="header" href="print.html#conditional-compilation" id="conditional-compilation"><h3>Conditional compilation</h3></a> <p>Sometimes one wants to have different compiler outputs from the same code, depending on build target, such as targeted operating system, or to enable release builds.</p> <p>Configuration options are boolean (on or off) and are named either with a single identifier (e.g. <code>foo</code>) or an identifier and a string (e.g. <code>foo = "bar"</code>; the quotes are required and spaces around the <code>=</code> are unimportant). Note that similarly-named options, such as <code>foo</code>, <code>foo="bar"</code> and <code>foo="baz"</code> may each be set or unset independently.</p> <p>Configuration options are either provided by the compiler or passed in on the command line using <code>--cfg</code> (e.g. <code>rustc main.rs --cfg foo --cfg 'bar="baz"'</code>). Rust code then checks for their presence using the <code>#[cfg(...)]</code> attribute:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { // The function is only included in the build when compiling for macOS #[cfg(target_os = "macos")] fn macos_only() { // ... } // This function is only included when either foo or bar is defined #[cfg(any(foo, bar))] fn needs_foo_or_bar() { // ... } // This function is only included when compiling for a unixish OS with a 32-bit // architecture #[cfg(all(unix, target_pointer_width = "32"))] fn on_32bit_unix() { // ... } // This function is only included when foo is not defined #[cfg(not(foo))] fn needs_not_foo() { // ... } #}</code></pre></pre> <p>This illustrates some conditional compilation can be achieved using the <code>#[cfg(...)]</code> attribute. <code>any</code>, <code>all</code> and <code>not</code> can be used to assemble arbitrarily complex configurations through nesting.</p> <p>The following configurations must be defined by the implementation:</p> <ul> <li><code>target_arch = "..."</code> - Target CPU architecture, such as <code>"x86"</code>, <code>"x86_64"</code> <code>"mips"</code>, <code>"powerpc"</code>, <code>"powerpc64"</code>, <code>"arm"</code>, or <code>"aarch64"</code>. This value is closely related to the first element of the platform target triple, though it is not identical.</li> <li><code>target_os = "..."</code> - Operating system of the target, examples include <code>"windows"</code>, <code>"macos"</code>, <code>"ios"</code>, <code>"linux"</code>, <code>"android"</code>, <code>"freebsd"</code>, <code>"dragonfly"</code>, <code>"bitrig"</code> , <code>"openbsd"</code> or <code>"netbsd"</code>. This value is closely related to the second and third element of the platform target triple, though it is not identical.</li> <li><code>target_family = "..."</code> - Operating system family of the target, e. g. <code>"unix"</code> or <code>"windows"</code>. The value of this configuration option is defined as a configuration itself, like <code>unix</code> or <code>windows</code>.</li> <li><code>unix</code> - See <code>target_family</code>.</li> <li><code>windows</code> - See <code>target_family</code>.</li> <li><code>target_env = ".."</code> - Further disambiguates the target platform with information about the ABI/libc. Presently this value is either <code>"gnu"</code>, <code>"msvc"</code>, <code>"musl"</code>, or the empty string. For historical reasons this value has only been defined as non-empty when needed for disambiguation. Thus on many GNU platforms this value will be empty. This value is closely related to the fourth element of the platform target triple, though it is not identical. For example, embedded ABIs such as <code>gnueabihf</code> will simply define <code>target_env</code> as <code>"gnu"</code>.</li> <li><code>target_endian = "..."</code> - Endianness of the target CPU, either <code>"little"</code> or <code>"big"</code>.</li> <li><code>target_pointer_width = "..."</code> - Target pointer width in bits. This is set to <code>"32"</code> for targets with 32-bit pointers, and likewise set to <code>"64"</code> for 64-bit pointers.</li> <li><code>target_has_atomic = "..."</code> - Set of integer sizes on which the target can perform atomic operations. Values are <code>"8"</code>, <code>"16"</code>, <code>"32"</code>, <code>"64"</code> and <code>"ptr"</code>.</li> <li><code>target_vendor = "..."</code> - Vendor of the target, for example <code>apple</code>, <code>pc</code>, or simply <code>"unknown"</code>.</li> <li><code>test</code> - Enabled when compiling the test harness (using the <code>--test</code> flag).</li> <li><code>debug_assertions</code> - Enabled by default when compiling without optimizations. This can be used to enable extra debugging code in development but not in production. For example, it controls the behavior of the standard library's <code>debug_assert!</code> macro.</li> </ul> <p>You can also set another attribute based on a <code>cfg</code> variable with <code>cfg_attr</code>:</p> <pre><code class="language-rust ignore">#[cfg_attr(a, b)] </code></pre> <p>This is the same as <code>#[b]</code> if <code>a</code> is set by <code>cfg</code>, and nothing otherwise.</p> <p>Lastly, configuration options can be used in expressions by invoking the <code>cfg!</code> macro: <code>cfg!(a)</code> evaluates to <code>true</code> if <code>a</code> is set, and <code>false</code> otherwise.</p> <a class="header" href="print.html#lint-check-attributes" id="lint-check-attributes"><h3>Lint check attributes</h3></a> <p>A lint check names a potentially undesirable coding pattern, such as unreachable code or omitted documentation, for the static entity to which the attribute applies.</p> <p>For any lint check <code>C</code>:</p> <ul> <li><code>allow(C)</code> overrides the check for <code>C</code> so that violations will go unreported,</li> <li><code>deny(C)</code> signals an error after encountering a violation of <code>C</code>,</li> <li><code>forbid(C)</code> is the same as <code>deny(C)</code>, but also forbids changing the lint level afterwards,</li> <li><code>warn(C)</code> warns about violations of <code>C</code> but continues compilation.</li> </ul> <p>The lint checks supported by the compiler can be found via <code>rustc -W help</code>, along with their default settings. [Compiler plugins][unstable book plugin] can provide additional lint checks.</p> <pre><code class="language-rust ignore">pub mod m1 { // Missing documentation is ignored here #[allow(missing_docs)] pub fn undocumented_one() -> i32 { 1 } // Missing documentation signals a warning here #[warn(missing_docs)] pub fn undocumented_too() -> i32 { 2 } // Missing documentation signals an error here #[deny(missing_docs)] pub fn undocumented_end() -> i32 { 3 } } </code></pre> <p>This example shows how one can use <code>allow</code> and <code>warn</code> to toggle a particular check on and off:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { #[warn(missing_docs)] pub mod m2{ #[allow(missing_docs)] pub mod nested { // Missing documentation is ignored here pub fn undocumented_one() -> i32 { 1 } // Missing documentation signals a warning here, // despite the allow above. #[warn(missing_docs)] pub fn undocumented_two() -> i32 { 2 } } // Missing documentation signals a warning here pub fn undocumented_too() -> i32 { 3 } } #}</code></pre></pre> <p>This example shows how one can use <code>forbid</code> to disallow uses of <code>allow</code> for that lint check:</p> <pre><code class="language-rust ignore">#[forbid(missing_docs)] pub mod m3 { // Attempting to toggle warning signals an error here #[allow(missing_docs)] /// Returns 2. pub fn undocumented_too() -> i32 { 2 } } </code></pre> <a class="header" href="print.html#inline-attribute" id="inline-attribute"><h3>Inline attribute</h3></a> <p>The inline attribute suggests that the compiler should place a copy of the function or static in the caller, rather than generating code to call the function or access the static where it is defined.</p> <p>The compiler automatically inlines functions based on internal heuristics. Incorrectly inlining functions can actually make the program slower, so it should be used with care.</p> <p><code>#[inline]</code> and <code>#[inline(always)]</code> always cause the function to be serialized into the crate metadata to allow cross-crate inlining.</p> <p>There are three different types of inline attributes:</p> <ul> <li><code>#[inline]</code> hints the compiler to perform an inline expansion.</li> <li><code>#[inline(always)]</code> asks the compiler to always perform an inline expansion.</li> <li><code>#[inline(never)]</code> asks the compiler to never perform an inline expansion.</li> </ul> <a class="header" href="print.html#derive" id="derive"><h3><code>derive</code></h3></a> <p>The <code>derive</code> attribute allows certain traits to be automatically implemented for data structures. For example, the following will create an <code>impl</code> for the <code>PartialEq</code> and <code>Clone</code> traits for <code>Foo</code>, the type parameter <code>T</code> will be given the <code>PartialEq</code> or <code>Clone</code> constraints for the appropriate <code>impl</code>:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { #[derive(PartialEq, Clone)] struct Foo<T> { a: i32, b: T, } #}</code></pre></pre> <p>The generated <code>impl</code> for <code>PartialEq</code> is equivalent to</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # struct Foo<T> { a: i32, b: T } impl<T: PartialEq> PartialEq for Foo<T> { fn eq(&self, other: &Foo<T>) -> bool { self.a == other.a && self.b == other.b } fn ne(&self, other: &Foo<T>) -> bool { self.a != other.a || self.b != other.b } } #}</code></pre></pre> <p>You can implement <code>derive</code> for your own type through <a href="procedural-macros.html">procedural macros</a>.</p> <a class="header" href="print.html#statements-and-expressions" id="statements-and-expressions"><h1>Statements and expressions</h1></a> <p>Rust is <em>primarily</em> an expression language. This means that most forms of value-producing or effect-causing evaluation are directed by the uniform syntax category of <em>expressions</em>. Each kind of expression can typically <em>nest</em> within each other kind of expression, and rules for evaluation of expressions involve specifying both the value produced by the expression and the order in which its sub-expressions are themselves evaluated.</p> <p>In contrast, statements in Rust serve <em>mostly</em> to contain and explicitly sequence expression evaluation.</p> <a class="header" href="print.html#statements" id="statements"><h1>Statements</h1></a> <p>A <em>statement</em> is a component of a <a href="expressions/block-expr.html">block</a>, which is in turn a component of an outer <a href="expressions.html">expression</a> or <a href="items/functions.html">function</a>.</p> <p>Rust has two kinds of statement: <a href="print.html#declaration-statements">declaration statements</a> and <a href="print.html#expression-statements">expression statements</a>.</p> <a class="header" href="print.html#declaration-statements" id="declaration-statements"><h2>Declaration statements</h2></a> <p>A <em>declaration statement</em> is one that introduces one or more <em>names</em> into the enclosing statement block. The declared names may denote new variables or new <a href="items.html">items</a>.</p> <p>The two kinds of declaration statements are item declarations and <code>let</code> statements.</p> <a class="header" href="print.html#item-declarations" id="item-declarations"><h3>Item declarations</h3></a> <p>An <em>item declaration statement</em> has a syntactic form identical to an <a href="items.html">item declaration</a> within a <a href="items/modules.html">module</a>. Declaring an item within a statement block restricts its scope to the block containing the statement. The item is not given a <a href="paths.html#canonical-paths">canonical path</a> nor are any sub-items it may declare. The exception to this is that associated items defined by <a href="items/implementations.html">implementations</a> are still accessible in outer scopes as long as the item and, if applicable, trait are accessible. It is otherwise identical in meaning to declaring the item inside a module.</p> <p>There is no implicit capture of the containing function's generic parameters, parameters, and local variables. For example, <code>inner</code> may not access <code>outer_var</code>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn outer() { let outer_var = true; fn inner() { /* outer_var is not in scope here */ } inner(); } #}</code></pre></pre> <a class="header" href="print.html#let-statements" id="let-statements"><h3><code>let</code> statements</h3></a> <p>A <em><code>let</code> statement</em> introduces a new set of variables, given by a pattern. The pattern may be followed by a type annotation, and/or an initializer expression. When no type annotation is given, the compiler will infer the type, or signal an error if insufficient type information is available for definite inference. Any variables introduced by a variable declaration are visible from the point of declaration until the end of the enclosing block scope.</p> <a class="header" href="print.html#expression-statements" id="expression-statements"><h2>Expression statements</h2></a> <p>An <em>expression statement</em> is one that evaluates an <a href="expressions.html">expression</a> and ignores its result. As a rule, an expression statement's purpose is to trigger the effects of evaluating its expression.</p> <p>An expression that consists of only a <a href="expressions/block-expr.html">block expression</a> or control flow expression, if used in a context where a statement is permitted, can omit the trailing semicolon. This can cause an ambiguity between it being parsed as a standalone statement and as a part of another expression; in this case, it is parsed as a statement.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # let mut v = vec![1, 2, 3]; v.pop(); // Ignore the element returned from pop if v.is_empty() { v.push(5); } else { v.remove(0); } // Semicolon can be omitted. [1]; // Separate expression statement, not an indexing expression. #}</code></pre></pre> <p>When the trailing semicolon is omitted, the result must be type <code>()</code>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { // bad: the block's type is i32, not () // Error: expected `()` because of default return type // if true { // 1 // } // good: the block's type is i32 if true { 1 } else { 2 }; #}</code></pre></pre> <a class="header" href="print.html#expressions" id="expressions"><h1>Expressions</h1></a> <p>An expression may have two roles: it always produces a <em>value</em>, and it may have <em>effects</em> (otherwise known as "side effects"). An expression <em>evaluates to</em> a value, and has effects during <em>evaluation</em>. Many expressions contain sub-expressions (operands). The meaning of each kind of expression dictates several things:</p> <ul> <li>Whether or not to evaluate the sub-expressions when evaluating the expression</li> <li>The order in which to evaluate the sub-expressions</li> <li>How to combine the sub-expressions' values to obtain the value of the expression</li> </ul> <p>In this way, the structure of expressions dictates the structure of execution. Blocks are just another kind of expression, so blocks, statements, expressions, and blocks again can recursively nest inside each other to an arbitrary depth.</p> <a class="header" href="print.html#expression-precedence" id="expression-precedence"><h2>Expression precedence</h2></a> <p>The precedence of Rust operators and expressions is ordered as follows, going from strong to weak. Binary Operators at the same precedence level are evaluated in the order given by their associativity.</p> <table><thead><tr><th> Operator/Expression </th><th> Associativity </th></tr></thead><tbody> <tr><td> Paths </td><td> </td></tr> <tr><td> Method calls </td><td> </td></tr> <tr><td> Field expressions </td><td> left to right </td></tr> <tr><td> Function calls, array indexing </td><td> </td></tr> <tr><td> <code>?</code> </td><td> </td></tr> <tr><td> Unary <code>-</code> <code>*</code> <code>!</code> <code>&</code> <code>&mut</code> </td><td> </td></tr> <tr><td> <code>as</code> <code>:</code> </td><td> left to right </td></tr> <tr><td> <code>*</code> <code>/</code> <code>%</code> </td><td> left to right </td></tr> <tr><td> <code>+</code> <code>-</code> </td><td> left to right </td></tr> <tr><td> <code><<</code> <code>>></code> </td><td> left to right </td></tr> <tr><td> <code>&</code> </td><td> left to right </td></tr> <tr><td> <code>^</code> </td><td> left to right </td></tr> <tr><td> <code>|</code> </td><td> left to right </td></tr> <tr><td> <code>==</code> <code>!=</code> <code><</code> <code>></code> <code><=</code> <code>>=</code> </td><td> Require parentheses </td></tr> <tr><td> <code>&&</code> </td><td> left to right </td></tr> <tr><td> <code>||</code> </td><td> left to right </td></tr> <tr><td> <code>..</code> <code>...</code> </td><td> Require parentheses </td></tr> <tr><td> <code><-</code> </td><td> right to left </td></tr> <tr><td> <code>=</code> <code>+=</code> <code>-=</code> <code>*=</code> <code>/=</code> <code>%=</code> <br> <code>&=</code> <code>|=</code> <code>^=</code> <code><<=</code> <code>>>=</code> </td><td> right to left </td></tr> <tr><td> <code>return</code> <code>break</code> closures </td><td> </td></tr> </tbody></table> <a class="header" href="print.html#place-expressions-and-value-expressions" id="place-expressions-and-value-expressions"><h2>Place Expressions and Value Expressions</h2></a> <p>Expressions are divided into two main categories: place expressions and value expressions. Likewise within each expression, sub-expressions may occur in either place context or value context. The evaluation of an expression depends both on its own category and the context it occurs within.</p> <p>A <em>place expression</em> is an expression that represents a memory location. These expressions are <a href="expressions/path-expr.html">paths</a> which refer to local variables, <a href="items/static-items.html">static variables</a>, <a href="expressions/operator-expr.html#the-dereference-operator">dereferences</a> (<code>*expr</code>), <a href="expressions/array-expr.html#array-and-slice-indexing-expressions">array indexing</a> expressions (<code>expr[expr]</code>), <a href="expressions/field-expr.html">field</a> references (<code>expr.f</code>) and parenthesized place expressions. All other expressions are value expressions.</p> <p>A <em>value expression</em> is an expression that represents an actual value.</p> <p>The left operand of an <a href="expressions/operator-expr.html#assignment-expressions">assignment</a> or <a href="expressions/operator-expr.html#compound-assignment-expressions">compound assignment</a> expression is a place expression context, as is the single operand of a unary <a href="expressions/operator-expr.html#borrow-operators">borrow</a>, and the operand of any <a href="print.html#implicit-borrows">implicit borrow</a>. The discriminant or subject of a <a href="expressions/match-expr.html">match expression</a> and right side of a <a href="statements.html#let-statements">let statement</a> is also a place expression context. All other expression contexts are value expression contexts.</p> <blockquote> <p>Note: Historically, place expressions were called <em>lvalues</em> and value expressions were called <em>rvalues</em>.</p> </blockquote> <a class="header" href="print.html#moved-and-copied-types" id="moved-and-copied-types"><h3>Moved and copied types</h3></a> <p>When a place expression is evaluated in a value expression context, or is bound by value in a pattern, it denotes the value held <em>in</em> that memory location. If the type of that value implements <a href="special-types-and-traits.html#copy"><code>Copy</code></a>, then the value will be copied. In the remaining situations if that type is <a href="special-types-and-traits.html#sized"><code>Sized</code></a>, then it may be possible to move the value. Only the following place expressions may be moved out of:</p> <ul> <li><a href="variables.html">Variables</a> which are not currently borrowed.</li> <li><a href="print.html#temporary-lifetimes">Temporary values</a>.</li> <li><a href="expressions/field-expr.html">Fields</a> of a place expression which can be moved out of and doesn't implement <a href="special-types-and-traits.html#drop"><code>Drop</code></a>.</li> <li>The result of <a href="expressions/operator-expr.html#the-dereference-operator">dereferencing</a> an expression with type <a href="../std/boxed/struct.Box.html"><code>Box<T></code></a> and that can also be moved out of.</li> </ul> <p>Moving out of a place expression that evaluates to a local variable, the location is deinitialized and cannot be read from again until it is reinitialized. In all other cases, trying to use a place expression in a value expression context is an error.</p> <a class="header" href="print.html#mutability" id="mutability"><h3>Mutability</h3></a> <p>For a place expression to be <a href="expressions/operator-expr.html#assignment-expressions">assigned</a> to, mutably <a href="expressions/operator-expr.html#borrow-operators">borrowed</a>, <a href="print.html#implicit-borrows">implicitly mutably borrowed</a>, or bound to a pattern containing <code>ref mut</code> it must be <em>mutable</em>. We call these <em>mutable place expressions</em>. In contrast, other place expressions are called <em>immutable place expressions</em>.</p> <p>The following expressions can be mutable place expression contexts:</p> <ul> <li>Mutable <a href="variables.html">variables</a>, which are not currently borrowed.</li> <li><a href="items/static-items.html#mutable-statics">Mutable <code>static</code> items</a>.</li> <li><a href="print.html#temporary-lifetimes">Temporary values</a>.</li> <li><a href="expressions/field-expr.html">Fields</a>, this evaluates the subexpression in a mutable place expression context.</li> <li><a href="expressions/operator-expr.html#the-dereference-operator">Dereferences</a> of a <code>*mut T</code> pointer.</li> <li>Dereference of a variable, or field of a variable, with type <code>&mut T</code>. Note: This is an exception to the requirement of the next rule.</li> <li>Dereferences of a type that implements <code>DerefMut</code>, this then requires that the value being dereferenced is evaluated is a mutable place expression context.</li> <li><a href="expressions/array-expr.html#array-and-slice-indexing-expressions">Array indexing</a> of a type that implements <code>DerefMut</code>, this then evaluates the value being indexed, but not the index, in mutable place expression context.</li> </ul> <a class="header" href="print.html#temporary-lifetimes" id="temporary-lifetimes"><h3>Temporary lifetimes</h3></a> <p>When using a value expression in most place expression contexts, a temporary unnamed memory location is created initialized to that value and the expression evaluates to that location instead, except if promoted to <code>'static</code>. Promotion of a value expression to a <code>'static</code> slot occurs when the expression could be written in a constant, borrowed, and dereferencing that borrow where the expression was the originally written, without changing the runtime behavior. That is, the promoted expression can be evaluated at compile-time and the resulting value does not contain <a href="interior-mutability.html">interior mutability</a> or <a href="destructors.html">destructors</a> (these properties are determined based on the value where possible, e.g. <code>&None</code> always has the type <code>&'static Option<_></code>, as it contains nothing disallowed). Otherwise, the lifetime of temporary values is typically</p> <ul> <li>the innermost enclosing statement; the tail expression of a block is considered part of the statement that encloses the block, or</li> <li>the condition expression or the loop conditional expression if the temporary is created in the condition expression of an <code>if</code> or in the loop conditional expression of a <code>while</code> expression.</li> </ul> <p>When a temporary value expression is being created that is assigned into a <a href="statements.html#let-statements"><code>let</code> declaration</a>, however, the temporary is created with the lifetime of the enclosing block instead, as using the enclosing <a href="statements.html#let-statements"><code>let</code> declaration</a> would be a guaranteed error (since a pointer to the temporary would be stored into a variable, but the temporary would be freed before the variable could be used). The compiler uses simple syntactic rules to decide which values are being assigned into a <code>let</code> binding, and therefore deserve a longer temporary lifetime.</p> <p>Here are some examples:</p> <ul> <li><code>let x = foo(&temp())</code>. The expression <code>temp()</code> is a value expression. As it is being borrowed, a temporary is created which will be freed after the innermost enclosing statement; in this case, the <code>let</code> declaration.</li> <li><code>let x = temp().foo()</code>. This is the same as the previous example, except that the value of <code>temp()</code> is being borrowed via autoref on a method-call. Here we are assuming that <code>foo()</code> is an <code>&self</code> method defined in some trait, say <code>Foo</code>. In other words, the expression <code>temp().foo()</code> is equivalent to <code>Foo::foo(&temp())</code>.</li> <li><code>let x = if foo(&temp()) {bar()} else {baz()};</code>. The expression <code>temp()</code> is a value expression. As the temporary is created in the condition expression of an <code>if</code>, it will be freed at the end of the condition expression; in this example before the call to <code>bar</code> or <code>baz</code> is made.</li> <li><code>let x = if temp().must_run_bar {bar()} else {baz()};</code>. Here we assume the type of <code>temp()</code> is a struct with a boolean field <code>must_run_bar</code>. As the previous example, the temporary corresponding to <code>temp()</code> will be freed at the end of the condition expression.</li> <li><code>while foo(&temp()) {bar();}</code>. The temporary containing the return value from the call to <code>temp()</code> is created in the loop conditional expression. Hence it will be freed at the end of the loop conditional expression; in this example before the call to <code>bar</code> if the loop body is executed.</li> <li><code>let x = &temp()</code>. Here, the same temporary is being assigned into <code>x</code>, rather than being passed as a parameter, and hence the temporary's lifetime is considered to be the enclosing block.</li> <li><code>let x = SomeStruct { foo: &temp() }</code>. As in the previous case, the temporary is assigned into a struct which is then assigned into a binding, and hence it is given the lifetime of the enclosing block.</li> <li><code>let x = [ &temp() ]</code>. As in the previous case, the temporary is assigned into an array which is then assigned into a binding, and hence it is given the lifetime of the enclosing block.</li> <li><code>let ref x = temp()</code>. In this case, the temporary is created using a ref binding, but the result is the same: the lifetime is extended to the enclosing block.</li> </ul> <a class="header" href="print.html#implicit-borrows" id="implicit-borrows"><h3>Implicit Borrows</h3></a> <p>Certain expressions will treat an expression as a place expression by implicitly borrowing it. For example, it is possible to compare two unsized [slices] for equality directly, because the <code>==</code> operator implicitly borrows it's operands:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # let c = [1, 2, 3]; # let d = vec![1, 2, 3]; let a: &[i32]; let b: &[i32]; # a = &c; # b = &d; // ... *a == *b; // Equivalent form: ::std::cmp::PartialEq::eq(&*a, &*b); #}</code></pre></pre> <p>Implicit borrows may be taken in the following expressions:</p> <ul> <li>Left operand in <a href="expressions/method-call-expr.html">method-call</a> expressions.</li> <li>Left operand in <a href="expressions/field-expr.html">field</a> expressions.</li> <li>Left operand in <a href="expressions/call-expr.html">call expressions</a>.</li> <li>Left operand in <a href="expressions/array-expr.html#array-and-slice-indexing-expressions">array indexing</a> expressions.</li> <li>Operand of the <a href="expressions/operator-expr.html#the-dereference-operator">dereference operator</a> (<code>*</code>).</li> <li>Operands of <a href="expressions/operator-expr.html#comparison-operators">comparison</a>.</li> <li>Left operands of the <a href="expressions/operator-expr.html#compound-assignment-expressions">compound assignment</a>.</li> </ul> <a class="header" href="print.html#constant-expressions" id="constant-expressions"><h2>Constant expressions</h2></a> <p>Certain types of expressions can be evaluated at compile time. These are called <em>constant expressions</em>. Certain places, such as in <a href="items/constant-items.html">constants</a> and <a href="items/static-items.html">statics</a>, require a constant expression, and are always evaluated at compile time. In other places, such as in <a href="statements.html#let-statements"><code>let</code> statements</a>, constant expressions may be evaluated at compile time. If errors, such as out of bounds <a href="expressions/array-expr.html#array-and-slice-indexing-expressions">array indexing</a> or <a href="expressions/operator-expr.html#overflow">overflow</a> occurs, then it is a compiler error if the value must be evaluated at compile time, otherwise it is just a warning, but the code will most likely panic when run.</p> <p>The following expressions are constant expressions, so long as any operands are also constant expressions and do not cause any <a href="destructors.html"><code>Drop::drop</code></a> calls to be ran.</p> <ul> <li><a href="expressions/literal-expr.html">Literals</a>.</li> <li><a href="expressions/path-expr.html">Paths</a> to <a href="items/functions.html">functions</a> and constants. Recursively defining constants is not allowed.</li> <li><a href="expressions/tuple-expr.html">Tuple expressions</a>.</li> <li><a href="expressions/array-expr.html">Array expressions</a>.</li> <li><a href="expressions/struct-expr.html">Struct</a> expressions.</li> <li><a href="expressions/enum-variant-expr.html">Enum variant</a> expressions.</li> <li><a href="expressions/block-expr.html">Block expressions</a>, including <code>unsafe</code> blocks, which only contain items and possibly a constant tail expression.</li> <li><a href="expressions/field-expr.html">Field</a> expressions.</li> <li>Index expressions, <a href="expressions/array-expr.html#array-and-slice-indexing-expressions">array indexing</a> or <a href="types.html#array-and-slice-types">slice</a> with a <code>usize</code>.</li> <li><a href="expressions/range-expr.html">Range expressions</a>.</li> <li><a href="expressions/closure-expr.html">Closure expressions</a> which don't capture variables from the environment.</li> <li>Built in <a href="expressions/operator-expr.html#negation-operators">negation</a>, <a href="expressions/operator-expr.html#arithmetic-and-logical-binary-operators">arithmetic, logical</a>, <a href="expressions/operator-expr.html#comparison-operators">comparison</a> or <a href="expressions/operator-expr.html#lazy-boolean-operators">lazy boolean</a> operators used on integer and floating point types, <code>bool</code> and <code>char</code>.</li> <li>Shared <a href="expressions/operator-expr.html#borrow-operators">borrow</a>s, except if applied to a type with <a href="interior-mutability.html">interior mutability</a>.</li> <li>The <a href="expressions/operator-expr.html#the-dereference-operator">dereference operator</a>.</li> <li><a href="expressions/operator-expr.html#grouped-expressions">Grouped</a> expressions.</li> <li><a href="expressions/operator-expr.html#type-cast-expressions">Cast</a> expressions, except pointer to address and function pointer to address casts.</li> </ul> <a class="header" href="print.html#overloading-traits" id="overloading-traits"><h2>Overloading Traits</h2></a> <p>Many of the following operators and expressions can also be overloaded for other types using traits in <code>std::ops</code> or <code>std::cmp</code>. These traits also exist in <code>core::ops</code> and <code>core::cmp</code> with the same names.</p> <a class="header" href="print.html#literal-expressions" id="literal-expressions"><h1>Literal expressions</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>LiteralExpression</em> :<br /> <a href="tokens.html#character-literals">CHAR_LITERAL</a><br /> | <a href="tokens.html#string-literals">STRING_LITERAL</a><br /> | <a href="tokens.html#raw-string-literals">RAW_STRING_LITERAL</a><br /> | <a href="tokens.html#byte-literals">BYTE_LITERAL</a><br /> | <a href="tokens.html#byte-string-literals">BYTE_STRING_LITERAL</a><br /> | <a href="tokens.html#raw-byte-string-literals">RAW_BYTE_STRING_LITERAL</a><br /> | <a href="tokens.html#integer-literals">INTEGER_LITERAL</a><br /> | <a href="tokens.html#floating-point-literals">FLOAT_LITERAL</a><br /> | <a href="tokens.html#boolean-literals">BOOLEAN_LITERAL</a></p> </blockquote> <p>A <em>literal expression</em> consists of one of the <a href="tokens.html#literals">literal</a> forms described earlier. It directly describes a number, character, string, or boolean value.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { "hello"; // string type '5'; // character type 5; // integer type #}</code></pre></pre> <a class="header" href="print.html#path-expressions" id="path-expressions"><h1>Path expressions</h1></a> <p>A <a href="paths.html">path</a> used as an expression context denotes either a local variable or an item. Path expressions that resolve to local or static variables are <a href="expressions.html#place-expressions-and-value-expressions">place expressions</a>, other paths are <a href="expressions.html#place-expressions-and-value-expressions">value expressions</a>. Using a <a href="items/static-items.html#mutable-statics"><code>static mut</code></a> variable requires an <a href="expressions/block-expr.html#unsafe-blocks"><code>unsafe</code> block</a>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # mod globals { # pub static STATIC_VAR: i32 = 5; # pub static mut STATIC_MUT_VAR: i32 = 7; # } # let local_var = 3; local_var; globals::STATIC_VAR; unsafe { globals::STATIC_MUT_VAR }; let some_constructor = Some::<i32>; let push_integer = Vec::<i32>::push; let slice_reverse = <[i32]>::reverse; #}</code></pre></pre> <a class="header" href="print.html#block-expressions" id="block-expressions"><h1>Block expressions</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>BlockExpression</em> :<br /> <code>{</code><br /> <a href="attributes.html"><em>InnerAttribute</em></a><sup>*</sup><br /> <a href="statements.html"><em>Statement</em></a><sup>*</sup><br /> <a href="expressions.html"><em>Expression</em></a><sup>?</sup><br /> <code>}</code></p> </blockquote> <p>A <em>block expression</em> is similar to a module in terms of the declarations that are possible, but can also contain <a href="statements.html">statements</a> and end with an <a href="expressions.html">expression</a>. Each block conceptually introduces a new namespace scope. Use items can bring new names into scopes and declared items are in scope for only the block itself.</p> <p>A block will execute each statement sequentially, and then execute the expression, if given. If the block doesn't end in an expression, its value is <code>()</code>:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x: () = { println!("Hello."); }; #}</code></pre></pre> <p>If it ends in an expression, its value and type are that of the expression:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x: i32 = { println!("Hello."); 5 }; assert_eq!(5, x); #}</code></pre></pre> <p>Blocks are always <a href="expressions.html#place-expressions-and-value-expressions">value expressions</a> and evaluate the last expression in value expression context. This can be used to force moving a value if really needed.</p> <a class="header" href="print.html#unsafe-blocks" id="unsafe-blocks"><h2><code>unsafe</code> blocks</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>UnsafeBlockExpression</em> :<br /> <code>unsafe</code> <em>BlockExpression</em></p> </blockquote> <p><em>See <a href="unsafe-blocks.html"><code>unsafe</code> block</a> for more information on when to use <code>unsafe</code></em></p> <p>A block of code can be prefixed with the <code>unsafe</code> keyword, to permit calling <code>unsafe</code> functions or dereferencing raw pointers within a safe function. Examples:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { unsafe { let b = [13u8, 17u8]; let a = &b[0] as *const u8; assert_eq!(*a, 13); assert_eq!(*a.offset(1), 17); } # unsafe fn f() -> i32 { 10 } let a = unsafe { f() }; #}</code></pre></pre> <a class="header" href="print.html#operator-expressions" id="operator-expressions"><h1>Operator expressions</h1></a> <p>Operators are defined for built in types by the Rust language. Many of the following operators can also be overloaded using traits in <code>std::ops</code> or <code>std::cmp</code>.</p> <a class="header" href="print.html#overflow" id="overflow"><h2>Overflow</h2></a> <p>Integer operators will panic when they overflow when compiled in debug mode. The <code>-C debug-assertions</code> and <code>-C overflow-checks</code> compiler flags can be used to control this more directly. The following things are considered to be overflow:</p> <ul> <li>When <code>+</code>, <code>*</code> or <code>-</code> create a value greater than the maximum value, or less than the minimum value that can be stored. This includes unary <code>-</code> on the smallest value of any signed integer type.</li> <li>Using <code>/</code> or <code>%</code>, where the left-hand argument is the smallest integer of a signed integer type and the right-hand argument is <code>-1</code>.</li> <li>Using <code><<</code> or <code>>></code> where the right-hand argument is greater than or equal to the number of bits in the type of the left-hand argument, or is negative.</li> </ul> <a class="header" href="print.html#grouped-expressions" id="grouped-expressions"><h2>Grouped expressions</h2></a> <p>An expression enclosed in parentheses evaluates to the result of the enclosed expression. Parentheses can be used to explicitly specify evaluation order within an expression.</p> <p>This operator cannot be overloaded.</p> <p>An example of a parenthesized expression:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x: i32 = 2 + 3 * 4; let y: i32 = (2 + 3) * 4; assert_eq!(x, 14); assert_eq!(y, 20); #}</code></pre></pre> <a class="header" href="print.html#borrow-operators" id="borrow-operators"><h2>Borrow operators</h2></a> <p>The <code>&</code> (shared borrow) and <code>&mut</code> (mutable borrow) operators are unary prefix operators. When applied to a <a href="expressions.html#place-expressions-and-value-expressions">place expression</a>, this expressions produces a reference (pointer) to the location that the value refers to. The memory location is also placed into a borrowed state for the duration of the reference. For a shared borrow (<code>&</code>), this implies that the place may not be mutated, but it may be read or shared again. For a mutable borrow (<code>&mut</code>), the place may not be accessed in any way until the borrow expires. <code>&mut</code> evaluates its operand in a mutable place expression context. If the <code>&</code> or <code>&mut</code> operators are applied to a <a href="expressions.html#place-expressions-and-value-expressions">value expression</a>, then a <a href="expressions.html#temporary-lifetimes">temporary value</a> is created.</p> <p>These operators cannot be overloaded.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { { // a temporary with value 7 is created that lasts for this scope. let shared_reference = &7; } let mut array = [-2, 3, 9]; { // Mutably borrows `array` for this scope. // `array` may only be used through `mutable_reference`. let mutable_reference = &mut array; } #}</code></pre></pre> <a class="header" href="print.html#the-dereference-operator" id="the-dereference-operator"><h2>The dereference operator</h2></a> <p>The <code>*</code> (dereference) operator is also a unary prefix operator. When applied to a <a href="types.html#pointer-types">pointer</a> it denotes the pointed-to location. If the expression is of type <code>&mut T</code> and <code>*mut T</code>, and is either a local variable, a (nested) field of a local variance or is a mutable <a href="expressions.html#place-expressions-and-value-expressions">place expression</a>, then the resulting memory location can be assigned to. Dereferencing a raw pointer requires <code>unsafe</code>.</p> <p>On non-pointer types <code>*x</code> is equivalent to <code>*std::ops::Deref::deref(&x)</code> in an <a href="expressions.html#mutability">immutable place expression context</a> and <code>*std::ops::Deref::deref_mut(&mut x)</code> in a mutable place expression context.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x = &7; assert_eq!(*x, 7); let y = &mut 9; *y = 11; assert_eq!(*y, 11); #}</code></pre></pre> <a class="header" href="print.html#the-question-mark-operator" id="the-question-mark-operator"><h2>The question mark operator</h2></a> <p>The question mark operator (<code>?</code>) unwraps valid values or returns errornous values, propagating them to the calling function. It is a unary postfix operator that can only be applied to the types <code>Result<T, E></code> and <code>Option<T></code>.</p> <p>When applied to values of the <code>Result<T, E></code> type, it propagates errors. If the value is <code>Err(e)</code>, then it will return <code>Err(From::from(e))</code> from the enclosing function or closure. If applied to <code>Ok(x)</code>, then it will unwrap the value to evaulate to <code>x</code>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # use std::num::ParseIntError; fn try_to_parse() -> Result<i32, ParseIntError> { let x: i32 = "123".parse()?; // x = 123 let y: i32 = "24a".parse()?; // returns an Err() immediately Ok(x + y) // Doesn't run. } let res = try_to_parse(); println!("{:?}", res); # assert!(res.is_err()) #}</code></pre></pre> <p>When applied to values of the <code>Option<T></code> type, it propagates <code>Nones</code>. If the value is <code>None</code>, then it will return <code>None</code>. If applied to <code>Some(x)</code>, then it will unwrap the value to evaluate to <code>x</code>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn try_option_some() -> Option<u8> { let val = Some(1)?; Some(val) } assert_eq!(try_option_some(), Some(1)); fn try_option_none() -> Option<u8> { let val = None?; Some(val) } assert_eq!(try_option_none(), None); #}</code></pre></pre> <p><code>?</code> cannot be overloaded.</p> <a class="header" href="print.html#negation-operators" id="negation-operators"><h2>Negation operators</h2></a> <p>These are the last two unary operators. This table summarizes the behavior of them on primitive types and which traits are used to overload these operators for other types. Remember that signed integers are always represented using two's complement. The operands of all of these operators are evaluated in <a href="expressions.html#place-expressions-and-value-expressions">value expression context</a> so are moved or copied.</p> <table><thead><tr><th> Symbol </th><th> Integer </th><th> <code>bool</code> </th><th> Floating Point </th><th> Overloading Trait </th></tr></thead><tbody> <tr><td> <code>-</code> </td><td> Negation* </td><td> </td><td> Negation </td><td> <code>std::ops::Neg</code> </td></tr> <tr><td> <code>!</code> </td><td> Bitwise NOT </td><td> Logical NOT </td><td> </td><td> <code>std::ops::Not</code> </td></tr> </tbody></table> <p>* Only for signed integer types.</p> <p>Here are some example of these operators</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x = 6; assert_eq!(-x, -6); assert_eq!(!x, -7); assert_eq!(true, !false); #}</code></pre></pre> <a class="header" href="print.html#arithmetic-and-logical-binary-operators" id="arithmetic-and-logical-binary-operators"><h2>Arithmetic and Logical Binary Operators</h2></a> <p>Binary operators expressions are all written with infix notation. This table summarizes the behavior of arithmetic and logical binary operators on primitive types and which traits are used to overload these operators for other types. Remember that signed integers are always represented using two's complement. The operands of all of these operators are evaluated in <a href="expressions.html#place-expressions-and-value-expressions">value expression context</a> so are moved or copied.</p> <table><thead><tr><th> Symbol </th><th> Integer </th><th> <code>bool</code> </th><th> Floating Point </th><th> Overloading Trait </th></tr></thead><tbody> <tr><td> <code>+</code> </td><td> Addition </td><td> </td><td> Addition </td><td> <code>std::ops::Add</code> </td></tr> <tr><td> <code>-</code> </td><td> Subtraction </td><td> </td><td> Subtraction </td><td> <code>std::ops::Sub</code> </td></tr> <tr><td> <code>*</code> </td><td> Multiplication </td><td> </td><td> Multiplication </td><td> <code>std::ops::Mul</code> </td></tr> <tr><td> <code>/</code> </td><td> Division </td><td> </td><td> Division </td><td> <code>std::ops::Div</code> </td></tr> <tr><td> <code>%</code> </td><td> Remainder </td><td> </td><td> Remainder </td><td> <code>std::ops::Rem</code> </td></tr> <tr><td> <code>&</code> </td><td> Bitwise AND </td><td> Logical AND </td><td> </td><td> <code>std::ops::BitAnd</code> </td></tr> <tr><td> <code>|</code> </td><td> Bitwise OR </td><td> Logical OR </td><td> </td><td> <code>std::ops::BitOr</code> </td></tr> <tr><td> <code>^</code> </td><td> Bitwise XOR </td><td> Logical XOR </td><td> </td><td> <code>std::ops::BitXor</code> </td></tr> <tr><td> <code><<</code> </td><td> Left Shift </td><td> </td><td> </td><td> <code>std::ops::Shl</code> </td></tr> <tr><td> <code>>></code> </td><td> Right Shift* </td><td> </td><td> </td><td> <code>std::ops::Shr</code> </td></tr> </tbody></table> <p>* Arithmetic right shift on signed integer types, logical right shift on unsigned integer types.</p> <p>Here are examples of these operators being used.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { assert_eq!(3 + 6, 9); assert_eq!(5.5 - 1.25, 4.25); assert_eq!(-5 * 14, -70); assert_eq!(14 / 3, 4); assert_eq!(100 % 7, 2); assert_eq!(0b1010 & 0b1100, 0b1000); assert_eq!(0b1010 | 0b1100, 0b1110); assert_eq!(0b1010 ^ 0b1100, 0b110); assert_eq!(13 << 3, 104); assert_eq!(-10 >> 2, -3); #}</code></pre></pre> <a class="header" href="print.html#comparison-operators" id="comparison-operators"><h2>Comparison Operators</h2></a> <p>Comparison operators are also defined both for primitive types and many type in the standard library. Parentheses are required when chaining comparison operators. For example, the expression <code>a == b == c</code> is invalid and may be written as <code>(a == b) == c</code>.</p> <p>Unlike arithmetic and logical operators, the traits for overloading the operators the traits for these operators are used more generally to show how a type may be compared and will likely be assumed to define actual comparisons by functions that use these traits as bounds. Many functions and macros in the standard library can then use that assumption (although not to ensure safety). Unlike the arithmetic and logical operators above, these operators implicitly take shared borrows of their operands, evaluating them in <a href="expressions.html#place-expressions-and-value-expressions">place expression context</a>:</p> <pre><code class="language-rust ignore">a == b; // is equivalent to ::std::cmp::PartialEq::eq(&a, &b); </code></pre> <p>This means that the operands don't have to be moved out of.</p> <table><thead><tr><th> Symbol </th><th> Meaning </th><th> Overloading method </th></tr></thead><tbody> <tr><td> <code>==</code> </td><td> Equal </td><td> <code>std::cmp::PartialEq::eq</code> </td></tr> <tr><td> <code>!=</code> </td><td> Not equal </td><td> <code>std::cmp::PartialEq::ne</code> </td></tr> <tr><td> <code>></code> </td><td> Greater than </td><td> <code>std::cmp::PartialOrd::gt</code> </td></tr> <tr><td> <code><</code> </td><td> Less than </td><td> <code>std::cmp::PartialOrd::lt</code> </td></tr> <tr><td> <code>>=</code> </td><td> Greater than or equal to </td><td> <code>std::cmp::PartialOrd::ge</code> </td></tr> <tr><td> <code><=</code> </td><td> Less than or equal to </td><td> <code>std::cmp::PartialOrd::le</code> </td></tr> </tbody></table> <p>Here are examples of the comparison operators being used.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { assert!(123 == 123); assert!(23 != -12); assert!(12.5 > 12.2); assert!([1, 2, 3] < [1, 3, 4]); assert!('A' <= 'B'); assert!("World" >= "Hello"); #}</code></pre></pre> <a class="header" href="print.html#lazy-boolean-operators" id="lazy-boolean-operators"><h2>Lazy boolean operators</h2></a> <p>The operators <code>||</code> and <code>&&</code> may be applied to operands of boolean type. The <code>||</code> operator denotes logical 'or', and the <code>&&</code> operator denotes logical 'and'. They differ from <code>|</code> and <code>&</code> in that the right-hand operand is only evaluated when the left-hand operand does not already determine the result of the expression. That is, <code>||</code> only evaluates its right-hand operand when the left-hand operand evaluates to <code>false</code>, and <code>&&</code> only when it evaluates to <code>true</code>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x = false || true; // true let y = false && panic!(); // false, doesn't evaluate `panic!()` #}</code></pre></pre> <a class="header" href="print.html#type-cast-expressions" id="type-cast-expressions"><h2>Type cast expressions</h2></a> <p>A type cast expression is denoted with the binary operator <code>as</code>.</p> <p>Executing an <code>as</code> expression casts the value on the left-hand side to the type on the right-hand side.</p> <p>An example of an <code>as</code> expression:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # fn sum(values: &[f64]) -> f64 { 0.0 } # fn len(values: &[f64]) -> i32 { 0 } fn average(values: &[f64]) -> f64 { let sum: f64 = sum(values); let size: f64 = len(values) as f64; sum / size } #}</code></pre></pre> <p><code>as</code> can be used to explicitly perform <a href="type-coercions.html">coercions</a>, as well as the following additional casts. Here <code>*T</code> means either <code>*const T</code> or <code>*mut T</code>.</p> <table><thead><tr><th> Type of <code>e</code> </th><th> <code>U</code> </th><th> Cast performed by <code>e as U</code> </th></tr></thead><tbody> <tr><td> Integer or Float type </td><td> Integer or Float type </td><td> Numeric cast </td></tr> <tr><td> C-like enum </td><td> Integer type </td><td> Enum cast </td></tr> <tr><td> <code>bool</code> or <code>char</code> </td><td> Integer type </td><td> Primitive to integer cast </td></tr> <tr><td> <code>u8</code> </td><td> <code>char</code> </td><td> <code>u8</code> to <code>char</code> cast </td></tr> <tr><td> <code>*T</code> </td><td> <code>*V</code> where <code>V: Sized</code> * </td><td> Pointer to pointer cast </td></tr> <tr><td> <code>*T</code> where <code>T: Sized</code> </td><td> Numeric type </td><td> Pointer to address cast </td></tr> <tr><td> Integer type </td><td> <code>*V</code> where <code>V: Sized</code> </td><td> Address to pointer cast </td></tr> <tr><td> <code>&[T; n]</code> </td><td> <code>*const T</code> </td><td> Array to pointer cast </td></tr> <tr><td> <a href="types.html#function-pointer-types">Function pointer</a> </td><td> <code>*V</code> where <code>V: Sized</code> </td><td> Function pointer to pointer cast </td></tr> <tr><td> Function pointer </td><td> Integer </td><td> Function pointer to address cast </td></tr> </tbody></table> <p>* or <code>T</code> and <code>V</code> are compatible unsized types, e.g., both slices, both the same trait object.</p> <a class="header" href="print.html#semantics" id="semantics"><h3>Semantics</h3></a> <ul> <li>Numeric cast <ul> <li>Casting between two integers of the same size (e.g. i32 -> u32) is a no-op</li> <li>Casting from a larger integer to a smaller integer (e.g. u32 -> u8) will truncate</li> <li>Casting from a smaller integer to a larger integer (e.g. u8 -> u32) will <ul> <li>zero-extend if the source is unsigned</li> <li>sign-extend if the source is signed</li> </ul> </li> <li>Casting from a float to an integer will round the float towards zero <ul> <li><strong><a href="https://github.com/rust-lang/rust/issues/10184">NOTE: currently this will cause Undefined Behavior if the rounded value cannot be represented by the target integer type</a></strong>. This includes Inf and NaN. This is a bug and will be fixed.</li> </ul> </li> <li>Casting from an integer to float will produce the floating point representation of the integer, rounded if necessary (rounding strategy unspecified)</li> <li>Casting from an f32 to an f64 is perfect and lossless</li> <li>Casting from an f64 to an f32 will produce the closest possible value (rounding strategy unspecified) <ul> <li><strong><a href="https://github.com/rust-lang/rust/issues/15536">NOTE: currently this will cause Undefined Behavior if the value is finite but larger or smaller than the largest or smallest finite value representable by f32</a></strong>. This is a bug and will be fixed.</li> </ul> </li> </ul> </li> <li>Enum cast <ul> <li>Casts an enum to its discriminant, then uses a numeric cast if needed.</li> </ul> </li> <li>Primitive to integer cast <ul> <li><code>false</code> casts to <code>0</code>, <code>true</code> casts to <code>1</code></li> <li><code>char</code> casts to the value of the code point, then uses a numeric cast if needed.</li> </ul> </li> <li><code>u8</code> to <code>char</code> cast <ul> <li>Casts to the <code>char</code> with the corresponding code point.</li> </ul> </li> </ul> <a class="header" href="print.html#assignment-expressions" id="assignment-expressions"><h2>Assignment expressions</h2></a> <p>An <em>assignment expression</em> consists of a <a href="expressions.html#place-expressions-and-value-expressions">place expression</a> followed by an equals sign (<code>=</code>) and a <a href="expressions.html#place-expressions-and-value-expressions">value expression</a>.</p> <p>Evaluating an assignment expression <a href="destructors.html">drops</a> the left-hand operand, unless it's an unitialized local variable or field of a local variable, and <a href="expressions.html#moved-and-copied-types">either copies or moves</a> its right-hand operand to its left-hand operand. The left-hand operand must be a place expression: using a value expression results in a compiler error, rather than promoting it to a temporary.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # let mut x = 0; # let y = 0; x = y; #}</code></pre></pre> <a class="header" href="print.html#compound-assignment-expressions" id="compound-assignment-expressions"><h2>Compound assignment expressions</h2></a> <p>The <code>+</code>, <code>-</code>, <code>*</code>, <code>/</code>, <code>%</code>, <code>&</code>, <code>|</code>, <code>^</code>, <code><<</code>, and <code>>></code> operators may be composed with the <code>=</code> operator. The expression <code>place_exp OP= value</code> is equivalent to <code>place_expr = place_expr OP val</code>. For example, <code>x = x + 1</code> may be written as <code>x += 1</code>. Any such expression always has the <a href="types.html#tuple-types"><code>unit</code> type</a>. These operators can all be overloaded using the trait with the same name as for the normal operation followed by 'Assign', for example, <code>std::ops::AddAssign</code> is used to overload <code>+=</code>. As with <code>=</code>, <code>place_expr</code> must be a <a href="expressions.html#place-expressions-and-value-expressions">place expression</a>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let mut x = 10; x += 4; assert_eq!(x, 14); #}</code></pre></pre> <a class="header" href="print.html#array-and-array-index-expressions" id="array-and-array-index-expressions"><h1>Array and array index expressions</h1></a> <a class="header" href="print.html#array-expressions" id="array-expressions"><h2>Array expressions</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>ArrayExpression</em> :<br /> <code>[</code> <code>]</code><br /> | <code>[</code> <a href="expressions.html"><em>Expression</em></a> ( <code>,</code> <a href="expressions.html"><em>Expression</em></a> )<sup>*</sup> <code>,</code><sup>?</sup> <code>]</code><br /> | <code>[</code> <a href="expressions.html"><em>Expression</em></a> <code>;</code> <a href="expressions.html"><em>Expression</em></a> <code>]</code></p> </blockquote> <p>An <em><a href="types.html#array-and-slice-types">array</a> expression</em> can be written by enclosing zero or more comma-separated expressions of uniform type in square brackets. This produces and array containing each of these values in the order they are written.</p> <p>Alternatively there can be exactly two expressions inside the brackets, separated by a semi-colon. The expression after the <code>;</code> must be a have type <code>usize</code> and be a <a href="expressions.html#constant-expressions">constant expression</a>, such as a <a href="tokens.html#literals">literal</a> or a <a href="items/constant-items.html">constant item</a>. <code>[a; b]</code> creates an array containing <code>b</code> copies of the value of <code>a</code>. If the expression after the semi-colon has a value greater than 1 then this requires that the type of <code>a</code> is <a href="special-types-and-traits.html#copy"><code>Copy</code></a>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { [1, 2, 3, 4]; ["a", "b", "c", "d"]; [0; 128]; // array with 128 zeros [0u8, 0u8, 0u8, 0u8,]; [[1, 0, 0], [0, 1, 0], [0, 0, 1]]; // 2D array #}</code></pre></pre> <a class="header" href="print.html#array-and-slice-indexing-expressions" id="array-and-slice-indexing-expressions"><h2>Array and slice indexing expressions</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>IndexExpression</em> :<br /> <a href="expressions.html"><em>Expression</em></a> <code>[</code> <a href="expressions.html"><em>Expression</em></a> <code>]</code></p> </blockquote> <p><a href="types.html#array-and-slice-types">Array and slice</a>-typed expressions can be indexed by writing a square-bracket-enclosed expression (the index) after them. When the array is mutable, the resulting <a href="expressions.html#place-expressions-and-value-expressions">memory location</a> can be assigned to. For other types an index expression <code>a[b]</code> is equivalent to <code>*std::ops::Index::index(&a, b)</code>, or <code>*std::opsIndexMut::index_mut(&mut a, b)</code> in a mutable place expression context. Just as with methods, Rust will also insert dereference operations on <code>a</code> repeatedly to find an implementation.</p> <p>Indices are zero-based, and are of type <code>usize</code> for arrays and slices. Array access is a <a href="expressions.html#constant-expressions">constant expression</a>, so bounds can be checked at compile-time for constant arrays with a constant index value. Otherwise a check will be performed at run-time that will put the thread in a <em>panicked state</em> if it fails.</p> <pre><pre class="playpen"><code class="language-rust should_panic"> # #![allow(unused_variables)] #fn main() { ([1, 2, 3, 4])[2]; // Evaluates to 3 let b = [[1, 0, 0], [0, 1, 0], [0, 0, 1]]; b[1][2]; // multidimensional array indexing let x = (["a", "b"])[10]; // warning: const index-expr is out of bounds let n = 10; let y = (["a", "b"])[n]; // panics let arr = ["a", "b"]; arr[10]; // panics #}</code></pre></pre> <p>The array index expression can be implemented for types other than arrays and slices by implementing the <a href="../std/ops/trait.Index.html">Index</a> and <a href="../std/ops/trait.IndexMut.html">IndexMut</a> traits.</p> <a class="header" href="print.html#tuple-and-tuple-indexing-expressions" id="tuple-and-tuple-indexing-expressions"><h1>Tuple and tuple indexing expressions</h1></a> <a class="header" href="print.html#tuple-expressions" id="tuple-expressions"><h2>Tuple expressions</h2></a> <p>Tuples are written by enclosing zero or more comma-separated expressions in parentheses. They are used to create <a href="types.html#tuple-types">tuple-typed</a> values.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { (0.0, 4.5); ("a", 4usize, true); (); #}</code></pre></pre> <p>You can disambiguate a single-element tuple from a value in parentheses with a comma:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { (0,); // single-element tuple (0); // zero in parentheses #}</code></pre></pre> <a class="header" href="print.html#tuple-indexing-expressions" id="tuple-indexing-expressions"><h2>Tuple indexing expressions</h2></a> <p><a href="types.html#tuple-types">Tuples</a> and <a href="items/structs.html">struct tuples</a> can be indexed using the number corresponding to the position of the field. The index must be written as a <a href="tokens.html#integer-literals">decimal literal</a> with no underscores or suffix. Tuple indexing expressions also differ from field expressions in that they can unambiguously be called as a function. In all other aspects they have the same behavior.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # struct Point(f32, f32); let pair = (1, 2); assert_eq!(pair.1, 2); let unit_x = Point(1.0, 0.0); assert_eq!(unit_x.0, 1.0); #}</code></pre></pre> <a class="header" href="print.html#struct-expressions" id="struct-expressions"><h1>Struct expressions</h1></a> <p>There are several forms of struct expressions. A <em>struct expression</em> consists of the <a href="paths.html">path</a> of a <a href="items/structs.html">struct item</a>, followed by a brace-enclosed list of zero or more comma-separated name-value pairs, providing the field values of a new instance of the struct. A field name can be any <a href="identifiers.html">identifier</a>, and is separated from its value expression by a colon. In the case of a tuple struct the field names are numbers corresponding to the position of the field. The numbers must be written in decimal, containing no underscores and with no leading zeros or integer suffix. A value of a <a href="items/unions.html">union</a> type can also be created using this syntax, except that it must specify exactly one field.</p> <p>Struct expressions can't be used directly in the head of a <a href="expressions/loop-expr.html">loop</a> or an <a href="expressions/if-expr.html#if-expressions">if</a>, <a href="expressions/if-expr.html#if-let-expressions">if let</a> or <a href="expressions/match-expr.html">match</a> expression. But struct expressions can still be in used inside parentheses, for example.</p> <p>A <em>tuple struct expression</em> consists of the path of a struct item, followed by a parenthesized list of one or more comma-separated expressions (in other words, the path of a struct item followed by a tuple expression). The struct item must be a tuple struct item.</p> <p>A <em>unit-like struct expression</em> consists only of the path of a struct item.</p> <p>The following are examples of struct expressions:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # struct Point { x: f64, y: f64 } # struct NothingInMe { } # struct TuplePoint(f64, f64); # mod game { pub struct User<'a> { pub name: &'a str, pub age: u32, pub score: usize } } # struct Cookie; fn some_fn<T>(t: T) {} Point {x: 10.0, y: 20.0}; NothingInMe {}; TuplePoint(10.0, 20.0); TuplePoint { 0: 10.0, 1: 20.0 }; // Results in the same value as the above line let u = game::User {name: "Joe", age: 35, score: 100_000}; some_fn::<Cookie>(Cookie); #}</code></pre></pre> <p>A struct expression forms a new value of the named struct type. Note that for a given <em>unit-like</em> struct type, this will always be the same value.</p> <p>A struct expression can terminate with the syntax <code>..</code> followed by an expression to denote a functional update. The expression following <code>..</code> (the base) must have the same struct type as the new struct type being formed. The entire expression denotes the result of constructing a new struct (with the same type as the base expression) with the given values for the fields that were explicitly specified and the values in the base expression for all other fields. Just as with all struct expressions, all of the fields of the struct must be <a href="visibility-and-privacy.html">visible</a>, even those not explicitly named.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # struct Point3d { x: i32, y: i32, z: i32 } let base = Point3d {x: 1, y: 2, z: 3}; Point3d {y: 0, z: 10, .. base}; #}</code></pre></pre> <a class="header" href="print.html#struct-field-init-shorthand" id="struct-field-init-shorthand"><h2>Struct field init shorthand</h2></a> <p>When initializing a data structure (struct, enum, union) with named (but not numbered) fields, it is allowed to write <code>fieldname</code> as a shorthand for <code>fieldname: fieldname</code>. This allows a compact syntax with less duplication.</p> <p>Example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # struct Point3d { x: i32, y: i32, z: i32 } # let x = 0; # let y_value = 0; # let z = 0; Point3d { x: x, y: y_value, z: z }; Point3d { x, y: y_value, z }; #}</code></pre></pre> <a class="header" href="print.html#enumeration-variant-expressions" id="enumeration-variant-expressions"><h1>Enumeration Variant expressions</h1></a> <p>Enumeration variants can be constructed similarly to structs, using a path to an enum variant instead of to a struct:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # enum Message { # Quit, # WriteString(String), # Move { x: i32, y: i32 }, # } let q = Message::Quit; let w = Message::WriteString("Some string".to_string()); let m = Message::Move { x: 50, y: 200 }; #}</code></pre></pre> <a class="header" href="print.html#call-expressions" id="call-expressions"><h1>Call expressions</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>CallExpression</em> :<br /> <a href="expressions.html"><em>Expression</em></a> <code>(</code> <em>CallParams</em><sup>?</sup> <code>)</code></p> <p><em>CallParams</em> :<br /> <a href="expressions.html"><em>Expression</em></a> ( <code>,</code> <a href="expressions.html"><em>Expression</em></a> )<sup>*</sup> <code>,</code><sup>?</sup></p> </blockquote> <p>A <em>call expression</em> consists of an expression followed by a parenthesized expression-list. It invokes a function, providing zero or more input variables. If the function eventually returns, then the expression completes. For <a href="types.html#function-item-types">non-function types</a>, the expression f(...) uses the method on one of the <a href="../std/ops/trait.Fn.html"><code>std::ops::Fn</code></a>, <a href="../std/ops/trait.FnMut.html"><code>std::ops::FnMut</code></a> or <a href="../std/ops/trait.FnOnce.html"><code>std::ops::FnOnce</code></a> traits, which differ in whether they take the type by reference, mutable reference, or take ownership respectively. An automatic borrow will be taken if needed. Rust will also automatically dereference <code>f</code> as required. Some examples of call expressions:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # fn add(x: i32, y: i32) -> i32 { 0 } let three: i32 = add(1i32, 2i32); let name: &'static str = (|| "Rust")(); #}</code></pre></pre> <a class="header" href="print.html#disambiguating-function-calls" id="disambiguating-function-calls"><h2>Disambiguating Function Calls</h2></a> <p>Rust treats all function calls as sugar for a more explicit, fully-qualified syntax. Upon compilation, Rust will desugar all function calls into the explicit form. Rust may sometimes require you to qualify function calls with trait, depending on the ambiguity of a call in light of in-scope items.</p> <blockquote> <p><strong>Note</strong>: In the past, the Rust community used the terms "Unambiguous Function Call Syntax", "Universal Function Call Syntax", or "UFCS", in documentation, issues, RFCs, and other community writings. However, the term lacks descriptive power and potentially confuses the issue at hand. We mention it here for searchability's sake.</p> </blockquote> <p>Several situations often occur which result in ambiguities about the receiver or referent of method or associated function calls. These situations may include:</p> <ul> <li>Multiple in-scope traits define methods with the same name for the same types</li> <li>Auto-<code>deref</code> is undesirable; for example, distinguishing between methods on a smart pointer itself and the pointer's referent</li> <li>Methods which take no arguments, like <code>default()</code>, and return properties of a type, like <code>size_of()</code></li> </ul> <p>To resolve the ambiguity, the programmer may refer to their desired method or function using more specific paths, types, or traits.</p> <p>For example,</p> <pre><pre class="playpen"><code class="language-rust">trait Pretty { fn print(&self); } trait Ugly { fn print(&self); } struct Foo; impl Pretty for Foo { fn print(&self) {} } struct Bar; impl Pretty for Bar { fn print(&self) {} } impl Ugly for Bar{ fn print(&self) {} } fn main() { let f = Foo; let b = Bar; // we can do this because we only have one item called `print` for `Foo`s f.print(); // more explicit, and, in the case of `Foo`, not necessary Foo::print(&f); // if you're not into the whole brevity thing <Foo as Pretty>::print(&f); // b.print(); // Error: multiple 'print' found // Bar::print(&b); // Still an error: multiple `print` found // necessary because of in-scope items defining `print` <Bar as Pretty>::print(&b); } </code></pre></pre> <p>Refer to <a href="https://github.com/rust-lang/rfcs/blob/master/text/0132-ufcs.md">RFC 132</a> for further details and motivations.</p> <a class="header" href="print.html#method-call-expressions" id="method-call-expressions"><h1>Method-call expressions</h1></a> <p>A <em>method call</em> consists of an expression (the <em>receiver</em>) followed by a single dot, an <a href="identifiers.html">identifier</a>, and a parenthesized expression-list. Method calls are resolved to methods on specific traits, either statically dispatching to a method if the exact <code>self</code>-type of the left-hand-side is known, or dynamically dispatching if the left-hand-side expression is an indirect <a href="types.html#trait-objects">trait object</a>.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let pi: Result<f32, _> = "3.14".parse(); let log_pi = pi.unwrap_or(1.0).log(2.72); # assert!(1.14 < log_pi && log_pi < 1.15) #}</code></pre></pre> <p>When looking up a method call, the receiver may be automatically dereferenced or borrowed in order to call a method. This requires a more complex lookup process than for other functions, since there may be a number of possible methods to call. The following procedure is used:</p> <p>The first step is to build a list of candidate receiver types. Obtain these by repeatedly <a href="expressions/operator-expr.html#the-dereference-operator">dereferencing</a> the receiver expression's type, adding each type encountered to the list, then finally attempting an [unsized coercion] at the end, and adding the result type if that is successful. Then, for each candidate <code>T</code>, add <code>&T</code> and <code>&mut T</code> to the list immediately after <code>T</code>.</p> <p>For instance, if the receiver has type <code>Box<[i32;2]></code>, then the candidate types will be <code>Box<[i32;2]></code>, <code>&Box<[i32;2]></code>, <code>&mut Box<[i32;2]></code>, <code>[i32; 2]</code> (by dereferencing), <code>&[i32; 2]</code>, <code>&mut [i32; 2]</code>, <code>[i32]</code> (by unsized coercion), <code>&[i32]</code>, and finally <code>&mut [i32]</code>.</p> <p>Then, for each candidate type <code>T</code>, search for a <a href="visibility-and-privacy.html">visible</a> method with a receiver of that type in the following places:</p> <ol> <li><code>T</code>'s inherent methods (methods implemented directly on <code>T</code>).</li> <li>Any of the methods provided by a <a href="visibility-and-privacy.html">visible</a> trait implemented by <code>T</code>. If <code>T</code> is a type parameter, methods provided by trait bounds on <code>T</code> are looked up first. Then all remaining methods in scope are looked up.</li> </ol> <blockquote> <p>Note: the lookup is done for each type in order, which can occasionally lead to surprising results. The below code will print "In trait impl!", because <code>&self</code> methods are looked up first, the trait method is found before the struct's <code>&mut self</code> method is found.</p> <pre><pre class="playpen"><code class="language-rust">struct Foo {} trait Bar { fn bar(&self); } impl Foo { fn bar(&mut self) { println!("In struct impl!") } } impl Bar for Foo { fn bar(&self) { println!("In trait impl!") } } fn main() { let mut f = Foo{}; f.bar(); } </code></pre></pre> </blockquote> <p>If this results in multiple possible candidates, then it is an error, and the receiver must be <a href="expressions/call-expr.html#disambiguating-function-calls">converted</a> to an appropriate receiver type to make the method call.</p> <p>This process does not take into account the mutability or lifetime of the receiver, or whether a method is <code>unsafe</code>. Once a method is looked up, if it can't be called for one (or more) of those reasons, the result is a compiler error.</p> <p>If a step is reached where there is more than one possible method, such as where generic methods or traits are considered the same, then it is a compiler error. These cases require a [disambiguating function call syntax] for method and function invocation.</p> <blockquote> <p>Warning: For <a href="types.html#trait-objects">trait objects</a>, if there is an inherent method of the same name as a trait method, it will give a compiler error when trying to call the method in a method call expression. Instead, you can call the method using [disambiguating function call syntax], in which case it calls the trait method, not the inherent method. There is no way to call the inherent method. Just don't define inherent methods on trait objects with the same name a trait method and you'll be fine.</p> </blockquote> <a class="header" href="print.html#field-access-expressions" id="field-access-expressions"><h1>Field access expressions</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>FieldExpression</em> :<br /> <a href="expressions.html"><em>Expression</em></a> <code>.</code> <a href="identifiers.html">IDENTIFIER</a></p> </blockquote> <p>A <em>field expression</em> consists of an expression followed by a single dot and an <a href="identifiers.html">identifier</a>, when not immediately followed by a parenthesized expression-list (the latter is always a <a href="expressions/method-call-expr.html">method call expression</a>). A field expression denotes a field of a <a href="items/structs.html">struct</a> or <a href="items/unions.html">union</a>. To call a function stored in a struct, parentheses are needed around the field expression.</p> <pre><code class="language-rust ignore">mystruct.myfield; foo().x; (Struct {a: 10, b: 20}).a; mystruct.method(); // Method expression (mystruct.function_field)() // Call expression containing a field expression </code></pre> <p>A field access is a <a href="expressions.html#place-expressions-and-value-expressions">place expression</a> referring to the location of that field. When the subexpression is <a href="expressions.html#mutability">mutable</a>, the field expression is also mutable.</p> <p>Also, if the type of the expression to the left of the dot is a pointer, it is automatically dereferenced as many times as necessary to make the field access possible. In cases of ambiguity, we prefer fewer autoderefs to more.</p> <p>Finally, the fields of a struct or a reference to a struct are treated as separate entities when borrowing. If the struct does not implement <a href="special-types-and-traits.html#drop"><code>Drop</code></a> and is stored in a local variable, this also applies to moving out of each of its fields. This also does not apply if automatic dereferencing is done though user defined types.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct A { f1: String, f2: String, f3: String } let mut x: A; # x = A { # f1: "f1".to_string(), # f2: "f2".to_string(), # f3: "f3".to_string() # }; let a: &mut String = &mut x.f1; // x.f1 borrowed mutably let b: &String = &x.f2; // x.f2 borrowed immutably let c: &String = &x.f2; // Can borrow again let d: String = x.f3; // Move out of x.f3 #}</code></pre></pre> <a class="header" href="print.html#closure-expressions" id="closure-expressions"><h1>Closure expressions</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>ClosureExpression</em> :<br /> <code>move</code><sup>?</sup><br /> ( <code>||</code> | <code>|</code> <a href="items/functions.html"><em>FunctionParameters</em></a><sup>?</sup> <code>|</code> )<br /> (<a href="expressions.html"><em>Expression</em></a> | <code>-></code> <a href="types.html"><em>TypeNoBounds</em></a> <a href="expressions/block-expr.html"><em>BlockExpression</em></a>)</p> </blockquote> <p>A <em>closure expression</em> defines a closure and denotes it as a value, in a single expression. A closure expression is a pipe-symbol-delimited (<code>|</code>) list of patterns followed by an expression. Type annotations may optionally be added for the type of the parameters or for the return type. If there is a return type, the expression used for the body of the closure must be a normal <a href="expressions/block-expr.html">block</a>. A closure expression also may begin with the <code>move</code> keyword before the initial <code>|</code>.</p> <p>A closure expression denotes a function that maps a list of parameters (<code>ident_list</code>) onto the expression that follows the <code>ident_list</code>. The patterns in the <code>ident_list</code> are the parameters to the closure. If a parameter's types is not specified, then the compiler infers it from context. Each closure expression has a unique anonymous type.</p> <p>Closure expressions are most useful when passing functions as arguments to other functions, as an abbreviation for defining and capturing a separate function.</p> <p>Significantly, closure expressions <em>capture their environment</em>, which regular <a href="items/functions.html">function definitions</a> do not. Without the <code>move</code> keyword, the closure expression infers how it captures each variable from its environment, preferring to capture by shared reference, effectively borrowing all outer variables mentioned inside the closure's body. If needed the compiler will infer that instead mutable references should be taken, or that the values should be moved or copied (depending on their type) from the environment. A closure can be forced to capture its environment by copying or moving values by prefixing it with the <code>move</code> keyword. This is often used to ensure that the closure's type is <code>'static</code>.</p> <p>The compiler will determine which of the <a href="types.html#closure-types">closure traits</a> the closure's type will implement by how it acts on its captured variables. The closure will also implement <a href="special-types-and-traits.html#send"><code>Send</code></a> and/or <a href="special-types-and-traits.html#sync"><code>Sync</code></a> if all of its captured types do. These traits allow functions to accept closures using generics, even though the exact types can't be named.</p> <p>In this example, we define a function <code>ten_times</code> that takes a higher-order function argument, and we then call it with a closure expression as an argument, followed by a closure expression that moves values from its environment.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn ten_times<F>(f: F) where F: Fn(i32) { for index in 0..10 { f(index); } } ten_times(|j| println!("hello, {}", j)); // With type annotations ten_times(|j: i32| -> () { println!("hello, {}", j) }); let word = "konnichiwa".to_owned(); ten_times(move |j| println!("{}, {}", word, j)); #}</code></pre></pre> <a class="header" href="print.html#loops" id="loops"><h1>Loops</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>LoopExpression</em> :<br /> <a href="print.html#loop-labels"><em>LoopLabel</em></a><sup>?</sup> (<br /> <a href="print.html#infinite-loops"><em>InfiniteLoopExpression</em></a><br /> | <a href="print.html#predicate-loops"><em>PredicateLoopExpression</em></a><br /> | <a href="print.html#predicate-pattern-loops"><em>PredicatePatternLoopExpression</em></a><br /> | <a href="print.html#iterator-loops"><em>IteratorLoopExpression</em></a><br /> )</p> </blockquote> <p>Rust supports four loop expressions:</p> <ul> <li>A <a href="print.html#infinite-loops"><code>loop</code> expression</a> denotes an infinite loop.</li> <li>A <a href="print.html#predicate-loops"><code>while</code> expression</a> loops until a predicate is false.</li> <li>A <a href="print.html#predicate-pattern-loops"><code>while let</code> expression</a> tests a refutable pattern.</li> <li>A <a href="print.html#iterator-loops"><code>for</code> expression</a> extracts values from an iterator, looping until the iterator is empty.</li> </ul> <p>All four types of loop support <a href="print.html#break-expressions"><code>break</code> expressions</a>, <a href="print.html#continue-expressions"><code>continue</code> expressions</a>, and <a href="print.html#loop-labels">labels</a>. Only <code>loop</code> supports <a href="print.html#break-and-loop-values">evaluation to non-trivial values</a>.</p> <a class="header" href="print.html#infinite-loops" id="infinite-loops"><h2>Infinite loops</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>InfiniteLoopExpression</em> :<br /> <code>loop</code> <a href="expressions/block-expr.html"><em>BlockExpression</em></a></p> </blockquote> <p>A <code>loop</code> expression repeats execution of its body continuously: <code>loop { println!("I live."); }</code>.</p> <p>A <code>loop</code> expression without an associated <code>break</code> expression is <a href="items/functions.html#diverging-functions">diverging</a>, and doesn't return anything. A <code>loop</code> expression containing associated <a href="print.html#break-expressions"><code>break</code> expression(s)</a> may terminate, and must have type compatible with the value of the <code>break</code> expression(s).</p> <a class="header" href="print.html#predicate-loops" id="predicate-loops"><h2>Predicate loops</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>PredicateLoopExpression</em> :<br /> <code>while</code> <a href="expressions.html"><em>Expression</em></a><sub>except struct expression</sub> <a href="expressions/block-expr.html"><em>BlockExpression</em></a></p> </blockquote> <p>A <code>while</code> loop begins by evaluating the boolean loop conditional expression. If the loop conditional expression evaluates to <code>true</code>, the loop body block executes, then control returns to the loop conditional expression. If the loop conditional expression evaluates to <code>false</code>, the <code>while</code> expression completes.</p> <p>An example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let mut i = 0; while i < 10 { println!("hello"); i = i + 1; } #}</code></pre></pre> <a class="header" href="print.html#predicate-pattern-loops" id="predicate-pattern-loops"><h2>Predicate pattern loops</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <a href="print.html#predicate-pattern-loops"><em>PredicatePatternLoopExpression</em></a> :<br /> <code>while</code> <code>let</code> <em>Pattern</em> <code>=</code> <a href="expressions.html"><em>Expression</em></a><sub>except struct expression</sub> <a href="expressions/block-expr.html"><em>BlockExpression</em></a></p> </blockquote> <p>A <code>while let</code> loop is semantically similar to a <code>while</code> loop but in place of a condition expression it expects the keyword <code>let</code> followed by a refutable pattern, an <code>=</code>, an expression and a block expression. If the value of the expression on the right hand side of the <code>=</code> matches the pattern, the loop body block executes then control returns to the pattern matching statement. Otherwise, the while expression completes.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let mut x = vec![1, 2, 3]; while let Some(y) = x.pop() { println!("y = {}", y); } #}</code></pre></pre> <a class="header" href="print.html#iterator-loops" id="iterator-loops"><h2>Iterator loops</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>IteratorLoopExpression</em> :<br /> <code>for</code> <em>Pattern</em> <code>in</code> <a href="expressions.html"><em>Expression</em></a><sub>except struct expression</sub> <a href="expressions/block-expr.html"><em>BlockExpression</em></a></p> </blockquote> <p>A <code>for</code> expression is a syntactic construct for looping over elements provided by an implementation of <code>std::iter::IntoIterator</code>. If the iterator yields a value, that value is given the specified name and the body of the loop is executed, then control returns to the head of the <code>for</code> loop. If the iterator is empty, the <code>for</code> expression completes.</p> <p>An example of a <code>for</code> loop over the contents of an array:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let v = &["apples", "cake", "coffee"]; for text in v { println!("I like {}.", text); } #}</code></pre></pre> <p>An example of a for loop over a series of integers:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let mut sum = 0; for n in 1..11 { sum += n; } assert_eq!(sum, 55); #}</code></pre></pre> <a class="header" href="print.html#loop-labels" id="loop-labels"><h2>Loop labels</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>LoopLabel</em> :<br /> <a href="tokens.html#symbols">LIFETIME_OR_LABEL</a> <code>:</code></p> </blockquote> <p>A loop expression may optionally have a <em>label</em>. The label is written as a lifetime preceding the loop expression, as in <code>'foo: loop { break 'foo; }</code>, <code>'bar: while false {}</code>, <code>'humbug: for _ in 0..0 {}</code>. If a label is present, then labeled <code>break</code> and <code>continue</code> expressions nested within this loop may exit out of this loop or return control to its head. See <a href="print.html#break-expressions">break expressions</a> and <a href="print.html#continue-expressions">continue expressions</a>.</p> <a class="header" href="print.html#break-expressions" id="break-expressions"><h2><code>break</code> expressions</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>BreakExpression</em> :<br /> <code>break</code> <a href="tokens.html#symbols">LIFETIME_OR_LABEL</a><sup>?</sup> <a href="expressions.html"><em>Expression</em></a><sup>?</sup></p> </blockquote> <p>When <code>break</code> is encountered, execution of the associated loop body is immediately terminated, for example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let mut last = 0; for x in 1..100 { if x > 12 { break; } last = x; } assert_eq!(last, 12); #}</code></pre></pre> <p>A <code>break</code> expression is normally associated with the innermost <code>loop</code>, <code>for</code> or <code>while</code> loop enclosing the <code>break</code> expression, but a <a href="print.html#loop-labels">label</a> can be used to specify which enclosing loop is affected. Example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { 'outer: loop { while true { break 'outer; } } #}</code></pre></pre> <p>A <code>break</code> expression is only permitted in the body of a loop, and has one of the forms <code>break</code>, <code>break 'label</code> or (<a href="print.html#break-and-loop-values">see below</a>) <code>break EXPR</code> or <code>break 'label EXPR</code>.</p> <a class="header" href="print.html#continue-expressions" id="continue-expressions"><h2><code>continue</code> expressions</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>ContinueExpression</em> :<br /> <code>continue</code> <a href="tokens.html#symbols">LIFETIME_OR_LABEL</a><sup>?</sup></p> </blockquote> <p>When <code>continue</code> is encountered, the current iteration of the associated loop body is immediately terminated, returning control to the loop <em>head</em>. In the case of a <code>while</code> loop, the head is the conditional expression controlling the loop. In the case of a <code>for</code> loop, the head is the call-expression controlling the loop.</p> <p>Like <code>break</code>, <code>continue</code> is normally associated with the innermost enclosing loop, but <code>continue 'label</code> may be used to specify the loop affected. A <code>continue</code> expression is only permitted in the body of a loop.</p> <a class="header" href="print.html#break-and-loop-values" id="break-and-loop-values"><h2><code>break</code> and loop values</h2></a> <p>When associated with a <code>loop</code>, a break expression may be used to return a value from that loop, via one of the forms <code>break EXPR</code> or <code>break 'label EXPR</code>, where <code>EXPR</code> is an expression whose result is returned from the <code>loop</code>. For example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let (mut a, mut b) = (1, 1); let result = loop { if b > 10 { break b; } let c = a + b; a = b; b = c; }; // first number in Fibonacci sequence over 10: assert_eq!(result, 13); #}</code></pre></pre> <p>In the case a <code>loop</code> has an associated <code>break</code>, it is not considered diverging, and the <code>loop</code> must have a type compatible with each <code>break</code> expression. <code>break</code> without an expression is considered identical to <code>break</code> with expression <code>()</code>.</p> <a class="header" href="print.html#range-expressions" id="range-expressions"><h1>Range expressions</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>RangeExpression</em> :<br /> <em>RangeExpr</em><br /> | <em>RangeFromExpr</em><br /> | <em>RangeToExpr</em><br /> | <em>RangeFullExpr</em></p> <p><em>RangeExpr</em> :<br /> <a href="expressions.html"><em>Expression</em></a> <code>..</code> <a href="expressions.html"><em>Expression</em></a></p> <p><em>RangeFromExpr</em> :<br /> <a href="expressions.html"><em>Expression</em></a> <code>..</code></p> <p><em>RangeToExpr</em> :<br /> <code>..</code> <a href="expressions.html"><em>Expression</em></a></p> <p><em>RangeFullExpr</em> :<br /> <code>..</code></p> </blockquote> <p>The <code>..</code> operator will construct an object of one of the <code>std::ops::Range</code> (or <code>core::ops::Range</code>) variants, according to the following table:</p> <table><thead><tr><th> Production </th><th> Syntax </th><th> Type </th><th> Range </th></tr></thead><tbody> <tr><td> <em>RangeExpr</em> </td><td> start<code>..</code>end </td><td> <a href="https://doc.rust-lang.org/std/ops/struct.Range.html">std::ops::Range</a> </td><td> start ≤ x < end </td></tr> <tr><td> <em>RangeFromExpr</em> </td><td> start<code>..</code> </td><td> <a href="https://doc.rust-lang.org/std/ops/struct.RangeFrom.html">std::ops::RangeFrom</a> </td><td> start ≤ x </td></tr> <tr><td> <em>RangeToExpr</em> </td><td> <code>..</code>end </td><td> <a href="https://doc.rust-lang.org/std/ops/struct.RangeTo.html">std::ops::RangeTo</a> </td><td> x < end </td></tr> <tr><td> <em>RangeFullExpr</em> </td><td> <code>..</code> </td><td> <a href="https://doc.rust-lang.org/std/ops/struct.RangeFull.html">std::ops::RangeFull</a> </td><td> - </td></tr> </tbody></table> <p>Examples:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { 1..2; // std::ops::Range 3..; // std::ops::RangeFrom ..4; // std::ops::RangeTo ..; // std::ops::RangeFull #}</code></pre></pre> <p>The following expressions are equivalent.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x = std::ops::Range {start: 0, end: 10}; let y = 0..10; assert_eq!(x, y); #}</code></pre></pre> <p>Ranges can be used in <code>for</code> loops:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { for i in 1..11 { println!("{}", i); } #}</code></pre></pre> <a class="header" href="print.html#if-and-if-let-expressions" id="if-and-if-let-expressions"><h1><code>if</code> and <code>if let</code> expressions</h1></a> <a class="header" href="print.html#if-expressions" id="if-expressions"><h2><code>if</code> expressions</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>IfExpression</em> :<br /> <code>if</code> <a href="expressions.html"><em>Expression</em></a><sub><em>except struct expression</em></sub> <a href="expressions/block-expr.html"><em>BlockExpression</em></a><br /> (<code>else</code> ( <a href="expressions/block-expr.html"><em>BlockExpression</em></a> | <em>IfExpression</em> | <em>IfLetExpression</em> ) )<sup>?</sup></p> </blockquote> <p>An <code>if</code> expression is a conditional branch in program control. The form of an <code>if</code> expression is a condition expression, followed by a consequent block, any number of <code>else if</code> conditions and blocks, and an optional trailing <code>else</code> block. The condition expressions must have type <code>bool</code>. If a condition expression evaluates to <code>true</code>, the consequent block is executed and any subsequent <code>else if</code> or <code>else</code> block is skipped. If a condition expression evaluates to <code>false</code>, the consequent block is skipped and any subsequent <code>else if</code> condition is evaluated. If all <code>if</code> and <code>else if</code> conditions evaluate to <code>false</code> then any <code>else</code> block is executed. An if expression evaluates to the same value as the executed block, or <code>()</code> if no block is evaluated. An <code>if</code> expression must have the same type in all situations.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # let x = 3; if x == 4 { println!("x is four"); } else if x == 3 { println!("x is three"); } else { println!("x is something else"); } let y = if 12 * 15 > 150 { "Bigger" } else { "Smaller" }; assert_eq!(y, "Bigger"); #}</code></pre></pre> <a class="header" href="print.html#if-let-expressions" id="if-let-expressions"><h2><code>if let</code> expressions</h2></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>IfLetExpression</em> :<br /> <code>if</code> <code>let</code> <em>Pattern</em> <code>=</code> <a href="expressions.html"><em>Expression</em></a><sub><em>except struct expression</em></sub> <a href="expressions/block-expr.html"><em>BlockExpression</em></a><br /> (<code>else</code> ( <a href="expressions/block-expr.html"><em>BlockExpression</em></a> | <em>IfExpression</em> | <em>IfLetExpression</em> ) )<sup>?</sup></p> </blockquote> <p>An <code>if let</code> expression is semantically similar to an <code>if</code> expression but in place of a condition expression it expects the keyword <code>let</code> followed by a refutable pattern, an <code>=</code> and an expression. If the value of the expression on the right hand side of the <code>=</code> matches the pattern, the corresponding block will execute, otherwise flow proceeds to the following <code>else</code> block if it exists. Like <code>if</code> expressions, <code>if let</code> expressions have a value determined by the block that is evaluated.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let dish = ("Ham", "Eggs"); // this body will be skipped because the pattern is refuted if let ("Bacon", b) = dish { println!("Bacon is served with {}", b); } else { // This block is evaluated instead. println!("No bacon will be served"); } // this body will execute if let ("Ham", b) = dish { println!("Ham is served with {}", b); } #}</code></pre></pre> <p><code>if</code> and <code>if let</code> expressions can be intermixed:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x = Some(3); let a = if let Some(1) = x { 1 } else if x == Some(2) { 2 } else if let Some(y) = x { y } else { -1 }; assert_eq!(a, 3); #}</code></pre></pre> <a class="header" href="print.html#match-expressions" id="match-expressions"><h1><code>match</code> expressions</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>MatchExpression</em> :<br /> <code>match</code> <a href="expressions.html"><em>Expression</em></a><sub><em>except struct expression</em></sub> <em>MatchBlock</em></p> <p><em>MatchBlock</em> :<br /> <code>{</code> <code>}</code><br /> | <code>{</code> (<code>|</code><sup>?</sup> <em>Pattern</em> (<code>|</code> <em>Pattern</em>)<sup>*</sup> (<code>if</code> <a href="expressions.html"><em>Expression</em></a>)<sup>?</sup> <code>=></code> (<a href="expressions/block-expr.html#block-expressions"><em>BlockExpression</em></a> <code>,</code><sup>?</sup> | <a href="expressions.html"><em>Expression</em></a> <code>,</code>))<sup>*</sup><br /> (<code>|</code><sup>?</sup> <em>Pattern</em> (<code>|</code> <em>Pattern</em>)<sup>*</sup> (<code>if</code> <a href="expressions.html"><em>Expression</em></a>)<sup>?</sup> <code>=></code> (<a href="expressions/block-expr.html#block-expressions"><em>BlockExpression</em></a> <code>,</code><sup>?</sup> | <a href="expressions.html"><em>Expression</em></a> <code>,</code><sup>?</sup>))<br /> <code>}</code></p> </blockquote> <p>A <code>match</code> expression branches on a <em>pattern</em>. The exact form of matching that occurs depends on the pattern. Patterns consist of some combination of literals, destructured arrays or enum constructors, structs and tuples, variable binding specifications, wildcards (<code>..</code>), and placeholders (<code>_</code>). A <code>match</code> expression has a <em>head expression</em>, which is the value to compare to the patterns. The type of the patterns must equal the type of the head expression.</p> <p>A <code>match</code> behaves differently depending on whether or not the head expression is a <a href="expressions.html#place-expressions-and-value-expressions">place expression or value expression</a>. If the head expression is a <a href="expressions.html#place-expressions-and-value-expressions">value expression</a>, it is first evaluated into a temporary location, and the resulting value is sequentially compared to the patterns in the arms until a match is found. The first arm with a matching pattern is chosen as the branch target of the <code>match</code>, any variables bound by the pattern are assigned to local variables in the arm's block, and control enters the block.</p> <p>When the head expression is a <a href="expressions.html#place-expressions-and-value-expressions">place expression</a>, the match does not allocate a temporary location; however, a by-value binding may copy or move from the memory location. When possible, it is preferable to match on place expressions, as the lifetime of these matches inherits the lifetime of the place expression rather than being restricted to the inside of the match.</p> <p>An example of a <code>match</code> expression:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x = 1; match x { 1 => println!("one"), 2 => println!("two"), 3 => println!("three"), 4 => println!("four"), 5 => println!("five"), _ => println!("something else"), } #}</code></pre></pre> <p>Patterns that bind variables default to binding to a copy or move of the matched value (depending on the matched value's type). This can be changed to bind to a reference by using the <code>ref</code> keyword, or to a mutable reference using <code>ref mut</code>.</p> <p>Patterns can be used to <em>destructure</em> structs, enums, and tuples. Destructuring breaks a value up into its component pieces. The syntax used is the same as when creating such values. When destructing a data structure with named (but not numbered) fields, it is allowed to write <code>fieldname</code> as a shorthand for <code>fieldname: fieldname</code>. In a pattern whose head expression has a <code>struct</code>, <code>enum</code> or <code>tupl</code> type, a placeholder (<code>_</code>) stands for a <em>single</em> data field, whereas a wildcard <code>..</code> stands for <em>all</em> the fields of a particular variant.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # enum Message { # Quit, # WriteString(String), # Move { x: i32, y: i32 }, # ChangeColor(u8, u8, u8), # } # let message = Message::Quit; match message { Message::Quit => println!("Quit"), Message::WriteString(write) => println!("{}", &write), Message::Move{ x, y: 0 } => println!("move {} horizontally", x), Message::Move{ .. } => println!("other move"), Message::ChangeColor { 0: red, 1: green, 2: _ } => { println!("color change, red: {}, green: {}", red, green); } }; #}</code></pre></pre> <p>Patterns can also dereference pointers by using the <code>&</code>, <code>&mut</code> and <code>box</code> symbols, as appropriate. For example, these two matches on <code>x: &i32</code> are equivalent:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # let x = &3; let y = match *x { 0 => "zero", _ => "some" }; let z = match x { &0 => "zero", _ => "some" }; assert_eq!(y, z); #}</code></pre></pre> <p>Subpatterns can also be bound to variables by the use of the syntax <code>variable @ subpattern</code>. For example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let x = 1; match x { e @ 1 ... 5 => println!("got a range element {}", e), _ => println!("anything"), } #}</code></pre></pre> <p>Multiple match patterns may be joined with the <code>|</code> operator. A range of values may be specified with <code>...</code>. For example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # let x = 2; let message = match x { 0 | 1 => "not many", 2 ... 9 => "a few", _ => "lots" }; #}</code></pre></pre> <p>Range patterns only work on <a href="types.html#textual-types"><code>char</code></a> and <a href="types.html#numeric-types">numeric types</a>. A range pattern may not be a sub-range of another range pattern inside the same <code>match</code>.</p> <p>Finally, match patterns can accept <em>pattern guards</em> to further refine the criteria for matching a case. Pattern guards appear after the pattern and consist of a bool-typed expression following the <code>if</code> keyword. A pattern guard may refer to the variables bound within the pattern they follow.</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # let maybe_digit = Some(0); # fn process_digit(i: i32) { } # fn process_other(i: i32) { } let message = match maybe_digit { Some(x) if x < 10 => process_digit(x), Some(x) => process_other(x), None => panic!(), }; #}</code></pre></pre> <a class="header" href="print.html#return-expressions" id="return-expressions"><h1><code>return</code> expressions</h1></a> <blockquote> <p><strong><sup>Syntax</sup></strong><br /> <em>ReturnExpression</em> :<br /> <code>return</code> <a href="expressions.html"><em>Expression</em></a><sup>?</sup></p> </blockquote> <p>Return expressions are denoted with the keyword <code>return</code>. Evaluating a <code>return</code> expression moves its argument into the designated output location for the current function call, destroys the current function activation frame, and transfers control to the caller frame.</p> <p>An example of a <code>return</code> expression:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn max(a: i32, b: i32) -> i32 { if a > b { return a; } return b; } #}</code></pre></pre> <a class="header" href="print.html#type-system" id="type-system"><h1>Type system</h1></a> <a class="header" href="print.html#types" id="types"><h1>Types</h1></a> <p>Every variable, item and value in a Rust program has a type. The <em>type</em> of a <em>value</em> defines the interpretation of the memory holding it.</p> <p>Built-in types are tightly integrated into the language, in nontrivial ways that are not possible to emulate in user-defined types. User-defined types have limited capabilities.</p> <a class="header" href="print.html#primitive-types" id="primitive-types"><h2>Primitive types</h2></a> <p>Some types are defined by the language, rather than as part of the standard library, these are called <em>primitive types</em>. Some of these are individual types:</p> <ul> <li>The boolean type <code>bool</code> with values <code>true</code> and <code>false</code>.</li> <li>The <a href="print.html#machine-types">machine types</a> (integer and floating-point).</li> <li>The <a href="print.html#machine-dependent-integer-types">machine-dependent integer types</a>.</li> <li>The <a href="print.html#textual-types">textual types</a> <code>char</code> and <code>str</code>.</li> </ul> <p>There are also some primitive constructs for generic types built in to the language:</p> <ul> <li><a href="print.html#tuple-types">Tuples</a></li> <li><a href="print.html#array-and-slice-types">Arrays</a></li> <li><a href="print.html#array-and-slice-types">Slices</a></li> <li><a href="print.html#function-pointer-types">Function pointers</a></li> <li><a href="print.html#pointer-types">References</a></li> <li><a href="print.html#raw-pointers-const-and-mut">Pointers</a></li> </ul> <a class="header" href="print.html#numeric-types" id="numeric-types"><h2>Numeric types</h2></a> <a class="header" href="print.html#machine-types" id="machine-types"><h3>Machine types</h3></a> <p>The machine types are the following:</p> <ul> <li> <p>The unsigned word types <code>u8</code>, <code>u16</code>, <code>u32</code> and <code>u64</code>, with values drawn from the integer intervals [0, 2^8 - 1], [0, 2^16 - 1], [0, 2^32 - 1] and [0, 2^64 - 1] respectively.</p> </li> <li> <p>The signed two's complement word types <code>i8</code>, <code>i16</code>, <code>i32</code> and <code>i64</code>, with values drawn from the integer intervals [-(2^(7)), 2^7 - 1], [-(2^(15)), 2^15 - 1], [-(2^(31)), 2^31 - 1], [-(2^(63)), 2^63 - 1] respectively.</p> </li> <li> <p>The IEEE 754-2008 <code>binary32</code> and <code>binary64</code> floating-point types: <code>f32</code> and <code>f64</code>, respectively.</p> </li> </ul> <a class="header" href="print.html#machine-dependent-integer-types" id="machine-dependent-integer-types"><h3>Machine-dependent integer types</h3></a> <p>The <code>usize</code> type is an unsigned integer type with the same number of bits as the platform's pointer type. It can represent every memory address in the process.</p> <p>The <code>isize</code> type is a signed integer type with the same number of bits as the platform's pointer type. The theoretical upper bound on object and array size is the maximum <code>isize</code> value. This ensures that <code>isize</code> can be used to calculate differences between pointers into an object or array and can address every byte within an object along with one byte past the end.</p> <a class="header" href="print.html#textual-types" id="textual-types"><h2>Textual types</h2></a> <p>The types <code>char</code> and <code>str</code> hold textual data.</p> <p>A value of type <code>char</code> is a <a href="http://www.unicode.org/glossary/#unicode_scalar_value">Unicode scalar value</a> (i.e. a code point that is not a surrogate), represented as a 32-bit unsigned word in the 0x0000 to 0xD7FF or 0xE000 to 0x10FFFF range. A <code>[char]</code> is effectively a UCS-4 / UTF-32 string.</p> <p>A value of type <code>str</code> is a Unicode string, represented as an array of 8-bit unsigned bytes holding a sequence of UTF-8 code points. Since <code>str</code> is a <a href="dynamically-sized-types.html">dynamically sized type</a>, it is not a <em>first-class</em> type, but can only be instantiated through a pointer type, such as <code>&str</code>.</p> <a class="header" href="print.html#tuple-types" id="tuple-types"><h2>Tuple types</h2></a> <p>A tuple <em>type</em> is a heterogeneous product of other types, called the <em>elements</em> of the tuple. It has no nominal name and is instead structurally typed.</p> <p>Tuple types and values are denoted by listing the types or values of their elements, respectively, in a parenthesized, comma-separated list.</p> <p>Because tuple elements don't have a name, they can only be accessed by pattern-matching or by using <code>N</code> directly as a field to access the <code>N</code>th element.</p> <p>An example of a tuple type and its use:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { type Pair<'a> = (i32, &'a str); let p: Pair<'static> = (10, "ten"); let (a, b) = p; assert_eq!(a, 10); assert_eq!(b, "ten"); assert_eq!(p.0, 10); assert_eq!(p.1, "ten"); #}</code></pre></pre> <p>For historical reasons and convenience, the tuple type with no elements (<code>()</code>) is often called ‘unit’ or ‘the unit type’.</p> <a class="header" href="print.html#array-and-slice-types" id="array-and-slice-types"><h2>Array, and Slice types</h2></a> <p>Rust has two different types for a list of items:</p> <ul> <li><code>[T; N]</code>, an 'array'</li> <li><code>[T]</code>, a 'slice'</li> </ul> <p>An array has a fixed size, and can be allocated on either the stack or the heap.</p> <p>A slice is a <a href="dynamically-sized-types.html">dynamically sized type</a> representing a 'view' into an array. To use a slice type it generally has to be used behind a pointer for example as</p> <ul> <li><code>&[T]</code>, a 'shared slice', often just called a 'slice', it doesn't own the data it points to, it borrows it.</li> <li><code>&mut [T]</code>, a 'mutable slice', mutably borrows the data it points to.</li> <li><code>Box<[T]></code>, a 'boxed slice'</li> </ul> <p>Examples:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { // A stack-allocated array let array: [i32; 3] = [1, 2, 3]; // A heap-allocated array, coerced to a slice let boxed_array: Box<[i32]> = Box::new([1, 2, 3]); // A (shared) slice into an array let slice: &[i32] = &boxed_array[..]; #}</code></pre></pre> <p>All elements of arrays and slices are always initialized, and access to an array or slice is always bounds-checked in safe methods and operators.</p> <blockquote> <p>Note: The <a href="../std/vec/struct.Vec.html"><code>Vec<T></code></a> standard library type provides a heap-allocated resizable array type.</p> </blockquote> <a class="header" href="print.html#struct-types" id="struct-types"><h2>Struct types</h2></a> <p>A <code>struct</code> <em>type</em> is a heterogeneous product of other types, called the <em>fields</em> of the type.<sup class="footnote-reference"><a href="print.html#structtype">1</a></sup></p> <p>New instances of a <code>struct</code> can be constructed with a <a href="expressions/struct-expr.html">struct expression</a>.</p> <p>The memory layout of a <code>struct</code> is undefined by default to allow for compiler optimizations like field reordering, but it can be fixed with the <code>#[repr(...)]</code> attribute. In either case, fields may be given in any order in a corresponding struct <em>expression</em>; the resulting <code>struct</code> value will always have the same memory layout.</p> <p>The fields of a <code>struct</code> may be qualified by <a href="visibility-and-privacy.html">visibility modifiers</a>, to allow access to data in a struct outside a module.</p> <p>A <em>tuple struct</em> type is just like a struct type, except that the fields are anonymous.</p> <p>A <em>unit-like struct</em> type is like a struct type, except that it has no fields. The one value constructed by the associated <a href="expressions/struct-expr.html">struct expression</a> is the only value that inhabits such a type.</p> <div class="footnote-definition" id="structtype"><sup class="footnote-definition-label">1</sup> <p><code>struct</code> types are analogous to <code>struct</code> types in C, the <em>record</em> types of the ML family, or the <em>struct</em> types of the Lisp family.</p> </div> <a class="header" href="print.html#enumerated-types" id="enumerated-types"><h2>Enumerated types</h2></a> <p>An <em>enumerated type</em> is a nominal, heterogeneous disjoint union type, denoted by the name of an <a href="items/enumerations.html"><code>enum</code> item</a>. <sup class="footnote-reference"><a href="print.html#enumtype">2</a></sup></p> <p>An <a href="items/enumerations.html"><code>enum</code> item</a> declares both the type and a number of <em>variants</em>, each of which is independently named and has the syntax of a struct, tuple struct or unit-like struct.</p> <p>New instances of an <code>enum</code> can be constructed in an <a href="expressions/enum-variant-expr.html">enumeration variant expression</a>.</p> <p>Any <code>enum</code> value consumes as much memory as the largest variant for its corresponding <code>enum</code> type, as well as the size needed to store a discriminant.</p> <p>Enum types cannot be denoted <em>structurally</em> as types, but must be denoted by named reference to an <a href="items/enumerations.html"><code>enum</code> item</a>.</p> <div class="footnote-definition" id="enumtype"><sup class="footnote-definition-label">2</sup> <p>The <code>enum</code> type is analogous to a <code>data</code> constructor declaration in ML, or a <em>pick ADT</em> in Limbo.</p> </div> <a class="header" href="print.html#union-types" id="union-types"><h2>Union types</h2></a> <p>A <em>union type</em> is a nominal, heterogeneous C-like union, denoted by the name of a <a href="items/unions.html"><code>union</code> item</a>.</p> <p>A union contains the value of any one of its fields. Since the accessing the wrong field can cause unexpected or undefined behaviour, <code>unsafe</code> is required to read from a union field or to write to a field that doesn't implement <a href="special-types-and-traits.html#copy"><code>Copy</code></a>.</p> <p>The memory layout of a <code>union</code> is undefined by default, but the <code>#[repr(...)]</code> attribute can be used to fix a layout.</p> <a class="header" href="print.html#recursive-types" id="recursive-types"><h2>Recursive types</h2></a> <p>Nominal types — <a href="print.html#struct-types">structs</a>, <a href="print.html#enumerated-types">enumerations</a> and <a href="print.html#union-types">unions</a> — may be recursive. That is, each <code>enum</code> variant or <code>struct</code> or <code>union</code> field may refer, directly or indirectly, to the enclosing <code>enum</code> or <code>struct</code> type itself. Such recursion has restrictions:</p> <ul> <li>Recursive types must include a nominal type in the recursion (not mere <a href="../grammar.html#type-definitions">type definitions</a>, or other structural types such as <a href="print.html#array-and-slice-types">arrays</a> or <a href="print.html#tuple-types">tuples</a>). So <code>type Rec = &'static [Rec]</code> is not allowed.</li> <li>The size of a recursive type must be finite; in other words the recursive fields of the type must be <a href="print.html#pointer-types">pointer types</a>.</li> <li>Recursive type definitions can cross module boundaries, but not module <em>visibility</em> boundaries, or crate boundaries (in order to simplify the module system and type checker).</li> </ul> <p>An example of a <em>recursive</em> type and its use:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { enum List<T> { Nil, Cons(T, Box<List<T>>) } let a: List<i32> = List::Cons(7, Box::new(List::Cons(13, Box::new(List::Nil)))); #}</code></pre></pre> <a class="header" href="print.html#pointer-types" id="pointer-types"><h2>Pointer types</h2></a> <p>All pointers in Rust are explicit first-class values. They can be moved or copied, stored into data structs, and returned from functions.</p> <a class="header" href="print.html#shared-references-" id="shared-references-"><h3>Shared references (<code>&</code>)</h3></a> <p>These point to memory <em>owned by some other value</em>. When a shared reference to a value is created it prevents direct mutation of the value. <a href="interior-mutability.html">Interior mutability</a> provides an exception for this in certain circumstances. As the name suggests, any number of shared references to a value may exit. A shared reference type is written <code>&type</code>, or <code>&'a type</code> when you need to specify an explicit lifetime. Copying a reference is a "shallow" operation: it involves only copying the pointer itself, that is, pointers are <code>Copy</code>. Releasing a reference has no effect on the value it points to, but referencing of a <a href="expressions.html#temporary-lifetimes">temporary value</a> will keep it alive during the scope of the reference itself.</p> <a class="header" href="print.html#mutable-references-mut" id="mutable-references-mut"><h3>Mutable references (<code>&mut</code>)</h3></a> <p>These also point to memory owned by some other value. A mutable reference type is written <code>&mut type</code> or <code>&'a mut type</code>. A mutable reference (that hasn't been borrowed) is the only way to access the value it points to, so is not <code>Copy</code>.</p> <a class="header" href="print.html#raw-pointers-const-and-mut" id="raw-pointers-const-and-mut"><h3>Raw pointers (<code>*const</code> and <code>*mut</code>)</h3></a> <p>Raw pointers are pointers without safety or liveness guarantees. Raw pointers are written as <code>*const T</code> or <code>*mut T</code>, for example <code>*const i32</code> means a raw pointer to a 32-bit integer. Copying or dropping a raw pointer has no effect on the lifecycle of any other value. Dereferencing a raw pointer is an <a href="unsafe-functions.html"><code>unsafe</code> operation</a>, this can also be used to convert a raw pointer to a reference by reborrowing it (<code>&*</code> or <code>&mut *</code>). Raw pointers are generally discouraged in Rust code; they exist to support interoperability with foreign code, and writing performance-critical or low-level functions.</p> <p>When comparing pointers they are compared by their address, rather than by what they point to. When comparing pointers to <a href="dynamically-sized-types.html">dynamically sized types</a> they also have their addition data compared.</p> <a class="header" href="print.html#smart-pointers" id="smart-pointers"><h3>Smart Pointers</h3></a> <p>The standard library contains additional 'smart pointer' types beyond references and raw pointers.</p> <a class="header" href="print.html#function-item-types" id="function-item-types"><h2>Function item types</h2></a> <p>When referred to, a function item, or the constructor of a tuple-like struct or enum variant, yields a zero-sized value of its <em>function item type</em>. That type explicitly identifies the function - its name, its type arguments, and its early-bound lifetime arguments (but not its late-bound lifetime arguments, which are only assigned when the function is called) - so the value does not need to contain an actual function pointer, and no indirection is needed when the function is called.</p> <p>There is no syntax that directly refers to a function item type, but the compiler will display the type as something like <code>fn(u32) -> i32 {fn_name}</code> in error messages.</p> <p>Because the function item type explicitly identifies the function, the item types of different functions - different items, or the same item with different generics - are distinct, and mixing them will create a type error:</p> <pre><pre class="playpen"><code class="language-rust compile_fail E0308"> # #![allow(unused_variables)] #fn main() { fn foo<T>() { } let x = &mut foo::<i32>; *x = foo::<u32>; //~ ERROR mismatched types #}</code></pre></pre> <p>However, there is a <a href="type-coercions.html">coercion</a> from function items to <a href="print.html#function-pointer-types">function pointers</a> with the same signature, which is triggered not only when a function item is used when a function pointer is directly expected, but also when different function item types with the same signature meet in different arms of the same <code>if</code> or <code>match</code>:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { # let want_i32 = false; # fn foo<T>() { } // `foo_ptr_1` has function pointer type `fn()` here let foo_ptr_1: fn() = foo::<i32>; // ... and so does `foo_ptr_2` - this type-checks. let foo_ptr_2 = if want_i32 { foo::<i32> } else { foo::<u32> }; #}</code></pre></pre> <p>All function items implement <a href="../std/ops/trait.Fn.html">Fn</a>, <a href="../std/ops/trait.FnMut.html">FnMut</a>, <a href="../std/ops/trait.FnOnce.html">FnOnce</a>, <a href="special-types-and-traits.html#copy">Copy</a>, <a href="special-types-and-traits.html#clone">Clone</a>, <a href="special-types-and-traits.html#send">Send</a>, and <a href="special-types-and-traits.html#sync">Sync</a>.</p> <a class="header" href="print.html#function-pointer-types" id="function-pointer-types"><h2>Function pointer types</h2></a> <p>Function pointer types, written using the <code>fn</code> keyword, refer to a function whose identity is not necessarily known at compile-time. They can be created via a coercion from both <a href="print.html#function-item-types">function items</a> and non-capturing <a href="print.html#closure-types">closures</a>.</p> <p>A function pointer type consists of a possibly-empty set of function-type modifiers (such as <code>unsafe</code> or <code>extern</code>), a sequence of input types and an output type.</p> <p>An example where <code>Binop</code> is defined as a function pointer type:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn add(x: i32, y: i32) -> i32 { x + y } let mut x = add(5,7); type Binop = fn(i32, i32) -> i32; let bo: Binop = add; x = bo(5,7); #}</code></pre></pre> <a class="header" href="print.html#closure-types" id="closure-types"><h2>Closure types</h2></a> <p>A <a href="expressions/closure-expr.html">closure expression</a> produces a closure value with a unique, anonymous type that cannot be written out.</p> <p>Depending on the requirements of the closure, its type implements one or more of the closure traits:</p> <ul> <li> <p><code>FnOnce</code> : The closure can be called once. A closure called as <code>FnOnce</code> can move out of its captured values.</p> </li> <li> <p><code>FnMut</code> : The closure can be called multiple times as mutable. A closure called as <code>FnMut</code> can mutate values from its environment. <code>FnMut</code> inherits from <code>FnOnce</code> (i.e. anything implementing <code>FnMut</code> also implements <code>FnOnce</code>).</p> </li> <li> <p><code>Fn</code> : The closure can be called multiple times through a shared reference. A closure called as <code>Fn</code> can neither move out from nor mutate captured variables, but read-only access to such values is allowed. Using <code>move</code> to capture variables by value is allowed so long as they aren't mutated or moved in the body of the closure. <code>Fn</code> inherits from <code>FnMut</code>, which itself inherits from <code>FnOnce</code>.</p> </li> </ul> <p>Closures that don't use anything from their environment, called <em>non-capturing closures</em>, can be coerced to function pointers (<code>fn</code>) with the matching signature. To adopt the example from the section above:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let add = |x, y| x + y; let mut x = add(5,7); type Binop = fn(i32, i32) -> i32; let bo: Binop = add; x = bo(5,7); #}</code></pre></pre> <a class="header" href="print.html#trait-objects" id="trait-objects"><h2>Trait objects</h2></a> <p>A <em>trait object</em> is an opaque value of another type that implements a set of traits. The set of traits is made up of an <a href="items/traits.html#object-safety">object safe</a> <em>base trait</em> plus any number of <a href="special-types-and-traits.html#auto-traits">auto traits</a>.</p> <p>Trait objects implement the base trait, its auto traits, and any super traits of the base trait.</p> <p>Trait objects are written as the path to the base trait followed by the list of auto traits followed optionally by a lifetime bound all separated by <code>+</code>. For example, given a trait <code>Trait</code>, the following are all trait objects: <code>Trait</code>, <code>Trait + Send</code>, <code>Trait + Send + Sync</code>, <code>Trait + 'static</code>, <code>Trait + Send + 'static</code>.</p> <p>Two trait object types alias each other if the base traits alias each other and if the sets of auto traits are the same and the lifetime bounds are the same. For example, <code>Trait + Send + UnwindSafe</code> is the same as <code>Trait + Unwindsafe + Send</code>.</p> <blockquote> <p>Warning: With two trait object types, even when the complete set of traits is the same, if the base traits differ, the type is different. For example, <code>Send + Sync</code> is a different type from <code>Sync + Send</code>. See <a href="https://github.com/rust-lang/rust/issues/33140">issue 33140</a>.</p> </blockquote> <blockquote> <p>Warning: Including the same auto trait multiple times is allowed, and each instance is considered a unique type. As such, <code>Trait + Send</code> is a distinct type than <code>Trait + Send + Send</code>. See <a href="https://github.com/rust-lang/rust/issues/47010">issue 47010</a>.</p> </blockquote> <p>Due to the opaqueness of which concrete type the value is of, trait objects are <a href="dynamically-sized-types.html">dynamically sized types</a>. Like all <abbr title="dynamically sized types">DSTs</abbr>, trait objects are used behind some type of pointer; for example <code>&SomeTrait</code> or <code>Box<SomeTrait></code>. Each instance of a pointer to a trait object includes:</p> <ul> <li>a pointer to an instance of a type <code>T</code> that implements <code>SomeTrait</code></li> <li>a <em>virtual method table</em>, often just called a <em>vtable</em>, which contains, for each method of <code>SomeTrait</code> that <code>T</code> implements, a pointer to <code>T</code>'s implementation (i.e. a function pointer).</li> </ul> <p>The purpose of trait objects is to permit "late binding" of methods. Calling a method on a trait object results in virtual dispatch at runtime: that is, a function pointer is loaded from the trait object vtable and invoked indirectly. The actual implementation for each vtable entry can vary on an object-by-object basis.</p> <p>An example of a trait object:</p> <pre><pre class="playpen"><code class="language-rust">trait Printable { fn stringify(&self) -> String; } impl Printable for i32 { fn stringify(&self) -> String { self.to_string() } } fn print(a: Box<Printable>) { println!("{}", a.stringify()); } fn main() { print(Box::new(10) as Box<Printable>); } </code></pre></pre> <p>In this example, the trait <code>Printable</code> occurs as a trait object in both the type signature of <code>print</code>, and the cast expression in <code>main</code>.</p> <a class="header" href="print.html#trait-object-lifetime-bounds" id="trait-object-lifetime-bounds"><h3>Trait Object Lifetime Bounds</h3></a> <p>Since a trait object can contain references, the lifetimes of those references need to be expressed as part of the trait object. The assumed lifetime of references held by a trait object is called its <em>default object lifetime bound</em>. These were defined in <a href="https://github.com/rust-lang/rfcs/blob/master/text/0599-default-object-bound.md">RFC 599</a> and amended in <a href="https://github.com/rust-lang/rfcs/blob/master/text/1156-adjust-default-object-bounds.md">RFC 1156</a>.</p> <p>For traits that themselves have no lifetime parameters:</p> <ul> <li>If there is a unique bound from the containing type then that is the default.</li> <li>If there is more than one bound from the containing type then an explicit bound must be specified.</li> <li>Otherwise the default bound is <code>'static</code>.</li> </ul> <pre><code class="language-rust ignore">// For the following trait... trait Foo { } // These two are the same as Box<T> has no lifetime bound on T Box<Foo> Box<Foo + 'static> // ...and so are these: impl Foo {} impl Foo + 'static {} // ...so are these, because &'a T requires T: 'a &'a Foo &'a (Foo + 'a) // std::cell::Ref<'a, T> also requires T: 'a, so these are the same std::cell::Ref<'a, Foo> std::cell::Ref<'a, Foo + 'a> // This is an error: struct TwoBounds<'a, 'b, T: ?Sized + 'a + 'b> TwoBounds<'a, 'b, Foo> // Error: the lifetime bound for this object type cannot // be deduced from context </code></pre> <p>The <code>+ 'static</code> and <code>+ 'a</code> refer to the default bounds of those kinds of trait objects, and also to how you can directly override them. Note that the innermost object sets the bound, so <code>&'a Box<Foo></code> is still <code>&'a Box<Foo + 'static></code>.</p> <p>For traits that have a single lifetime <em>bound</em> of their own then, instead of infering 'static as the default bound, the bound on the trait is used instead</p> <pre><code class="language-rust ignore">// For the following trait... trait Bar<'a>: 'a { } // ...these two are the same: Box<Bar<'a>> Box<Bar<'a> + 'a> // ...and so are these: impl<'a> Foo<'a> {} impl<'a> Foo<'a> + 'a {} // This is still an error: struct TwoBounds<'a, 'b, T: ?Sized + 'a + 'b> TwoBounds<'a, 'b, Foo<'c>> </code></pre> <a class="header" href="print.html#type-parameters-1" id="type-parameters-1"><h2>Type parameters</h2></a> <p>Within the body of an item that has type parameter declarations, the names of its type parameters are types:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn to_vec<A: Clone>(xs: &[A]) -> Vec<A> { if xs.is_empty() { return vec![]; } let first: A = xs[0].clone(); let mut rest: Vec<A> = to_vec(&xs[1..]); rest.insert(0, first); rest } #}</code></pre></pre> <p>Here, <code>first</code> has type <code>A</code>, referring to <code>to_vec</code>'s <code>A</code> type parameter; and <code>rest</code> has type <code>Vec<A></code>, a vector with element type <code>A</code>.</p> <a class="header" href="print.html#self-types" id="self-types"><h2>Self types</h2></a> <p>The special type <code>Self</code> has a meaning within traits and impls: it refers to the implementing type. For example, in:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { pub trait From<T> { fn from(T) -> Self; } impl From<i32> for String { fn from(x: i32) -> Self { x.to_string() } } #}</code></pre></pre> <p>The notation <code>Self</code> in the impl refers to the implementing type: <code>String</code>. In another example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { trait Printable { fn make_string(&self) -> String; } impl Printable for String { fn make_string(&self) -> String { (*self).clone() } } #}</code></pre></pre> <p>The notation <code>&self</code> is a shorthand for <code>self: &Self</code>.</p> <a class="header" href="print.html#dynamically-sized-types" id="dynamically-sized-types"><h1>Dynamically Sized Types</h1></a> <p>Most types have a fixed size that is known at compile time and implement the trait <a href="special-types-and-traits.html#sized"><code>Sized</code></a>. A type with a size that is known only at run-time is called a <em>dynamically sized type</em> (<em>DST</em>) or, informally, an unsized type. <a href="types.html#array-and-slice-types">Slices</a> and <a href="types.html#trait-objects">trait objects</a> are two examples of <abbr title="dynamically sized types">DSTs</abbr>. Such types can only be used in certain cases:</p> <ul> <li><a href="types.html#pointer-types">Pointer types</a> to <abbr title="dynamically sized types">DSTs</abbr> are sized but have twice the size of pointers to sized types <ul> <li>Pointers to slices also store the number of elements of the slice.</li> <li>Pointers to trait objects also store a pointer to a vtable.</li> </ul> </li> <li><abbr title="dynamically sized types">DSTs</abbr> can be provided as type arguments when a bound of <code>?Sized</code>. By default any type parameter has a <code>Sized</code> bound.</li> <li>Traits may be implemented for <abbr title="dynamically sized types">DSTs</abbr>. Unlike type parameters<code>Self: ?Sized</code> by default in trait definitions.</li> <li>Structs may contain a <abbr title="dynamically sized type">DST</abbr> as the last field, this makes the struct itself a <abbr title="dynamically sized type">DST</abbr>.</li> </ul> <p>Notably: <a href="variables.html">variables</a>, function parameters, <a href="items/constant-items.html">const</a> and <a href="items/static-items.html">static</a> items must be <code>Sized</code>.</p> <a class="header" href="print.html#type-layout" id="type-layout"><h1>Type Layout</h1></a> <p>The layout of a type is its size, alignment, and the relative offsets of its fields. For enums, how the discriminant is laid out and interpreted is also part of type layout.</p> <p>Type layout can be changed with each compilation. Instead of trying to document exactly what is done, we only document what is guaranteed today.</p> <a class="header" href="print.html#size-and-alignment" id="size-and-alignment"><h2>Size and Alignment</h2></a> <p>All values have an alignment and size.</p> <p>The <em>alignment</em> of a value specifies what addresses are valid to store the value at. A value of alignment <code>n</code> must only be stored at an address that is a multiple of n. For example, a value with an alignment of 2 must be stored at an even address, while a value with an alignment of 1 can be stored at any address. Alignment is measured in bytes, and must be at least 1, and always a power of 2. The alignment of a value can be checked with the <a href="../std/mem/fn.align_of_val.html"><code>align_of_val</code></a> function.</p> <p>The <em>size</em> of a value is the offset in bytes between successive elements in an array with that item type including alignment padding. The size of a value is always a multiple of its alignment. The size of a value can be checked with the <a href="../std/mem/fn.size_of_val.html"><code>size_of_val</code></a> function.</p> <p>Types where all values have the same size and alignment known at compile time implement the <a href="../std/marker/trait.Sized.html"><code>Sized</code></a> trait and can be checked with the <a href="../std/mem/fn.size_of.html"><code>size_of</code></a> and <a href="../std/mem/fn.align_of.html"><code>align_of</code></a> functions. Types that are not <a href="../std/marker/trait.Sized.html"><code>Sized</code></a> are known as <a href="dynamically-sized-types.html">dynamically sized types</a>. Since all values of a <code>Sized</code> type share the same size and alignment, we refer to those shared values as the size of the type and the alignment of the type respectively.</p> <a class="header" href="print.html#primitive-data-layout" id="primitive-data-layout"><h2>Primitive Data Layout</h2></a> <p>The size of most primitives is given in this table.</p> <table><thead><tr><th>Type </th><th> <code>size_of::\<Type>()</code></th></tr></thead><tbody> <tr><td>bool </td><td> 1</td></tr> <tr><td>u8 </td><td> 1</td></tr> <tr><td>u16 </td><td> 2</td></tr> <tr><td>u32 </td><td> 4</td></tr> <tr><td>u64 </td><td> 8</td></tr> <tr><td>i8 </td><td> 1</td></tr> <tr><td>i16 </td><td> 2</td></tr> <tr><td>i32 </td><td> 4</td></tr> <tr><td>i64 </td><td> 8</td></tr> <tr><td>f32 </td><td> 4</td></tr> <tr><td>f64 </td><td> 8</td></tr> <tr><td>char </td><td> 4</td></tr> </tbody></table> <p><code>usize</code> and <code>isize</code> have a size big enough to contain every address on the target platform. For example, on a 32 bit target, this is 4 bytes and on a 64 bit target, this is 8 bytes.</p> <p>Most primitives are generally aligned to their size, although this is platform-specific behavior. In particular, on x86 u64 and f64 are only aligned to 32 bits.</p> <a class="header" href="print.html#pointers-and-references-layout" id="pointers-and-references-layout"><h2>Pointers and References Layout</h2></a> <p>Pointers and references have the same layout. Mutability of the pointer or reference does not change the layout.</p> <p>Pointers to sized types have the same size and alignment as <code>usize</code>.</p> <p>Pointers to unsized types are sized. The size and alignemnt is guaranteed to be at least equal to the size and alignment of a pointer.</p> <blockquote> <p>Note: Though you should not rely on this, all pointers to <abbr title="Dynamically Sized Types">DSTs</abbr> are currently twice the size of the size of <code>usize</code> and have the same alignment.</p> </blockquote> <a class="header" href="print.html#array-layout" id="array-layout"><h2>Array Layout</h2></a> <p>Arrays are laid out so that the <code>nth</code> element of the array is offset from the start of the array by <code>n * the size of the type</code> bytes. An array of <code>[T; n]</code> has a size of <code>size_of::<T>() * n</code> and the same alignment of <code>T</code>.</p> <a class="header" href="print.html#slice-layout" id="slice-layout"><h2>Slice Layout</h2></a> <p>Slices have the same layout as the section of the array they slice.</p> <blockquote> <p>Note: This is about the raw <code>[T]</code> type, not pointers (<code>&[T]</code>, <code>Box<[T]></code>, etc.) to slices.</p> </blockquote> <a class="header" href="print.html#tuple-layout" id="tuple-layout"><h2>Tuple Layout</h2></a> <p>Tuples do not have any guarantes about their layout.</p> <p>The exception to this is the unit tuple (<code>()</code>) which is guaranteed as a zero-sized type to have a size of 0 and an alignment of 1.</p> <a class="header" href="print.html#trait-object-layout" id="trait-object-layout"><h2>Trait Object Layout</h2></a> <p>Trait objects have the same layout as the value the trait object is of.</p> <blockquote> <p>Note: This is about the raw trait object types, not pointers (<code>&Trait</code>, <code>Box<Trait></code>, etc.) to trait objects.</p> </blockquote> <a class="header" href="print.html#closure-layout" id="closure-layout"><h2>Closure Layout</h2></a> <p>Closures have no layout guarantees.</p> <a class="header" href="print.html#representations" id="representations"><h2>Representations</h2></a> <p>All user-defined composite types (<code>struct</code>s, <code>enum</code>, and <code>union</code>s) have a <em>representation</em> that specifies what the layout is for the type.</p> <p>The possible representations for a type are the default representation, <code>C</code>, the primitive representations, and <code>packed</code>. Multiple representations can be applied to a single type.</p> <p>The representation of a type can be changed by applying the [<code>repr</code> attribute] to it. The following example shows a struct with a <code>C</code> representation.</p> <pre><code>#[repr(C)] struct ThreeInts { first: i16, second: i8, third: i32 } </code></pre> <blockquote> <p>Note: As a consequence of the representation being an attribute on the item, the representation does not depend on generic parameters. Any two types with the same name have the same representation. For example, <code>Foo<Bar></code> and <code>Foo<Baz></code> both have the same representation.</p> </blockquote> <p>The representation of a type does not change the layout of its fields. For example, a struct with a <code>C</code> representation that contains a struct <code>Inner</code> with the default representation will not change the layout of Inner.</p> <a class="header" href="print.html#the-default-representation" id="the-default-representation"><h3>The Default Representation</h3></a> <p>Nominal types without a <code>repr</code> attribute have the default representation. Informally, this representation is also called the <code>rust</code> representation.</p> <p>There are no guarantees of data layout made by this representation.</p> <a class="header" href="print.html#the-c-representation" id="the-c-representation"><h3>The <code>C</code> Representation</h3></a> <p>The <code>C</code> representation is designed for dual purposes. One purpose is for creating types that are interoptable with the C Language. The second purpose is to create types that you can soundly performing operations that rely on data layout such as reinterpreting values as a different type.</p> <p>Because of this dual purpose, it is possible to create types that are not useful for interfacing with the C programming language.</p> <p>This representation can be applied to structs, unions, and enums.</p> <a class="header" href="print.html#reprc-structs" id="reprc-structs"><h4>#[repr(C)] Structs</h4></a> <p>The alignment of the struct is the alignment of the most-aligned field in it.</p> <p>The size and offset of fields is determined by the following algorithm.</p> <p>Start with a current offset of 0 bytes.</p> <p>For each field in declaration order in the struct, first determine the size and alignment of the field. If the current offset is not a multiple of the field's alignment, then add padding bytes to the current offset until it is a multiple of the field's alignment. The offset for the field is what the current offset is now. Then increase the current offset by the size of the field.</p> <p>Finally, the size of the struct is the current offset rounded up to the nearest multiple of the struct's alignment.</p> <p>Here is this algorithm described in psudeocode.</p> <pre><code class="language-rust ignore">struct.alignment = struct.fields().map(|field| field.alignment).max(); let current_offset = 0; for field in struct.fields_in_declaration_order() { // Increase the current offset so that it's a multiple of the alignment // of this field. For the first field, this will always be zero. // The skipped bytes are called padding bytes. current_offset += field.alignment % current_offset; struct[field].offset = current_offset; current_offset += field.size; } struct.size = current_offset + current_offset % struct.alignment; </code></pre> <blockquote> <p>Note: This algorithm can produce zero-sized structs. This differs from C where structs without data still have a size of one byte.</p> </blockquote> <a class="header" href="print.html#reprc-unions" id="reprc-unions"><h4>#[repr(C)] Unions</h4></a> <p>A union declared with <code>#[repr(C)]</code> will have the same size and alignment as an equivalent C union declaration in the C language for the target platform. The union will have a size of the maximum size of all of its fields rounded to its alignment, and an alignment of the maximum alignment of all of its fields. These maximums may come from different fields.</p> <pre><code>#[repr(C)] union Union { f1: u16, f2: [u8; 4], } assert_eq!(std::mem::size_of::<Union>(), 4); // From f2 assert_eq!(std::mem::align_of::<Union>(), 2); // From f1 #[repr(C)] union SizeRoundedUp { a: u32, b: [u16; 3], } assert_eq!(std::mem::size_of::<SizeRoundedUp>(), 8); // Size of 6 from b, // rounded up to 8 from // alignment of a. assert_eq!(std::mem::align_of::<SizeRoundedUp>(), 4); // From a </code></pre> <a class="header" href="print.html#reprc-enums" id="reprc-enums"><h4>#[repr(C)] Enums</h4></a> <p>For <a href="items/enumerations.html#custom-discriminant-values-for-field-less-enumerations">C-like enumerations</a>, the <code>C</code> representation has the size and alignment of the default <code>enum</code> size and alignment for the target platform's C ABI.</p> <blockquote> <p>Note: The enum representation in C is implementation defined, so this is really a "best guess". In particular, this may be incorrect when the C code of interest is compiled with certain flags.</p> </blockquote> <blockquote> <p>Warning: There are crucial differences between an <code>enum</code> in the C language and Rust's C-like enumerations with this representation. An <code>enum</code> in C is mostly a <code>typedef</code> plus some named constants; in other words, an object of an <code>enum</code> type can hold any integer value. For example, this is often used for bitflags in <code>C</code>. In contrast, Rust’s C-like enumerations can only legally hold the discrimnant values, everything else is undefined behaviour. Therefore, using a C-like enumeration in FFI to model a C <code>enum</code> is often wrong.</p> </blockquote> <p>It is an error for <a href="items/enumerations.html#zero-variant-enums">zero-variant enumerations</a> to have the <code>C</code> representation.</p> <p>For all other enumerations, the layout is unspecified.</p> <p>Likewise, combining the <code>C</code> representation with a primitive representation, the layout is unspecified.</p> <a class="header" href="print.html#primitive-representations" id="primitive-representations"><h3>Primitive representations</h3></a> <p>The <em>primitive representations</em> are the representations with the same names as the primitive integer types. That is: <code>u8</code>, <code>u16</code>, <code>u32</code>, <code>u64</code>, <code>usize</code>, <code>i8</code>, <code>i16</code>, <code>i32</code>, <code>i64</code>, and <code>isize</code>.</p> <p>Primitive representations can only be applied to enumerations.</p> <p>For <a href="items/enumerations.html#custom-discriminant-values-for-field-less-enumerations">C-like enumerations</a>, they set the size and alignment to be the same as the primitive type of the same name. For example, a C-like enumeration with a <code>u8</code> representation can only have discriminants between 0 and 255 inclusive.</p> <p>It is an error for <a href="items/enumerations.html#zero-variant-enums">zero-variant enumerations</a> to have a primitive representation.</p> <p>For all other enumerations, the layout is unspecified.</p> <p>Likewise, combining two primitive representations together is unspecified.</p> <a class="header" href="print.html#the-align-representation" id="the-align-representation"><h3>The <code>align</code> Representation</h3></a> <p>The <code>align</code> representation can be used on <code>struct</code>s and <code>union</code>s to raise the alignment of the type to a given value.</p> <p>Alignment is specified as a parameter in the form of <code>#[repr(align(x))]</code>. The alignment value must be a power of two of type <code>u32</code>. The <code>align</code> representation can raise the alignment of a type to be greater than it's primitive alignment, it cannot lower the alignment of a type.</p> <p>The <code>align</code> and <code>packed</code> representations cannot be applied on the same type and a <code>packed</code> type cannot transitively contain another <code>align</code>ed type.</p> <a class="header" href="print.html#the-packed-representation" id="the-packed-representation"><h3>The <code>packed</code> Representation</h3></a> <p>The <code>packed</code> representation can only be used on <code>struct</code>s and <code>union</code>s.</p> <p>It modifies the representation (either the default or <code>C</code>) by removing any padding bytes and forcing the alignment of the type to <code>1</code>.</p> <p>The <code>align</code> and <code>packed</code> representations cannot be applied on the same type and a <code>packed</code> type cannot transitively contain another <code>align</code>ed type.</p> <blockquote> <p>Warning: Dereferencing an unaligned pointer is [undefined behaviour] and it is possible to <a href="https://github.com/rust-lang/rust/issues/27060">safely create unaligned pointers to <code>packed</code> fields</a>. Like all ways to create undefined behavior in safe Rust, this is a bug.</p> </blockquote> <a class="header" href="print.html#interior-mutability" id="interior-mutability"><h1>Interior Mutability</h1></a> <p>Sometimes a type needs be mutated while having multiple aliases. In Rust this is achieved using a pattern called <em>interior mutability</em>. A type has interior mutability if its internal state can be changed through a <a href="types.html#shared-references-">shared reference</a> to it. This goes against the usual <a href="behavior-considered-undefined.html">requirement</a> that the value pointed to by a shared reference is not mutated.</p> <p><a href="../std/cell/struct.UnsafeCell.html"><code>std::cell::UnsafeCell<T></code></a> type is the only allowed way in Rust to disable this requirement. When <code>UnsafeCell<T></code> is immutably aliased, it is still safe to mutate, or obtain a mutable reference to, the <code>T</code> it contains. As with all other types, it is undefined behavior to have multiple <code>&mut UnsafeCell<T></code> aliases.</p> <p>Other types with interior mutability can be created by using <code>UnsafeCell<T></code> as a field. The standard library provides a variety of types that provide safe interior mutability APIs. For example, <a href="../std/cell/struct.RefCell.html"><code>std::cell::RefCell<T></code></a> uses run-time borrow checks to ensure the usual rules around multiple references. The <a href="../std/sync/atomic/index.html"><code>std::sync::atomic</code></a> module contains types that wrap a value that is only accessed with atomic operations, allowing the value to be shared and mutated across threads.</p> <a class="header" href="print.html#subtyping" id="subtyping"><h1>Subtyping</h1></a> <p>Subtyping is implicit and can occur at any stage in type checking or inference. Subtyping in Rust is very restricted and occurs only due to variance with respect to lifetimes and between types with higher ranked lifetimes. If we were to erase lifetimes from types, then the only subtyping would be due to type equality.</p> <p>Consider the following example: string literals always have <code>'static</code> lifetime. Nevertheless, we can assign <code>s</code> to <code>t</code>:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn bar<'a>() { let s: &'static str = "hi"; let t: &'a str = s; } #}</code></pre></pre> <p>Since <code>'static</code> "lives longer" than <code>'a</code>, <code>&'static str</code> is a subtype of <code>&'a str</code>.</p> <a class="header" href="print.html#type-coercions" id="type-coercions"><h1>Type coercions</h1></a> <p>Coercions are defined in <a href="https://github.com/rust-lang/rfcs/blob/master/text/0401-coercions.md">RFC 401</a>. <a href="https://github.com/rust-lang/rfcs/blob/master/text/1558-closure-to-fn-coercion.md">RFC 1558</a> then expanded on that. A coercion is implicit and has no syntax.</p> <a class="header" href="print.html#coercion-sites" id="coercion-sites"><h2>Coercion sites</h2></a> <p>A coercion can only occur at certain coercion sites in a program; these are typically places where the desired type is explicit or can be derived by propagation from explicit types (without type inference). Possible coercion sites are:</p> <ul> <li> <p><code>let</code> statements where an explicit type is given.</p> <p>For example, <code>42</code> is coerced to have type <code>i8</code> in the following:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { let _: i8 = 42; #}</code></pre></pre> </li> <li> <p><code>static</code> and <code>const</code> statements (similar to <code>let</code> statements).</p> </li> <li> <p>Arguments for function calls</p> <p>The value being coerced is the actual parameter, and it is coerced to the type of the formal parameter.</p> <p>For example, <code>42</code> is coerced to have type <code>i8</code> in the following:</p> <pre><pre class="playpen"><code class="language-rust">fn bar(_: i8) { } fn main() { bar(42); } </code></pre></pre> <p>For method calls, the receiver (<code>self</code> parameter) can only take advantage of <a href="print.html#unsized-coercions">unsized coercions</a>.</p> </li> <li> <p>Instantiations of struct or variant fields</p> <p>For example, <code>42</code> is coerced to have type <code>i8</code> in the following:</p> <pre><pre class="playpen"><code class="language-rust">struct Foo { x: i8 } fn main() { Foo { x: 42 }; } </code></pre></pre> </li> <li> <p>Function results, either the final line of a block if it is not semicolon-terminated or any expression in a <code>return</code> statement</p> <p>For example, <code>42</code> is coerced to have type <code>i8</code> in the following:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { fn foo() -> i8 { 42 } #}</code></pre></pre> </li> </ul> <p>If the expression in one of these coercion sites is a coercion-propagating expression, then the relevant sub-expressions in that expression are also coercion sites. Propagation recurses from these new coercion sites. Propagating expressions and their relevant sub-expressions are:</p> <ul> <li> <p>Array literals, where the array has type <code>[U; n]</code>. Each sub-expression in the array literal is a coercion site for coercion to type <code>U</code>.</p> </li> <li> <p>Array literals with repeating syntax, where the array has type <code>[U; n]</code>. The repeated sub-expression is a coercion site for coercion to type <code>U</code>.</p> </li> <li> <p>Tuples, where a tuple is a coercion site to type <code>(U_0, U_1, ..., U_n)</code>. Each sub-expression is a coercion site to the respective type, e.g. the zeroth sub-expression is a coercion site to type <code>U_0</code>.</p> </li> <li> <p>Parenthesized sub-expressions (<code>(e)</code>): if the expression has type <code>U</code>, then the sub-expression is a coercion site to <code>U</code>.</p> </li> <li> <p>Blocks: if a block has type <code>U</code>, then the last expression in the block (if it is not semicolon-terminated) is a coercion site to <code>U</code>. This includes blocks which are part of control flow statements, such as <code>if</code>/<code>else</code>, if the block has a known type.</p> </li> </ul> <a class="header" href="print.html#coercion-types" id="coercion-types"><h2>Coercion types</h2></a> <p>Coercion is allowed between the following types:</p> <ul> <li> <p><code>T</code> to <code>U</code> if <code>T</code> is a subtype of <code>U</code> (<em>reflexive case</em>)</p> </li> <li> <p><code>T_1</code> to <code>T_3</code> where <code>T_1</code> coerces to <code>T_2</code> and <code>T_2</code> coerces to <code>T_3</code> (<em>transitive case</em>)</p> <p>Note that this is not fully supported yet</p> </li> <li> <p><code>&mut T</code> to <code>&T</code></p> </li> <li> <p><code>*mut T</code> to <code>*const T</code></p> </li> <li> <p><code>&T</code> to <code>*const T</code></p> </li> <li> <p><code>&mut T</code> to <code>*mut T</code></p> </li> <li> <p><code>&T</code> or <code>&mut T</code> to <code>&U</code> if <code>T</code> implements <code>Deref<Target = U></code>. For example:</p> <pre><pre class="playpen"><code class="language-rust">use std::ops::Deref; struct CharContainer { value: char, } impl Deref for CharContainer { type Target = char; fn deref<'a>(&'a self) -> &'a char { &self.value } } fn foo(arg: &char) {} fn main() { let x = &mut CharContainer { value: 'y' }; foo(x); //&mut CharContainer is coerced to &char. } </code></pre></pre> </li> <li> <p><code>&mut T</code> to <code>&mut U</code> if <code>T</code> implements <code>DerefMut<Target = U></code>.</p> </li> <li> <p>TyCtor(<code>T</code>) to TyCtor(<code>U</code>), where TyCtor(<code>T</code>) is one of</p> <ul> <li><code>&T</code></li> <li><code>&mut T</code></li> <li><code>*const T</code></li> <li><code>*mut T</code></li> <li><code>Box<T></code></li> </ul> <p>and where <code>T</code> can obtained from <code>U</code> by <a href="print.html#unsized-coercions">unsized coercion</a>.</p> <!--In the future, coerce_inner will be recursively extended to tuples and structs. In addition, coercions from sub-traits to super-traits will be added. See [RFC 401] for more details.--> </li> <li> <p>Non capturing closures to <code>fn</code> pointers</p> </li> </ul> <a class="header" href="print.html#unsized-coercions" id="unsized-coercions"><h3>Unsized Coercions</h3></a> <p>The following coercions are called <code>unsized coercions</code>, since they relate to converting sized types to unsized types, and are permitted in a few cases where other coercions are not, as described above. They can still happen anywhere else a coercion can occur.</p> <p>Two traits, [<code>Unsize</code>] and [<code>CoerceUnsized</code>], are used to assist in this process and expose it for library use. The compiler following coercions are built-in and, if <code>T</code> can be coerced to <code>U</code> with one of the, then the compiler will provide an implementation of <code>Unsize<U></code> for <code>T</code>:</p> <ul> <li> <p><code>[T; n]</code> to <code>[T]</code>.</p> </li> <li> <p><code>T</code> to <code>U</code>, when <code>U</code> is a trait object type and either <code>T</code> implements <code>U</code> or <code>T</code> is a trait object for a subtrait of <code>U</code>.</p> </li> <li> <p><code>Foo<..., T, ...></code> to <code>Foo<..., U, ...></code>, when:</p> <ul> <li><code>Foo</code> is a struct.</li> <li><code>T</code> implements <code>Unsize<U></code>.</li> <li>The last field of <code>Foo</code> has a type involving <code>T</code>.</li> <li>If that field has type <code>Bar<T></code>, then <code>Bar<T></code> implements <code>Unsized<Bar<U>></code>.</li> <li>T is not part of the type of any other fields.</li> </ul> </li> </ul> <p>Additionally, a type <code>Foo<T></code> can implement <code>CoerceUnsized<Foo<U>></code> when <code>T</code> implements <code>Unsize<U></code> or <code>CoerceUnsized<Foo<U>></code>. This allows it to provide a unsized coercion to <code>Foo<U></code>.</p> <blockquote> <p>Note: While the definition of the unsized coercions and their implementation has been stabilized, the traits themselves are not yet stable and therefore can't be used directly in stable Rust.</p> </blockquote> <a class="header" href="print.html#destructors" id="destructors"><h1>Destructors</h1></a> <p>When an <a href="glossary.html#initialized">initialized</a> <a href="variables.html">variable</a> in Rust goes out of scope or a <a href="expressions.html#temporary-lifetimes">temporary</a> is no longer needed its <em>destructor</em> is run. <a href="expressions/operator-expr.html#assignment-expressions">Assignment</a> also runs the destructor of its left-hand operand, unless it's an unitialized variable. If a <a href="types.html#struct-types">struct</a> variable has been partially initialized, only its initialized fields are dropped.</p> <p>The destrutor of a type consists of</p> <ol> <li>Calling its <a href="../std/ops/trait.Drop.html"><code>std::ops::Drop::drop</code></a> method, if it has one.</li> <li>Recursively running the destructor of all of its fields. <ul> <li>The fields of a <a href="types.html#struct-types">struct</a>, <a href="types.html#tuple-types">tuple</a> or <a href="types.html#enumerated-types">enum variant</a> are dropped in declaration order. *</li> <li>The elements of an <a href="types.html#array-and-slice-types">array</a> or owned <a href="types.html#array-and-slice-types">slice</a> are dropped from the first element to the last. *</li> <li>The captured values of a <a href="types.html#closure-types">closure</a> are dropped in an unspecified order.</li> <li><a href="types.html#trait-objects">Trait objects</a> run the destructor of the underlying type.</li> <li>Other types don't result in any further drops.</li> </ul> </li> </ol> <p>* This order was stabilized in <a href="https://github.com/rust-lang/rfcs/blob/master/text/1857-stabilize-drop-order.md">RFC 1857</a>.</p> <p>Variables are dropped in reverse order of declaration. Variables declared in the same pattern drop in an unspecified ordered.</p> <p>If a destructor must be run manually, such as when implementing your own smart pointer, <a href="../std/ptr/fn.drop_in_place.html"><code>std::ptr::drop_in_place</code></a> can be used.</p> <p>Some examples:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { struct ShowOnDrop(&'static str); impl Drop for ShowOnDrop { fn drop(&mut self) { println!("{}", self.0); } } { let mut overwritten = ShowOnDrop("Drops when overwritten"); overwritten = ShowOnDrop("drops when scope ends"); } # println!(""); { let declared_first = ShowOnDrop("Dropped last"); let declared_last = ShowOnDrop("Dropped first"); } # println!(""); { // Tuple elements drop in forwards order let tuple = (ShowOnDrop("Tuple first"), ShowOnDrop("Tuple second")); } # println!(""); loop { // Tuple expression doesn't finish evaluating so temporaries drop in reverse order: let partial_tuple = (ShowOnDrop("Temp first"), ShowOnDrop("Temp second"), break); } # println!(""); { let moved; // No destructor run on assignment. moved = ShowOnDrop("Drops when moved"); // drops now, but is then uninitialized moved; let uninitialized: ShowOnDrop; // Only first element drops let mut partially_initialized: (ShowOnDrop, ShowOnDrop); partially_initialized.0 = ShowOnDrop("Partial tuple first"); } #}</code></pre></pre> <a class="header" href="print.html#not-running-destructors" id="not-running-destructors"><h2>Not running destructors</h2></a> <p>Not running destructors in Rust is safe even if it has a type that isn't <code>'static</code>. <a href="../std/mem/union.ManuallyDrop.html"><code>std::mem::ManuallyDrop</code></a> provides a wrapper to prevent a variable or field from being dropped automatically.</p> <a class="header" href="print.html#special-types-and-traits" id="special-types-and-traits"><h1>Special types and traits</h1></a> <p>Certain types and traits that exist in <a href="../std/index.html">the standard library</a> are known to the Rust compiler. This chapter documents the special features of these types and traits.</p> <a class="header" href="print.html#boxt" id="boxt"><h2><code>Box<T></code></h2></a> <p><a href="../std/boxed/struct.Box.html"><code>Box<T></code></a> has a few special features that Rust doesn't currently allow for user defined types.</p> <ul> <li>The <a href="expressions/operator-expr.html#the-dereference-operator">dereference operator</a> for <code>Box<T></code> produces a place which can be moved from. This means that the <code>*</code> operator and the destructor of <code>Box<T></code> are built-in to the language.</li> <li><a href="items/traits.html#associated-functions-and-methods">Methods</a> can take <code>Box<Self></code> as a receiver.</li> <li>A trait may be implemented for <code>Box<T></code> in the same crate as <code>T</code>, which the <a href="items/implementations.html#trait-implementation-coherence">orphan rules</a> prevent for other generic types.</li> </ul> <a class="header" href="print.html#unsafecellt" id="unsafecellt"><h2><code>UnsafeCell<T></code></h2></a> <p><a href="../std/cell/struct.UnsafeCell.html"><code>std::cell::UnsafeCell<T></code></a> is used for <a href="interior-mutability.html">interior mutability</a>. It ensures that the compiler doesn't perform optimisations that are incorrect for such types. It also ensures that <a href="items/static-items.html"><code>static</code> items</a> which have a type with interior mutability aren't placed in memory marked as read only.</p> <a class="header" href="print.html#phantomdatat" id="phantomdatat"><h2><code>PhantomData<T></code></h2></a> <p><a href="../std/marker/struct.PhantomData.html"><code>std::marker::PhantomData<T></code></a> is a zero-sized, minimum alignment, type that is considered to own a <code>T</code> for the purposes of <a href="../nomicon/subtyping.html">variance</a>, <a href="../nomicon/dropck.html">drop check</a> and <a href="print.html#auto-traits">auto traits</a>.</p> <a class="header" href="print.html#operator-traits" id="operator-traits"><h2>Operator Traits</h2></a> <p>The traits in <a href="../std/ops/index.html"><code>std::ops</code></a> and <a href="../std/cmp/index.html"><code>std::cmp</code></a> are used to overload <a href="expressions/operator-expr.html">operators</a>, <a href="expressions/array-expr.html#array-and-slice-indexing-expressions">indexing expressions</a> and <a href="expressions/call-expr.html">call expressions</a>.</p> <a class="header" href="print.html#deref-and-derefmut" id="deref-and-derefmut"><h2><code>Deref</code> and <code>DerefMut</code></h2></a> <p>As well as overloading the unary <code>*</code> operator, <a href="../std/ops/trait.Deref.html"><code>Deref</code></a> and <a href="../std/ops/trait.DerefMut.html"><code>DerefMut</code></a> are also used in <a href="expressions/method-call-expr.html">method resolution</a> and <a href="type-coercions.html#coercion-types">deref coercions</a>.</p> <a class="header" href="print.html#drop" id="drop"><h2><code>Drop</code></h2></a> <p>The <a href="../std/ops/trait.Drop.html"><code>Drop</code></a> trait provides a <a href="destructors.html">destructor</a>, to be run whenever a value of this type is to be destroyed.</p> <a class="header" href="print.html#copy" id="copy"><h2><code>Copy</code></h2></a> <p>The <a href="../std/marker/trait.Copy.html"><code>Copy</code></a> trait changes the semantics of a type implementing it. Values whose type implements <code>Copy</code> are copied rather than moved upon assignment. <code>Copy</code> cannot be implemented for types which implement <code>Drop</code>, or which have fields that are not <code>Copy</code>. <code>Copy</code> is implemented by the compiler for</p> <ul> <li><a href="types.html#numeric-types">Numeric types</a></li> <li><code>char</code> and <code>bool</code></li> <li><a href="types.html#tuple-types">Tuples</a> of <code>Copy</code> types</li> <li><a href="types.html#array-and-slice-types">Arrays</a> of <code>Copy</code> types</li> <li><a href="types.html#shared-references-">Shared references</a></li> <li><a href="types.html#raw-pointers-const-and-mut">Raw pointers</a></li> <li><a href="types.html#function-pointer-types">Function pointers</a> and <a href="types.html#function-item-types">function item types</a></li> </ul> <a class="header" href="print.html#clone" id="clone"><h2><code>Clone</code></h2></a> <p>The <a href="../std/clone/trait.Clone.html"><code>Clone</code></a> trait is a supertrait of <code>Copy</code>, so it also needs compiler generated implementations. It is implemented by the compiler for the following types:</p> <ul> <li>Types with a built-in <code>Copy</code> implementation (see above)</li> <li><a href="types.html#tuple-types">Tuples</a> of <code>Clone</code> types</li> <li><a href="types.html#array-and-slice-types">Arrays</a> of <code>Clone</code> types</li> </ul> <a class="header" href="print.html#send" id="send"><h2><code>Send</code></h2></a> <p>The <a href="../std/marker/trait.Send.html"><code>Send</code></a> trait indicates that a value of this type is safe to send from one thread to another.</p> <a class="header" href="print.html#sync" id="sync"><h2><code>Sync</code></h2></a> <p>The <a href="../std/marker/trait.Sync.html"><code>Sync</code></a> trait indicates that a value of this type is safe to share between multiple threads. This trait must be implemented for all types used in immutable <a href="items/static-items.html"><code>static</code> items</a>.</p> <a class="header" href="print.html#auto-traits" id="auto-traits"><h2>Auto traits</h2></a> <p>The <a href="../std/marker/trait.Send.html"><code>Send</code></a>, <a href="../std/marker/trait.Sync.html"><code>Sync</code></a>, <a href="../std/panic/trait.UnwindSafe.html"><code>UnwindSafe</code></a> and <a href="../std/panic/trait.RefUnwindSafe.html"><code>RefUnwindSafe</code></a> traits are <em>auto traits</em>. Auto traits have special properties.</p> <p>First, auto traits are automatically implemented using the following rules:</p> <ul> <li><code>&T</code>, <code>&mut T</code>, <code>*const T</code>, <code>*mut T</code>, <code>[T; n]</code> and <code>[T]</code> implement the trait if <code>T</code> does.</li> <li>Function item types and function pointers automatically implement the trait.</li> <li>Structs, enums, unions and tuples implement the trait if all of their fields do.</li> <li>Closures implement the trait if the types of all of their captures do. A closure that captures a <code>T</code> by shared reference and a <code>U</code> by value implements any auto traits that both <code>&T</code> and <code>U</code> do.</li> </ul> <p>Auto traits can also have negative implementations, shown as <code>impl !AutoTrait for T</code> in the standard library documentation, that override the automatic implementations. For example <code>*mut T</code> has a negative implementation of <code>Send</code>, and so <code>*mut T</code> and <code>(*mut T,)</code> are not <code>Send</code>. Finally, auto traits may be added as a bound to any <a href="types.html#trait-objects">trait object</a>: <code>Box<Debug + Send + UnwindSafe></code> is a valid type.</p> <a class="header" href="print.html#sized" id="sized"><h2><code>Sized</code></h2></a> <p>The <a href="../std/marker/trait.Sized.html"><code>Sized</code></a> trait indicates that the size of this type is known at compile-time; that is, it's not a <a href="dynamically-sized-types.html">dynamically sized type</a>. <a href="types.html#type-parameters">Type parameters</a> are <code>Sized</code> by default. <code>Sized</code> is always implemented automatically by the compiler, not by <a href="items/implementations.html">implementation items</a>.</p> <a class="header" href="print.html#memory-model" id="memory-model"><h1>Memory model</h1></a> <p>A Rust program's memory consists of a static set of <em>items</em> and a <em>heap</em>. Immutable portions of the heap may be safely shared between threads, mutable portions may not be safely shared, but several mechanisms for effectively-safe sharing of mutable values, built on unsafe code but enforcing a safe locking discipline, exist in the standard library.</p> <p>Allocations in the stack consist of <em>variables</em>, and allocations in the heap consist of <em>boxes</em>.</p> <a class="header" href="print.html#memory-allocation-and-lifetime" id="memory-allocation-and-lifetime"><h1>Memory allocation and lifetime</h1></a> <p>The <em>items</em> of a program are those functions, modules and types that have their value calculated at compile-time and stored uniquely in the memory image of the rust process. Items are neither dynamically allocated nor freed.</p> <p>The <em>heap</em> is a general term that describes boxes. The lifetime of an allocation in the heap depends on the lifetime of the box values pointing to it. Since box values may themselves be passed in and out of frames, or stored in the heap, heap allocations may outlive the frame they are allocated within. An allocation in the heap is guaranteed to reside at a single location in the heap for the whole lifetime of the allocation - it will never be relocated as a result of moving a box value.</p> <a class="header" href="print.html#memory-ownership" id="memory-ownership"><h2>Memory ownership</h2></a> <p>When a stack frame is exited, its local allocations are all released, and its references to boxes are dropped.</p> <a class="header" href="print.html#variables" id="variables"><h1>Variables</h1></a> <p>A <em>variable</em> is a component of a stack frame, either a named function parameter, an anonymous <a href="expressions.html#temporary-lifetimes">temporary</a>, or a named local variable.</p> <p>A <em>local variable</em> (or <em>stack-local</em> allocation) holds a value directly, allocated within the stack's memory. The value is a part of the stack frame.</p> <p>Local variables are immutable unless declared otherwise. For example: <code>let mut x = ...</code>.</p> <p>Function parameters are immutable unless declared with <code>mut</code>. The <code>mut</code> keyword applies only to the following parameter. For example: <code>|mut x, y|</code> and <code>fn f(mut x: Box<i32>, y: Box<i32>)</code> declare one mutable variable <code>x</code> and one immutable variable <code>y</code>.</p> <p>Methods that take either <code>self</code> or <code>Box<Self></code> can optionally place them in a mutable variable by prefixing them with <code>mut</code> (similar to regular arguments). For example:</p> <pre><pre class="playpen"><code class="language-rust"> # #![allow(unused_variables)] #fn main() { trait Changer: Sized { fn change(mut self) {} fn modify(mut self: Box<Self>) {} } #}</code></pre></pre> <p>Local variables are not initialized when allocated. Instead, the entire frame worth of local variables are allocated, on frame-entry, in an uninitialized state. Subsequent statements within a function may or may not initialize the local variables. Local variables can be used only after they have been initialized; this is enforced by the compiler.</p> <a class="header" href="print.html#linkage" id="linkage"><h1>Linkage</h1></a> <p>The Rust compiler supports various methods to link crates together both statically and dynamically. This section will explore the various methods to link Rust crates together, and more information about native libraries can be found in the <a href="../book/ffi.html">FFI section of the book</a>.</p> <p>In one session of compilation, the compiler can generate multiple artifacts through the usage of either command line flags or the <code>crate_type</code> attribute. If one or more command line flags are specified, all <code>crate_type</code> attributes will be ignored in favor of only building the artifacts specified by command line.</p> <ul> <li> <p><code>--crate-type=bin</code>, <code>#[crate_type = "bin"]</code> - A runnable executable will be produced. This requires that there is a <code>main</code> function in the crate which will be run when the program begins executing. This will link in all Rust and native dependencies, producing a distributable binary.</p> </li> <li> <p><code>--crate-type=lib</code>, <code>#[crate_type = "lib"]</code> - A Rust library will be produced. This is an ambiguous concept as to what exactly is produced because a library can manifest itself in several forms. The purpose of this generic <code>lib</code> option is to generate the "compiler recommended" style of library. The output library will always be usable by rustc, but the actual type of library may change from time-to-time. The remaining output types are all different flavors of libraries, and the <code>lib</code> type can be seen as an alias for one of them (but the actual one is compiler-defined).</p> </li> <li> <p><code>--crate-type=dylib</code>, <code>#[crate_type = "dylib"]</code> - A dynamic Rust library will be produced. This is different from the <code>lib</code> output type in that this forces dynamic library generation. The resulting dynamic library can be used as a dependency for other libraries and/or executables. This output type will create <code>*.so</code> files on linux, <code>*.dylib</code> files on osx, and <code>*.dll</code> files on windows.</p> </li> <li> <p><code>--crate-type=staticlib</code>, <code>#[crate_type = "staticlib"]</code> - A static system library will be produced. This is different from other library outputs in that the Rust compiler will never attempt to link to <code>staticlib</code> outputs. The purpose of this output type is to create a static library containing all of the local crate's code along with all upstream dependencies. The static library is actually a <code>*.a</code> archive on linux and osx and a <code>*.lib</code> file on windows. This format is recommended for use in situations such as linking Rust code into an existing non-Rust application because it will not have dynamic dependencies on other Rust code.</p> </li> <li> <p><code>--crate-type=cdylib</code>, <code>#[crate_type = "cdylib"]</code> - A dynamic system library will be produced. This is used when compiling Rust code as a dynamic library to be loaded from another language. This output type will create <code>*.so</code> files on Linux, <code>*.dylib</code> files on macOS, and <code>*.dll</code> files on Windows.</p> </li> <li> <p><code>--crate-type=rlib</code>, <code>#[crate_type = "rlib"]</code> - A "Rust library" file will be produced. This is used as an intermediate artifact and can be thought of as a "static Rust library". These <code>rlib</code> files, unlike <code>staticlib</code> files, are interpreted by the Rust compiler in future linkage. This essentially means that <code>rustc</code> will look for metadata in <code>rlib</code> files like it looks for metadata in dynamic libraries. This form of output is used to produce statically linked executables as well as <code>staticlib</code> outputs.</p> </li> <li> <p><code>--crate-type=proc-macro</code>, <code>#[crate_type = "proc-macro"]</code> - The output produced is not specified, but if a <code>-L</code> path is provided to it then the compiler will recognize the output artifacts as a macro and it can be loaded for a program. If a crate is compiled with the <code>proc-macro</code> crate type it will forbid exporting any items in the crate other than those functions tagged <code>#[proc_macro_derive]</code> and those functions must also be placed at the crate root. Finally, the compiler will automatically set the <code>cfg(proc_macro)</code> annotation whenever any crate type of a compilation is the <code>proc-macro</code> crate type.</p> </li> </ul> <p>Note that these outputs are stackable in the sense that if multiple are specified, then the compiler will produce each form of output at once without having to recompile. However, this only applies for outputs specified by the same method. If only <code>crate_type</code> attributes are specified, then they will all be built, but if one or more <code>--crate-type</code> command line flags are specified, then only those outputs will be built.</p> <p>With all these different kinds of outputs, if crate A depends on crate B, then the compiler could find B in various different forms throughout the system. The only forms looked for by the compiler, however, are the <code>rlib</code> format and the dynamic library format. With these two options for a dependent library, the compiler must at some point make a choice between these two formats. With this in mind, the compiler follows these rules when determining what format of dependencies will be used:</p> <ol> <li> <p>If a static library is being produced, all upstream dependencies are required to be available in <code>rlib</code> formats. This requirement stems from the reason that a dynamic library cannot be converted into a static format.</p> <p>Note that it is impossible to link in native dynamic dependencies to a static library, and in this case warnings will be printed about all unlinked native dynamic dependencies.</p> </li> <li> <p>If an <code>rlib</code> file is being produced, then there are no restrictions on what format the upstream dependencies are available in. It is simply required that all upstream dependencies be available for reading metadata from.</p> <p>The reason for this is that <code>rlib</code> files do not contain any of their upstream dependencies. It wouldn't be very efficient for all <code>rlib</code> files to contain a copy of <code>libstd.rlib</code>!</p> </li> <li> <p>If an executable is being produced and the <code>-C prefer-dynamic</code> flag is not specified, then dependencies are first attempted to be found in the <code>rlib</code> format. If some dependencies are not available in an rlib format, then dynamic linking is attempted (see below).</p> </li> <li> <p>If a dynamic library or an executable that is being dynamically linked is being produced, then the compiler will attempt to reconcile the available dependencies in either the rlib or dylib format to create a final product.</p> <p>A major goal of the compiler is to ensure that a library never appears more than once in any artifact. For example, if dynamic libraries B and C were each statically linked to library A, then a crate could not link to B and C together because there would be two copies of A. The compiler allows mixing the rlib and dylib formats, but this restriction must be satisfied.</p> <p>The compiler currently implements no method of hinting what format a library should be linked with. When dynamically linking, the compiler will attempt to maximize dynamic dependencies while still allowing some dependencies to be linked in via an rlib.</p> <p>For most situations, having all libraries available as a dylib is recommended if dynamically linking. For other situations, the compiler will emit a warning if it is unable to determine which formats to link each library with.</p> </li> </ol> <p>In general, <code>--crate-type=bin</code> or <code>--crate-type=lib</code> should be sufficient for all compilation needs, and the other options are just available if more fine-grained control is desired over the output format of a Rust crate.</p> <a class="header" href="print.html#static-and-dynamic-c-runtimes" id="static-and-dynamic-c-runtimes"><h2>Static and dynamic C runtimes</h2></a> <p>The standard library in general strives to support both statically linked and dynamically linked C runtimes for targets as appropriate. For example the <code>x86_64-pc-windows-msvc</code> and <code>x86_64-unknown-linux-musl</code> targets typically come with both runtimes and the user selects which one they'd like. All targets in the compiler have a default mode of linking to the C runtime. Typically targets are linked dynamically by default, but there are exceptions which are static by default such as:</p> <ul> <li><code>arm-unknown-linux-musleabi</code></li> <li><code>arm-unknown-linux-musleabihf</code></li> <li><code>armv7-unknown-linux-musleabihf</code></li> <li><code>i686-unknown-linux-musl</code></li> <li><code>x86_64-unknown-linux-musl</code></li> </ul> <p>The linkage of the C runtime is configured to respect the <code>crt-static</code> target feature. These target features are typically configured from the command line via flags to the compiler itself. For example to enable a static runtime you would execute:</p> <pre><code class="language-notrust">rustc -C target-feature=+crt-static foo.rs </code></pre> <p>whereas to link dynamically to the C runtime you would execute:</p> <pre><code class="language-notrust">rustc -C target-feature=-crt-static foo.rs </code></pre> <p>Targets which do not support switching between linkage of the C runtime will ignore this flag. It's recommended to inspect the resulting binary to ensure that it's linked as you would expect after the compiler succeeds.</p> <p>Crates may also learn about how the C runtime is being linked. Code on MSVC, for example, needs to be compiled differently (e.g. with <code>/MT</code> or <code>/MD</code>) depending on the runtime being linked. This is exported currently through the <code>target_feature</code> attribute (note this is a nightly feature):</p> <pre><code class="language-rust ignore">#[cfg(target_feature = "crt-static")] fn foo() { println!("the C runtime should be statically linked"); } #[cfg(not(target_feature = "crt-static"))] fn foo() { println!("the C runtime should be dynamically linked"); } </code></pre> <p>Also note that Cargo build scripts can learn about this feature through <a href="http://doc.crates.io/environment-variables.html#environment-variables-cargo-sets-for-build-scripts">environment variables</a>. In a build script you can detect the linkage via:</p> <pre><pre class="playpen"><code class="language-rust">use std::env; fn main() { let linkage = env::var("CARGO_CFG_TARGET_FEATURE").unwrap_or(String::new()); if linkage.contains("crt-static") { println!("the C runtime will be statically linked"); } else { println!("the C runtime will be dynamically linked"); } } </code></pre></pre> <p>To use this feature locally, you typically will use the <code>RUSTFLAGS</code> environment variable to specify flags to the compiler through Cargo. For example to compile a statically linked binary on MSVC you would execute:</p> <pre><code class="language-ignore notrust">RUSTFLAGS='-C target-feature=+crt-static' cargo build --target x86_64-pc-windows-msvc </code></pre> <a class="header" href="print.html#unsafety" id="unsafety"><h1>Unsafety</h1></a> <p>Unsafe operations are those that potentially violate the memory-safety guarantees of Rust's static semantics.</p> <p>The following language level features cannot be used in the safe subset of Rust:</p> <ul> <li>Dereferencing a <a href="types.html#pointer-types">raw pointer</a>.</li> <li>Reading or writing a <a href="items/static-items.html#mutable-statics">mutable static variable</a>.</li> <li>Reading a field of a <a href="items/unions.html"><code>union</code></a>, or writing to a field of a union that isn't <a href="special-types-and-traits.html#copy"><code>Copy</code></a>.</li> <li>Calling an unsafe function (including an intrinsic or foreign function).</li> <li>Implementing an unsafe trait.</li> </ul> <a class="header" href="print.html#unsafe-functions" id="unsafe-functions"><h1>Unsafe functions</h1></a> <p>Unsafe functions are functions that are not safe in all contexts and/or for all possible inputs. Such a function must be prefixed with the keyword <code>unsafe</code> and can only be called from an <code>unsafe</code> block or another <code>unsafe</code> function.</p> <a class="header" href="print.html#unsafe-blocks-1" id="unsafe-blocks-1"><h1>Unsafe blocks</h1></a> <p>A block of code can be prefixed with the <code>unsafe</code> keyword, to permit calling <code>unsafe</code> functions or dereferencing raw pointers within a safe function.</p> <p>When a programmer has sufficient conviction that a sequence of potentially unsafe operations is actually safe, they can encapsulate that sequence (taken as a whole) within an <code>unsafe</code> block. The compiler will consider uses of such code safe, in the surrounding context.</p> <p>Unsafe blocks are used to wrap foreign libraries, make direct use of hardware or implement features not directly present in the language. For example, Rust provides the language features necessary to implement memory-safe concurrency in the language but the implementation of threads and message passing is in the standard library.</p> <p>Rust's type system is a conservative approximation of the dynamic safety requirements, so in some cases there is a performance cost to using safe code. For example, a doubly-linked list is not a tree structure and can only be represented with reference-counted pointers in safe code. By using <code>unsafe</code> blocks to represent the reverse links as raw pointers, it can be implemented with only boxes.</p> <a class="header" href="print.html#behavior-considered-undefined" id="behavior-considered-undefined"><h2>Behavior considered undefined</h2></a> <p>Rust code, including within <code>unsafe</code> blocks and <code>unsafe</code> functions is incorrect if it exhibits any of the behaviors in the following list. It is the programmer's responsibility when writing <code>unsafe</code> code that it is not possible to let <code>safe</code> code exhibit these behaviors.</p> <ul> <li>Data races.</li> <li>Dereferencing a null or dangling raw pointer.</li> <li>Unaligned pointer reading and writing outside of <a href="https://doc.rust-lang.org/std/ptr/fn.read_unaligned.html"><code>read_unaligned</code></a> and <a href="https://doc.rust-lang.org/std/ptr/fn.write_unaligned.html"><code>write_unaligned</code></a>.</li> <li>Reads of <a href="http://llvm.org/docs/LangRef.html#undefined-values">undef</a> (uninitialized) memory.</li> <li>Breaking the <a href="http://llvm.org/docs/LangRef.html#pointer-aliasing-rules">pointer aliasing rules</a> on accesses through raw pointers; a subset of the rules used by C.</li> <li><code>&mut T</code> and <code>&T</code> follow LLVM’s scoped <a href="http://llvm.org/docs/LangRef.html#noalias">noalias</a> model, except if the <code>&T</code> contains an <a href="https://doc.rust-lang.org/std/cell/struct.UnsafeCell.html"><code>UnsafeCell<U></code></a>.</li> <li>Mutating non-mutable data — that is, data reached through a shared reference or data owned by a <code>let</code> binding), unless that data is contained within an <a href="https://doc.rust-lang.org/std/cell/struct.UnsafeCell.html"><code>UnsafeCell<U></code></a>.</li> <li>Invoking undefined behavior via compiler intrinsics: <ul> <li>Indexing outside of the bounds of an object with <a href="https://doc.rust-lang.org/std/primitive.pointer.html#method.offset"><code>offset</code></a> with the exception of one byte past the end of the object.</li> <li>Using <a href="https://doc.rust-lang.org/std/ptr/fn.copy_nonoverlapping.html"><code>std::ptr::copy_nonoverlapping_memory</code></a>, a.k.a. the <code>memcpy32</code>and <code>memcpy64</code> intrinsics, on overlapping buffers.</li> </ul> </li> <li>Invalid values in primitive types, even in private fields and locals: <ul> <li>Dangling or null references and boxes.</li> <li>A value other than <code>false</code> (<code>0</code>) or <code>true</code> (<code>1</code>) in a <code>bool</code>.</li> <li>A discriminant in an <code>enum</code> not included in the type definition.</li> <li>A value in a <code>char</code> which is a surrogate or above <code>char::MAX</code>.</li> <li>Non-UTF-8 byte sequences in a <code>str</code>.</li> </ul> </li> </ul> <a class="header" href="print.html#behavior-not-considered-unsafe" id="behavior-not-considered-unsafe"><h2>Behavior not considered <code>unsafe</code></h2></a> <p>The Rust compiler does not consider the following behaviors <em>unsafe</em>, though a programmer may (should) find them undesirable, unexpected, or erroneous.</p> <a class="header" href="print.html#deadlocks" id="deadlocks"><h5>Deadlocks</h5></a> <a class="header" href="print.html#leaks-of-memory-and-other-resources" id="leaks-of-memory-and-other-resources"><h5>Leaks of memory and other resources</h5></a> <a class="header" href="print.html#exiting-without-calling-destructors" id="exiting-without-calling-destructors"><h5>Exiting without calling destructors</h5></a> <a class="header" href="print.html#exposing-randomized-base-addresses-through-pointer-leaks" id="exposing-randomized-base-addresses-through-pointer-leaks"><h5>Exposing randomized base addresses through pointer leaks</h5></a> <a class="header" href="print.html#integer-overflow" id="integer-overflow"><h5>Integer overflow</h5></a> <p>If a program contains arithmetic overflow, the programmer has made an error. In the following discussion, we maintain a distinction between arithmetic overflow and wrapping arithmetic. The first is erroneous, while the second is intentional.</p> <p>When the programmer has enabled <code>debug_assert!</code> assertions (for example, by enabling a non-optimized build), implementations must insert dynamic checks that <code>panic</code> on overflow. Other kinds of builds may result in <code>panics</code> or silently wrapped values on overflow, at the implementation's discretion.</p> <p>In the case of implicitly-wrapped overflow, implementations must provide well-defined (even if still considered erroneous) results by using two's complement overflow conventions.</p> <p>The integral types provide inherent methods to allow programmers explicitly to perform wrapping arithmetic. For example, <code>i32::wrapping_add</code> provides two's complement, wrapping addition.</p> <p>The standard library also provides a <code>Wrapping<T></code> newtype which ensures all standard arithmetic operations for <code>T</code> have wrapping semantics.</p> <p>See <a href="https://github.com/rust-lang/rfcs/blob/master/text/0560-integer-overflow.md">RFC 560</a> for error conditions, rationale, and more details about integer overflow.</p> <a class="header" href="print.html#influences" id="influences"><h1>Influences</h1></a> <p>Rust is not a particularly original language, with design elements coming from a wide range of sources. Some of these are listed below (including elements that have since been removed):</p> <ul> <li>SML, OCaml: algebraic data types, pattern matching, type inference, semicolon statement separation</li> <li>C++: references, RAII, smart pointers, move semantics, monomorphization, memory model</li> <li>ML Kit, Cyclone: region based memory management</li> <li>Haskell (GHC): typeclasses, type families</li> <li>Newsqueak, Alef, Limbo: channels, concurrency</li> <li>Erlang: message passing, thread failure, <strike>linked thread failure</strike>, <strike>lightweight concurrency</strike></li> <li>Swift: optional bindings</li> <li>Scheme: hygienic macros</li> <li>C#: attributes</li> <li>Ruby: <strike>block syntax</strike></li> <li>NIL, Hermes: <strike>typestate</strike></li> <li><a href="http://www.unicode.org/reports/tr31/">Unicode Annex #31</a>: identifier and pattern syntax</li> </ul> <a class="header" href="print.html#as-yet-undocumented-features" id="as-yet-undocumented-features"><h1>As-yet-undocumented Features</h1></a> <p>Several accepted, stabilized, and implemented RFCs lack documentation in this reference, The Book, <em>Rust by Example</em>, or some combination of those three. Until we have written reference documentation for these features, we provide links to other sources of information about them. Therefore, expect this list to shrink!</p> <ul> <li><a href="https://github.com/rust-lang/rfcs/pull/40"><code>libstd</code> facade</a></li> <li><a href="https://github.com/rust-lang/rfcs/pull/48">Trait reform</a> – some partial documentation exists (the use of <code>Self</code>), but not for everything: e.g. coherence and orphan rules.</li> <li><a href="https://github.com/rust-lang/rfcs/pull/49">Attributes on <code>match</code> arms</a> – the underlying idea is documented in the [Attributes] section, but the applicability to internal items is never specified.</li> <li><a href="https://github.com/rust-lang/rfcs/pull/131">Flexible target specification</a> - Some---but not all---flags are documented in <a href="attributes.html#conditional-compilation">Conditional compilation</a></li> <li>[Require parentheses for chained comparisons]</li> <li><a href="https://github.com/rust-lang/rfcs/pull/1717"><code>dllimport</code></a> - one element mentioned but not explained at <a href="attributes.html#ffi-attributes">FFI attributes</a></li> <li><a href="https://github.com/rust-lang/rfcs/pull/1721">define <code>crt_link</code></a></li> <li><a href="https://github.com/rust-lang/rfcs/pull/1725">define <code>unaligned_access</code></a></li> </ul> <a class="header" href="print.html#glossary" id="glossary"><h1>Glossary</h1></a> <a class="header" href="print.html#abstract-syntax-tree" id="abstract-syntax-tree"><h3>Abstract Syntax Tree</h3></a> <p>An ‘abstract syntax tree’, or ‘AST’, is an intermediate representation of the structure of the program when the compiler is compiling it.</p> <a class="header" href="print.html#alignment" id="alignment"><h3>Alignment</h3></a> <p>The alignment of a value specifies what addresses values are preferred to start at. Always a power of two. References to a value must be aligned. <a href="type-layout.html#size-and-alignment">More</a>.</p> <a class="header" href="print.html#arity" id="arity"><h3>Arity</h3></a> <p>Arity refers to the number of arguments a function or operator takes. For some examples, <code>f(2, 3)</code> and <code>g(4, 6)</code> have arity 2, while <code>h(8, 2, 6)</code> has arity 3. The <code>!</code> operator has arity 1.</p> <a class="header" href="print.html#array" id="array"><h3>Array</h3></a> <p>An array, sometimes also called a fixed-size array or an inline array, is a value describing a collection of elements, each selected by an index that can be computed at run time by the program. It occupies a contiguous region of memory.</p> <a class="header" href="print.html#associated-item" id="associated-item"><h3>Associated Item</h3></a> <p>An associated item is an item that is associated with another item. Associated items are defined in <a href="items/implementations.html">implementations</a> and declared in <a href="items/traits.html">traits</a>. Only functions, constants, and type aliases can be associated.</p> <a class="header" href="print.html#bound" id="bound"><h3>Bound</h3></a> <p>Bounds are constraints on a type or trait. For example, if a bound is placed on the argument a function takes, types passed to that function must abide by that constraint.</p> <a class="header" href="print.html#combinator" id="combinator"><h3>Combinator</h3></a> <p>Combinators are higher-order functions that apply only functions and earlier defined combinators to provide a result from its arguments. They can be used to manage control flow in a modular fashion.</p> <a class="header" href="print.html#dispatch" id="dispatch"><h3>Dispatch</h3></a> <p>Dispatch is the mechanism to determine which specific version of code is actually run when it involves polymorphism. Two major forms of dispatch are static dispatch and dynamic dispatch. While Rust favors static dispatch, it also supports dynamic dispatch through a mechanism called ‘trait objects’.</p> <a class="header" href="print.html#dynamically-sized-type" id="dynamically-sized-type"><h3>Dynamically Sized Type</h3></a> <p>A dynamically sized type (DST) is a type without a statically known size or alignment.</p> <a class="header" href="print.html#expression" id="expression"><h3>Expression</h3></a> <p>An expression is a combination of values, constants, variables, operators and functions that evaluate to a single value, with or without side-effects.</p> <p>For example, <code>2 + (3 * 4)</code> is an expression that returns the value 14.</p> <a class="header" href="print.html#initialized" id="initialized"><h3>Initialized</h3></a> <p>A variable is initialized if it has been assigned a value and hasn't since been moved from. All other memory locations are assumed to be initialized. Only unsafe Rust can create such a memory without initializing it.</p> <a class="header" href="print.html#nominal-types" id="nominal-types"><h3>Nominal Types</h3></a> <p>Types that can be referred to by a path directly. Specifically <a href="items/enumerations.html">enums</a>, <a href="items/structs.html">structs</a>, <a href="items/unions.html">unions</a>, and <a href="types.html#trait-objects">trait objects</a>.</p> <a class="header" href="print.html#object-safe-traits" id="object-safe-traits"><h3>Object Safe Traits</h3></a> <p><a href="items/traits.html">Traits</a> that can be used as <a href="types.html#trait-objects">trait objects</a>. Only traits that follow specific <a href="items/traits.html#object-safety">rules</a> are object safe.</p> <a class="header" href="print.html#prelude" id="prelude"><h3>Prelude</h3></a> <p>Prelude, or The Rust Prelude, is a small collection of items - mostly traits - that are imported into every module of every crate. The traits in the prelude are pervasive.</p> <a class="header" href="print.html#size" id="size"><h3>Size</h3></a> <p>The size of a value has two definitions.</p> <p>The first is that it is how much memory must be allocated to store that value.</p> <p>The second is that it is the offset in bytes between successive elements in an array with that item type.</p> <p>It is a multiple of the alignment, including zero. The size can change depending on compiler version (as new optimizations are made) and target platform (similar to how <code>usize</code> varies per-platform).</p> <p><a href="type-layout.html#size-and-alignment">More</a>.</p> <a class="header" href="print.html#slice" id="slice"><h3>Slice</h3></a> <p>A slice is dynamically-sized view into a contiguous sequence, written as <code>[T]</code>.</p> <p>It is often seen in its borrowed forms, either mutable or shared. The shared slice type is <code>&[T]</code>, while the mutable slice type is <code>&mut [T]</code>, where <code>T</code> represents the element type.</p> <a class="header" href="print.html#statement" id="statement"><h3>Statement</h3></a> <p>A statement is the smallest standalone element of a programming language that commands a computer to perform an action.</p> <a class="header" href="print.html#string-literal" id="string-literal"><h3>String literal</h3></a> <p>A string literal is a string stored directly in the final binary, and so will be valid for the <code>'static</code> duration.</p> <p>Its type is <code>'static</code> duration borrowed string slice, <code>&'static str</code>.</p> <a class="header" href="print.html#string-slice" id="string-slice"><h3>String slice</h3></a> <p>A string slice is the most primitive string type in Rust, written as <code>str</code>. It is often seen in its borrowed forms, either mutable or shared. The shared string slice type is <code>&str</code>, while the mutable string slice type is <code>&mut str</code>.</p> <p>Strings slices are always valid UTF-8.</p> <a class="header" href="print.html#trait" id="trait"><h3>Trait</h3></a> <p>A trait is a language item that is used for describing the functionalities a type must provide. It allows a type to make certain promises about its behavior.</p> <p>Generic functions and generic structs can use traits to constrain, or bound, the types they accept.</p> </main> <nav class="nav-wrapper" aria-label="Page navigation"> <!-- Mobile navigation buttons --> <div style="clear: both"></div> </nav> </div> </div> <nav class="nav-wide-wrapper" aria-label="Page navigation"> </nav> </div> <!-- Local fallback for Font Awesome --> <script> if (getComputedStyle(document.querySelector(".fa")).fontFamily !== "FontAwesome") { var link = document.createElement('link'); link.rel = 'stylesheet'; link.type = 'text/css'; link.href = '_FontAwesome/css/font-awesome.css'; document.head.insertBefore(link, document.head.firstChild) } </script> <script> document.addEventListener('DOMContentLoaded', function() { window.print(); }) </script> <script src="highlight.js"></script> <script src="book.js"></script> <!-- Custom JS script --> </body> </html>