Types

Grammar

D is statically typed. Every expression has a type. Types constrain the values an expression can hold, and determine the semantics of operations on those values.

Type:
    TypeCtorsopt BasicType TypeSuffixesopt

TypeCtors:
    TypeCtor
    TypeCtor TypeCtors

TypeCtor:
    const
    immutable
    inout
    shared

BasicType:
    FundamentalType
    . QualifiedIdentifier
    QualifiedIdentifier
    Typeof
    Typeof . QualifiedIdentifier
    TypeCtor ( Type )
    Vector
    TraitsExpression
    MixinType

Vector:
    __vector ( VectorBaseType )

VectorBaseType:
    Type

FundamentalType:
    bool
    byte
    ubyte
    short
    ushort
    int
    uint
    long
    ulong
    cent
    ucent
    char
    wchar
    dchar
    float
    double
    real
    ifloat
    idouble
    ireal
    cfloat
    cdouble
    creal
    void

TypeSuffixes:
    TypeSuffix TypeSuffixesopt

TypeSuffix:
    *
    [ ]
    [ AssignExpression ]
    [ AssignExpression .. AssignExpression ]
    [ Type ]
    delegate Parameters MemberFunctionAttributesopt
    function Parameters FunctionAttributesopt

QualifiedIdentifier:
    Identifier
    Identifier . QualifiedIdentifier
    TemplateInstance
    TemplateInstance . QualifiedIdentifier
    Identifier [ AssignExpression ]
    Identifier [ AssignExpression ] . QualifiedIdentifier

• Basic Data Types are leaf types.
• Derived Data Types build on leaf types.
• User-Defined Types are aggregates of basic and derived types.

Basic Data Types

Endianness of basic types is part of the ABI

NOTE: Complex and imaginary types ifloat, idouble, ireal, cfloat, cdouble, and creal have been deprecated in favor of std.complex.Complex.

Derived Data Types

• Pointers
• Static Arrays
• Dynamic Arrays
• Associative Arrays
• Function Types
• Delegate Types
• Type Sequences

int* p; // pointer
int[2] sa; // static array
int[] da; // dynamic array/slice

int[string] aa; // associative array
void function() fp; // function pointer

import std.meta : AliasSeq;
AliasSeq!(int, string) tsi; // type sequence instance

Pointers

A pointer to type T has a value which is a reference (address) to another object of type T. It is commonly called a pointer to T and its type is T*. To access the object value, use the * dereference operator:

int* p;

assert(p == null);
p = new int(5);
assert(p != null);

assert(*p == 5);
(*p)++;
assert(*p == 6);

If a pointer contains a null value, it is not pointing to a valid object.

When a pointer to T is dereferenced, it must either contain a null value, or point to a valid object of type T.

To set a pointer to point at an existing object, use the & address of operator:

int i = 2;
int* p = &i;

assert(p == &i);
assert(*p == 2);
*p = 4;
assert(i == 4);

See also Pointer Arithmetic.

User-Defined Types

• Enums
• Structs and Unions
• Classes
• Interfaces

Type Conversions

Pointer Conversions

Pointers implicitly convert to void*.

Casting between pointers and non-pointers is allowed. Some pointer casts are disallowed in @safe code.

Implicit Conversions

Implicit conversions are used to automatically convert types as required. The rules for integers are detailed in the next sections.

An enum can be implicitly converted to its base type, but going the other way requires an explicit conversion.

• All types implicitly convert to noreturn.
• Static and dynamic arrays implicitly convert to void arrays.
• Function pointers and delegates can convert to covariant types.

Class Conversions

A derived class can be implicitly converted to its base class, but going the other way requires an explicit cast. For example:

class Base {}
class Derived : Base {}
Base bd = new Derived();              // implicit conversion
Derived db = cast(Derived)new Base(); // explicit conversion

A dynamic array, say x, of a derived class can be implicitly converted to a dynamic array, say y, of a base class iff elements of x and y are qualified as being either both const or both immutable.

class Base {}
class Derived : Base {}
const(Base)[] ca = (const(Derived)[]).init; // `const` elements
immutable(Base)[] ia = (immutable(Derived)[]).init; // `immutable` elements

A static array, say x, of a derived class can be implicitly converted to a static array, say y, of a base class iff elements of x and y are qualified as being either both const or both immutable or both mutable (neither const nor immutable).

class Base {}
class Derived : Base {}
Base[3] ma = (Derived[3]).init; // mutable elements
const(Base)[3] ca = (const(Derived)[3]).init; // `const` elements
immutable(Base)[3] ia = (immutable(Derived)[3]).init; // `immutable` elements

Integer Promotions

Integer Promotions are conversions of the following types:

If an enum has as a base type one of the types in the left column, it is converted to the type in the right column.

Integer promotion applies to each operand of a binary expression:

void fun()
{
    byte a;
    auto b = a + a;
    static assert(is(typeof(b) == int));
    // error: can't implicitly convert expression of type int to byte:
    //byte c = a + a;

    ushort d;
    // error: can't implicitly convert expression of type int to ushort:
    //d = d * d;
    int e = d * d; // OK
    static assert(is(typeof(int() * d) == int));

    dchar f;
    static assert(is(typeof(f - f) == uint));
}

Usual Arithmetic Conversions

The usual arithmetic conversions convert operands of binary operators to a common type. The operands must already be of arithmetic types. The following rules are applied in order, looking at the base type:

1. If either operand is real, the other operand is converted to real.
2. Else if either operand is double, the other operand is converted to double.
3. Else if either operand is float, the other operand is converted to float.
4. Else the integer promotions above are done on each operand, followed by:
   1. If both are the same type, no more conversions are done.
   2. If both are signed or both are unsigned, the smaller type is converted to the larger.
   3. If the signed type is larger than the unsigned type, the unsigned type is converted to the signed type.
   4. The signed type is converted to the unsigned type.

Example: Signed and unsigned conversions:

int i;
uint u;
static assert(is(typeof(i + u) == uint));
static assert(is(typeof(short() + u) == uint));
static assert(is(typeof(ulong() + i) == ulong));
static assert(is(typeof(long() - u) == long));
static assert(is(typeof(long() * ulong()) == ulong));

Example: Floating point:

float f;
static assert(is(typeof(f + ulong()) == float));

double d;
static assert(is(typeof(f * d) == double));
static assert(is(typeof(real() / d) == real));

Enum Operations

If one or both of the operand types is an enum after undergoing the above conversions, the result type is determined as follows:

1. If the operands are the same type, the result will be of that type.
2. If one operand is an enum and the other is the base type of that enum, the result is the base type.
3. If the two operands are different enums, the result is the closest base type common to both. A base type being closer means there is a shorter sequence of conversions to base type to get there from the original type.

enum E { a, b, c }
enum F { x, y }

void test()
{
    E e = E.a;
    e = e | E.c;
    //e = e + 4; // error, can't assign int to E
    int i = e + 4;
    e += 4; // OK, see below

    F f;
    //f = e | f; // error, can't assign int to F
    i = e | f;
}

Integer Type Conversions

An integer of type I implicitly converts to another integer type J when J.sizeof >= I.sizeof.

void f(byte b, ubyte ub, short s)
{
    b = ub; // OK, bit pattern same
    ub = b; // OK, bit pattern same
    s = b; // OK, widening conversion
    b = s; // error, implicit narrowing
}

Integer values cannot be implicitly converted to another type that cannot represent the integer bit pattern after integral promotion. For example:

ubyte  u1 = -1; // error, -1 cannot be represented in a ubyte
ushort u2 = -1; // error, -1 cannot be represented in a ushort
uint   u3 = -1; // ok, -1 can be represented in an int, which can be converted to a uint
ulong  u4 = -1; // ok, -1 can be represented in a long, which can be converted to a ulong

Floating Point Type Conversions

• Integral types implicitly convert to floating point types.
• Floating point types cannot be implicitly converted to integral types.

void f(int i, float f)
{
    f = i; // OK
    i = f; // error
}

• Complex or imaginary floating point types cannot be implicitly converted to non-complex floating point types.
• Non-complex floating point types cannot be implicitly converted to imaginary floating point types.

Value Range Propagation

Besides type-based implicit conversions, D allows certain integer expressions to implicitly convert to a narrower type after integer promotion. This works by analysing the minimum and maximum possible range of values for each expression. If that range of values matches or is a subset of a narrower target type's value range, implicit conversion is allowed. If a subexpression is known at compile-time, that can further narrow the range of values.

void fun(char c, int i, ubyte b)
{
    // min is c.min + 100 > short.min
    // max is c.max + 100 < short.max
    short s = c + 100; // OK

    ubyte j = i & 0x3F; // OK, 0 ... 0x3F
    //ubyte k = i & 0x14A; // error, 0x14A > ubyte.max
    ushort k = i & 0x14A; // OK

    k = i & b; // OK, 0 ... b.max
    //b = b + b; // error, b.max + b.max > b.max
    s = b + b; // OK, 0 ... b.max + b.max
}

Note the implementation does not track the range of possible values for mutable variables:

void fun(int i)
{
    ushort s = i & 0xff; // OK
    // s is now assumed to be s.min ... s.max, not 0 ... 0xff
    //ubyte b = s; // error
    ubyte b = s & 0xff; // OK

    const int c = i & 0xff;
    // c's range is fixed and known
    b = c; // OK
}

• For more information, see the dmc article.
• See also: https://en.wikipedia.org/wiki/Value_range_analysis.

bool

The bool type is a byte-size type that can only hold the value true or false.

The only operators that can accept operands of type bool are: & |, ^, &=, |=, ^=, !, &&, ||, and ?:.

A bool value can be implicitly converted to any integral type, with false becoming 0 and true becoming 1.

The numeric literals 0 and 1 can be implicitly converted to the bool values false and true, respectively. Casting an expression to bool means testing !=0 for arithmetic types, and !=null for pointers or references.

• Interpreting a value with a byte representation other than 0 or 1 as bool (e.g. an overlapped union field).
• Reading a void-initialized bool.

byte i = 2;
bool b = cast(bool) i; // OK, same as `i != 0`
assert(b);

bool* p = cast(bool*) &i; // unsafe cast
// `*p` holds 0x2, an invalid bool value
// reading `*p` is undefined behavior

Function Types

A function type has the form:

StorageClassesopt Type Parameters FunctionAttributesopt

Function types are not included in the Type grammar. A function type e.g. int(int) can be aliased. A function type is only used for type tests or as the target type of a pointer.

Instantiating a function type is illegal. Instead, a pointer to function or delegate can be used. Those have these type forms respectively:

Type function Parameters FunctionAttributesopt
Type delegate Parameters MemberFunctionAttributesopt

void f(int);
alias Fun = void(int);
static assert(is(typeof(f) == Fun));
static assert(is(Fun* == void function(int)));

See Function Pointers.

Delegates

Delegates are an aggregate of two pieces of data, either:

• An object reference and a pointer to a non-static member function.
• A pointer to a closure and a pointer to a nested function. The object reference forms the this pointer when the function is called.

Delegates are declared and initialized similarly to function pointers:

int delegate(int) dg; // dg is a delegate to a function

class OB
{
    int member(int);
}

void f(OB o)
{
    dg = &o.member; // dg is a delegate to object o and member function member
}

Delegates cannot be initialized with static member functions or non-member functions.

Delegates are called analogously to function pointers:

fp(3);   // call func(3)
dg(3);   // call o.member(3)

The equivalent of member function pointers can be constructed using anonymous lambda functions:

class C
{
    int a;
    int foo(int i) { return i + a; }
}

// mfp is the member function pointer
auto mfp = function(C self, int i) { return self.foo(i); };
auto c = new C();  // create an instance of C
mfp(c, 1);  // and call c.foo(1)

typeof

Typeof:
    typeof ( Expression )
    typeof ( return )

typeof is a way to specify a type based on the type of an expression. For example:

void func(int i)
{
    typeof(i) j;       // j is of type int
    typeof(3 + 6.0) x; // x is of type double
    typeof(1)* p;      // p is of type pointer to int
    int[typeof(p)] a;  // a is of type int[int*]

    writeln(typeof('c').sizeof); // prints 1
    double c = cast(typeof(1.0))j; // cast j to double
}

Expression is not evaluated, it is used purely to generate the type:

void func()
{
    int i = 1;
    typeof(++i) j; // j is declared to be an int, i is not incremented
    writeln(i);  // prints 1
}

If Expression is a ValueSeq it will produce a TypeSeq containing the types of each element.

Special cases:

1. typeof(return) will, when inside a function scope, give the return type of that function.
2. typeof(this) will generate the type of what this would be in a non-static member function, even if not in a member function.
3. Analogously, typeof(super) will generate the type of what super would be in a non-static member function.

class A { }

class B : A
{
    typeof(this) x;  // x is declared to be a B
    typeof(super) y; // y is declared to be an A
}

struct C
{
    static typeof(this) z;  // z is declared to be a C

    typeof(super) q; // error, no super struct for C
}

typeof(this) r;   // error, no enclosing struct or class

If the expression is a Property Function, typeof gives its return type.

struct S
{
    @property int foo() { return 1; }
}
typeof(S.foo) n;  // n is declared to be an int

If the expression is a Template, typeof gives the type void.

template t {}
static assert(is(typeof(t) == void));

Mixin Types

MixinType:
    mixin ( ArgumentList )

Each AssignExpression in the ArgumentList is evaluated at compile time, and the result must be representable as a string. The resulting strings are concatenated to form a string. The text contents of the string must be compilable as a valid Type, and is compiled as such.

void test(mixin("int")* p) // int* p
{
    mixin("int")[] a;      // int[] a;
    mixin("int[]") b;      // int[] b;
}

Aliased Types

size_t

size_t is an alias to one of the unsigned integral basic types, and represents a type that is large enough to represent an offset into all addressable memory.

ptrdiff_t

ptrdiff_t is an alias to the signed integral basic type the same size as size_t.

string

A string is a special case of an array.

noreturn

noreturn is the bottom type which can implicitly convert to any type, including void. A value of type noreturn will never be produced and the compiler can optimize such code accordingly.

A function that never returns has the return type noreturn. This can occur due to an infinite loop or always throwing an exception.

noreturn abort(const(char)[] message);

int example(int i)
{
    if (i < 0)
    {
        // abort does not return, so it doesn't need to produce an int
        int val = abort("less than zero");
    }
    // ternary expression's common type is still int
    return i != 0 ? 1024 / i : abort("calculation went awry.");
}

noreturn is defined as typeof(*null). This is because dereferencing a null literal halts execution.