Revision f2c68fb4009ad058aa5c7d3dfeb2c3b989638614 authored by swift-ci on 17 March 2023, 03:05:27 UTC, committed by GitHub on 17 March 2023, 03:05:27 UTC
Add REQUIRES: assert in one the VariadicGenerics tests.
LangRef.html
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
"http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
<title>Swift Language Reference Manual</title>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<meta name="author" content="Chris Lattner">
<meta name="description"
content="Swift Language Reference Manual.">
<link rel="stylesheet" href="_static/swift.css" type="text/css">
<script type="text/javascript" src="toc.js"></script>
</head>
<body>
<h1>Swift Language Reference</h1>
<p>
<!-- The Table of Contents is automatically inserted in this <div>.
Do not delete this <div>. -->
<div id="nav"></div>
</p>
<!-- ********************************************************************* -->
<h2 id="decl">Declarations</h2>
<!-- ********************************************************************* -->
<pre class="grammar">
decl ::= <a href="#decl-class">decl-class</a>
decl ::= <a href="#decl-constructor">decl-constructor</a>
decl ::= <a href="#decl-deinit">decl-deinit</a>
decl ::= <a href="#decl-extension">decl-extension</a>
decl ::= <a href="#decl-func">decl-func</a>
decl ::= <a href="#decl-import">decl-import</a>
decl ::= <a href="#decl-enum">decl-enum</a>
decl ::= <a href="#decl-enum-element">decl-enum-element</a>
decl ::= <a href="#decl-protocol">decl-protocol</a>
decl ::= <a href="#decl-struct">decl-struct</a>
decl ::= <a href="#decl-typealias">decl-typealias</a>
decl ::= <a href="#decl-let">decl-let</a>
decl ::= <a href="#decl-var">decl-var</a>
decl ::= <a href="#decl-subscript">decl-subscript</a>
</pre>
<!-- ===================================================================== -->
<h3 id="decl-top-level">Module-Scope Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
top-level ::= <a href="#brace-item-list">brace-item</a>*
</pre>
<p>The top level of a swift source file is grammatically identical to the
contents of a func decl. Some declarations, however, are restricted to
module scope.
</p>
<!-- _____________________________________________________________________ -->
<h4 id="brace-item-list">Brace Enclosed Items</h4>
<pre class="grammar">
brace-item-list ::= '{' brace-item* '}'
brace-item ::= <a href="#decl">decl</a>
brace-item ::= <a href="#expr">expr</a>
brace-item ::= <a href="#stmt">stmt</a>
</pre>
<p>The brace item list provides a sequencing operation which evaluates the
members of its body in order. Function bodies and the bodies of control
flow statements use braces. Also, the <a
href="#decl-top-level">source file</a> itself is effectively a
brace item list, but without the braces.
</p>
<!-- ===================================================================== -->
<h3 id="decl-import">import Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
decl-import ::= <a href="#attribute-list">attribute-list</a> 'import' import-kind? import-path
import-kind ::= 'typealias'
import-kind ::= 'struct'
import-kind ::= 'class'
import-kind ::= 'enum'
import-kind ::= 'protocol'
import-kind ::= 'var'
import-kind ::= 'func'
import-path ::= <a href="#any-identifier">any-identifier</a> ('.' <a href="#any-identifier">any-identifier</a>)*
</pre>
<p>'import' declarations allow named values and types to be accessed with
local names, even when they are defined in other modules and namespaces. See
the section on <a href="#namebind">name binding</a> for more
information on how these work. import declarations are only allowed at
module scope.</p>
<p>'import' directives only impact a single source file: imports in one
swift file do not affect name lookup in another file. import directives can
only occur at the top level of a file, not within a function or namespace.</p>
<p>An import without an explicit import-kind names a module; all of the
module's members are imported into the current scope. The module's name is
also imported into the current scope in order to allow qualified access to
the module's members, which can be useful for disambiguation.</p>
<p>If an import-kind is provided, the last element of the import path is
taken to be the name of a decl <em>within</em> the module named by the rest of
the path. Only that name is introduced into the current scope; the name of the
module itself is <em>not</em> accessible, nor any other decls within the
module.</p>
<p>Different import-kinds perform different filters on the decls within a
module:</p>
<ul>
<li><code>typealias</code> can be used to import any concrete type (struct,
class, enum, or another typealias). It cannot be used to import protocols,
which are often used for more than just their existential type.</li>
<li><code>struct</code>, <code>class</code>, <code>enum</code> can be used
to import any type whose <i>canonical type</i> is a struct, class,
or enum, respectively. (This allows "Int" to be imported as a struct, for
example, even though its definition in the standard library may be a
typealias for another struct type.)</li>
<li><code>protocol</code> is used to import a protocol</li>
<li><code>var</code> is used to import a module-scoped variable</li>
<li><code>func</code> will import all overloads of a function</li>
</ul>
<pre class="example">
<i>// Import all of the top level symbols and types in a module.</i>
import Swift
<i>// Import a single type.</i>
import typealias Swift.BufferedStream
<i>// Import all addition overloads.</i>
import func Swift.+
</pre>
<!-- ===================================================================== -->
<h3 id="decl-extension">extension Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
decl-extension ::= 'extension' <a href="#type-identifier">type-identifier</a> <a href="#inheritance">inheritance</a>? '{' <a href="#decl">decl</a>* '}'
</pre>
<p>'extension' declarations allow adding member declarations to existing
types, even in other source files and modules. There are different
semantic rules for each type that is extended.
</p>
<!-- _____________________________________________________________________ -->
<h4 id="decl-extension-enum-struct"><a href="#decl-enum">enum</a>, <a
href="#decl-struct">struct</a>, and <a href="#decl-class">class</a>
declaration extensions</h4>
<p>FIXME: Write this section.</p>
<!-- ===================================================================== -->
<h3 id="decl-let">let Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
decl-let ::= <a href="#attribute-list">attribute-list</a> 'let' <a href="#pattern">pattern</a> initializer? (',' pattern initializer?)*
initializer ::= '=' <a href="#expr">expr</a>
</pre>
<p>'let' declarations define an immutable binding between an initializer value
and a name.</p>
<p>Here are some examples of 'let' declarations:</p>
<pre class="example">
<i>// Simple examples.</i>
let a = 4
let c : Int = 42
<i>// This decodes the tuple return value into independently named parts</i>
<i>// and both 'val' and 'err' are in scope after this line.</i>
let (val, err) = foo()
// let declarations require an initializer (though the type is optional).
let b : Int <i>// error: let requires an initializer</i>
// Let bindings of classes make the binding immutable, not the object.
class Rocket {
func blastOff() { ... }
}
let rocket = Rocket()
rocket.blastOff() // okay
rocket = Rocket() <i>// error, cannot change a let binding</i>
</pre>
<!-- "var Declarations" was converted to ReST -->
<!-- ===================================================================== -->
<h3 id="decl-func">func Declarations</h3>
<!-- ===================================================================== -->
<!-- "func Declarations" was converted to ReST -->
<!-- _____________________________________________________________________ -->
<h4 id="func-signature">Function signatures</h4>
<pre class="grammar">
func-signature ::= func-arguments func-signature-result?
func-arguments ::= <a href="#pattern-tuple">pattern-tuple</a>+
func-arguments ::= selector-tuple
<a id="selector-tuple">selector-tuple</a> ::= '(' <a href="#pattern-tuple">pattern-tuple-element</a> ')' (<a href="#identifier">identifier-or-any</a> '(' pattern-tuple-element ')')+
func-signature-result ::= '->' <a href="#type">type</a>
</pre>
<p>A function signature specifies one or more sets of parameter
patterns, plus an optional result type.</p>
<p>When a result type is not written, it is implicitly the empty tuple type,
<tt>()</tt>.</p>
<p>In the body of the function described by a particular signature,
all the variables bound by all of the parameter patterns are in
scope, and the function must return a value of the result type.</p>
<p>An outermost pattern in a function signature must be <a
href="#fully_typed_types">fully-typed</a> and irrefutable. If a result type is
given, it must also be fully-typed.</p>
<p>The type of a function with signature <tt>(P<sub>0</sub>)(P<sub>1</sub>)...(P<sub><i>n</i></sub>) -> R</tt>
is <tt>T<sub>0</sub> -> T<sub>1</sub> -> ... -> T<sub><i>n</i></sub> -> R</tt>,
where <tt>T<sub><i>i</i></sub></tt> is the bottom-up type of the pattern
<tt>P<sub><i>i</i></sub></tt>. This is called "currying". The
behavior of all the intermediate functions (those which do not
return <tt>R</tt>) is to capture their arguments, plus any
arguments from prior patterns, and returns a function which takes
the next set of arguments. When the "uncurried" function is
called (the one taking <tt>T<sub><i>n</i></sub></tt> and returning
<tt>R</tt>), all of the arguments are then available and the
function body is finally evaluated as normal.</p>
<p>A function declared with a selector-style signature
<tt>func(a<sub>0</sub>:T<sub>0</sub>) name<sub>1</sub>(a<sub>1</sub>:T<sub>1</sub>) ... name<sub><i>n</i></sub>(a<sub><i>n</i></sub>:T<sub><i>n</i></sub>) -> R</tt>
has the type <tt>(_:T<sub>0</sub>, name<sub>1</sub>:T<sub>1</sub>, ... name<sub><i>n</i></sub>:T<sub><i>n</i></sub>) -> R</tt>,
that is, the names of the fields in the argument tuple are the
<tt>name<sub><i>n</i></sub></tt> identifiers preceding each argument
pattern. However, in the body of a function
described by a signature, those arguments will be bound using the
corresponding
<tt>a<sub><i>n</i></sub></tt> patterns inside
the arguments. This allows for Cocoa-style keyword function
names such as <tt>doThing(x, withThing:y)</tt> to be defined without
requiring that an awkward keyword name be the same as the
variable name.
<p>Here are some examples of func definitions:</p>
<pre class="example">
<i>// Implicitly returns (), aka <a href="#stdlib-Void">Void</a></i>
func a() {}
<i>// Same as 'a'</i>
func a1() -> Void {}
<i>// Function pointers to a function expression.</i>
var a2 = func ()->() {}
var a3 = func () {}
var a4 = func {}
<i>// Really simple function</i>
func c(_ arg : Int) -> Int { return arg+4 }
<i>// Simple operators.</i>
func [infix_left=190] + (lhs : Int, rhs : Int) -> Int
func [infix_left=160] == (lhs : Int, rhs : Int) -> Bool
<i>// Curried function with multiple return values:</i>
func d(_ a : Int) (b : Int) -> (res1 : Int, res2 : Int) {
return (a,b)
}
<i>// A more realistic example on a trivial type.</i>
struct bankaccount {
amount : Int
static func bankaccount() -> bankaccount {
// Custom 'constructor' logic goes here.
}
func deposit(_ arg : Int) {
amount = amount + arg
}
static func someMetatypeMethod() {}
}
<i>// Dot syntax on metatype.</i>
bankaccount.someMetatypeMethod()
<i>// A function with selector-style signature.</i>
enum PersonOfInterest {
case ColonelMustard
case MissScarlet
}
enum Room {
case Conservatory
case Ballroom
}
enum Weapon {
case Candlestick
case LeadPipe
}
func accuseSuspect(_ suspect:PersonOfInterest)
inRoom(room:Room)
withWeapon(weapon:Weapon) {
print("It was \(suspect) in the \(room) with the \(weapon)")
}
<i>// Calling a selector-style function.</i>
accuseSuspect(.ColonelMustard, inRoom:.Ballroom, withWeapon:.LeadPipe)
</pre>
<!-- ===================================================================== -->
<h3 id="decl-typealias">typealias Declarations</h3>
<!-- ===================================================================== -->
<div class="commentary">
We use the keyword "typealias" instead of "typedef" because it really is an
alias for an existing type, not a "definition" of a new type.
</div>
<pre class="grammar">
decl-typealias ::= typealias-head '=' <a href="#type">type</a>
<a name="typealias-head"></a>typealias-head ::= 'typealias' <a href="#identifier">identifier</a> <a href="#inheritance">inheritance</a>?
</pre>
<p>'typealias' makes a named alias of a type, like a typedef in C. From that
point on, the alias may be used in all situations the specified name is. If an <a href="#inheritance">inheritance</a> clause is provided, it specifies protocols to which the aliased type shall conform.</p>
<p>Here are some examples of type aliases:</p>
<pre class="example">
<i>// location is an alias for a tuple of ints.</i>
typealias location = (x : Int, y : Int)
<i>// pair_fn is a function that takes two ints and returns a tuple.</i>
typealias pair_fn = (Int) -> (Int) -> (first : Int, second : Int)
</pre>
<!-- ===================================================================== -->
<h3 id="decl-enum">enum Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
decl-enum ::= <a href="#attribute-list">attribute-list</a> 'enum' <a href="#identifier">identifier</a> <a href="#generic-params">generic-params</a>? <a href="#inheritance">inheritance</a>? enum-body
enum-body ::= '{' decl* '}'
decl-enum-element ::= <a href="#attribute-list">attribute-list</a> 'case' enum-case (',' enum-case)*
enum-case ::= <a href="#identifier">identifier</a> <a href="#type-tuple">type-tuple</a>? ('->' <a href="#type">type</a>)?
</pre>
<p>An <tt>enum</tt> declaration creates a <a href="#type-enum">enum type</a>.
Here are some examples of enum declarations:</p>
<pre class="example">
<i>// Declares three enums.</i>
enum DataSearchFlags {
case None
case Backward
case Anchored
}
<i>// Shorthand for the above.</i>
enum DataSearchFlags {
case None, Backward, Anchored
}
<i>// Declare discriminated union with enum decl.</i>
enum SomeInts {
case None
case One(Int)
case Two(Int, Int)
}
func f1(_ searchpolicy : DataSearchFlags) <i>// DataSearchFlags is a valid type name</i>
func test1() {
f1(DataSearchFlags.None) <i>// Use of constructor with qualified identifier</i>
f1(.None) <i>// Use of constructor with context sensitive type inference</i>
<i>// "None" has no type argument, so the constructor's type is "DataSearchFlags".</i>
var a : DataSearchFlags = .None
}
enum SomeMoreInts {
case None <i>// Doesn't conflict with previous "None".</i>
case One(Int)
case Two(Int, Int)
}
func f2(_ a : SomeMoreInts)
func test2() {
<i>// Constructors for enum element can be used in the obvious way.</i>
f2(.None)
f2(.One(4))
f2(.Two(1, 2))
<i>// Constructor for None has type "SomeMoreInts".</i>
var a : SomeMoreInts = SomeMoreInts.None
<i>// Constructor for One has type "(Int) -> SomeMoreInts".</i>
var b : (Int) -> SomeMoreInts = SomeMoreInts.One
<i>// Constructor for Two has type "(Int,Int) -> SomeMoreInts".</i>
var c : (Int,Int) -> SomeMoreInts = SomeMoreInts.Two
}
</pre>
<!-- ===================================================================== -->
<h3 id="decl-struct">struct Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
decl-struct ::= <a href="#attribute-list">attribute-list</a> 'struct' <a href="#identifier">identifier</a> <a href="#generic-params">generic-params</a>? <a href="#inheritance">inheritance</a>? '{' decl-struct-body '}'
decl-struct-body ::= <a href="#decl">decl</a>*
</pre>
<p>A struct declares a simple value type that can contain data members and
have methods.</p>
<p>The body of a 'struct' is a list of decls. Stored (non-computed) 'var'
decls declare members with storage in the struct. Other declarations act
like they would in an <a href="#decl-extension">extension</a> of the
struct type.</p>
<p>Here are a few simple examples:</p>
<pre class="example">
struct S1 {
var a : Int, b : Int
}
struct S2 {
var a : Int
func f() -> Int { return b }
var b : Int
}
</pre>
<p>Here are some more realistic examples of structs:</p>
<pre class="example">
struct Point { x : Int, y : Int }
struct Size { width : Int, height : Int }
struct Rect {
origin : Point,
size : Size
typealias CoordinateType = Int
func area() -> Int { return size.width*size.height }
}
func test4() {
var a : Point
var b = Point.Point(1, 2) // Silly but fine.
var c = Point(y = 1, x = 2) // Using metatype.
var x1 = Rect(a, Size(42, 123))
var x2 = Rect(size = Size(width = 42, height=123), origin = a)
var x1_area = x1.width*x1.height
var x1_area2 = x1.area()
}
</pre>
<div class="commentary">
Structs do not support inheritance due to undesirable ripple effects across
the design of the language. For example, method dispatch would arguably need
to become virtual, not static. The storage of the type would arguably need
to become indirected so that an array of T could be implemented sanely
(because we don't know if T is actually a T, or a subclass of T). We'd need
to store the "isa"/vtable in the struct so that virtual method dispatch
could be implemented, and this has additional storage costs. None of these
tradeoffs make sense for the intended use cases we have in mind for structs
(Ints, Floats, Points, Rects, UUIDs, IP addresses, C struct interop, etc,
etc). Said differently: we're trying to force a hard wall
between types that need indirect access by their nature and those types
that need direct access by their nature. The former are called classes. The
latter are called structs.
</div>
<!-- ===================================================================== -->
<h3 id="decl-class">class Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
decl-class ::= <a href="#attribute-list">attribute-list</a> 'class' <a href="#identifier">identifier</a> <a href="#generic-params">generic-params</a>? <a href="#inheritance">inheritance</a>? '{' decl-class-body '}'
decl-class-body ::= <a href="#decl">decl</a>*
</pre>
<p>A class declares a reference type referring to an object which can contain
data members and have methods. Classes support single inheritance;
a parent class should be listed as the first type in the
inheritance list.</p>
<p>The body of a 'class' is a list of decls. Stored (non-computed) 'var' decls
declare members with storage in the class. Non-type 'var' and 'func'
decls declare instance members;type 'var' and 'func' decls declare
members of the class itself. Both class and instance members can
be overridden by a subclass.</p>
<p>Type declarations inside a class act essentially the same way as type
declarations outside a class.</p>
<p>FIXME: For the moment, see classes.rst for more details on the
class system.</p>
<p>FIXME: Add a reference to the section on generics.</p>
<p>The only way to create a new instance of a class is with a
call to one of the class's constructors.</p>
<p>Here is a simple example:</p>
<pre class="example">
class C1 {
var a : Int
var b : Int
}
</pre>
<!-- ===================================================================== -->
<h3 id="decl-protocol">Protocol Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
decl-protocol ::= <a href="#attribute-list">attribute-list</a> 'protocol' <a href="#identifier">identifier</a> <a href="#inheritance">inheritance</a>? '{' protocol-member* '}'
</pre>
<p>A protocol declaration describes an abstract interface implemented by
another type. It consists of a set of declarations, which may be instance
methods or properties. A type <i>conforms</i> to a protocol if it
provides declarations that correspond to each of the declarations in
a protocol.</p>
<p>Here are some examples of protocols:</p>
<pre class="example">
protocol Document {
var title : String
}
</pre>
<!-- _____________________________________________________________________ -->
<h4 id="protocol-member-func">'func' protocol elements</h4>
<pre class="grammar">
protocol-member ::= <a href="#decl-func">decl-func</a>
</pre>
<p>'func' members of a protocol define a value of function type that may be
accessed with dot syntax on a value of the protocol's type. The function
gets an implicit "self" argument of the protocol type or (for a type
function) of the metatype of the protocol.</p>
<!-- _____________________________________________________________________ -->
<h4 id="protocol-member-var">'var' protocol elements</h4>
<pre class="grammar">
protocol-member ::= <a href="#decl-var">decl-var</a>
</pre>
<p>'var' members of a protocol define "property" values that may be accessed
with dot syntax on a value of the protocol's type. The actual
variables may have no storage, and will always be accessed by a getter
and setter. Thus, the variables shall have neither an initializer
nor a getter/setter clause.</p>
<!-- _____________________________________________________________________ -->
<h4 id="protocol-member-subscript">'subscript' protocol elements</h4>
<pre class="grammar">
protocol-member ::= <a href="#subscript-head">subscript-head</a>
</pre>
<p>'subscript' members of a protocol define subscripting operations
that may be accessed with the subscript operator ('[]') applied to a
value of the protocol's type. </p>
<div class="commentary">
TODO: There is currently no way to express a requirement for a
read-only or write-only subscript operation or variable. We may
end up doing this with some kind of 'const' or 'immutable'
attribute.
</div>
<!-- _____________________________________________________________________ -->
<h4 id="protocol-member-typealias">'typealias' protocol elements (associated types)</h4>
<pre class="grammar">
protocol-member ::= <a href="#typealias-head">typealias-head</a> ('=' <a href="#type">type</a>)?
</pre>
<p>'typealias' members of a protocol define associated types, which
are types used within the description of a protocol (typically in
the inputs and outputs of 'func' members) that vary from one
conforming type to another. When an associated type has an <a
href="#inheritance">inheritance</a> clause, any type meant to
satisfy the associated type requirement must conform to each of the
protocols specified within that inheritance clause. If a type is
provided after the '=', it is a default definition for the
associated type that will be used as the type witness if the type
witness cannot be determined in any other way.</p>
<pre class="example">
protocol SequenceType {
typename Iterator : IteratorProtocol
func makeIterator() -> Iterator
}
</pre>
<!-- "subscript Declarations" was converted to ReST -->
<!-- ===================================================================== -->
<h3 id="decl-constructor">constructor Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
decl-constructor ::= <a href="#attribute-list">attribute-list</a> 'init' <a href="#generic-params">generic-params</a>? constructor-signature <a href="#brace-item-list">brace-item-list</a>
constructor-signature ::= <a href="#pattern-tuple">pattern-tuple</a> constructor-result?
constructor-signature ::= <a href="#identifier">identifier-or-any</a> <a href="#selector-tuple">selector-tuple</a> constructor-result?
constructor-result ::= '->' 'Self'
</pre>
<p>'constructor' declares a constructor for a class, struct, or enum. Such
a declaration is used whenever an object is constructed. Specifically,
for classes, it is used when a new expression is written, and for structs
and enums, it is used for function application when the "function"
is a metatype.</p>
<p>FIXME: We haven't decided the precise rules for when constructors are
implicitly declared. Default construction doesn't work right for structs
or enums. We haven't decided what the restrictions are if a member
isn't default-constructible.</p>
<p>A simple example:</p>
<pre class="example">
struct X {
var member : Int
init(x : Int) {
member = x
}
}
var a = X(10)
</pre>
<p>If a class is derived from a superclass, it must explicitly invoke a
superclass constructor using the <tt>super.init</tt> syntax.
<tt>super.init</tt> may only be used in a subclass constructor;
it is not valid in a struct, enum, or root class constructor. Additionally,
<tt>super.init</tt> may only be referenced exactly once per derived
constructor. An example:</p>
<pre class="example">
class View {
var bounds : Rect
init(bounds:Rect) {
self.bounds = bounds
}
}
class Button : View {
var onClick : () -> ()
init(bounds:Rect, onClick:() -> ()) {
super.init(bounds)
self.onClick = onClick
}
}
</pre>
<!-- ===================================================================== -->
<h3 id="decl-deinit">deinitializer Declarations</h3>
<!-- ===================================================================== -->
<pre class="grammar">
decl-deinit ::= <a href="#attribute-list">attribute-list</a> 'deinit' <a href="#brace-item-list">brace-item-list</a>
</pre>
<p>'deinit' declares a deinitializer for a class. This function is called
when there are no longer any references to a class object, just before it
is destroyed. Note that deinitializers can only be declared for classes,
and cannot be declared in extensions. Subclass deinitializers implicitly
invoke their superclass deinitializers after executing.</p>
<p>FIXME: We haven't really decided the precise rules here, but it's probably
a fatal error to either throw an exception or stash a reference to 'self'
in a deinitializer. Not sure what happens when we cause the reference count
of another object to reach zero inside a deinitializer. We might eventually
allow deinitializers in extensions once we have ivars in extensions.</p>
<p>A simple example:</p>
<pre class="example">
class X {
var fd : Int
deinit {
close(fd)
}
}
</pre>
<!-- ===================================================================== -->
<h3 id="attribute-list">Attribute Lists</h3>
<!-- ===================================================================== -->
<pre class="grammar">
attribute-list ::= /*empty*/
attribute-list ::= attribute-list-clause attribute-list
attribute-list-clause ::= '@' attribute
attribute-list-clause ::= '@' attribute ','? attribute-list-clause
attribute ::= attribute-infix
attribute ::= attribute-resilience
attribute ::= attribute-inout
attribute ::= attribute-autoclosure
attribute ::= attribute-noreturn
</pre>
<p>An attribute list is written as a sequence of attributes, each of which has
a leading '@' sign. Attributes can be optionally comma separated.
Attributes may not be repeated within a list.</p>
<!-- _____________________________________________________________________ -->
<h4 id="attribute-infix">Infix Attributes</h4>
<pre class="grammar">
attribute-infix ::= 'infix_left' '=' <a href="#integer_literal">integer_literal</a>
attribute-infix ::= 'infix_right' '=' <a href="#integer_literal">integer_literal</a>
attribute-infix ::= 'infix '=' <a href="#integer_literal">integer_literal</a>
</pre>
<p>The infix attributes may only be applied to the declaration of a
function of binary operator type whose name is an
<a href="#operator"><tt>operator</tt></a>. The name indicates the
associativity of the operator, either left associative, right associative, or
non-associative.</p>
<p>FIXME: Implement these restrictions.</p>
<!-- _____________________________________________________________________ -->
<h4 id="attribute-resilience">Resilience Attribute</h4>
<pre class="grammar">
attribute-resilience ::= 'resilient'
attribute-resilience ::= 'fragile'
attribute-resilience ::= 'born_fragile'
</pre>
<p>See the resilience design.</p>
<!-- _____________________________________________________________________ -->
<h4 id="attribute-inout"><tt>inout</tt> Attribute</h4>
<pre class="grammar">
attribute-inout ::= 'inout'
</pre>
<p><tt>inout</tt> is only valid in a <tt>type</tt> that
appears within either a <a href="#pattern"><tt>pattern</tt></a> of
a <tt>function-signature</tt> or the input type of a function
type.
</p>
<p><tt>inout</tt> indicates that the argument will be passed as an "in-out"
parameter. The caller must pass an lvalue decorated with the <tt>&</tt>
prefix operator as the argument. Semantically, the value of the argument
is passed "in" to the callee to a local value, and that local value is
stored back "out" to the lvalue when the callee exits. This is normally
indistinguishable from pass-by-reference semantics.</p>
<p><tt>inout</tt> differs from traditional pass-by-reference when closures
are involved. If a closure captures an <tt>inout</tt> parameter, the
<em>local value</em> is captured, not the reference. The local value at
the time of function exit is written back to the lvalue.
If the closure outlives the lifetime of the call, the local value lives
independent of the original lvalue; further mutations within the closure
do not affect the lvalue that was passed as the byref argument.
For example, the following code:
<pre class=example>
func foo(x: inout Int) -> () -> Int {
func bar() -> Int {
x += 1
return x
}
// Call 'bar' once while the inout is active.
bar()
return bar
}
var x = 219
var f = foo(&x)
// x is updated to the value of foo's local x at function exit.
print("global x = \(x)")
// These calls only update the captured local 'x', which is now independent
// of the inout parameter.
print("local x = \(f())")
print("local x = \(f())")
print("local x = \(f())")
print("global x = \(x)")
</pre>
will print:
<pre class=example>
global x = 220
local x = 221
local x = 222
local x = 223
global x = 220
</pre>
<p>The type being annotated must be <a href="#materializable">materializable</a>.
The type after annotation is never materializable.
<p>FIXME: we probably need a const-like variant, which permits
r-values (and avoids writeback when the l-value is not physical).
We may also need some way of representing <q>this will be
consumed by the nth curry</q>.
</p>
<!-- _____________________________________________________________________ -->
<h4 id="attribute-autoclosure">autoclosure Attribute</h4>
<pre class="grammar">
attribute-autoclosure ::= 'autoclosure'
</pre>
<p>The <tt>autoclosure</tt> attribute modifies a <a
href="#type-function">function type</a>, changing the behavior of any
assignment into (or initialization of) a value with the function type.
Instead of requiring that the rvalue and lvalue have the same function type,
an "auto closing" function type requires its initializer expression to have
the same type as the function's result type, and it implicitly binds a
closure over this expression. This is typically useful for function arguments
that want to capture computation that can be run lazily.</p>
<p><tt>autoclosure</tt> is only valid in a <tt>type</tt> of a
syntactic function type that is defined to take a syntactic empty tuple.
</p>
<pre class="example">
<i>// An auto closure value. This captures an implicit closure over the</i>
<i>// specified expression, instead of the expression itself.</i>
var a : @autoclosure () -> Int = 4
<i>// Definition of an 'assert' function. Assertions and logging routines</i>
<i>// often want to conditionally evaluate their argument.</i>
func assert(_ condition : @autoclosure () -> Bool)
<i>// Definition of the || operator - it captures its right hand side as</i>
<i>// an autoclosure so it can short-circuit evaluate it.</i>
func [infix_left=110] || (lhs: Bool, rhs: @autoclosure ()->Bool) -> Bool
<i>// Example uses of these functions:</i>
assert(i < j)
if (a == 0 || b == 42) { ... }
</pre>
<!-- _____________________________________________________________________ -->
<h4 id="attribute-noreturn">No Return Attribute</h4>
<pre class="grammar">
attribute-noreturn ::= 'noreturn'
</pre>
<p>Attribute <tt>noreturn</tt> is only valid in the attribute list of a
function declaration or in the attribute list of a <tt>type</tt>
that describes a syntactic function type.
</p>
<p><tt>noreturn</tt> indicates to the compiler that the function will not
return to the caller. This attribute should be used to suppress the
uninitialized variable, missing return warnings and errors. The compiler is
also allowed to more aggressively optimize the code in presence of this
attribute.
</p>
<p>If a function with no a <tt>noreturn</tt> attribute contains a
<tt>return</tt> statement, an error will be raised.
</p>
<!-- ********************************************************************* -->
<h2 id="type">Types</h2>
<!-- ********************************************************************* -->
<pre class="grammar">
type ::= <a href="#attribute-list">attribute-list</a> <a href="#type-function">type-function</a>
type ::= <a href="#attribute-list">attribute-list</a> <a href="#type-array">type-array</a>
type-simple ::= <a href="#type-identifier">type-identifier</a>
type-simple ::= <a href="#type-tuple">type-tuple</a>
type-simple ::= <a href="#type-composition">type-composition</a>
type-simple ::= <a href="#type-metatype">type-metatype</a>
type-simple ::= <a href="#type-optional">type-optional</a>
</pre>
<p>Swift has a small collection of core datatypes that are built into the
compiler. Most user-facing datatypes are defined by the
<a href="#stdlib">standard library</a> or declared as a user defined
types.</p>
<!-- _____________________________________________________________________ -->
<h3>Metatypes</h3>
<p id="metatype">Each type has a corresponding <i>metatype</i>, with the same
name as the type, that is injected into the standard name lookup scope when
a type is <a href="#decl">declared</a>. This allows access to '<a
href="#decl-func">static functions</a>' through dot syntax. For example:</p>
<pre class="example">
// Declares a type 'foo' as well as its metatype.
struct foo {
static func bar() {}
}
// Declares x to be of type foo. A reference to a name in type context
// refers to the type itself.
var x : foo
// Accesses a static function on the foo metatype. In a value context, the
// name of its type refers to its metatype.
foo.bar()
</pre>
<!-- _____________________________________________________________________ -->
<h3 id="fully_typed_types">Fully-Typed Types</h3>
<p>A type may be <i>fully-typed</i>. A type is fully-typed <i>unless</i> one
of the following conditions hold:</p>
<ol>
<li>It is a function type whose result or input type is not
fully-typed.</li>
<li>It is a tuple type with an element that is not
fully-typed. A tuple element is fully-typed unless it has no
explicit type (which is permitted for defaultable elements) or its
explicit type is not fully-typed. In other words, a type is
fully-typed unless it syntactically contains a tuple element with
no explicit type annotation.</li>
</ol>
<p>A type being 'fully-typed' informally means that the type is specified
directly from its type annotation without needing contextual or other
information to resolve its type.</p>
<!-- _____________________________________________________________________ -->
<h3>Materializable Types</h3>
<p id="materializable">A type may be <i>materializable</i>. A type
is materializable unless it is 1) annotated with
a <a href="#attribute-inout"><tt>inout</tt></a> attribute or 2) a
tuple with a non-materializable element type. In general, variables
must have materializable type.</p>
<!-- ===================================================================== -->
<h3 id="type-identifier">Named Types</h3>
<!-- ===================================================================== -->
<pre class="grammar">
type-identifier ::= type-identifier-component ('.' type-identifier-component)*
type-identifier-component ::= <a href="#identifier">identifier</a> <a href="#generic-args">generic-args</a>?
</pre>
<p>Named types may be used simply by using their name. Named types are
introduced by <a href="#decl-typealias">typealias</a> declarations or
through type declarations that expand to one.</p>
<pre class="example">
typealias location = (x : Int, y : Int)
var x : location <i>// use of a named type.</i>
</pre>
<p>Type names may use dot syntax to refer to names types declared in other
modules or types nested within other types.</p>
<pre class="example">
<i>// Direct reference to a member of another module.</i>
var x : Swift.Int
</pre>
<p>Each component of a named type may be followed by a list of generic
parameters for that component enclosed in angle brackets <tt><></tt>.
<pre class="example">
<i>// A generic class definition.</i>
class Dict<K, V> { }
<i>// A variable of a generic instance type.</i>
var map : Dict<String, Int>
</pre>
<!-- ===================================================================== -->
<h3 id="type-tuple">Tuple Types</h3>
<!-- ===================================================================== -->
<div class="commentary">
Tuples are everywhere in Swift: even the argument list of a function is a
tuple of those arguments.
</div>
<pre class="grammar">
type-tuple ::= '(' type-tuple-body? ')'
type-tuple-body ::= type-tuple-element (',' type-tuple-element)* '...'?
type-tuple-element ::= identifier ':' <a href="#type">type</a>
type-tuple-element ::= <a href="#type">type</a>
</pre>
<p>Syntactically, tuple types are simply a (possibly empty) list of
elements enclosed in parentheses. A tuple type with a single, anonymous
element is exactly that type: the parentheses are treated as
grouping parentheses.</p>
<p>Tuples are the low-level form of data aggregation in Swift, and are used as
the building block of <a href="#type-function">function</a> argument lists,
multiple return values, <a href="#decl-enum">enum</a> bodies, etc. Because
tuples are widely accessible and available everywhere in the language,
aggregate data access and transformation is uniform and powerful.</p>
<p>Each element of a tuple contains an optional name followed by a type.</p>
<p>If the tuple body ends with '...', the tuple is a varargs tuple. The type
of the last element is changed from T to T[], and there are special rules
for converting an expression to varargs tuple type.</p>
<pre class="example">
<i>// Variable definitions.</i>
var a : ()
var b : (Int, Int)
var c : (x : (), y : Int)
<i>// Tuple type inferred from an initializers:</i>
var m = () <i>// Type = ()</i>
var n = (x: 1, y: 2) <i>// Type = (x : Int, y : Int)</i>
var o = (1, 2, 3) <i>// Type = (Int, Int, Int)</i>
<i>// Function argument and result is a tuple type.</i>
func foo(_ x : Int, y : Int) -> (val : Int, err : Int)
<i>// enum and struct declarations with tuple values.</i>
struct S {
var (a : Int, b : Int)
}
enum Vertex {
case Point2(x : Int, y : Int)
case Point3(x : Int, y : Int, z : Int)
case Point4(w : Int, x : Int, y : Int, z : Int)
}
</pre>
<!-- ===================================================================== -->
<h3 id="type-function">Function Types</h3>
<!-- ===================================================================== -->
<pre class="grammar">
type-function ::= <a href="#type">type-tuple</a> '->' <a href="#type">type</a>
</pre>
<p>Function types have a single input and single result type, separated by
an arrow. Because each of the types is allowed to be a tuple, we trivially
support multiple arguments and multiple results. "Function" types are
more properly known as a "closure" type, because they can embody any
context captured when the function value was formed.</p>
<p>The result type of a function type must
be <a href="#materializable">materializable</a>. The argument type of a
function is always required to be parenthesized (a tuple). The behavior
of function types may be modified with the <a
href="#attribute-autoclosure"><tt>autoclosure</tt> attribute</a>.</p>
<p>Because of the grammar structure, a nested function type like
"(a) -> (b) -> c" is parsed as "(a) -> ((b) -> c)". This means
that if
you declare this that you can pass it one argument to get a function that
"takes b and returns c" or you can pass two arguments to "get a c". This
is known as <a href="http://en.wikipedia.org/wiki/Currying">currying</a>.
For example:
</p>
<pre class="example">
<i>// A simple function that takes a tuple and returns Int:</i>
var a : (a : Int, b : Int) -> Int
<i>// A simple function that returns multiple values:</i>
var a : (a : Int, b : Int) -> (val: Int, err: Int)
<i>// Declare a function that returns a function:</i>
var x : (Int) -> (Int) -> Int
<i>// y has type (Int) -> Int</i>
var y = x(1)
<i>// z1 and z2 both has type Int, and both have the same value (assuming
// the function had no side effects).</i>
var z1 = x(1)(2)
var z2 = y(2)
<i>// An auto closure value. This captures an implicit closure over the</i>
<i>// specified expression, instead of the expression itself.</i>
var a : @autoclosure () -> Int = 4
</pre>
<!-- ===================================================================== -->
<h3 id="type-enum">Enum Types</h3>
<!-- ===================================================================== -->
<div class="commentary">
'enum' types are known as <a
href="http://en.wikipedia.org/wiki/Algebraic_data_type">algebraic data
types</a> (ADTs) by the broader programming language community.
We name them 'enum' after C enums, because ADTs
fulfill many of the same roles as enums in the C tradition.
</div>
<p>an enum type is a simple discriminated union: the runtime representation
of a value of enum type only has one of the specified elements at a time.</p>
<p>All of the element types of an enum type must
be <a href="#materializable">materializable</a>.</p>
<p>an enum type is defined by a <a href="#decl-enum">enum decl</a>.
<p>Values of enum type may not be default initialized unless the user
provides a no-argument constructor.</p>
<p>The enum metatype has a member corresponding to each declared element.
For elements with a declared type, this member is a function which can
construct an enum containing that element. For elements without a
declared type, the member is simply an enum value for that element. A
enum value has no accessible members except those explicitly defined
by the user.</p>
<p>A reference to a member of the enum metatype can be shortened using <a
href="#expr-delayed-identifier">delayed identifier resolution</a>
with <a href="#typecheck_context">context sensitive type inference</a>.
</p>
<p>The enum's value can be tested and accessed by pattern-matching the enum
against a <a href="#pattern-enum-element">enum element pattern</a>.
<p>TODO: Should attributes be allowed on enum elements?
TODO: Eventually, with generics we'll have equality and inequality operators.
Enum decls should be able to implicitly define these for their types.
TODO: Need pattern matching and element extraction.
</p>
<!-- ===================================================================== -->
<h3 id="type-array">Array Types</h3>
<!-- ===================================================================== -->
<div class="commentary">
Note that array types are parsed inside-out, with the first
bounds clause being the outermost one. This little oddity is required
for the bounds of nested arrays to correspond in sequence to subscript
indexes. That is, given an array "x : Int[5][7][11][13]" and a
chained subscript expression of the form "x[i][j][k][l]", we really
want "i" to be bounded by 5, "j" by 7, and so on. This is probably
the only case where C's rule of "declaration follows use" really makes
sense. There's precedent for this in many languages, including Java and C#.
</div>
<pre class="grammar">
type-array ::= <a href="#type">type-simple</a>
type-array ::= <a href="#type">type-array</a> '[' ']'
type-array ::= <a href="#type">type-array</a> '[' <a href="#expr">expr</a> ']'
</pre>
<p>Array types include a base type and an optional size. Array types indicate
a linear sequence of elements stored consecutively memory. Array elements may
be efficiently indexed in constant time. All array indexes are bounds checked
and out of bound accesses are diagnosed with either a compile time or
runtime failure (TODO: runtime failure mode not specified).</p>
<p>While they look syntactically very similar, an array type with a size has
very different semantics than an array without. In the former case, the type
indicates a declaration of actual storage space. In the later case, the type
indicates storage space allocated elsewhere of runtime-specified size.
</p>
<p>For an array with a size, the size must be more than zero (no
indices would be valid). For now, the array size must be a literal integer.
TODO: Define a notion like C's integer-constant-expression for how constant
folding works.</p>
<p>The element type of an array type must
be <a href="#materializable">materializable</a>.</p>
<p>FIXME: Int[][] not valid because the element type isn't sized. We need
some constraint to reject this, or do we?</p>
<p>Some example array types:</p>
<pre class="example">
<i>// A simple array declaration:</i>
var a : Int[4]
<i>// A reference to another array:</i>
var b : Int[] = a
<i>// Declare a two dimensional array:</i>
var c : Int[4][4]
<i>// Declare a reference to another array, two dimensional:</i>
var d : Int[4][]
<i>// Declare an array of function pointers:</i>
var array_fn_ptrs : (: (Int) -> Int)[42]
var g = array_fn_ptrs[12](4)
<i>// Without parens, this is a function that returns a fixed size array:</i>
var fn_returning_array : (Int) -> Int[42]
var h : Int[42] = fn_returning_array(4)
<i>// You can even have arrays of tuples and other things, these work right
// through composition:</i>
var array_of_tuples : (a : Int, b : Int)[42]
var tuple_of_arrays : (a : Int[42], b : Int[42])
array_of_tuples[12].a = array_of_tuples[13].b
tuple_of_arrays.a[12] = array_of_tuples.b[13]
</pre>
<!-- _____________________________________________________________________ -->
<h3 id="type-metatype">Metatype Types</h3>
<pre class="grammar">
type-metatype ::= type-simple '.' 'Type'
</pre>
<p>Every type has an associated metatype <tt>type(of: T)</tt>. A value of the metatype
type is a reference to a global object which describes the type.
Most metatype types are singleton and therefore require no
storage, but metatypes associated with <a href="#decl-class">class
types</a> follow the same subtyping rules as their associated
class types and therefore are not singleton.</p>
<!-- _____________________________________________________________________ -->
<h3 id="type-optional">Optional Types</h3>
<div class="commentary">
Similar constructs exist in Haskell (<a
href="http://hackage.haskell.org/packages/archive/base/latest/doc/html/Data-Maybe.html">Maybe</a>),
the Boost library (<a
href="http://www.boost.org/doc/libs/1_54_0/libs/optional/doc/html/index.html">Optional</a>),
and C++14 (<a href="http://en.cppreference.com/w/cpp/utility/optional">optional</a>).
</div>
<pre class="grammar">
type-optional ::= type-simple '?'-postfix
</pre>
<p>An optional type is syntactic sugar for the library type
Optional<T>. This is a <a href="#decl-enum">enum</a> with two
cases: None and Some, used to represent a value that may or may not be
present.</p>
<p>Swift provides a number of special, builtin behaviors involving
this library type:
<ul>
<li>There is an implicit conversion from any type <code>T</code> to the
corresponding optional type <code>T?</code>.</li>
<li><code>weak</code> variables must have type <code>T?</code>
and automatically become <code>None</code> when the referent begins
deallocation.</li>
<li>The <a href="#expr-optional">optional-chaining operator</a> works
on values of type <code>T?</code>.</li>
<li>Several other expressions generate values of type
<code>T?</code>.</li>
</ul>
To support these intrinsic use cases, the library is required to
provide functions with these exact signatures:
<ul>
<li><code>func _doesOptionalHaveValueAsBool<T>(v : T?) -> Bool</code></li>
<li><code>func _diagnoseUnexpectedNilOptional()</code></li>
<li><code>func _getOptionalValue<T>(v : T?) -> T</code></li>
</ul>
</p>
<p>Since optional types are part of the
<a href="#type-simple">type-simple</a> grammar, it is not possible to write
<code>T[]?</code> for an optional array. Use <code>(T[]?)</code> instead.
</p>
<p>Some example optional types:</p>
<pre class="example">
<i>// A simple optional declaration:</i>
var a : Int? // equivalent to Optional<Int>
<i>// An empty optional:</i>
var b : Int? = .None
<i>// Declare an array of optionals:</i>
var c : [Int?] = [10, nil, 42]
</pre>
<!-- _____________________________________________________________________ -->
<h3 id="type-composition">Protocol Composition Types</h3>
<pre class="grammar">
type-composition ::= <a href="#type-identifier">type-identifier</a> ('&' <a href="#type-identifier">type-identifier</a>)*
</pre>
<p>A protocol composition type composes together a number of
protocols to describe a type that meets the requirements of each of
those protocols. A protocol composition type <code>A & B</code>
is similar to an explicitly-defined protocol that inherits both
<code>A</code> and <code>B</code></p>
<pre class="example">
protocol C : A, B { }
</pre>
<p>but without the need to introduce a new name.</p>
<div class="commentary">
If we drop implicit conformance to protocols, protocol composition
types become much more important, because they allow you to give a
name to a composition without requiring types to explicitly
conform to that name.
</div>
<p>Each of the types named in the <code>type-composition</code> shall
refer to either a protocol or to a protocol composition. The empty
protocol composition is the keyword <code>Any</code> and every
type conforms to it.
<pre class="example">
<i>// A value that represents any type</i>
var any : Any = 17
<i>// A value that conforms to both the Document and Enumerator protocols</i>
var doc : Document & Enumerator
doc.isEmpty() <i>// uses Enumerator.isEmpty()</i>
doc.title = "Hello" <i>// uses Document.title</i>
</pre>
<!-- _____________________________________________________________________ -->
<h3 id="inheritance">Type Inheritance</h3>
<pre class="grammar">
inheritance ::= ':' <a href="#type-identifier">type-identifier</a> (',' <a href="#type-identifier">type-identifier</a>)*
</pre>
<p>A named type (e.g., a class, struct, enum, or protocol) can
"inherit" some set of protocols, which implies that any object of
that type conforms to each of those protocols. When a protocol
inherits other protocols, the set of requirements from all of those
protocols is effectivel aggregated into the protocol, and a type that
conforms to the current protocol shall conform to each of the
protocols that it inherits.</p>
<p>When a non-protocol type inherits a protocol, it is specifying
explicitly that it conforms to that protocol. The program is
ill-formed if the type does not conform to the protocol.</p>
<pre class="example">
protocol VersionedDocument : Document { <i>// every VersionedDocument is a Document</i>
func bumpVersion()
}
func print(_ doc : Document) { <i>/* ... */</i> }
var myDocument : VersionedDocument;
print(myDocument) <i>// okay: a VersionedDocument is a Document</i>
class StoredHTML : VersionedDocument { <i>// okay: StoredHTML conforms to VersionedDocument</i>
var Title : String
func bumpVersion()
}
</pre>
<!-- 'Patterns' converted to ReST. -->
<!-- ********************************************************************* -->
<h2 id="expr">Expressions</h2>
<!-- ********************************************************************* -->
<div class="commentary">
Support for user-defined operators causes some amount of parsing
to be delayed until after name resolution has occurred. Other
restrictions and disambiguations in the grammar permit the parser
to decide all other aspects of parsing, such as where statements
must be divided.<br><br>
Semicolons in C are generally just clutter. Swift generally tries
to define away the need for them.
</div>
<pre class="grammar">
expr ::= expr-basic
expr ::= <a
href="#expr-trailing-closure">expr-trailing-closure</a> <a href="#expr-cast">expr-cast</a>?
expr-basic ::= expr-sequence <a href="#expr-cast">expr-cast</a>?
expr-sequence ::= <a href="#expr-unary">expr-unary</a> <a href="#expr-binary">expr-binary</a>*
expr-primary ::= <a href="#expr-literal">expr-literal</a>
expr-primary ::= <a href="#expr-identifier">expr-identifier</a>
expr-primary ::= <a href="#expr-super">expr-super</a>
expr-primary ::= <a href="#expr-closure">expr-closure</a>
expr-primary ::= <a href="#expr-anon-closure-arg">expr-anon-closure-arg</a>
expr-primary ::= <a href="#expr-paren">expr-paren</a>
expr-primary ::= <a href="#expr-delayed-identifier">expr-delayed-identifier</a>
expr-postfix ::= expr-primary
expr-postfix ::= expr-postfix <a href="#operator">operator-postfix</a>
expr-postfix ::= <a href="#expr-new">expr-new</a>
expr-postfix ::= <a href="#expr-dot">expr-dot</a>
expr-postfix ::= <a href="#expr-metatype">expr-metatype</a>
expr-postfix ::= <a href="#expr-init">expr-init</a>
expr-postfix ::= <a href="#expr-subscript">expr-subscript</a>
expr-postfix ::= <a href="#expr-call">expr-call</a>
expr-postfix ::= <a href="#expr-optional">expr-optional</a>
expr-force-value ::= <a href="#expr-force-value">expr-force-value</a>
</pre>
<p>At the top level of the expression grammar, expressions are a
sequence of unary expressions joined by binary operators. When
parsing an expr, a binary operator immediately following an
expr-unary continues the expression, and the program is ill-formed
if it is not then followed by another expr-unary. This resolves
an ambiguity which could otherwise arise in statement contexts due
to semicolon elision.</p>
<pre class="example">
5 !- +~123 -+- ~+6
(foo)(())
bar(49+1)
baz()
</pre>
<p>A unary or binary expression may optionally be followed by a
<a href="#expr-cast">cast operator</a>.
<!-- ===================================================================== -->
<h3 id="expr-binary">Binary Operators</h3>
<!-- ===================================================================== -->
<div class="commentary">
Should this use the expr-identifier production to allow qualified
identifiers? This would allow "foo Swift.+ bar". Is ADL or something
like it enough?<br><br>
</div>
<pre class="grammar">
expr-binary ::= op-binary-or-ternary <a href="#expr-unary">expr-unary</a> expr-cast?
op-binary-or-ternary ::= <a href="#operator">operator-binary</a>
op-binary-or-ternary ::= '='
op-binary-or-ternary ::= '?'-infix <a href="#expr">expr-sequence</a> ':'
expr-cast ::= 'is' <a href="#type">type</a>
expr-cast ::= 'as' <a href="#type">type</a>
</pre>
<p>Infix binary expressions are not formed during parsing. Instead,
they are formed after name resolution by building a tree from an
operator-delimited sequence of unary expressions. Precedence and
associativity are determined by the <a href="#attribute-infix">infix</a>
attribute on the resolved names, which must fully agree.</p>
<p>If an operator is used as a binary operator, but name resolution
does not find at least one function of binary operator type, the
expression is ill-formed.</p>
<p>A simple example is:</p>
<pre class="example">
4 + 5 * 123
</pre>
<!-- ===================================================================== -->
<h3 id="expr-binary-builtin">Builtin Binary Operators</h3>
<!-- ===================================================================== -->
<p>In addition to user-defined operators, a handful of builtin operators are
defined that parse inside binary expressions with predefined precedence and
associativity.
<h4 id="expr-assign">Assignment operator</h4>
<p>The assignment operator <tt>a = b</tt> updates the value of <tt>a</tt> with
the value of <tt>b</tt>. Its precedence is hardcoded as if declared as
follows:</p>
<pre class="example">
// Not valid Swift code
infix operator = {
precedence 90
associativity right
}
</pre>
The left-hand operand must be an lvalue, or a tuple of lvalues. Assigning to
a tuple of lvalues performs destructuring reassignment.
<pre class="example">
var (a, b) = (1, 2)
// Swap two values.
(a, b) = (b, a)
// Reassign two values.
(a, b) = (11, 22)
// Reassign two values by destructuring a tuple.
var tuple = (111, 222)
(a, b) = tuple
</pre>
<p>An assignment expression evaluates to void. Unlike C, productions such as
these are invalid:</p>
<pre class="example">
// Error: x = y doesn't return Bool
if x = y { }
// Error: (y = z) doesn't return Int
var x, y, z : Int
x = y = z
</pre>
<h4 id="expr-ternary">Ternary operator</h4>
<p>The ternary operator <tt>a ? b : c</tt> conditionally evaluates its middle
or right operand based on the value of its left operand. Its precedence is
hardcoded as if the middle <tt>? b :</tt> subexpression were a binary operator
declared as follows:</p>
<pre class="example">
// Not valid Swift code
infix operator ?...: {
precedence 100
associativity right
}
</pre>
<p>The subexpression to the left of the
'?' is evaluated, and is converted to 'Bool' using the result's
'boolValue' property if it is not already 'Bool'. If the condition is
true, the subexpression to the right of '?' is evaluated, and its result
becomes the result of the expression. If the
condition is false, the subexpression to the right of ':' is evaluated, and
its result becomes the result of the expression. Only one of the
'?' or ':' subexpressions will be evaluated. The results of the
'?' and ':' subexpressions must be implicitly convertible to a common type,
which becomes the type of the ternary expression.
<pre class="example">
x += b ? y : z
x += a ? b ? y : z : w
for i in 1...101 {
print(i % 15 ? "fizzbuzz"
: i % 3 == 0 ? "fizz"
: i % 5 == 0 ? "buzz"
: "\(i)")
}
</pre>
<h4 id="expr-cast">Cast operators</h4>
<p>Cast expressions influence the types of their subexpressions. They can appear
at the end of a binary operator sequence; their left operand is parsed as if
the cast operators were declared as follows:</p>
<pre class="example">
// Not valid Swift code
infix operator as {
precedence 95
associativity none
}
</pre>
<p>The right operand of all operators is parsed as a type.</p>
<ul>
<li><tt>x as T</tt> will try to cast the value of the expression
<tt>x</tt> to the type <tt>T</tt>. If the type of <tt>x</tt> is
implicitly convertible to <tt>T</tt>, the conversion is performed
and the result of the expression is of type <tt>T</tt>. Otherwise,
the result of the expression is of type <tt>T?</tt>. In this case,
the type of the operand is checked at runtime, and if it is
castable to <tt>T</tt>, the <tt>Optional</tt> result contains
the result of the cast. If the cast fails, the result contains
<tt>.None</tt>. The latter is only permissible when
<tt>T</tt> is a subtype of the compile-time type of <tt>x</tt>.
An example:
<pre class="example">
var b: B = D()
var d: D? = b as D
var b2 = d! as B
</pre>
<li><tt>x is T</tt> will query the type of the value of <tt>x</tt> at runtime.
<tt>T</tt> must be a subtype of the compile-time type of <tt>x</tt>.
If the runtime value of <tt>x</tt> is <tt>T</tt>, the <tt>is</tt> expression
evaluates to true; otherwise, it evaluates to false.
<pre class="example">
if b is D {
var d = (b as D)!
}
</pre>
</ul>
<p><tt>as</tt> and <tt>is</tt> both parse a type for their
right-hand argument. They must be parenthesized if followed by subsequent
operators:
<pre class="example">
(b as D)?.derivedMethod()
((B as D) as D2)
(b is D) ? (b as D)! : D()
</pre>
<!-- ===================================================================== -->
<h3 id="expr-unary">Unary Operators</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-unary ::= <a href="#operator">operator-prefix</a>* <a href="#expr">expr-postfix</a>
</pre>
<p>If an operator is used as a unary operator, but name resolution
does not find at least one function that takes a single argument, the
expression is ill-formed.</p>
<p>Simple examples:</p>
<pre class="example">
i = -j
</pre>
<!-- ===================================================================== -->
<h3 id="expr-literal">Literals</h3>
<!-- ===================================================================== -->
<div class="commentary">
The type of a literal is inferred from its context, to allow things like "4"
to be compatible with any width integer type without 'promotion' rules or
casting. In ambiguous cases like "var x = 4", the literals are forced to
a default type specified by the standard library.
</div>
<pre class="grammar">
expr-literal ::= <a href="#integer_literal">integer_literal</a>
expr-literal ::= <a href="#floating_literal">floating_literal</a>
expr-literal ::= <a href="#string_literal">string_literal</a>
expr-literal ::= expr-array
expr-literal ::= expr-dictionary
expr-literal ::= '__FILE__'
expr-literal ::= '__LINE__'
expr-literal ::= '__COLUMN__'
expr-array ::= '[' expr (',' expr)* ','? ']'
expr-array ::= '[' ']'
expr-dictionary ::= '[' expr ':' expr (',' expr ':' expr)* ','? ']'
expr-dictionary ::= '[' ':' ']'
</pre>
<p>Numeric literals are either integer, floating point, character, or string
depending on its lexical form. The type of the literal is inferred
based on its context. If there is no contextual type information for an
expression, all unresolved types are inferred to 'IntegerLiteralType'
type, to 'FloatLiteralType', and to
'StringLiteralType', respectively.
If a literal is used and these types are not defined, then the code is
malformed.</p>
<p>A literal is compatible with its inferred type if that type implements an
informal protocol required by literals. This informal protocol requires
that the type have an unambiguous "type" function defined whose
result type is the same as the inferred type, and that takes a single
argument that is either itself literal compatible, or is a <a
href="#builtin">builtin</a> integer type.</p>
<p>The '<tt>__FILE__</tt>', '<tt>__LINE__</tt>', and '<tt>__COLUMN__</tt>'
magic identifiers expand to a literal representation of their position in
the source code. '<tt>__FILE__</tt>' expands to a string literal;
'<tt>__LINE__</tt>' and '<tt>__COLUMN__</tt>' each expand to an integer
literal.</p>
<pre class="example">
<i>// File foo.swift</i>
var file = __FILE__ <i>// file : String = "foo.swift"</i>
var line = __LINE__ <i>// line : Int = 4</i>
var col = __COLUMN__ <i>// column : Int = 11</i>
</pre>
<p>If '<tt>__FILE__</tt>', '<tt>__LINE__</tt>', and/or '<tt>__COLUMN__</tt>'
are used as default argument values in a function declaration, they
instead expand to the source location of each function call that
instantiates the default argument.</p>
<pre class="example">
func log(_ message:String,
file:String = __FILE__,
line:Int = __LINE__) {
print("\(file):\(line): \(message)")
}
log("Orders received")
doIt()
log("Job's finished")
</pre>
<!-- ===================================================================== -->
<h3 id="expr-identifier">Identifiers</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-identifier ::= <a href="#identifier">identifier</a> <a href="#generic-args">generic-args</a>?
</pre>
<p>A raw identifier refers to a value found via <a
href="#namebind_value_lookup_unqual">unqualified value lookup</a>, and has
the type of the declaration returned by name lookup and overload
resolution. Value declarations are installed with <a
href="#decl-var">var</a> and the syntactic sugar forms like <a
href="decl-func">func</a> declarations.</p>
<p>If an identifier refers to a generic type, an instance of that generic may
be referenced by following the identifier with a list of type parameters
enclosed in angle brackets <tt><></tt>:</p>
<pre class="example">
<i>// A generic struct.</i>
struct Dict<K,V> {
init() {}
static func fromKeysAndValues(_ keys:K[], values:T[]) -> Dict<K,V> {}
}
<i>// Construct an instance of the generic struct.</i>
var foo = Dict<String, Int>()
<i>// Invoke a type method of an instance of the generic struct.</i>
var bar = Dict<String, Int>.fromKeysAndValues(
["zim", "zang", "zung"],
[ 123, 456, 789 ])
</pre>
<h4 id="expr-generic-disambiguation">Generic disambiguation</h4>
<p>Note that <tt><</tt> and <tt>></tt> are used as both angle brackets in
<a href="#expr-identifier">generic identifiers</a> and as characters in
<a href="#expr-binary">binary operator</a> names. Because of this, there are
potential parsing ambiguities. Swift uses a context-free heuristic to
determine whether to parse an expression involving <tt><</tt> and <tt>></tt>
as a generic parameter list or a binary operator:
<ul>
<li>When an <a href="#identifier">identifier</a> is followed by <tt><</tt>,
Swift attempts to parse starting from the <tt><</tt> as a
<a href="#type-identifier">generic parameter list</a>.
<li>If it succeeds in parsing a generic parameter list, it looks at the
token after the closing <tt>></tt>. If it sees one of the following tokens:
<blockquote>
<tt>( [ { } ] ) . , ;</tt>
</blockquote>
then the expression is parsed as a generic parameter list.
<li>If Swift cannot parse a generic parameter list after the <tt><</tt>,
or the matching <tt>></tt> is not followed by one of the above tokens,
the <tt><</tt> is parsed as an operator character.
</ul>
<p>These rules assume that, in most cases, generic type names will be used
in constructor expressions as in <tt>Foo<T>(x)</tt> or to access type
members as in <tt>Foo<T>.bar()</tt>. Referring to a generic metatype as a
value in an expression may require parentheses around the type name.
<pre class="example">
<i>// An operator that operates on metatypes.</i>
infix func +-+ <T, U>(t:T.Type, u:U.Type) -> Foo { }
var foo = (Dict<String, Int>) +-+ (Array<UnicodeScalar>)
print(foo)
</pre>
<p>On the other hand, some expressions involving <tt><</tt> and
<tt>></tt> operators may misparse as generic arguments as well. These
can also be corrected by adding or removing parentheses.
<pre class="example">
func foo(_ x:Bool, y:Bool)
var a,b,c,d,e : Int
foo(a < b, c > (d + e)) // ERROR: Misparses as (a<b,c>)(d + e)
foo((a < b), c > (d + e)) // Force parsing as (a < b), (c > (d + e))
foo(a < b, c > d + e) // Also parses as (a < b), (c > (d + e))
</pre>
<!-- ===================================================================== -->
<h3 id="expr-super">Super</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-super ::= expr-super-method
expr-super ::= expr-super-subscript
expr-super ::= expr-super-constructor
expr-super-method ::= 'super' '.' <a href="#expr-identifier">expr-identifier</a>
expr-super-subscript ::= 'super' '[' <a href="#expr">expr</a> ']'
expr-super-constructor ::= 'super' '.' 'init'
</pre>
<p>The keyword <tt>super</tt> is used to refer to superclass members from
a subclass method. This can be used to access members of a superclass
overridden by the subclass. The following forms are allowed:
<ul>
<li>A superclass property or method can be accessed with the form
<tt>super.name</tt>.</li>
<li>A superclass subscript accessor can be accessed with the form
<tt>super[index]</tt>.</li>
<li>Within a constructor, a superclass constructor can be accessed with the
form <tt>super.init</tt>.</li>
</ul>
<p><tt>super</tt> expressions are invalid outside of a subclass method.
<tt>super.init</tt> is invalid outside of a subclass constructor.
<tt>super.init</tt> furthermore may only be called once per derived
constructor, and must be called before the derived constructor accesses
<tt>self</tt> or any instance variables.
</p>
<!-- ===================================================================== -->
<h3 id="expr-closure">Closure Expression</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-closure ::= '{' closure-signature? <a href="#brace-item-list">brace-item-list</a> '}'
closure-signature ::= <a href="#pattern-tuple">pattern-tuple</a> <a href="#func-signature">func-signature-result</a>? 'in'
closure-signature ::= <a href="#identifier">identifier</a> (',' <a href="#identifier">identifier</a>*) <a href="#func-signature">func-signature-result</a>? 'in'
</pre>
<p>A closure defines an anonymous function as an expression. Like a
<a href="#decl-func">func</a> declaration, a closure has parameters,
a return type, and some number of statements that are executed when
the closure is called. Like local functions, closures can capture
values from its enclosing function and closure scopes. Closures are
often used in lieu of local functions when the function name would
only be used once, to be called by some other function. As a syntax
optimization, when the closure contains only a single expression, it's
value is used as the result of the closure. Thus, the closure <code>{
5 }</code> is equivalent to <code>{ return 5 }</code>.</p>
<p>Unlike <a href="#decl-func">func</a>
declarations, the return type, parameter types, and even the <a
href="#expr-anon-closure-arg">names of parameters</a> can be
omitted from the definition of the closure, making it a concise
syntax for small closures. In such cases, the context in which the
closure is used must provide information about the parameter and
return types. In the special case where the closure consists of only
a single expression, that expression participates in the
type checking of its context. </p>
<pre class="example">
<i>// Takes a closure that it calls to determine an ordering relation.</i>
func magic(_ val : Int, predicate : (a : Int, b : Int) -> Bool)
func f() {
<i>// Compare one way. Closure is inferred to return Bool and take two ints</i>
<i>// from the argument context. This same information infers that $0 and $1</i>
<i>// both have type 'Int'.</i>
magic(42, { $0 < $1 })
<i>// Compare the other way.</i>
magic(42, { $1 < $0 })
<i>// Provide parameter names, but infer the types.</i>
magic(42, { x, y in y < x })
<i>// Provide parameter names and types.</i>
magic(42, { (x : Int, y : Int) in y < x })
<i>// Provide parameter names and types, and return type, with multiple statements.</i>
magic(42, { (x : Int, y : Int) -> Bool in
print("Comparing \(x) to \(y).\n")
return y < x
})
<i>// Error, not enough context to infer the type of $0.</i>
var x = { $0 }
}
</pre>
<!-- ===================================================================== -->
<h3 id="expr-anon-closure-arg">Anonymous Closure Arguments</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-anon-closure-arg ::= <a href="#dollarident">dollarident</a>
</pre>
<p>A use of an identifier whose name fits the "$[0-9]+" regular
expression is a reference to an anonymous closure argument that is formed when
the containing expression is <a href="#typecheck_anon">coerced into a closure
context</a>. All other dollar identifiers are invalid.</p>
<p>This can only be used in the body of a closure (<a
href="#expr-closure">expr-closure</a>) that does not have explicitly-specified parameters.
</p>
<!-- ===================================================================== -->
<h3 id="expr-delayed-identifier">Delayed Identifier Resolution</h3>
<!-- ===================================================================== -->
<div class="commentary">
The ".bar" syntax was picked because it is related to the syntax of a fully
qualified "foo.bar" reference.
</div>
<pre class="grammar">
expr-delayed-identifier ::= '.' <a href="#identifier">identifier</a> <a href="#expr-paren">expr-paren</a>?
</pre>
<p>A delayed identifier expression refers to a case of an <a
href="type-enum">enum</a> type or a type member of a nominal
type, without knowing which type it is referring to. The
expression is resolved to a member of a concrete type through
context-sensitive type inference. When it is followed by an <a
href="#expr-paren">expr-paren</a>, the member must either be an
enum case that carries a value or a (type) member function.</p>
<pre class="example">
enum Direction { case Up, Down }
func search(_ val : Int, direction : Direction)
func f() {
search(42, .Up)
search(17, .Down)
}
</pre>
<!-- ===================================================================== -->
<h3 id="expr-paren">Parenthesized Expressions</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-paren ::= '(' ')'
expr-paren ::= '(' expr-paren-element (',' expr-paren-element)* ')'
expr-paren-element ::= (<a href="#identifier">identifier</a> ':')? <a href="#expr">expr</a>
</pre>
<p>Parentheses expressions contain an (optionally empty) list of optionally
named values. Parentheses in an expression context denote one of two
things: 1) grouping parentheses, or 2) a tuple literal.</p>
<p>Grouping parentheses occur when there is exactly one value in the list and
that value does not have a name. In this case, the type of the parenthesis
expression is the type of the single value.</p>
<p>All other cases are tuple literals. The type of the expression is a tuple
type whose elements and order match that of the initializer. If there are
any named elements, those elements become names for the tuple type. A
parenthesis expression with no value has a type of the empty tuple.
</p>
<p>Some examples:</p>
<pre class="example">
<i>// Simple grouping parenthesis.</i>
var a = (4) <i>// Type = Int</i>
var b = (4+a) <i>// Type = Int</i>
<i>// Tuple literals.</i>
var c = () <i>// Type = ()</i>
var d = (4, 5) <i>// Type = (Int, Int)</i>
var e = (c, d) <i>// Type = ((), (Int, Int))</i>
var f = (x : 4, y : 5) <i>// Type = (x : Int, y : Int)</i>
var g = (4, y : 5, 6) <i>// Type = (Int, y : Int, Int)</i>
<i>// Named arguments to functions.</i>
func foo(_ a : Int, b : Int)
foo(b = 4, a = 1)
</pre>
<!-- ===================================================================== -->
<h3 id="expr-dot">Dot Expressions</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-dot ::= <a href="#expr">expr-postfix</a> '.' <a href="#integer_literal">integer_literal</a>
</pre>
<p>If the base expression has <a href="#type-tuple">tuple type</a>, then the
magic identifier "[0-9]+" accesses the specified anonymous member of the
tuple. Otherwise, this form is invalid.</p>
<pre class="grammar">
expr-dot ::= <a href="#expr">expr-postfix</a> '.' <a href="#expr-identifier">expr-identifier</a>
</pre>
<p>If the base expression has <a href="#type-tuple">tuple type</a> and if the
identifier is the name of a field in the tuple, then this is a reference to
the specified field.</p>
<p>Otherwise, <a href="#namebind_value_lookup_dot">dot name lookup</a> is
performed, and this expression is treated as function application. This
allows looking up members in modules, metatypes, etc.</p>
<!-- ===================================================================== -->
<h3 id="expr-init">Initializer Expressions</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-init ::= <a href="#expr">expr-postfix</a> '.' 'init'
</pre>
<p>An initializer reference refers to a set of initializers of the
base expression. The base expression must be the <code>self</code>
parameter of an initializer, which is used to delegate the
initialization of the object to another initializer.</p>
<pre class="example">
class X {
init() {
self.init(5) // delegate to initializer below
}
init(value: Int) { /* ... */ }
}
</pre>
<!-- ===================================================================== -->
<h3 id="expr-subscript">Subscript Expressions</h3>
<!-- ===================================================================== -->
<div class="commentary">
There is no "built-in" semantics for subscripting. Rather, all
subscripting semantics is implemented via subscript declarations
in the library.
<br/>We require that the '[' not be the first token on a line, so that
a statement can begin with an array expression.
</div>
<pre class="grammar">
expr-subscript ::= <a href="#expr">expr-postfix</a> '[' <a href="#expr">expr</a> ']'
</pre>
<p>A subscript expression invokes a <a
href="#decl-subscript">subscript getter or setter</a> on the type
of the <tt>expr-postfix</tt>. The <tt>expr</tt> is used as the
subscript argument, which will be provided to either the getter or
setter depending on whether the subscript expression is used as an
rvalue (reading) or lvalue (writing), respectively. A subscript
expression that resolves to a subscript declaration with no setter
cannot be modified.</p>
<!-- ===================================================================== -->
<h3 id="expr-new">New Expressions</h3>
<!-- ===================================================================== -->
<div class="commentary">
It's not really clear what the behavior of multiple bounds should be.
<br/><br/>We should probably allow an initializer. The semantics would be
to evaluate that constructor for each element constructed.
</div>
<pre class="grammar">
expr-new ::= 'new' <a href="#type">type-identifier</a> expr-new-bounds
expr-new-bounds ::= expr-new-bound
expr-new-bounds ::= expr-new-bounds expr-new-bound
expr-new-bound ::= '[' <a href="#expr">expr?</a> ']'
</pre>
<p>Allocates and initializes a new array of objects. The first clause must
be an expression; subsequent bounds, if present, must be constant under
the <a href="#type-array">usual rules for array types</a>. The opening
square bracket must be on the same line as the type name.</p>
<!-- ===================================================================== -->
<h3 id="expr-call">Function Application</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-call ::= <a href="#expr">expr-postfix</a> <a href="#expr-paren">expr-paren</a>
</pre>
<p>The leading <tt>'('</tt> of the <tt>expr-paren</tt> must not be
the first token on a line. This greatly reduces the likelihood of
confusion from semicolon elision, without requiring feedback from
the typechecker or more aggressive whitespace sensitivity.</p>
<p>If the <tt>expr-postfix</tt> refers to a (possibly
parenthesized) name of a type, the <tt>expr-paren</tt> is first
coerced to the type named by <tt>expr-postfix</tt>. If that coercion
fails, then the <tt>expr-postfix</tt> refers to the set of
constructors for that type.</p>
<p>Simple examples:</p>
<pre class="example">
<i>// Application of an empty tuple to the function f.</i>
f()
<i>// Application of 4 to the function f.</i>
g(4)
<i>// Application of 4 to the function returned by h().</i>
var h : (Int) -> (Int) -> Int
...
h()(4)
<i>// Two separate statements</i>
i()
(j <+ 2)()
</pre>
<!-- ===================================================================== -->
<h3 id="expr-trailing-closure">Trailing Closures</h3>
<!-- ===================================================================== -->
<div class="commentary">
It is possible to model trailing closures as simply another way to
perform a function call, forgoing the syntactic transformation for
<a href="#expr-call">expr-call</a>, if functions meant to be used
with trailing closures are written as curried functions, e.g.,
<pre>
func map<T, U>(_ array : T[])(fn : (T) -> U) -> U[] { ... }
</pre>
There are two problems with this (admittedly simpler) design.
First, functions imported from C, C++, and Objective-C won't ever
be written in this curried syntax, so we would have to implement
redundant entry points to enable this syntax. Second, this design
forces the idea of currying front and center for Swift programmers
who otherwise wouldn't care, for mostly theoretical reasons.
</div>
<pre class="grammar">
expr-trailing-closure ::= <a href="#expr">expr-postfix</a> <a href="#expr-closure">expr-closure</a>+
</pre>
<p>A postfix expression followed by a closure will be invoked with
the closure as its argument. This syntax is referred to as a
"trailing" closure, because the closure itself is outside the
parentheses used to call the expression. Trailing closures are
syntactic sugar that eliminates the awkwardness of closing a
function call with "})", where the "}" ends the closure and the ")"
ends the call.</p>
<p>Trailing closures use a simple syntactic translation, making them
purely syntactic sugar. If the postfix expression preceding the
trailing closure is an <a href="#expr-call">expr-call</a>, the
closure is added to the end of the <a
href="#expr-paren">expr-paren</a> of that call. Otherwise, the
postfix expression is (implicitly) called with the trailing closure
as its only argument.</p>
<pre class="example">
dispatch_async(q) {
print("Whenever you get around to it\n")
}
</pre>
<!-- ===================================================================== -->
<h3 id="expr-optional">Optional Chaining</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-optional ::= <a href="#expr-postfix">expr-postfix</a> '?'-postfix
</pre>
<p>The optional-chaining operator provides a convenient syntax for
dereferencing, calling, or subscripting optional values.</p>
<p>Informally, the operator attempts to strip one level
of <code>Optional</code> from its operand, and if that fails, all
the following postfix operators are skipped and just evaluate
to <code>None</code>.</p>
<p>More formally:
<ul>
<li>The operand must be convertible to type <code>T?</code> for
some type <code>T</code>. As a special rule, the expression
is ill-formed if the operand is converted to optional type by
the implicit conversion from <code>T</code>
to <code>T?</code>.</li>
<li>The expression itself then has type <code>T</code> (and is
an r-value).</li>
<li>A <code>postfix-expression</code> <i>E1</i> is said to
<i>directly chain</i> to a <code>postfix-expression</code>
<i>E2</i> if <i>E1</i> is syntactically
the <code>postfix-expression</code> base of <code>E2</code>;
note that this does not include any syntactic nesting,
e.g. via parentheses. <i>E1</i> <i>chains</i> to <i>E2</i>
if they are the same expression or <i>E1</i> directly chains
to an expression which chains to <i>E2</i>. This relation has
a maximum, called the <i>largest chained expression</i>.</li>
<li>The largest chained expression of an <code>expr-optional</code>
must be convertible to an r-value of type <code>U?</code> for
some type <code>U</code>. Note that a single expression may
be the largest chained expression of multiple
<code>expr-optional</code>s.</li>
<li>If the operand evaluates to <code>Some(x) : T?</code> for some
value <code>x : T</code>, the expression yields
<code>x</code>.</li>
<li>If the operand evaluates to <code>None : T?</code>, evaluation
of all the chained expressions immediately terminates, and
the largest chained expression yields the value
<code>None : U?</code>.</li>
</ul></p>
<!-- ===================================================================== -->
<h3 id="expr-force-value">Forcing an expression's value</h3>
<!-- ===================================================================== -->
<pre class="grammar">
expr-force-value ::= <a href="#expr-postfix">expr-postfix</a> '!'
</pre>
<p>The postfix '!' forces an optional value to its stored value
(i.e., the <code>x</code> in <code>.Some(x)</code>), failing at
runtime if the optional is <code>.None</code>.
<!-- ********************************************************************* -->
<h2 id = "stmt">Statements</h2>
<!-- ********************************************************************* -->
<div class="commentary">
Statements can only exist in contexts that are themselves a stmt.
Statements have no type, they just induce control flow changes. We choose
to use constructs that will be familiar to a broad range of C/Java
programmers.
</div>
<pre class="grammar">
stmt ::= <a href="#stmt-semicolon">stmt-semicolon</a>
stmt ::= <a href="#stmt-if">stmt-if</a>
stmt ::= <a href="#stmt-while">stmt-while</a>
stmt ::= <a href="#stmt-repeat-while">stmt-repeat-while</a>
stmt ::= <a href="#stmt-for-c-style">stmt-for-c-style</a>
stmt ::= <a href="#stmt-for-each">stmt-for-each</a>
stmt ::= <a href="#stmt-switch">stmt-switch</a>
stmt ::= stmt-control-transfer
stmt-control-transfer ::= <a href="#stmt-return">stmt-return</a>
stmt-control-transfer ::= <a href="#stmt-break">stmt-break</a>
stmt-control-transfer ::= <a href="#stmt-continue">stmt-continue</a>
stmt-control-transfer ::= <a href="#stmt-fallthrough">stmt-fallthrough</a>
</pre>
<p>Statements provide the control flow constructs of function bodies and
top-level code.</p>
<pre class="example">
<i>// A function with some statements.</i>
func fib(_ v : Int) -> Int {
if v < 2 {
return v
}
return fib(v-1)+fib(v-2)
}
</pre>
<!-- ===================================================================== -->
<h3 id="stmt-semicolon">Semicolon Statement</h3>
<!-- ===================================================================== -->
<div class="commentary">
Allowing semicolons as statements causes us to allow semicolons as statement
separators as well. This, in turn, means that we don't reject code that has
semicolons after each statement, which will be common when people first
start getting used to Swift.
</div>
<pre class="grammar">
stmt-semicolon ::= ';'
</pre>
<p>The semicolon statement has no effect.</p>
<!-- ===================================================================== -->
<h3 id="stmt-return">'return' Statement</h3>
<!-- ===================================================================== -->
<pre class="grammar">
stmt-return ::= 'return' <a href="#expr">expr</a>
stmt-return ::= 'return'
</pre>
<p>The return statement sets the return value of the current <a
href="#decl-func">func declaration</a> or <a href="#expr-closure">closure
expression</a> and transfers control out of the function. It sets the
return value by converting the specified expression result (or '()' if
none is specified) to the return type of the 'func'.
</p>
<p>The stmt-return grammar is ambiguous: "{ return 4 }" could be parsed as
{"return" "4"} or as a single statement. Ambiguity here is resolved toward
the first production, because control flow can't transfer to an
subexpression.</p>
<!-- 'break' and 'continue' converted to ReST. -->
<!-- ===================================================================== -->
<h3 id="stmt-if">'if' Statement</h3>
<!-- ===================================================================== -->
<div class="commentary">
We require braces around the body of an 'if' for two reasons: first, it
eliminates the need for parentheses around the condition by making them
visually distinctive. Second, it will eliminate all the dithering about
whether and when people should, or should not, use braces for if bodies.
</div>
<pre class="grammar">
stmt-if ::= 'if' <a href="#expr">expr-basic</a> <a href="#brace-item-list">brace-item-list</a> stmt-if-else?
stmt-if-else ::= 'else' <a href="#brace-item-list">brace-item-list</a>
stmt-if-else ::= 'else' stmt-if
</pre>
<p>'if' statements provide a simple control transfer operations that evaluates
the condition, gets the 'boolValue' property of the result if the result
not a 'Bool', then determines the direction of the branch based on the result.
(Internally, the standard library type 'Bool' has a boolValue property that
returns a 'Builtin.Int1'.) It is an error if the type of the expression is
context-dependent or some non-Bool type.
</p>
<p>Some examples include:</p>
<pre class="example">
if true {
/*...*/
}
if X == 4 {
} else {
}
if X == 4 {
} else if X == 5 {
} else {
}
</pre>
<!-- ===================================================================== -->
<h3 id="stmt-while">'while' Statement</h3>
<!-- ===================================================================== -->
<pre class="grammar">
stmt-while ::= 'while' <a href="#expr">expr-basic</a> <a href="#brace-item-list">brace-item-list</a>
</pre>
<p>'while' statements provide simple loop construct which (on each iteration
of the loop) evaluates the condition, gets the 'boolValue' property of
the result if the result not a 'Bool', then determines whether to keep
looping. (Internally, the standard library type 'Bool' has a boolValue
property that yields a 'Builtin.Int1'.) It is an error if the type of
the expression is context-dependent or some non-Bool type.
</p>
<p>Some examples include:</p>
<pre class="example">
while true {
/*...*/
}
while X == 4 {
X = 3
}
</pre>
<!-- ===================================================================== -->
<h3 id="stmt-repeat-while">'repeat-while' Statement</h3>
<!-- ===================================================================== -->
<pre class="grammar">
stmt-repeat-while ::= 'repeat' <a href="#brace-item-list">brace-item-list</a> 'while' '<a href="#expr">expr</a>
</pre>
<p>'repeat-while' statements provide simple loop construct which (on each
iteration of the loop) evaluates the body, then evaluates the condition,
getting the 'boolValue' property of the result if the result not a 'Bool',
then determines whether to keep looping. (Internally, the standard library
type 'Bool' has a boolValue property that yields a 'Builtin.Int1'). It is
an error if the type of the expression is context-dependent or some non-Bool
type.
</p>
<p>Some examples include:</p>
<pre class="example">
repeat {
/*...*/
} while true
repeat {
X = 3
} while X == 4
</pre>
<!-- ===================================================================== -->
<h3 id="stmt-for-c-style">C-Style 'for' Statement</h3>
<!-- ===================================================================== -->
<pre class="grammar">
stmt-for-c-style ::= 'for' stmt-for-c-style-init? ';' <a href="#expr">expr</a>? ';' stmt-for-c-style-inc? <a href="#brace-item-list">brace-item-list</a>
stmt-for-c-style ::= 'for' '(' stmt-for-c-style-init? ';' <a href="#expr">expr</a>? ';' stmt-for-c-style-inc? ')' <a href="#brace-item-list">brace-item-list</a>
stmt-for-c-style-init ::= <a href="#decl-var">decl-var</a>
stmt-for-c-style-init ::= expr (',' expr)*
stmt-for-c-style-inc ::= expr-basic (',' expr-basic)*
</pre>
<p>C-Style 'for' statements provide simple loop construct which evaluates the
first part (the initializer) before entering the loop, then evaluates the
second condition as a logic value to determines whether to keep looping.
The third condition is executed at the end of the loop. All three are
evaluated in a new scope that surrounds the for statement.
</p>
<p>Some examples include:</p>
<pre class="example">
for i = 0; i != 10; ++i {
/*...*/
}
for (i = 0; i != 10; ++i) {
/*...*/
}
for var (i,j) = (0,1); i != 10; ++i {
/*...*/
}
</pre>
<!-- ===================================================================== -->
<h3 id="stmt-for-each">'for-each' Statement</h3>
<!-- ===================================================================== -->
<pre class="grammar">
stmt-for-each ::= 'for' <a href="#pattern">pattern</a> 'in' <a href="#expr">expr-basic</a> <a href="#brace-item-list">brace-item-list</a>
</pre>
<p>Enumerator-based 'for' statements provide enumeration over the values in a
container. The <tt>expr</tt> is either a container or an enumerator; and
respectively, it either conforms to the formal Enumeration or formal Enumerator
protocol.
<p>Note that each iteration of the loop declares a distinct variable for each
variable in the pattern. For example, in a loop like "for i in 0...10",
if i is captured inside the loop, each iteration captures a different "i",
so there would be a total of ten versions generated each time the loop
runs.</p>
<p>Some examples include:</p>
<pre class="example">
for i in 0...100 {
print(String(i));
}
</pre>
<!-- 'switch' and 'fallthrough' converted to ReST. -->
<!-- ********************************************************************* -->
<h2>Protocols</h2>
<!-- ********************************************************************* -->
<!-- ********************************************************************* -->
<h2>Objects</h2>
<!-- ********************************************************************* -->
<!-- ********************************************************************* -->
<h2>Generics</h2>
<!-- ********************************************************************* -->
<!-- ===================================================================== -->
<h3 id="generic-params">Generic Parameters</h3>
<!-- ===================================================================== -->
<pre class="grammar">
generic-params ::= '<' generic-param (',' generic-param)* where-clause? '>'
generic-param ::= identifier
generic-param ::= identifier ':' <a href="#type-identifier">type-identifier</a>
generic-param ::= identifier ':' <a href="#type-composition">type-composition</a>
<a id="where-clause">where-clause</a> ::= 'where' requirement (',' requirement) *
requirement ::= conformance-requirement
::= same-type-requirement
conformance-requirement ::= <a href="#type-identifier">type-identifier</a> ':' <a href="#type-identifier">type-identifier</a>
conformance-requirement ::= <a href="#type-identifier">type-identifier</a> ':' <a href="#type-composition">type-composition</a>
same-type-requirement ::= <a href="#type-identifier">type-identifier</a> '==' <a href="#type-identifier">type-identifier</a>
</pre>
<p>A generic function or type is parameterized by a given set of
generic parameters. The generic parameters each have a name as well
as some set of requirements that specify the capabilities that any
corresponding generic argument might have. For example, the generic
parameter <code>T : CustomStringConvertible</code> requires that any generic
argument substituted for the generic parameter <code>T</code>
conform to the protocol <code>CustomStringConvertible</code>. Similarly, a generic
parameter <code>U : SomeClass</code> requires that any generic
argument substituted for the generic parameter <code>U</code>
inherit from the class <code>SomeClass</code>.
<p>Additional requirements on generic parameters and associated types
of generic parameters can be introduced via the "where" clause,
which can include additional protocol-conformance requirements
(e.g., the generic parameter list <code><T where T :
CustomStringConvertible></code>, which is equivalent to <code><T :
CustomStringConvertible></code>), as well as same-type requirements that
require two types to be identical (e.g., <code><T : Collection, U
: Collection where T.Element == U.Element></code>).
<!-- ===================================================================== -->
<h3 id="generic-args">Generic Arguments</h3>
<!-- ===================================================================== -->
<pre class="grammar">
generic-args ::= '<' generic-arg (',' generic-arg)* '>'
generic-arg ::= <a href="#type">type</a>
</pre>
<p>Generic argument lists specify the generic arguments to be provided
to a generic type or function, which replace the generic parameters
of that type or function to produce a specialized version of that
type or function. For example, given a generic class:
<pre class="example">
class Dictionary<Key : Hashable, Value> { /* ... */ }
</pre>
<p>The type <code>Dictionary<String, Int></code>, replaces the
generic parameter <code>Key</code> with <code>String</code> and the
generic parameter <code>Value</code> with <code>Int</code>. Each
generic argument must satisfy all of the requirements of its
corresponding generic parameter (e.g., <code>String</code> must
conform to the <code>Hashable</code> protocol), and all generic
arguments, when taken together, must satisfy the additional
requirements specified in the <code>where</code> clause.
<!-- ********************************************************************* -->
<h2 id="namebind">Name Binding</h2>
<!-- ********************************************************************* -->
<p>Name binding in swift is performed in different ways depending on what
language entity is being considered:</p>
<p>Value names (for <a
href="#decl-var">var</a> and <a href="#decl-func">func</a> declarations) and
type names (for <a href="#decl-typealias">typealias</a>, <a
href="#decl-enum">enum</a>, and <a href="#decl-struct">struct</a>
declarations) follow the same <a href="#namebind_scope">scope</a> and
<a href="#namebind_typevalue_lookup">name lookup</a> rules as described below.
</p>
<p>tuple element names</p>
<p>scope within enum decls</p>
<p>Context sensitive member references are resolved <a
href="#typecheck_context">during type checking</a>.</p>
<h3 id="namebind_scope">Scopes for Type and Value Names</h3>
<h3 id="namebind_value_lookup_unqual">Name Lookup Unqualified Value Names</h3>
<h3 id="namebind_value_lookup_dot">"dot" Name Lookup Value Names</h3>
<h3 id="namebind_typevalue_lookup">Name Lookup for Type and Value Names</h3>
<p>Basic algo:</p>
<ul>
<li>Search the current scope tree for a local name. Local names cannot be
forward referenced.</li>
<li>Bind to names defined in the current component, including the current
module. TODO: is this a good thing? We could require explicit
imports if we wanted to.</li>
<li>Bind to identifiers that are imported with an import directive. Imports
are searched in order of introduction (top-down). The location of an
import directive in a file (e.g. between func decls) does not affect name
lookup, but the order of imports w.r.t. each other does.</li>
</ul>
<p>Shadowing: Given a ValueDecl D1 in the current module and a ValueDecl D2
in an imported module with the same name and a member of the same type (if
relevant): 1. If D1 is a TypeDecl, D2 is shadowed. 2. If neither D1 nor D2
is a TypeDecl, and they have the same type, D2 is shadowed. If a
declaration in an imported module is shadowed by any declaration in the
current module, it is not found by unqualified global lookup or lookup for
members of a type.</p>
<h3 id="namebind_dot">Name Lookup for Dot Expressions</h3>
<p>
<a href="#expr-dot">Dot Expressions</a> bind to name of tuple elements.
</p>
<!-- ********************************************************************* -->
<h2 id="typecheck">Type Checking</h2>
<!-- ********************************************************************* -->
<p>
Binary expressions, function application, etc.
</p>
<h3 id="typecheck_conversions">Standard Conversions</h3>
<!--
Consider foo(4, 5) when foo is declared to take ((Int,Int=3), Int=6). This
could be parsed as either ((4,5), 6) or ((4,3),5), but the later one is
the "right" answer.
-->
<h3 id="typecheck_anon">Anonymous Argument Resolution</h3>
<h3 id="typecheck_context">Context Sensitive Type Resolution</h3>
<!-- *********************************************************************** -->
<hr>
<address>
<a href="http://jigsaw.w3.org/css-validator/check/referer"><img
src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
<a href="http://validator.w3.org/check/referer"><img
src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
<a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
</address>
</body>
</html>
Computing file changes ...
