Functions

Function Declarations

FuncDeclaration:
    StorageClassesopt BasicType FuncDeclarator FunctionBody
    AutoFuncDeclaration

AutoFuncDeclaration:
    StorageClasses Identifier FuncDeclaratorSuffix FunctionBody

FuncDeclarator:
    TypeSuffixesopt Identifier FuncDeclaratorSuffix

FuncDeclaratorSuffix:
    Parameters MemberFunctionAttributesopt
    TemplateParameters Parameters MemberFunctionAttributesopt Constraintopt

• A FuncDeclaration with a function body is called a function definition.
• A FuncDeclaration without a function body is called a function prototype.
• A FuncDeclaration with TemplateParameters defines a function template.

Function Parameters

Parameters:
    ( ParameterListopt )

ParameterList:
    Parameter
    Parameter , ParameterListopt
    VariadicArgumentsAttributesopt ...

Parameter:
    ParameterDeclaration
    ParameterDeclaration ...
    ParameterDeclaration = AssignExpression
    ParameterDeclaration = AssignExpression ...

ParameterDeclaration:
    ParameterAttributesopt BasicType Declarator
    ParameterAttributesopt Type

ParameterAttributes:
    ParameterStorageClass
    UserDefinedAttribute
    ParameterAttributes ParameterStorageClass
    ParameterAttributes UserDefinedAttribute

ParameterStorageClass:
    auto
    TypeCtor
    final
    in
    lazy
    out
    ref
    return
    scope

VariadicArgumentsAttributes:
    VariadicArgumentsAttribute
    VariadicArgumentsAttribute VariadicArgumentsAttributes

VariadicArgumentsAttribute:
    const
    immutable
    return
    scope
    shared

See also: parameter storage classes.

Function Attributes

FunctionAttributes:
    FunctionAttribute
    FunctionAttribute FunctionAttributes

FunctionAttribute:
    FunctionAttributeKwd
    Property
    AtAttribute

MemberFunctionAttributes:
    MemberFunctionAttribute
    MemberFunctionAttribute MemberFunctionAttributes

MemberFunctionAttribute:
    const
    immutable
    inout
    return
    scope
    shared
    FunctionAttribute

Function Bodies

FunctionBody:
    SpecifiedFunctionBody
    ShortenedFunctionBody
    MissingFunctionBody

SpecifiedFunctionBody:
    doopt BlockStatement
    FunctionContractsopt InOutContractExpression doopt BlockStatement
    FunctionContractsopt InOutStatement do BlockStatement

ShortenedFunctionBody:
    InOutContractExpressionsopt => AssignExpression ;

Examples:

int hasSpecifiedBody() { return 1; }
int hasShortenedBody() => 1; // equivalent

The ShortenedFunctionBody form implies a return statement. This syntax also applies for function literals.

Function Prototypes

MissingFunctionBody:
    ;
    FunctionContractsopt InOutContractExpression ;
    FunctionContractsopt InOutStatement

Function declarations with a MissingFunctionBody, e.g.:

int foo();

that are not declared as abstract are expected to have their implementations elsewhere, and that implementation will be provided at the link step. This enables an implementation of a function to be completely hidden from the user of it, and the implementation may be in another language such as C, assembler, etc. Typically a function prototype would have non-extern(D) linkage.

A function prototype can have extern(D) linkage. This is useful for D interface files.

Function Contracts

FunctionContracts:
    FunctionContract
    FunctionContract FunctionContracts

FunctionContract:
    InOutContractExpression
    InOutStatement

InOutContractExpressions:
    InOutContractExpression
    InOutContractExpression InOutContractExpressions

InOutContractExpression:
    InContractExpression
    OutContractExpression

InOutStatement:
    InStatement
    OutStatement

Function Contracts specify the preconditions and postconditions of a function. They are used in Contract Programming.

Preconditions and postconditions do not affect the type of the function.

Preconditions

InContractExpression:
    in ( AssertArguments )

InStatement:
    in BlockStatement

An InContractExpression is a precondition.

• The first AssignExpression of the AssertArguments must evaluate to true. If it does not, the precondition has failed.

• The second AssignExpression, if present, must be implicitly convertible to type const(char)[].

An InStatement is also a precondition. Any AssertExpression appearing in an InStatement will be an InContractExpression.

Preconditions must semantically be satisfied before the function starts executing. If a precondition fails, the program enters an Invalid State.

The expression form is:

in (expression)
in (expression, "failure string")
{
    ...function body...
}

The block statement form is:

in
{
    ...contract preconditions...
}
do
{
    ...function body...
}

Postconditions

OutContractExpression:
    out ( ; AssertArguments )
    out ( Identifier ; AssertArguments )

OutStatement:
    out BlockStatement
    out ( Identifier ) BlockStatement

An OutContractExpression is a postcondition.

• The first AssignExpression of the AssertArguments must evaluate to true. If it does not, the postcondition has failed.
• The second AssignExpression, if present, must be implicitly convertible to type const(char)[].

An OutStatement is also a postcondition. Any AssertExpression appearing in an OutStatement will be an OutContractExpression.

Postconditions must semantically be satisfied after the function finishes executing. If a postcondition fails, the program enters an Invalid State.

• Any mutable input arguments each have the expected value/range of values.
• Any return value has a correct value/range of values.

The expression form is:

out (identifier; expression)
out (identifier; expression, "failure string")
out (; expression)
out (; expression, "failure string")
{
    ...function body...
}

The block statement form is:

out
{
    ...contract postconditions...
}
out (identifier)
{
    ...contract postconditions...
}
do
{
    ...function body...
}

The optional identifier in either type of postcondition is set to the return value of the function, and can be accessed from within the postcondition. It is implicitly const.

Example

int fun(ref int a, int b)
in (a > 0)
in (b >= 0, "b cannot be negative!")
out (r; r > 0, "return must be positive")
out (; a != 0)
{
    // function body
}

int fun(ref int a, int b)
in
{
    assert(a > 0);
    assert(b >= 0, "b cannot be negative!");
}
out (r)
{
    assert(r > 0, "return must be positive");
    assert(a != 0);
}
do
{
    // function body
}

The two functions are identical semantically.

In, Out and Inheritance

If a function in a derived class overrides a function from its super class, then only the preconditions of one of the function and its overridden functions must be satisfied. Overriding functions then becomes a process of loosening the preconditions.

A function without preconditions means its precondition is always satisfied. Therefore if any function in an inheritance hierarchy has no preconditions, then any preconditions on functions overriding it have no meaningful effect.

Conversely, all of the postconditions of the function and its overridden functions must to be satisfied. Adding overriding functions then becomes a processes of tightening the postconditions.

Function Return Values

At least one return statement is required if the function specifies a return type that is not void, unless:

• the function executes an infinite loop
• the function executes an assert(0) statement
• the function evaluates an expression of type noreturn
• the function contains inline assembler code

Function return values not marked as ref are considered to be rvalues. This means they cannot be passed by reference to other functions.

Pure Functions

Pure functions are annotated with the pure attribute. Pure functions cannot directly access global or static mutable state. Pure functions can only call pure functions.

Pure functions can:

• Modify the local state of the function.
• Throw exceptions.

int x;
immutable int y;

pure int foo(int i)
{
    i++;     // ok, modifying local state
    //x = i; // error, modifying global state
    //i = x; // error, reading mutable global state
    i = y;   // ok, reading immutable global state
    throw new Exception("failed"); // ok
}

A pure function can override an impure function, but cannot be overridden by an impure function. I.e. it is covariant with an impure function.

Strong vs Weak Purity

A weakly pure function has parameters with mutable indirections. Program state can be modified transitively through the matching argument.

pure size_t foo(int[] arr)
{
    arr[] += 1;
    return arr.length;
}
int[] a = [1, 2, 3];
foo(a);
assert(a == [2, 3, 4]);

A strongly pure function has no parameters with mutable indirections and cannot modify any program state external to the function.

struct S { double x; }

pure size_t foo(immutable(int)[] arr, int num, S val)
{
    //arr[num] = 1; // compile error
    num = 2;        // has no side effect on the caller side
    val.x = 3.14;   // ditto
    return arr.length;
}

A strongly pure function can call a weakly pure function.

Special Cases

A pure function can:

• read and write the floating point exception flags
• read and write the floating point mode flags, as long as those flags are restored to their initial state upon function entry

Debugging

A pure function can perform impure operations in statements that are in a ConditionalStatement controlled by a DebugCondition.

pure int foo(int i)
{
    debug writeln("i = ", i); // ok, impure code allowed in debug statement
    ...
}

Nested Functions

Nested functions inside a pure function are implicitly marked as pure.

pure int foo(int x, immutable int y)
{
    int bar()
    // implicitly marked as pure, to be "weakly pure"
    // since hidden context pointer to foo stack context is mutable
    {
        x = 10;     // can access states in enclosing scope
                    // through the mutable context pointer
        return x;
    }
    pragma(msg, typeof(&bar));  // int delegate() pure

    int baz() immutable
    // qualifies hidden context pointer with immutable,
    // and has no other parameters, therefore "strongly pure"
    {
        //return x; // error, cannot access mutable data
                    // through the immutable context pointer
        return y;   // ok
    }

    // can call pure nested functions
    return bar() + baz();
}

Pure Factory Functions

A pure factory function is a strongly pure function that returns a result that has only mutable indirections. All mutable memory returned by the call cannot be referenced by any other part of the program, i.e. it is newly allocated by the function. The mutable references of the result similarly cannot refer to any object that existed before the function call. This allows the result to be implicitly cast from anything to immutable or const shared, and from shared and const shared to (unshared) const. For example:

struct List { int payload; List* next; }

pure List* make(int a, int b)
{
    auto result = new List(a, null);
    result.next = new List(b, null);
    return result;
}

void main()
{
    auto list = make(1, 2);
    pragma(msg, typeof(list));       // List*

    immutable ilist = make(1, 2);
    pragma(msg, typeof(ilist));      // immutable List*
    pragma(msg, typeof(ilist.next)); // immutable List*
}

All references in make's result refer to List objects created by make, and no other part of the program refers to any of these objects. Hence the result can initialize an immutable variable.

This does not affect any Exception or Error thrown from the function.

Optimization

int a(int) pure; // no mutable indirections
int b(const Object) pure; // depends on argument passed
immutable(Object) c(immutable Object) pure; // always memoizable

void g();

void f(int n, const Object co, immutable Object io)
{
    const int x = a(n);
    g(); // `n` does not change between calls to `a`
    int i = a(n); // same as `i = x`

    const int y = b(co);
    // `co` may have mutable indirection
    g(); // may change fields of `co` through another reference
    i = b(co); // call is not memoizable, result may differ

    const int z = b(io);
    i = b(io); // same as `i = z`
}

Such a function may still have behavior inconsistent with memoization by e.g. using casts or by changing behavior depending on the address of its parameters. An implementation is currently not required to enforce validity of memoization in all cases.

If a function throws an Exception or an Error, the assumptions related to memoization do not carry to the thrown exception.

Pure destructors do not benefit of special elision.

Nothrow Functions

Nothrow functions can only throw exceptions derived from class Error.

Nothrow functions are covariant with throwing ones.

Ref Functions

A ref function returns by reference (instead of by value). The return value of a ref function must be an lvalue (whereas the return value of a non-ref function can be an rvalue, too). An expression formed by calling a ref function is an lvalue (whereas an expression formed by calling a non-ref function is an rvalue).

int *p;

ref int foo()
{
    p = new int(2);
    return *p;
}

void main()
{
    int i = foo();
    assert(i == 2);

    foo() = 3;     // reference returns can be lvalues
    assert(*p == 3);
}

Returning a reference to an expired function context is not allowed. This includes local variables, temporaries and parameters that are part of an expired function context.

ref int sun()
{
    int i;
    return i;  // error, escaping a reference to local variable i
}

A ref parameter may not be returned by ref, unless it is return ref.

ref int moon(ref int i)
{
    return i; // error
}

Auto Functions

Auto functions have their return type inferred from any ReturnStatements in the function body.

An auto function is declared without a return type. Auto functions can use any valid StorageClass, not just auto.

If there are multiple ReturnStatements, the types of them must be implicitly convertible to a common type. If there are no ReturnStatements, the return type is inferred to be void.

auto foo(int x) { return x + 3; }          // inferred to be int
pure bar(int x) { return x; return 2.5; }  // inferred to be double

Auto Ref Functions

Auto ref functions can infer their return type just as auto functions do. In addition, they become ref functions if all of these apply:

• All expressions returned from the function are lvalues
• No local variables are returned
• Any parameters returned are reference parameters
• Each returned expression must implicitly convert to an lvalue of the deduced return type

  auto ref f1(int x)     { return x; }  // value return
  auto ref f2()          { return 3; }  // value return
  auto ref f3(ref int x) { return x; }  // ref return
  auto ref f4(out int x) { return x; }  // ref return
  auto ref f5()
  {
      static int x;
      return x; // ref return
  }
  

The ref-ness of a function is determined from all ReturnStatements in the function body:

auto ref f1(ref int x) { return 3; return x; }  // ok, value return
auto ref f2(ref int x) { return x; return 3; }  // ok, value return
auto ref f3(ref int x, ref double y)
{
    return x; return y;
    // The return type is deduced to be double, but cast(double)x is not an lvalue,
    // so f3 has a value return.
}

Auto ref functions can have an explicit return type.

auto ref int bar(ref int x) { return x; }  // ok, ref return
auto ref int foo(double x) { return x; }   // error, cannot convert double to int

Inout Functions

For extensive information see inout type qualifier.

Optional Parentheses

If a function call passes no explicit argument, i.e. it would syntactically use (), then these parentheses may be omitted, similar to a getter invocation of a property function. A UFCS call can also omit empty parentheses.

void foo() {}   // no arguments
void fun(int x = 10) {}
void bar(int[] arr) {}

void main()
{
    foo();      // OK
    foo;        // also OK
    fun;        // OK

    int[] arr;
    arr.bar();  // UFCS call
    arr.bar;    // also OK
}

Due to ambiguity, parentheses are required to call a delegate or a function pointer:

void main()
{
    int function() fp;

    assert(fp == 6);    // Error, incompatible types int function() and int
    assert(*fp == 6);   // Error, incompatible types int() and int

    int delegate() dg;
    assert(dg == 6);    // Error, incompatible types int delegate() and int
}

If a function returns a delegate or a function pointer, any parentheses apply first to the function call, not the result. Two sets of parentheses are required to call the result directly:

int getNum() { return 6; }
int function() getFunc() { return &getNum; }

void main()
{
    int function() fp;

    fp = getFunc;   // implicit call
    assert(fp() == 6);

    fp = getFunc(); // explicit call
    assert(fp() == 6);

    int x = getFunc()();
    assert(x == 6);
}

struct S
{
    int getNum() { return 6; }
    int delegate() getDel() return { return &getNum; }
}

void main()
{
    S s;
    int delegate() dg;

    dg = s.getDel;   // implicit call
    assert(dg() == 6);

    dg = s.getDel(); // explicit call
    assert(dg() == 6);

    int y = s.getDel()();
    assert(y == 6);
}

Property Functions

WARNING: The definition and usefulness of property functions is being reviewed, and the implementation is currently incomplete. Using property functions is not recommended until the definition is more certain and implementation more mature.

Properties are functions that can be syntactically treated as if they were fields or variables. Properties can be read from or written to. A property is read by calling a method or function with no arguments; a property is written by calling a method or function with its argument being the value it is set to.

Simple getter and setter properties can be written using UFCS. These can be enhanced with the additon of the @property attribute to the function, which adds the following behaviors:

• @property functions cannot be overloaded with non-@property functions with the same name.
• @property functions can only have zero, one or two parameters.
• @property functions cannot have variadic parameters.
• For the expression typeof(exp) where exp is an @property function, the type is the return type of the function, rather than the type of the function.
• For the expression __traits(compiles, exp) where exp is an @property function, a further check is made to see if the function can be called.
• @property are mangled differently, meaning that @property must be consistently used across different compilation units.
• The ObjectiveC interface recognizes @property setter functions as special and modifies them accordingly.

A simple property would be:

struct Foo
{
    @property int data() { return m_data; } // read property

    @property int data(int value) { return m_data = value; } // write property

  private:
    int m_data;
}

To use it:

int test()
{
    Foo f;

    f.data = 3;        // same as f.data(3);
    return f.data + 3; // same as return f.data() + 3;
}

The absence of a read method means that the property is write-only. The absence of a write method means that the property is read-only. Multiple write methods can exist; the correct one is selected using the usual function overloading rules.

In all the other respects, these methods are like any other methods. They can be static, have different linkages, have their address taken, etc.

The built in properties .sizeof, .alignof, and .mangleof may not be declared as fields or methods in structs, unions, classes or enums.

If a property function has no parameters, it works as a getter. If has exactly one parameter, it works as a setter.

Virtual Functions

Virtual functions are class member functions that are called indirectly through a function pointer table, called a vtbl[], rather than directly. Member functions that are virtual can be overridden in a derived class:

class A
{
    void foo(int x) {}
}

class B : A
{
    override void foo(int x) {}
    //override void foo() {} // error, no foo() in A
}

void test()
{
    A a = new B();
    a.foo(1);   // calls B.foo(int)
}

The override attribute is required when overriding a function. This is useful for catching errors when a base class's member function has its parameters changed, and all derived classes need to have their overriding functions updated.

The final method attribute prevents a subclass from overriding the method.

The following are not virtual:

• Struct and union member functions
• final member functions
• static member functions
• Member functions which are private or package
• Member template functions

Example:

class A
{
    int def() { ... }
    final int foo() { ... }
    final private int bar() { ... }
    private int abc() { ... }
}

class B : A
{
    override int def() { ... }  // ok, overrides A.def
    override int foo() { ... }  // error, A.foo is final
    int bar() { ... }  // ok, A.bar is final private, but not virtual
    int abc() { ... }  // ok, A.abc is not virtual, B.abc is virtual
}

void test()
{
    A a = new B;
    a.def();    // calls B.def
    a.foo();    // calls A.foo
    a.bar();    // calls A.bar
    a.abc();    // calls A.abc
}

Member functions with Objective-C linkage are virtual even if marked with final or static, and can be overridden.

Covariance

An overriding function may be covariant with the overridden function. A covariant function has a type that is implicitly convertible to the type of the overridden function.

class A { }
class B : A { }

class Foo
{
    A test() { return null; }
}

class Bar : Foo
{
    // overrides and is covariant with Foo.test()
    override B test() { return null; }
}

Calling Base Class Methods

To directly call a member function of a base class Base, write Base. before the function name. This avoids dynamic dispatch through a function pointer. For example:

class B
{
    int foo() { return 1; }
}
class C : B
{
    override int foo() { return 2; }

    void test()
    {
        assert(B.foo() == 1);  // translated to this.B.foo(), and
                               // calls B.foo statically.
        assert(C.foo() == 2);  // calls C.foo statically, even if
                               // the actual instance of 'this' is D.
    }
}
class D : C
{
    override int foo() { return 3; }
}
void main()
{
    auto d = new D();
    assert(d.foo() == 3);    // calls D.foo
    assert(d.B.foo() == 1);  // calls B.foo
    assert(d.C.foo() == 2);  // calls C.foo
    d.test();
}

Base class methods can also be called through the super reference.

Overload Sets and Overriding

When doing overload resolution, the functions in the base class are not considered, as they are not in the same Overload Set:

class A
{
    int foo(int x) { ... }
    int foo(long y) { ... }
}

class B : A
{
    override int foo(long x) { ... }
}

void test()
{
    B b = new B();
    b.foo(1);  // calls B.foo(long), since A.foo(int) is not considered

    A a = b;
    a.foo(1);  // issues runtime error (instead of calling A.foo(int))
}

To include the base class's functions in the overload resolution process, use an AliasDeclaration:

class A
{
    int foo(int x) { ... }
    int foo(long y) { ... }
}

class B : A
{
    alias foo = A.foo;
    override int foo(long x) { ... }
}

void test()
{
    A a = new B();
    a.foo(1);      // calls A.foo(int)
    B b = new B();
    b.foo(1);      // calls A.foo(int)
}

If such an AliasDeclaration is not used, the derived class's functions completely override all the functions of the same name in the base class, even if the types of the parameters in the base class functions are different. It is illegal if, through implicit conversions to the base class, those other functions do get called:

class A
{
    void set(long i) { }
    void set(int i)  { }
}
class B : A
{
    override void set(long i) { }
}

void test()
{
    A a = new B;
    a.set(3);   // error, use of A.set(int) is hidden by B
                // use 'alias set = A.set;' to introduce base class overload set
}

Default Values

A function parameter's default value is not inherited:

class A
{
    void foo(int x = 5) { ... }
}

class B : A
{
    void foo(int x = 7) { ... }
}

class C : B
{
    void foo(int x) { ... }
}

void test()
{
    A a = new A();
    a.foo();       // calls A.foo(5)

    B b = new B();
    b.foo();       // calls B.foo(7)

    C c = new C();
    c.foo();       // error, need an argument for C.foo
}

Inherited Attributes

An overriding function inherits any unspecified FunctionAttributes from the attributes of the overridden function.

class B
{
    void foo() pure nothrow @safe {}
}
class D : B
{
    override void foo() {}
}
void main()
{
    auto d = new D();
    pragma(msg, typeof(&d.foo));
    // prints "void delegate() pure nothrow @safe" in compile time
}

Restrictions

The attributes @disable and deprecated are not allowed on overriding functions.

class B
{
    void foo() {}
}

class D : B
{
    @disable override void foo() {}  // error, can't apply @disable to overriding function
}

Inline Functions

The compiler makes the decision whether to inline a function or not. This decision may be controlled by pragma(inline).

Function Overloading

Function overloading occurs when two or more functions in the same scope have the same name. The function selected is the one that is the best match to the arguments. The matching levels are:

1. No match
2. Match with implicit conversions
3. Match with qualifier conversion (if the argument type is qualifier-convertible to the parameter type)
4. Exact match

Named arguments are resolved for a candidate according to Matching Arguments to Parameters. If this fails (for example, because the overload does not have a parameter matching a named argument), the level is no match. Other than that, named arguments do not affect the matching level.

Each argument (including any this reference) is compared against the function's corresponding parameter to determine the match level for that argument. The match level for a function is the worst match level of each of its arguments.

Literals do not match ref or out parameters.

scope parameter storage class does not affect function overloading.

If two or more functions have the same match level, then partial ordering is used to disambiguate to find the best match. Partial ordering finds the most specialized function. If neither function is more specialized than the other, then it is an ambiguity error. Partial ordering is determined for functions f and g by taking the parameter types of f, constructing a list of arguments by taking the default values of those types, and attempting to match them against g. If it succeeds, then g is at least as specialized as f. For example:

class A { }
class B : A { }
class C : B { }
void foo(A);
void foo(B);

void test()
{
    C c;
    /* Both foo(A) and foo(B) match with implicit conversions (level 2).
     * Applying partial ordering rules,
     * foo(B) cannot be called with an A, and foo(A) can be called
     * with a B. Therefore, foo(B) is more specialized, and is selected.
     */
    foo(c); // calls foo(B)
}

A function with a variadic argument is considered less specialized than a function without.

A static member function can be overloaded with a member function. The struct, class or union of the static member function is inferred from the type of the this argument.

struct S {
    void eggs(int);
    static void eggs(long);
}
S s;
s.eggs(0);  // calls void eggs(int);
S.eggs(0);  // error: need `this`
s.eggs(0L); // calls static void eggs(long);
S.eggs(0L); // calls static void eggs(long);

struct T {
    void bacon(int);
    static void bacon(int);
}
T t;
t.bacon(0);  // error: ambiguous
T.bacon(0);  // error: ambiguous

Overload Sets

Functions declared at the same scope overload against each other, and are called an Overload Set. An example of an overload set are functions defined at module level:

module A;
void foo() { }
void foo(long i) { }

A.foo() and A.foo(long) form an overload set. A different module can also define another overload set of functions with the same name:

module B;
class C { }
void foo(C) { }
void foo(int i) { }

and A and B can be imported by a third module, C. Both overload sets, the A.foo overload set and the B.foo overload set, are found when searching for symbol foo. An instance of foo is selected based on it matching in exactly one overload set:

import A;
import B;

void bar(C c , long i)
{
    foo();    // calls A.foo()
    foo(i);  // calls A.foo(long)
    foo(c);   // calls B.foo(C)
    foo(1,2); // error, does not match any foo
    foo(1);   // error, matches A.foo(long) and B.foo(int)
    A.foo(1); // calls A.foo(long)
}

Even though B.foo(int) is a better match than A.foo(long) for foo(1), it is an error because the two matches are in different overload sets.

Overload sets can be merged with an alias declaration:

import A;
import B;

alias foo = A.foo;
alias foo = B.foo;

void bar(C c)
{
    foo();    // calls A.foo()
    foo(1L);  // calls A.foo(long)
    foo(c);   // calls B.foo(C)
    foo(1,2); // error, does not match any foo
    foo(1);   // calls B.foo(int)
    A.foo(1); // calls A.foo(long)
}

Function Parameters

Parameter Storage Classes

Parameter storage classes are in, out, ref, lazy, return and scope. Parameters can also take the type constructors const, immutable, shared and inout.

in, out, ref and lazy are mutually exclusive. The first three are used to denote input, output and input/output parameters, respectively. For example:

int read(in char[] input, ref size_t count, out int errno);

void main()
{
    size_t a = 42;
    int b;
    int r = read("Hello World", a, b);
}

read has three parameters. input will only be read and no reference to it will be retained. count may be read and written to, and errno will be set to a value from within the function.

The argument "Hello World" gets bound to parameter input, a gets bound to count and b to errno.

In Parameters

The parameter is an input to the function. Input parameters behave as if they have the const scope storage classes. Input parameters may also be passed by reference by the compiler.

Unlike ref parameters, in parameters can bind to both lvalues and rvalues (such as literals).

Types that would trigger a side effect if passed by value (such as types with copy constructor, postblit, or destructor), and types which cannot be copied (e.g. if their copy constructor is marked as @disable) will always be passed by reference. Dynamic arrays, classes, associative arrays, function pointers, and delegates will always be passed by value.

Ref and Out Parameters

By default, parameters take rvalue arguments. A ref parameter takes an lvalue argument, so changes to its value will operate on the caller's argument.

void inc(ref int x)
{
    x += 1;
}

void seattle()
{
    int z = 3;
    inc(z);
    assert(z == 4);
}

A ref parameter can also be returned by reference, see Return Ref Parameters.

An out parameter is similar to a ref parameter, except it is initialized with x.init upon function invocation.

void zero(out int x)
{
    assert(x == 0);
}

void two(out int x)
{
    x = 2;
}

void tacoma()
{
    int a = 3;
    zero(a);
    assert(a == 0);

    int y = 3;
    two(y);
    assert(y == 2);
}

For dynamic array and class object parameters, which are always passed by reference, out and ref apply only to the reference and not the contents.

Lazy Parameters

An argument to a lazy parameter is not evaluated before the function is called. The argument is only evaluated if/when the parameter is evaluated within the function. Hence, a lazy argument can be executed 0 or more times.

import std.stdio : writeln;

void main()
{
    int x;
    3.times(writeln(x++));
    writeln("-");
    writeln(x);
}

void times(int n, lazy void exp)
{
    while (n--)
        exp();
}

prints to the console:

0
1
2
−
3

A lazy parameter cannot be an lvalue.

The underlying delegate of the lazy parameter may be extracted by using the & operator:

void test(lazy int dg)
{
    int delegate() dg_ = &dg;
    assert(dg_() == 7);
    assert(dg == dg_());
}

void main()
{
    int a = 7;
    test(a);
}

A lazy parameter of type void can accept an argument of any type.

See Also: Lazy Variadic Functions

Default Arguments

Function parameter declarations can have default values:

void foo(int x, int y = 3)
{
    ...
}
...
foo(4);   // same as foo(4, 3);

Default parameters are resolved and semantically checked in the context of the function declaration.

module m;
private immutable int b;
pure void g(int a = b) {}

import m;
int b;
pure void f()
{
    g();  // ok, uses m.b
}

The attributes of the AssignExpression are applied where the default expression is used.

module m;
int b;
pure void g(int a = b) {}

import m;
enum int b = 3;
pure void f()
{
    g();  // error, cannot access mutable global `m.b` in pure function
}

See also: function type aliases with default values.

Return Ref Parameters

Return ref parameters are used with ref functions to ensure that the returned reference will not outlive the matching argument's lifetime.

ref int identity(return ref int x) {
    return x; // pass-through function that does nothing
}

ref int fun() {
    int x;
    return identity(x); // Error: escaping reference to local variable x
}

ref int gun(return ref int x) {
    return identity(x); // OK
}

Returning the address of a ref variable is also checked in @safe code.

int* pluto(ref int i) @safe
{
    return &i; // error: returning &i escapes a reference to parameter i
}

int* mars(return ref int i) @safe
{
    return &i;  // OK with -preview=dip1000
}

void f(ref scope int* p, return ref int i) @safe
{
    p = &i; // OK with -preview=dip1000
}

void main() @safe
{
    int i;
    int* p;
    f(p, i); // OK, lifetime of p is shorter than i
    *p = 5;
    assert(i == 5);

    int j;
    //f(p, j); // error, lifetime of p is longer than j
}

struct S
{
    private int* p;

    void f(return ref int i) scope @safe
    {
        p = &i; // OK with -preview=dip1000
    }
}

void main() @safe
{
    int i;
    S s;
    s.f(i); // OK, lifetime of `s` is shorter than `i`
    *s.p = 2;
    assert(i == 2);
}

If there are multiple return ref parameters, the lifetime of the return value is the smallest lifetime of the corresponding arguments.

Neither the type of the return ref parameter(s) nor the type of the return value is considered when determining the lifetime of the return value.

It is not an error if the return type does not contain any indirections.

int mercury(return ref int i)
{
    return i; // ok
}

Template functions, auto functions, nested functions and lambdas can deduce the return attribute.

@safe:

ref int templateFunction()(ref int i)
{
    return i; // ok
}

ref auto autoFunction(ref int i)
{
    return i; // ok
}

void uranus()
{
    ref int nestedFunction(ref int i)
    {
        return i; // ok
    }
    auto lambdaFunction =
        (ref int i)
        {
            return &i; // ok
        };
}

Struct Return Methods

Struct non-static methods can be marked with the return attribute to ensure a returned reference will not outlive the struct instance.

struct S
{
    private int x;
    ref int get() return { return x; }
}

ref int escape()
{
    S s;
    return s.get(); // Error: escaping reference to local variable s
}

The hidden this ref-parameter then becomes return ref.

The return attribute can also be used to limit the lifetime of the returned value, even when the method is not ref:

struct S
{
    private int i;
    int* get() return @safe => &i;
}

void f() @safe
{
    int* p;
    {
        S s;
        int *q = s.get(); // OK, q has shorter lifetime than s
        p = s.get(); // error, p has longer lifetime
        p = (new S).get(); // OK, heap allocated S
    }
}

Scope Parameters

A scope parameter of reference type must not escape the function call (e.g. by being assigned to a global variable). It has no effect for non-reference types. scope escape analysis is only done for @safe functions. For other functions scope semantics must be manually enforced.

@safe:

int* gp;
void thorin(scope int*);
void gloin(int*);
int* balin(scope int* q, int* r)
{
     gp = q; // error, q escapes to global gp
     gp = r; // ok

     thorin(q); // ok, q does not escape thorin()
     thorin(r); // ok

     gloin(q); // error, gloin() escapes q
     gloin(r); // ok that gloin() escapes r

     return q; // error, cannot return 'scope' q
     return r; // ok
}

Return Scope Parameters

Parameters marked as return scope that contain indirections can only escape those indirections via the function's return value.

@safe:

int* gp;
void thorin(scope int*);
void gloin(int*);
int* balin(return scope int* p)
{
     gp = p; // error, p escapes to global gp
     thorin(p); // ok, p does not escape thorin()
     gloin(p); // error, gloin() escapes p
     return p; // ok
}

Class references are considered pointers that are subject to scope.

@safe:

class C { }
C gp;
void thorin(scope C);
void gloin(C);
C balin(return scope C p, scope C q, C r)
{
     gp = p; // error, p escapes to global gp
     gp = q; // error, q escapes to global gp
     gp = r; // ok

     thorin(p); // ok, p does not escape thorin()
     thorin(q); // ok
     thorin(r); // ok

     gloin(p); // error, gloin() escapes p
     gloin(q); // error, gloin() escapes q
     gloin(r); // ok that gloin() escapes r

     return p; // ok
     return q; // error, cannot return 'scope' q
     return r; // ok
}

return scope can be applied to the this of class and interface member functions.

class C
{
    C bofur() return scope { return this; }
}

Template functions, auto functions, nested functions and lambdas can deduce the return scope attribute.

Ref Return Scope Parameters

It is not possible to have both return ref and return scope semantics for the same parameter. When a parameter is passed by ref and has both the return and scope storage classes, it gets return scope semantics if and only if the return and scope keywords appear adjacent to each other, in that order. Specifying a return ref and scope parameter enables returning a reference to a scope pointer. In all other cases, the parameter has return ref semantics and regular scope semantics.

U xerxes(       ref return scope V v) // (1) ref and return scope
U sargon(return ref        scope V v) // (2) return ref and scope

struct S
{
    // note: in struct member functions, the implicit `this` parameter
    // is passed by `ref`

    U xerxes() return scope;        // return scope
    U sargon()        scope return; // return ref, `return` comes after `scope`
    U xerxes() return const scope;  // return ref, `return` and `scope` are not adjacent
}

Example of combinations of return scope, return ref, and scope semantics:

@safe:

int* globalPtr;

struct S
{
    int  val;
    int* ptr;

    this(return scope ref int* p) { ptr = p; }

    // note: `this` is passed by `ref` in structs

    int* retRefA() scope return // return-ref, scope
    {
        globalPtr = this.ptr; // disallowed, `this` is `scope`
        return &this.val; // allowed, `return` means `return ref`
    }

    ref int retRefB() scope return // return-ref, scope
    {
        globalPtr = this.ptr; // disallowed, `this` is `scope`
        return  this.val; // allowed, `return` means `return ref`
    }

    int* retScopeA() return scope // ref, return-scope
    {
        return &this.val; // disallowed, escaping a reference to `this`
        return this.ptr;  // allowed, returning a `return scope` pointer
    }

    ref int retScopeB() return scope // ref, return-scope
    {
        return this.val;  // disallowed, escaping a reference to `this`
        return *this.ptr; // allowed, returning a `return scope` pointer
    }

    ref int* retRefScopeC() scope return // return-ref, scope
    {
        return this.ptr; // allowed, returning a reference to a scope pointer
    }
}

int* retRefA(return ref scope S s)
{
    globalPtr = s.ptr; // disallowed, `s` is `scope`
    return &s.val; // allowed, returning a reference to `return ref s`
}

ref int retRefB(return ref scope S s)
{
    globalPtr = s.ptr; // disallowed, `s` is `scope`
    return s.val;
}

int* retScopeA(ref return scope S s)
{
    return &s.val; // disallowed, escaping a reference to `s`
    return s.ptr;  // allowed, returning a `return scope` pointer
}

ref int retScopeB(ref return scope S s)
{
    return s.val;  // disallowed, escaping a reference to `s`
    return *s.ptr; // allowed, returning a `return scope` pointer
}

ref int* retRefScopeC(return ref scope int* p)
{
    return p; // allowed, returning a reference to a scope pointer
}

Inferred scope parameters in pure functions

• The function is pure, hence the argument cannot be assigned to a global variable
• The function is nothrow, hence the argument cannot be assigned to a thrown Exception object
• None of the other parameters have mutable indirections, hence the argument cannot be assigned to a longer-lived variable

• If the function returns by ref or has a return type that contains pointers, the argument could be returned, so it is treated as return scope
• Otherwise, the argument cannot escape the function, so it is treated as scope

@safe:

int dereference(int* x) pure nothrow;
int* identity(int* x) pure nothrow;
int* identityThrow(int* x) pure;
void assignToRef(int* x, ref int* escapeHatch) pure nothrow;
void assignToPtr(int* x, int** escapeHatch) pure nothrow;
void cannotAssignTo(int* x, const ref int* noEscapeHatch) pure nothrow;

int* globalPtr;

int* test(scope int* ptr)
{
    int result = dereference(ptr); // allowed, treated as `scope`
    int* ptr2 = identity(ptr); // allowed, treated as `return scope`
    int* ptr3 = identityThrow(ptr); // not allowed, can throw an `Exception`
    assignToRef(ptr, globalPtr); // not allowed, mutable second parameter
    assignToPtr(ptr, &globalPtr); // not allowed, mutable second parameter
    cannotAssignTo(ptr, globalPtr); // allowed

    return ptr2; // not allowed, ptr2 is inferred `scope` now
}

User-Defined Attributes for Parameters

Variadic Functions

Variadic Functions take a variable number of arguments. There are three forms:

1. C-style variadic functions
2. Variadic functions with type info
3. Typesafe variadic functions

C-style Variadic Functions

A C-style variadic function is declared with a parameter ... as the last function parameter. It has non-D linkage, such as extern (C).

To access the variadic arguments, import the standard library module core.stdc.stdarg.

import core.stdc.stdarg;

extern (C) void dry(int x, int y, ...); // C-style Variadic Function

void spin()
{
    dry(3, 4);      // ok, no variadic arguments
    dry(3, 4, 6.8); // ok, one variadic argument
    dry(2);         // error, no argument for parameter y
}

There must be at least one non-variadic parameter declared.

extern (C) int def(...); // error, must have at least one parameter

C-style variadic functions match the C calling convention for variadic functions, and can call C Standard library functions like printf.

extern (C) int printf(const(char)*, ...);

void main()
{
    printf("hello world\n");
}

C-style variadic functions cannot be marked as @safe.

void wash()
{
    rinse(3, 4, 5);   // first variadic argument is 5
}

import core.stdc.stdarg;
extern (C) void rinse(int x, int y, ...)
{
    va_list args;
    va_start(args, y); // y is the last named parameter
    int z;
    va_arg(args, z);   // z is set to 5
    va_end(args);
}

D-style Variadic Functions

D-style variadic functions have D linkage and ... as the last parameter.

... can be the only parameter.

If there are parameters preceding the ... parameter, there must be a comma separating them from the ....

int abc(char c, ...);   // one required parameter: c
int def(...);           // no required parameters
int ghi(int i ...);     // a typesafe variadic function
//int boo(, ...);       // error

Two hidden arguments are passed to the function:

• void* _argptr
• TypeInfo[] _arguments

_argptr is a reference to the first of the variadic arguments. To access the variadic arguments, import core.vararg. Use _argptr in conjunction with core.va_arg:

import core.vararg;

void test()
{
    foo(3, 4, 5);   // first variadic argument is 5
}

@system void foo(int x, int y, ...)
{
    int z = va_arg!int(_argptr); // z is set to 5 and _argptr is advanced
                                 // to the next argument
}

_arguments gives the number of arguments and the typeid of each, enabling type safety to be checked at run time.

import std.stdio;

void main()
{
    Foo f = new Foo();
    Bar b = new Bar();

    writefln("%s", f);
    printargs(1, 2, 3L, 4.5, f, b);
}

class Foo { int x = 3; }
class Bar { long y = 4; }

import core.vararg;

@system void printargs(int x, ...)
{
    writefln("%d arguments", _arguments.length);
    for (int i = 0; i < _arguments.length; i++)
    {
        writeln(_arguments[i]);

        if (_arguments[i] == typeid(int))
        {
            int j = va_arg!(int)(_argptr);
            writefln("\t%d", j);
        }
        else if (_arguments[i] == typeid(long))
        {
            long j = va_arg!(long)(_argptr);
            writefln("\t%d", j);
        }
        else if (_arguments[i] == typeid(double))
        {
            double d = va_arg!(double)(_argptr);
            writefln("\t%g", d);
        }
        else if (_arguments[i] == typeid(Foo))
        {
            Foo f = va_arg!(Foo)(_argptr);
            writefln("\t%s", f);
        }
        else if (_arguments[i] == typeid(Bar))
        {
            Bar b = va_arg!(Bar)(_argptr);
            writefln("\t%s", b);
        }
        else
            assert(0);
    }
}

0x00870FE0
5 arguments
int
        2
long
        3
double
        4.5
Foo
        0x00870FE0
Bar
        0x00870FD0

D-style variadic functions cannot be marked as @safe.

Typesafe Variadic Functions

A typesafe variadic function has D linkage and a variadic parameter declared as either an array or a class. The array or class is constructed from the arguments, and is passed as an array or class object.

For dynamic arrays:

int sum(int[] ar ...) // typesafe variadic function
{
    int s;
    foreach (int x; ar)
        s += x;
    return s;
}

import std.stdio;

void main()
{
    writeln(stan());  // 6
    writeln(ollie()); // 15
}

int stan()
{
    return sum(1, 2, 3) + sum(); // returns 6+0
}

int ollie()
{
    int[3] ii = [4, 5, 6];
    return sum(ii);             // returns 15
}

For static arrays, the number of arguments must match the array dimension.

int sum(int[3] ar ...) // typesafe variadic function
{
    int s;
    foreach (int x; ar)
        s += x;
    return s;
}

int frank()
{
    return sum(2, 3);    // error, need 3 values for array
    return sum(1, 2, 3); // returns 6
}

int dave()
{
    int[3] ii = [4, 5, 6];
    int[] jj = ii;
    return sum(ii); // returns 15
    return sum(jj); // error, type mismatch
}

For class objects:

int tesla(int x, C c ...)
{
    return x + c.x;
}

class C
{
    int x;
    string s;

    this(int x, string s)
    {
        this.x = x;
        this.s = s;
    }
}

void edison()
{
    C g = new C(3, "abc");
    tesla(1, c);         // ok, since c is an instance of C
    tesla(1, 4, "def");  // ok
    tesla(1, 5);         // error, no matching constructor for C
}

The lifetime of the variadic class object or array instance ends at the end of the function.

C orville(C c ...)
{
    return c;   // error, c instance contents invalid after return
}

int[] wilbur(int[] a ...)
{
    return a;       // error, array contents invalid after return
    return a[0..1]; // error, array contents invalid after return
    return a.dup;   // ok, since copy is made
}

For other types, the argument is passed by value.

int neil(int i ...)
{
    return i;
}

void buzz()
{
    neil(3);    // returns 3
    neil(3, 4); // error, too many arguments
    int[] x;
    neil(x);    // error, type mismatch
}

Lazy Variadic Functions

If the variadic parameter of a function is an array of delegates with no parameters, then each of the arguments whose type does not match that of the delegate is converted to a delegate of that type.

void hal(scope int delegate()[] dgs ...);

void dave()
{
    int delegate() dg;
    hal(1, 3+x, dg, cast(int delegate())null);   // (1)
    hal( { return 1; }, { return 3+x; }, dg, null ); // same as (1)
}

The variadic delegate array differs from using a lazy variadic array. With the former each array element access would evaluate every array element. With the latter, only the element being accessed would be evaluated.

import std.stdio;

void main()
{
    int x;
    ming(++x, ++x);

    int y;
    flash(++y, ++y);
}

// lazy variadic array
void ming(lazy int[] arr...)
{
    writeln(arr[0]); // 1
    writeln(arr[1]); // 4
}

// variadic delegate array
void flash(scope int delegate()[] arr ...)
{
    writeln(arr[0]()); // 1
    writeln(arr[1]()); // 2
}

Hidden Parameters

• Non-static member functions all have a hidden parameter called the this reference, which refers to the object for which the function is called.
• D-style variadic functions have hidden parameters.
• Functions with Objective-C linkage have an additional hidden, unnamed, parameter which is the selector it was called with.

Ref Scope Return Cases

Definitions

Classification

A parameter must be in one of the following states:

Mapping Syntax Onto Classification

The juxtaposition of return immediately preceding scope means ReturnScope. Otherwise, return and ref in any position means ReturnRef.

Member Functions

Member functions are rewritten as if the this parameter is the first parameter of a non-member function,

struct S {
    X foo();
}

is treated as:

X foo(ref S);

and:

class C {
    X foo()
}

is treated as:

X foo(P)

P and ref

The rules account for switching between ref and P, such as:

int* foo(return ref int i) { return &i; }
ref int foo(int* p) { return *p; }

Covariance

Covariance means a parameter with restrictions can be converted to a parameter with fewer restrictions. This is deducible from the description of each state.

For example, scope matches all non-ref parameters, and ref scope matches all ref parameters.

Local Variables

Local variables are declared within the scope of a function. Function parameters are included.

A local variable cannot be read without first assigning it a value.

The address of or reference to a local non-static variable cannot be returned from the function.

A local variable and a label in the same function cannot have the same name.

A local variable cannot hide another local variable in the same function.

ref double func(int x)
{
    int x;       // error, hides previous definition of x
    double y;
    {
        char y;  // error, hides previous definition of y
        int z;
    }
    {
        wchar z; // Ok, previous z is out of scope
    }
  z:             // error, z is a local variable and a label
    return y;    // error, returning ref to local
}

Local Static Variables

Local variables in functions declared as static, shared static or __gshared are statically allocated rather than being allocated on the stack. The lifetime of __gshared and shared static variables begins when the function is first executed and ends when the program ends. The lifetime of static variables begins when the function is first executed within the thread and ends when that thread terminates.

import std.stdio : writeln;

void foo()
{
    static int n;
    if (++n == 100)
        writeln("called 100 times");
}

The initializer for a static variable must be evaluatable at compile time. There are no static constructors or static destructors for static local variables.

Although static variable name visibility follows the usual scoping rules, the names of them must be unique within a particular function.

void main()
{
    { static int x; }
    { static int x; } // error
    { int i; }
    { int i; } // ok
}

Nested Functions

Functions may be nested within other functions:

int bar(int a)
{
    int foo(int b)
    {
        int abc() { return 1; }

        return b + abc();
    }
    return foo(a);
}

void test()
{
    int i = bar(3); // i is assigned 4
}

Nested functions can be accessed only if the name is in scope.

void foo()
{
    void A()
    {
        B(); // error, B() is forward referenced
        C(); // error, C undefined
    }
    void B()
    {
        A(); // ok, in scope
        void C()
        {
            void D()
            {
                A();      // ok
                B();      // ok
                C();      // ok
                D();      // ok
            }
        }
    }
    A(); // ok
    B(); // ok
    C(); // error, C undefined
}

int bar(int a)
{
    int foo(int b) { return b + 1; }
    int abc(int b) { return foo(b); }   // ok
    return foo(a);
}

void test()
{
    int i = bar(3);     // ok
    int j = bar.foo(3); // error, bar.foo not visible
}

Nested functions have access to the variables and other symbols defined by the lexically enclosing function. This access includes both the ability to read and write them.

int bar(int a)
{
    int c = 3;

    int foo(int b)
    {
        b += c;       // 4 is added to b
        c++;          // bar.c is now 5
        return b + c; // 12 is returned
    }
    c = 4;
    int i = foo(a); // i is set to 12
    return i + c;   // returns 17
}

void test()
{
    int i = bar(3); // i is assigned 17
}

This access can span multiple nesting levels:

int bar(int a)
{
    int c = 3;

    int foo(int b)
    {
        int abc()
        {
            return c;   // access bar.c
        }
        return b + c + abc();
    }
    return foo(3);
}

Static nested functions cannot access any stack variables of any lexically enclosing function, but can access static variables. This is analogous to how static member functions behave.

int bar(int a)
{
    int c;
    static int d;

    static int foo(int b)
    {
        b = d;          // ok
        b = c;          // error, foo() cannot access frame of bar()
        return b + 1;
    }
    return foo(a);
}

Functions can be nested within member functions:

struct Foo
{
    int a;

    int bar()
    {
        int c;

        int foo()
        {
            return c + a;
        }
        return 0;
    }
}

Nested functions always have the D function linkage type.

Declaration Order

Unlike module level declarations, declarations within function scope are processed in order. This means that two nested functions cannot mutually call each other:

void test()
{
    void foo() { bar(); } // error, bar not defined
    void bar() { foo(); } // ok
}

There are several workarounds for this limitation:

• Declare the functions to be static members of a nested struct:

void test()
{
    static struct S
    {
        static void foo() { bar(); } // ok
        static void bar() { foo(); } // ok
    }

    S.foo();  // compiles (but note the infinite runtime loop)
}

• Declare one or more of the functions to be function templates even if they take no specific template arguments:

void test()
{
    void foo()() { bar(); } // ok (foo is a function template)
    void bar()   { foo(); } // ok
}

• Declare the functions inside of a mixin template:

mixin template T()
{
    void foo() { bar(); } // ok
    void bar() { foo(); } // ok
}

void main()
{
    mixin T!();
}

• Use a delegate:

void test()
{
    void delegate() fp;
    void foo() { fp(); }
    void bar() { foo(); }
    fp = &bar;
}

Nested functions cannot be overloaded.

Function Pointers, Delegates and Closures

Function Pointers

A function pointer is declared with the function keyword:

void f(int);
void function(int) fp = &f; // fp is a pointer to a function taking an int

A function pointer can point to a static nested function:

int function() fp;  // fp is a pointer to a function returning an int

void test()
{
    static int a = 7;
    static int foo() { return a + 3; }

    fp = &foo;
}

void main()
{
    assert(!fp);
    test();
    int i = fp();
    assert(i == 10);
}

int abc(int x)   { return x + 1; }
uint def(uint y) { return y + 1; }

int function(int)   fp1 = &abc;
uint function(uint) fp2 = &def;
// Do not rely on fp1 and fp2 being different values; the compiler may merge
// them.

Delegates & Closures

A delegate can be set to a non-static nested function:

int delegate() dg;

void test()
{
    int a = 7;
    int foo() { return a + 3; }

    dg = &foo;
    int i = dg(); // i is set to 10
}

void main()
{
    test();
    int i = dg(); // ok, test.a is in a closure and still exists
    assert(i == 10);
}

The stack variables referenced by a nested function are still valid even after the function exits (NOTE this is different from D 1.0). This combining of the environment and the function is called a dynamic closure.

Those referenced stack variables that make up the closure are allocated on the GC heap, unless:

• The closure is passed to a scope parameter.
• The closure is an initializer for a scope variable.
• The closure is assigned to a scope variable.

@nogc:
void f(scope int delegate());
void g(int delegate());

void main()
{
    int i;
    int h() { return i; }
    h(); // OK
    scope x = &h; // OK
    x(); // OK
    //auto y = &h; // error, can't allocate closure in @nogc function
    f(&h); // OK
    //g(&h); // error

    // delegate literals
    f(() => i); // OK
    scope d = () => i; // OK
    d = () => i + 1; // OK
    f(d);
    //g(() => i); // error, can't allocate closure in @nogc function
}

Method Delegates

Delegates to non-static nested functions contain two pieces of data: the pointer to the stack frame of the lexically enclosing function (called the context pointer) and the address of the function. This is analogous to struct/class non-static member function delegates consisting of a this pointer and the address of the member function. Both forms of delegates are indistinguishable, and are the same type.

A delegate can be set to a particular object and method using &obj.method:

struct Foo
{
    int a;
    int get() { return a; }
}

int add1(int delegate() dg)
{
    return dg() + 1;
}

void main()
{
    Foo f = {7};
    int delegate() dg = &f.get; // bind to an instance of Foo and a method
    assert(dg.ptr == &f);
    assert(dg.funcptr == &Foo.get);

    int i = add1(dg);
    assert(i == 8);

    int x = 27;
    int abc() { return x; }

    i = add1(&abc);
    assert(i == 28);
}

The .ptr property of a delegate will return the context pointer value as a void*.

The .funcptr property of a delegate will return the function pointer value as a function type.

Initialization

Function pointers are zero-initialized by default. They can be initialized to the address of any function (including a function literal). Initialization with the address of a function that requires a context pointer is not allowed in @safe functions.

struct S
{
    static int sfunc();
    int member();   // has hidden `this` reference parameter
}

@safe void sun()
{
    int function() fp = &S.sfunc;
    fp(); // Ok
    fp = &S.member; // error
}

@system void moon()
{
    int function() fp = &S.member; // Ok because @system
    fp(); // undefined behavior
}

Delegates are zero-initialized by default. They can be initialized by taking the address of a non-static member function, but a context pointer must be supplied. They can be initialized by taking the address of a non-static nested function or function literal, where the context pointer will be set to point to the stack frame, closure, or null.

Delegates cannot be initialized by taking the address of a global function, a static member function, or a static nested function.

struct S
{
    static int sfunc();
    int member() { return 1; }
}

void main()
{
    S s;
    int delegate() dg = &s.member; // Ok, s supplies context pointer
    assert(dg() == 1);

    //dg = &S.sfunc;  // error
    //dg = &S.member; // error

    int moon() { return 2; }
    dg = &moon;     // Ok
    assert(dg() == 2);

    static int mars() { return 3; }
    //dg = &mars;     // error

    dg = () { return 4; }; // Ok
    assert(dg() == 4);
}

The last assignment uses a FunctionLiteral, which is inferred as a delegate.

Anonymous Functions and Anonymous Delegates

See FunctionLiterals.

main() Function

For console programs, main() serves as the entry point. It gets called after all the module initializers are run, and after any unittests are run. After it returns, all the module destructors are run. main() must be declared as follows:

        MainFunction:
            MainReturnDecl main() MainFunctionBody
            MainReturnDecl main(string[] Identifier) MainFunctionBody

        MainReturnDecl:
            void
            int
            noreturn
            auto

        MainFunctionBody:
            ShortenedFunctionBody
            SpecifiedFunctionBody
        

• If main returns void, the OS will receive a zero value on success.
• If main returns void or noreturn, the OS will receive a non-zero value on abnormal termination, such as an uncaught exception.
• If main is declared as auto, the inferred return type must be one of void, int and noreturn.

If the string[] parameter is declared, the parameter will hold arguments passed to the program by the OS. The first argument is typically the executable name, followed by any command-line arguments.

The main function must have D linkage.

Attributes may be added as needed, e.g. @safe, @nogc, nothrow, etc.

extern(C) main() Function

Programs may define an extern(C) main function as an alternative to the standard entry point. This form is required for BetterC programs.

A C main function must be declared as follows:

        CMainFunction:
            extern (C) MainReturnDecl main(CmainParametersopt) BlockStatement

        CmainParameters:
            int Identifier, char** Identifier
            int Identifier, char** Identifier, char** Identifier
        

When defined, the first two parameters denote a C-style array (length + pointer) that holds the arguments passed to the program by the OS. The third parameter is a POSIX extension called environ and holds information about the current environment variables.

This function takes the place of the C main function and is executed immediately without any setup or teardown associated with a D main function. Programs reliant on module constructors, module destructors, or unittests need to manually perform (de)initialization using the appropriate runtime functions.

pragma(mangle, "main")
int myMain(int a, int b, int c)
{
    return 0;
}

Function Templates

Functions can have compile time arguments in the form of a template. See function templates.

Compile Time Function Execution (CTFE)

In contexts where a compile time value is required, functions can be used to compute those values. This is called Compile Time Function Execution, or CTFE.

These contexts are:

• initialization of a static variable or a manifest constant
• static initializers of struct/class members
• dimension of a static array
• argument for a template value parameter
• static if
• static foreach
• static assert
• mixin statement
• pragma argument
• __traits argument

enum eval(alias arg) = arg;

int square(int i)
{
    return i * i;
}

void main()
{
    import std.stdio;

    static j = square(3);      // CTFE
    writeln(j);
    assert(square(4) == 16);       // run time
    static assert(square(3) == 9); // CTFE
    writeln(eval!(square(5))); // CTFE
}

The function must have a SpecifiedFunctionBody.

CTFE is subject to the following restrictions:

1. Expressions may not reference any global or local static variables.
2. AsmStatements are not permitted
3. Non-portable casts (eg, from int[] to float[]), including casts which depend on endianness, are not permitted. Casts between signed and unsigned types are permitted.
4. Reinterpretation of overlapped fields in a union is not permitted.

Pointers are permitted in CTFE, provided they are used safely:

• Pointer arithmetic is permitted only on pointers which point to static or dynamic array elements. A pointer may also point to the first element past the array, although such pointers cannot be dereferenced. Pointer arithmetic on pointers which are null, or which point to a non-array, is not allowed.
• Ordered comparison (<, <=, >, >=) between two pointers is permitted when both pointers point to the same array, or when at least one pointer is null.
• Pointer comparisons between discontiguous memory blocks are illegal, unless two such comparisons are combined using && or || to yield a result which is independent of the ordering of memory blocks. Each comparison must consist of two pointer expressions compared with <, <=, >, or >=, and may optionally be negated with !. For example, the expression (p1 > q1 && p2 <= q2) is permitted when p1, p2 are expressions yielding pointers to memory block P, and q1, q2 are expressions yielding pointers to memory block Q, even when P and Q are unrelated memory blocks. It returns true if [p1..p2] lies inside [q1..q2], and false otherwise. Similarly, the expression (p1 < q1 || p2 > q2) is true if [p1..p2] lies outside [q1..q2], and false otherwise.
• Equality comparisons (==, !=, is, !is) are permitted between all pointers, without restriction.
• Any pointer may be cast to void* and from void* back to its original type. Casting between pointer and non-pointer types is illegal.

The above restrictions apply only to expressions which are actually executed. For example:

static int y = 0;

int countTen(int x)
{
    if (x > 10)
        ++y;    // access static variable
    return x;
}

static assert(countTen(6) == 6);    // OK
static assert(countTen(12) == 12);  // invalid, modifies y.

The __ctfe boolean pseudo-variable evaluates to true during CTFE but false otherwise.

Non-recoverable errors (such as assert failures) are illegal.

String Mixins and Compile Time Function Execution

All functions that execute in CTFE must also be executable at run time. The compile time evaluation of a function does the equivalent of running the function at run time. The semantics of a function cannot depend on compile time values of the function. For example:

int foo(string s)
{
    return mixin(s);
}

const int x = foo("1");

No-GC Functions

No-GC functions are functions marked with the @nogc attribute. Those functions do not allocate memory on the GC heap. These operations are not allowed in No-GC functions:

1. constructing an array on the heap
2. resizing an array by writing to its .length property
3. array concatenation
4. array appending
5. constructing an associative array
6. indexing an associative array
   Rationale: Indexing may throw a RangeError if the specified key is not present.
7. allocating an object with new on the heap
   Note: new declarations of class types in function scopes are compatible with @nogc if used to initialize a scope variable, as they result in allocations on the stack.
8. calling functions that are not @nogc, unless the call is in a ConditionalStatement controlled by a DebugCondition

@nogc void foo()
{
    auto a = ['a'];    // (1) error, allocates
    a.length = 1;      // (2) error, array resizing allocates
    a = a ~ a;         // (3) error, arrays concatenation allocates
    a ~= 'c';          // (4) error, appending to arrays allocates

    auto aa = ["x":1]; // (5) error, allocates
    aa["abc"];         // (6) error, indexing may allocate and throws

    auto p = new int;  // (7) error, operator new allocates
    scope auto p = new GenericClass(); // (7) Ok
    bar();             // (8) error, bar() may allocate
    debug bar();       // (8) Ok
}
void bar() { }

No-GC functions can only use a closure if it is scope - see Delegates & Closures.

@nogc int delegate() foo()
{
    int n;              // error, variable n cannot be allocated on heap
    return (){ return n; } // since `n` escapes `foo()`, a closure is required
}

@nogc affects the type of the function. A @nogc function is covariant with a non-@nogc function.

void function() fp;
void function() @nogc gp;  // pointer to @nogc function

void foo();
@nogc void bar();

void test()
{
    fp = &foo; // ok
    fp = &bar; // ok, it's covariant
    gp = &foo; // error, not contravariant
    gp = &bar; // ok
}

Function Safety

Safe Functions

Safe functions are marked with the @safe attribute. @safe can be inferred, see Function Attribute Inference.

Safe functions have safe interfaces. An implementation must enforce this by restricting the function's body to operations that are known to be safe, except for calls to @trusted functions.

The following restrictions are enforced by the compiler in safe functions:

• No casting from a pointer type T to any type U with pointers, except when:
• T implicitly converts to U
• U implements class or interface T, and both types are extern(D)
• T.opCast!U is @safe
• Both types are dynamic arrays, or both types are raw pointers, and:
  • The target element type is not a pointer type
  • The target element type is not mutable when the source element type is a pointer type
  • Either the target element type is a dynamic array, or the target element type is no larger than the source element type
  • Any source element type modifiers implicitly convert to the target element type modifiers
  • Neither element type is a function type
  • The target element type is not mutable when the source type is void[]
  • The target element type only stores safe values
  • The source element type only stores safe values when the target element type is mutable
  • Neither element type is opaque.
• No casting from any non-pointer type to a pointer type.
• No pointer arithmetic (including pointer indexing & slicing).
• Cannot access union fields that:
• Have pointers or references overlapping with other types
• Have invariants overlapping with other types
• Overlap with fields that could become unsafe values
• Calling any System Functions.
• No catching of exceptions that are not derived from class Exception.
• Inline assembler must be marked as @trusted.
• No explicit casting (except with a matching @safe opCast) of:
• mutable objects to immutable (except basic data types)
• immutable objects to mutable (except basic data types)
• thread local objects to shared
• shared objects to thread local
• Cannot access @system or __gshared variables.
• Cannot use void initializers for types containing:
• Pointers/reference types
• Types with invariants
• Unsafe values

Functions nested inside safe functions default to being safe functions.

Safe functions are covariant with trusted or system functions.

Safe External Functions

External functions don't have a function body visible to the compiler:

@safe extern (C) void play();

Trusted Functions

Trusted functions are marked with the @trusted attribute.

Like safe functions, trusted functions have safe interfaces. Unlike safe functions, this is not enforced by restrictions on the function body. Instead, it is the responsibility of the programmer to ensure that the interface of a trusted function is safe.

Example:

immutable(int)* f(int* p) @trusted
{
    version (none) p[2] = 13;
    // Invalid. p[2] is out of bounds. This line would exhibit undefined
    // behavior.

    version (none) p[1] = 13;
    // Invalid. In this program, p[1] happens to be in-bounds, so the
    // line would not exhibit undefined behavior, but a trusted function
    // is not allowed to rely on this.

    version (none) return cast(immutable) p;
    // Invalid. @safe code still has mutable access and could trigger
    // undefined behavior by overwriting the value later on.

    int* p2 = new int;
    *p2 = 42;
    return cast(immutable) p2;
    // Valid. After f returns, no mutable aliases of p2 can exist.
}

void main() @safe
{
    int[2] a = [10, 20];
    int* mp = &a[0];
    immutable(int)* ip = f(mp);
    assert(a[1] == 20); // Guaranteed. f cannot access a[1].
    assert(ip !is mp); // Guaranteed. f cannot introduce unsafe aliasing.
}

Trusted functions may call safe, trusted, or system functions.

Trusted functions are covariant with safe or system functions.

System Functions

System functions are functions not marked with @safe or @trusted and are not nested inside @safe functions. System functions may be marked with the @system attribute. A function being system does not mean it actually is unsafe, it just means that its safety must be manually verified.

System functions are not covariant with trusted or safe functions.

System functions can call safe and trusted functions.

Safe Interfaces

When a function call's arguments, any context and accessible globals each have safe values with safe aliasing, that function has a safe interface when:

1. it cannot exhibit undefined behavior, and
2. it cannot create unsafe values that are accessible from @safe code (e.g., via return values, global variables, or ref parameters), and
3. it cannot introduce unsafe aliasing that is accessible from @safe code.

Functions that meet these requirements may be @safe or @trusted. Function that do not meet these requirements can only be @system.

Examples:

• C's free does not have a safe interface:

  extern (C) @system void free(void* ptr);
  

  because free(p) invalidates p, making its value unsafe. free can only be @system.
• C's strlen and memcpy do not have safe interfaces:

  extern (C) @system size_t strlen(char* s);
  extern (C) @system void* memcpy(void* dst, void* src, size_t nbytes);
  

  because they iterate pointers based on unverified assumptions (strlen assumes that s is zero-terminated; memcpy assumes that the memory objects pointed to by dst and src are at least nbytes big). Any function that traverses a C string passed as an argument can only be @system. Any function that trusts a separate parameter for array bounds can only be @system.
• C's malloc does have a safe interface:

  extern (C) @trusted void* malloc(size_t sz);
  

  It does not exhibit undefined behavior for any input. It returns either a valid pointer, which is safe, or null which is also safe. It returns a pointer to a fresh allocation, so it cannot introduce any unsafe aliasing.
  Note: The implementation of malloc is most likely @system code.
• A D version of memcpy can have a safe interface:

  @safe void memcpy(E)(E[] src, E[] dst)
  {
      import std.math : min;
      foreach (i; 0 .. min(src.length, dst.length))
      {
          dst[i] = src[i];
      }
  }
  

  because the rules for safe values ensure that the lengths of the arrays are correct.

Safe Values

A variable or field marked as @system does not hold a safe value, regardless of its type.

For a bool, only 0 and 1 are safe values.

For all other basic data types, all possible bit patterns are safe.

A pointer is a safe value when it is one of:

1. null
2. it points to a memory object that is live and the pointed to value in that memory object is safe.

Examples:

int* n = null; /* n is safe because dereferencing null is a well-defined
    crash. */
int* x = cast(int*) 0xDEADBEEF; /* x is (most likely) unsafe because it
    is not a valid pointer and cannot be dereferenced. */

import core.stdc.stdlib: malloc, free;
int* p1 = cast(int*) malloc(int.sizeof); /* p1 is safe because the
    pointer is valid and *p1 is safe regardless of its actual value. */
free(p1); /* This makes p1 unsafe. */
int** p2 = &p1; /* While it can be dereferenced, p2 is unsafe because p1
    is unsafe. */
p1 = null; /* This makes p1 and p2 safe. */

A dynamic array is safe when:

1. its pointer is safe, and
2. its length is in-bounds with the corresponding memory object, and
3. all its elements are safe.

Examples:

int[] f() @system
{
    bool b = true; /* b is initialized safe */
    *(cast(ubyte*) &b) = 0xAA; /* b is now unsafe because it's not 0 or 1 */
    int[3] a;
    int[] d1 = a[0 .. 3]; /* d1 is safe. */
    int[] d2 = a.ptr[0 .. 4]; /* d2 is unsafe because it goes beyond a's
        bounds. */
    int*[] d3 = [cast(int*) 0xDEADBEEF]; /* d3 is unsafe because the
        element is unsafe. */
    return d1; /* Up to here, d1 was safe, but its pointer becomes
        invalid when the function returns, so the returned dynamic array
        is unsafe. */
}

A static array is safe when all its elements are safe. Regardless of the element type, a static array with length zero is always safe.

An associative array is safe when all its keys and elements are safe.

A struct/union instance is safe when:

1. the values of its accessible fields are safe, and
2. it does not introduce unsafe aliasing with unions.

Examples:

void fun()
{
    struct S { int* p; }
    S s1 = S(new int); /* s1 is safe. */
    S s2 = S(cast(int*) 0xDEADBEEF); /* s2 is unsafe, because s2.p is
        unsafe. */

    union U { int* p; size_t x; }
    U u = U(new int); /* Even though both u.p and u.x are safe, u is unsafe
        because of unsafe aliasing. */
}

A class reference is safe when it is null or:

1. it refers to a valid class instance of the class type or a type derived from the class type, and
2. the values of the instance's accessible fields are safe, and
3. it does not introduce unsafe aliasing with unions.

A function pointer is safe when it is null or it refers to a valid function that has the same or a covariant signature.

A delegate is safe when:

1. its .funcptr property is null or refers to a function that matches or is covariant with the delegate type, and
2. its .ptr property is null or refers to a memory object that is in a form expected by the function.

Safe Aliasing

When one memory location is accessible with two different types, that aliasing is considered safe if:

1. both types are const or immutable; or
2. one of the types is mutable while the other is a const-qualified basic data type; or
3. both types are mutable basic data types; or
4. one of the types is a static array type with length zero; or
5. one of the types is a static array type with non-zero length, and aliasing of the array's element type and the other type is safe; or
6. both types are pointer types, and aliasing of the target types is safe, and the target types have the same size.

All other cases of aliasing are considered unsafe.

Examples:

void f1(ref ubyte x, ref float y) @safe { x = 0; y = float.init; }
union U1 { ubyte x; float y; } // safe aliasing

void test1()
{
    U1 u1;
    f1(u1.x, u1.y); // Ok
}

void f2(ref int* x, ref int y) @trusted { x = new int; y = 0xDEADBEEF; }
union U2 { int* x; int y; } // unsafe aliasing

void test2()
{
    U2 u2;
    version (none) f1(u2.x, u2.y); // not safe
}

Function Attribute Inference

FunctionLiterals, Auto Functions, Auto Ref Functions, nested functions and function templates, since their function bodies are always present, infer the following attributes unless specifically overridden:

• pure
• nothrow
• @safe
• @nogc
• return ref parameters
• scope parameters
• return scope parameters
• ref return scope parameters

Attribute inference is not done for other functions, even if the function body is present.

The inference is done by determining if the function body follows the rules of the particular attribute.

Cyclic functions (i.e. functions that wind up directly or indirectly calling themselves) are inferred as being impure, throwing, and @system.

If a function attempts to test itself for those attributes, then the function is inferred as not having those attributes.

Uniform Function Call Syntax (UFCS)

A free function can be called like a member function when both:

• The member function does not (or cannot) exist for the object expression
• The free function's first parameter type matches the object expression

The object expression can be any type. This is called a UFCS function call.

void sun(T, int);

void moon(T t)
{
    t.sun(1);
    // If `T` does not have a member function `sun`,
    // `t.sun(1)` is interpreted as if it were written `sun(t, 1)`
}

A more complex example:

stdin.byLine(KeepTerminator.yes)
    .map!(a => a.idup)
    .array
    .sort
    .copy(stdout.lockingTextWriter());

is the equivalent of:

copy(sort(array(map!(a => a.idup)(byLine(stdin, KeepTerminator.yes)))), lockingTextWriter(stdout));

UFCS works with @property functions:

@property prop(X thisObj);
@property prop(X thisObj, int value);

X obj;
obj.prop;      // if X does not have member prop, reinterpret as prop(obj);
obj.prop = 1;  // similarly, reinterpret as prop(obj, 1);

Functions declared in a local scope are not found when searching for a matching UFCS function. Neither are other local symbols, although local imports are searched:

module a;

void foo(X);
alias boo = foo;

void main()
{
    void bar(X);     // bar declared in local scope
    import b : baz;  // void baz(X);

    X obj;
    obj.foo();    // OK, calls a.foo;
    //obj.bar();  // NG, UFCS does not see nested functions
    obj.baz();    // OK, calls b.baz, because it is declared at the
                  // top level scope of module b

    import b : boo = baz;
    obj.boo();    // OK, calls aliased b.baz instead of a.boo (== a.foo),
                  // because the declared alias name 'boo' in local scope
                  // overrides module scope name
}

Member functions are not found when searching for a matching UFCS function.

class C
{
    void mfoo(X);           // member function
    static void sbar(X);    // static member function
    import b : ibaz = baz;  // void baz(X);

    void test()
    {
        X obj;
        //obj.mfoo();  // NG, UFCS does not see member functions
        //obj.sbar();  // NG, UFCS does not see static member functions
        obj.ibaz();    // OK, ibaz is an alias of baz which is declared at
                       //     the top level scope of module b
    }
}

Otherwise, UFCS function lookup proceeds normally.

int front(int[] arr) { return arr[0]; }

void main()
{
    int[] a = [1,2,3];
    auto x = a.front();   // call .front by UFCS

    auto front = x;       // front is now a variable
    auto y = a.front();   // Error, front is not a function
}

class C
{
    int[] arr;
    int front()
    {
        return arr.front(); // Error, C.front is not callable
                            // using argument types (int[])
    }
}