AttributeSpecifier: Attribute : Attribute DeclarationBlock Attribute: LinkageAttribute AlignAttribute DeprecatedAttribute VisibilityAttribute Pragma static extern abstract final override synchronized auto scope const immutable inout shared __gshared AtAttribute nothrow pure ref return AtAttribute: @ disable @ nogc Property @ safe @ system @ trusted UserDefinedAttribute Property: @ property DeclarationBlock: DeclDef { DeclDefsopt }
Attributes are a way to modify one or more declarations. The general forms are:
attribute declaration; // affects the declaration attribute: // affects all declarations until the end of // the current scope declaration; declaration; ... attribute { // affects all declarations in the block declaration; declaration; ... }
LinkageAttribute: extern ( LinkageType ) extern ( C++, IdentifierList ) LinkageType: C C++ D Windows System Objective-C
D provides an easy way to call C functions and operating system API functions, as compatibility with both is essential. The LinkageType is case sensitive, and is meant to be extensible by the implementation (they are not keywords). C and D must be supplied, the others are what makes sense for the implementation. C++ offers limited compatibility with C++. Objective-C offers limited compatibility with Objective-C, see the Interfacing to Objective-C documentation for more information. System is the same as Windows on Windows platforms, and C on other platforms. Implementation Note: for Win32 platforms, Windows should exist.
C function calling conventions are specified by:
extern (C): int foo(); // call foo() with C conventionsNote that extern(C) can be provided for all types of declarations, including struct or class, even though there is no corresponding match on the C side. In that case, the attribute is ignored. This behavior applies for nested functions and nested variables as well. However, for static member methods and static nested functions, adding extern(C) will change the calling convention, but not the mangling.
D conventions are:
extern (D):
Windows API conventions are:
extern (Windows): void *VirtualAlloc( void *lpAddress, uint dwSize, uint flAllocationType, uint flProtect );
The Windows convention is distinct from the C convention only on Win32 platforms, where it is equivalent to the stdcall convention.
Note that a lone extern declaration is used as a storage class.
The linkage form extern (C++, IdentifierList) creates C++ declarations that reside in C++ namespaces. The IdentifierList specifies the namespaces.
extern (C++, N) { void foo(); }
refers to the C++ declaration:
namespace N { void foo(); }
and can be referred to with or without qualification:
foo(); N.foo();
Namespaces create a new named scope that is imported into its enclosing scope.
extern (C++, N) { void foo(); void bar(); } extern (C++, M) { void foo(); } bar(); // ok foo(); // error - N.foo() or M.foo() ? M.foo(); // ok
Multiple identifiers in the IdentifierList create nested namespaces:
extern (C++, N.M) { extern (C++) { extern (C++, R) { void foo(); } } } N.M.R.foo();
refers to the C++ declaration:
namespace N { namespace M { namespace R { void foo(); } } }
AlignAttribute: align align ( AssignExpression )
Specifies the alignment of:
align by itself sets it to the default, which matches the default member alignment of the companion C compiler.
struct S { align: byte a; // placed at offset 0 int b; // placed at offset 4 long c; // placed at offset 8 } auto sz = S.sizeof; // 16
AssignExpression specifies the alignment which matches the behavior of the companion C compiler when non-default alignments are used. It must be a positive power of 2.
A value of 1 means that no alignment is done; fields are packed together.
struct S { align (1): byte a; // placed at offset 0 int b; // placed at offset 1 long c; // placed at offset 5 } auto sz = S.sizeof; // 16
The alignment for the fields of an aggregate does not affect the alignment of the aggregate itself - that is affected by the alignment setting outside of the aggregate.
align (2) struct S { align (1): byte a; // placed at offset 0 int b; // placed at offset 1 long c; // placed at offset 5 } auto sz = S.sizeof; // 14
Setting the alignment of a field aligns it to that power of 2, regardless of the size of the field.
struct S { align (4): byte a; // placed at offset 0 byte b; // placed at offset 4 short c; // placed at offset 8 } auto sz = S.sizeof; // 12
Do not align references or pointers that were allocated using NewExpression on boundaries that are not a multiple of size_t. The garbage collector assumes that pointers and references to GC allocated objects will be on size_t byte boundaries.
The AlignAttribute is reset to the default when entering a function scope or a non-anonymous struct, union, class, and restored when exiting that scope. It is not inherited from a base class.
DeprecatedAttribute: deprecated deprecated ( AssignExpression )
It is often necessary to deprecate a feature in a library, yet retain it for backwards compatibility. Such declarations can be marked as deprecated, which means that the compiler can be instructed to produce an error if any code refers to deprecated declarations:
deprecated { void oldFoo(); } oldFoo(); // Deprecated: function test.oldFoo is deprecated
Optionally a string literal or manifest constant can be used to provide additional information in the deprecation message.
deprecated("Don't use bar") void oldBar(); oldBar(); // Deprecated: function test.oldBar is deprecated - Don't use bar
Calling CTFE-able functions or using manifest constants is also possible.
import std.format; enum Message = format("%s and all its members are obsolete", Foobar.stringof); deprecated(Message) class Foobar {} auto f = new Foobar(); // Deprecated: class test.Foobar is deprecated - Foobar // and all its members are obsolete deprecated(format("%s is also obsolete", "This class")) class BarFoo {} auto bf = new BarFoo(); // Deprecated: class test.BarFoo is deprecated - This // class is also obsolete
Implementation Note: The compiler should have a switch specifying if deprecated should be ignored, cause a warning, or cause an error during compilation.
VisibilityAttribute: private package package ( IdentifierList ) protected public export
Visibility is an attribute that is one of private, package, protected, public, or export. They may be referred to as protection attributes in documents predating DIP22.
Symbols with private visibility can only be accessed from within the same module. Private member functions are implicitly final and cannot be overridden.
package extends private so that package members can be accessed from code in other modules that are in the same package. If no identifier is provided, this applies to the innermost package only, or defaults to private if a module is not nested in a package.
package may have an optional parameter in the form of a dot-separated identifier list which is resolved as the qualified package name. The package must be either the module's parent package or one of its anscestors. If this optional parameter is present, the symbol will be visible in the specified package and all of its descendants.
protected only applies inside classes (and templates as they can be mixed in) and means that a symbol can only be seen by members of the same module, or by a derived class. If accessing a protected instance member through a derived class member function, that member can only be accessed for the object instance which can be implicitly cast to the same type as ‘this’. protected module members are illegal.
public means that any code within the executable can see the member. It is the default visibility attribute.
export means that any code outside the executable can access the member. export is analogous to exporting definitions from a DLL.
Visibility participates in symbol name lookup.
The const attribute changes the type of the declared symbol from T to const(T), where T is the type specified (or inferred) for the introduced symbol in the absence of const.
const int foo = 7; static assert(is(typeof(foo) == const(int))); const { double bar = foo + 6; } static assert(is(typeof(bar) == const(double))); class C { const void foo(); const { void bar(); } void baz() const; } pragma(msg, typeof(C.foo)); // const void() pragma(msg, typeof(C.bar)); // const void() pragma(msg, typeof(C.baz)); // const void() static assert(is(typeof(C.foo) == typeof(C.bar)) && is(typeof(C.bar) == typeof(C.baz)));
The immutable attribute modifies the type from T to immutable(T), the same way as const does.
The inout attribute modifies the type from T to inout(T), the same way as const does.
The shared attribute modifies the type from T to shared(T), the same way as const does.
By default, non-immutable global declarations reside in thread local storage. When a global variable is marked with the __gshared attribute, its value is shared across all threads.
int foo; // Each thread has its own exclusive copy of foo. __gshared int bar; // bar is shared by all threads.
__gshared may also be applied to member variables and local variables. In these cases, __gshared is equivalent to static, except that the variable is shared by all threads rather than being thread local.
class Foo { __gshared int bar; } int foo() { __gshared int bar = 0; return bar++; // Not thread safe. }
Unlike the shared attribute, __gshared provides no safe-guards against data races or other multi-threaded synchronization issues. It is the responsibility of the programmer to ensure that access to variables marked __gshared is synchronized correctly.
__gshared is disallowed in safe mode.
A reference to a declaration marked with the @disable attribute causes a compile time error. This can be used to explicitly disallow certain operations or overloads at compile time rather than relying on generating a runtime error.
@disable void foo() { }
void main() { foo(); /* error, foo is disabled */ }
Disabling struct no-arg constructor disallows default construction of the struct.
Disabling struct postblit makes the struct not copyable.
See Function Safety.
See No-GC Functions.
See Property Functions.
See Nothrow Functions.
See Pure Functions.
See Ref Functions.
The override attribute applies to virtual functions. It means that the function must override a function with the same name and parameters in a base class. The override attribute is useful for catching errors when a base class's member function gets its parameters changed, and all derived classes need to have their overriding functions updated.
class Foo { int bar(); int abc(int x); } class Foo2 : Foo { override { int bar(char c); // error, no bar(char) in Foo int abc(int x); // ok } }
The static attribute applies to functions and data. It means that the declaration does not apply to a particular instance of an object, but to the type of the object. In other words, it means there is no this reference. static is ignored when applied to other declarations.
class Foo { static int bar() { return 6; } int foobar() { return 7; } } ... Foo f = new Foo; Foo.bar(); // produces 6 Foo.foobar(); // error, no instance of Foo f.bar(); // produces 6; f.foobar(); // produces 7;
Static functions are never virtual.
Static data has one instance per thread, not one per object.
Static does not have the additional C meaning of being local to a file. Use the private attribute in D to achieve that. For example:
module foo; int x = 3; // x is global private int y = 4; // y is local to module foo
The auto attribute is used when there are no other attributes and type inference is desired.
auto i = 6.8; // declare i as a double
For functions, the auto attribute means return type inference. See Auto Functions.
The scope attribute is used for local variables and for class declarations. For class declarations, the scope attribute creates a scope class. For local declarations, scope implements the RAII (Resource Acquisition Is Initialization) protocol. This means that the destructor for an object is automatically called when the reference to it goes out of scope. The destructor is called even if the scope is exited via a thrown exception, thus scope is used to guarantee cleanup.
If there is more than one scope variable going out of scope at the same point, then the destructors are called in the reverse order that the variables were constructed.
scope cannot be applied to globals, statics, data members, ref or out parameters. Arrays of scopes are not allowed, and scope function return values are not allowed. Assignment to a scope, other than initialization, is not allowed. Rationale: These restrictions may get relaxed in the future if a compelling reason to appears.
An abstract member function must be overridden by a derived class. Only virtual member functions may be declared abstract; non-virtual member functions and free-standing functions cannot be declared abstract.
Classes become abstract if any of its virtual member functions are declared abstract or if they are defined within an abstract attribute. Note that an abstract class may also contain non-virtual member functions.
Classes defined within an abstract attribute or with abstract member functions cannot be instantiated directly. They can only be instantiated as a base class of another, non-abstract, class.
Member functions declared as abstract can still have function bodies. This is so that even though they must be overridden, they can still provide ‘base class functionality’, e.g. through super.foo() in a derived class. Note that the class is still abstract and cannot be instantiated directly.
UserDefinedAttribute: @ ( ArgumentList ) @ Identifier @ Identifier ( ArgumentListopt ) @ TemplateInstance @ TemplateInstance ( ArgumentListopt )
User-Defined Attributes (UDA) are compile-time expressions that can be attached to a declaration. These attributes can then be queried, extracted, and manipulated at compile time. There is no runtime component to them.
A user-defined attribute looks like:@(3) int a;
@("string", 7) int b; enum Foo; @Foo int c; struct Bar { int x; } @Bar(3) int d;
If there are multiple UDAs in scope for a declaration, they are concatenated:
@(1)
{
@(2) int a; // has UDAs (1, 2)
@("string") int b; // has UDAs (1, "string")
}
UDAs can be extracted into an expression tuple using __traits:
@('c') string s;
pragma(msg, __traits(getAttributes, s)); // prints tuple('c')
If there are no user-defined attributes for the symbol, an empty tuple is returned. The expression tuple can be turned into a manipulatable tuple:
enum EEE = 7; @("hello") struct SSS { } @(3) { @(4) @EEE @SSS int foo; } alias TP = __traits(getAttributes, foo); pragma(msg, TP); // prints tuple(3, 4, 7, (SSS)) pragma(msg, TP[2]); // prints 7
Of course the tuple types can be used to declare things:
TP[3] a; // a is declared as an SSS
The attribute of the type name is not the same as the attribute of the variable:
pragma(msg, __traits(getAttributes, typeof(a))); // prints tuple("hello")
Of course, the real value of UDAs is to be able to create user-defined types with specific values. Having attribute values of basic types does not scale. The attribute tuples can be manipulated like any other tuple, and can be passed as the argument list to a template.
Whether the attributes are values or types is up to the user, and whether later attributes accumulate or override earlier ones is also up to how the user interprets them.
UDAs cannot be attached to template parameters.