object - D Programming Language (original) (raw)

Forms the symbols available to all D programs. Includes Object, which is the root of the class object hierarchy. This module is implicitly imported.

alias noreturn = noreturn;

All D class objects inherit from Object.

Convert Object to a human readable string.

nothrow @trusted size_t toHash();

Compute hash function for Object.

Compare with another Object obj.

Returns:

this < obj	< 0
this == obj	0
this > obj	> 0

Test whether this is equal to o. The default implementation only compares by identity (using the is operator). Generally, overrides and overloads for opEquals should attempt to compare objects by their contents. A class will most likely want to add an overload that takes your specific type as the argument and does the content comparison. Then you can override this and forward it to your specific typed overload with a cast. Remember to check for null on the typed overload.

Examples:

class Child { int contents; bool opEquals(const Child c) const @safe pure nothrow @nogc { if (c is null) return false; return this.contents == c.contents; }

  override bool opEquals(Object o)

{ return this.opEquals(cast(Child) o); } }

static Object factory(string classname);

Create instance of class specified by the fully qualified name classname. The class must either have no constructors or have a default constructor.

Example

module foo.bar;

class C { this() { x = 10; } int x; }

void main() { auto c = cast(C)Object.factory("foo.bar.C"); assert(c !is null && c.x == 10); }

bool opEquals(LHS, RHS)(LHS lhs, RHS rhs)
if ((is(LHS : const(Object)) || is(LHS : shared(const(Object)))) && (is(RHS : const(Object)) || is(RHS : shared(const(Object)))));

Implementation for class opEquals override. Calls the class-defined methods after a null check. Please note this is not nogc right now, even if your implementation is, because of the typeinfo name string compare. This is because of dmd's dll implementation. However, it can infer to @safe if your class' opEquals is.

Examples:

If aliased to the same object or both null => equal

class F { int flag; this(int flag) { this.flag = flag; } }

F f; assert(f == f); f = new F(1); assert(f == f);

Examples:

If either is null => non-equal

class F { int flag; this(int flag) { this.flag = flag; } } F f; assert(!(new F(0) == f)); assert(!(f == new F(0)));

Examples:

If same exact type => one call to method opEquals This test passes @safe because it defines a new opEquals with @safe

class F { int flag;

this(int flag)
{
    this.flag = flag;
}

bool opEquals(const F o) const @safe nothrow pure
{
    return flag == o.flag;
}

}

F f; assert(new F(0) == new F(0)); assert(!(new F(0) == new F(1)));

Examples:

General case => symmetric calls to method opEquals

int fEquals, gEquals;

class Base { int flag; this(int flag) { this.flag = flag; } }

class F : Base { this(int flag) { super(flag); }

bool opEquals(const Base o) @safe
{
    fEquals++;
    return flag == o.flag;
}

}

class G : Base { this(int flag) { super(flag); }

bool opEquals(const Base o) @safe
{
    gEquals++;
    return flag == o.flag;
}

}

assert(new F(1) == new G(1)); assert(fEquals == 1); assert(gEquals == 1);

Examples:

This test shows an example for a comprehensive inheritance equality chain too.

static class Base { int member;

this(int member) pure @safe nothrow @nogc
{
    this.member = member;
}

override bool opEquals(Object rhs) const
{
    return this.opEquals(cast(Base) rhs);
}

bool opEquals(const Base rhs) const @nogc pure nothrow @safe
{
    if (rhs is null)
        return false;
    return this.member == rhs.member;
}

}

bool testThroughBase() nothrow @safe { Base b1 = new Base(0); Base b2 = new Base(0); assert(b1 == b2); Base b3 = new Base(1); assert(b1 != b3); return true; }

static assert(testThroughBase());

bool testThroughObject() { Object o1 = new Base(0); Object o2 = new Base(0); assert(o1 == o2); Object o3 = new Base(1); assert(o1 != o3); return true; }

static assert(testThroughObject());

static class Child : Base { int member2;

this(int member, int member2) pure @safe nothrow @nogc
{
    super(member);
    this.member2 = member2;
}

    override bool opEquals(Object rhs) const
{
    return this.opEquals(cast(Child) rhs);
}
override bool opEquals(const Base rhs) const
{
    return this.opEquals(cast(const Child) rhs);
}
    bool opEquals(const Child rhs) const @nogc pure nothrow @safe
{
    if (rhs is null)
        return false;
                            return Base.opEquals(rhs) && this.member2 == rhs.member2;
}

}

bool testThroughChild() { Child a = new Child(0, 0); Child b = new Child(0, 1); assert(a != b);

Base ba = a;
Base bb = b;
assert(ba != bb);

Object oa = a;
Object ob = b;
assert(oa != ob);

return true;

}

static assert(testThroughChild());

void setSameMutex(shared Object ownee, shared Object owner);

Makes ownee use owner's mutex. This will initialize owner's mutex if it hasn't been set yet.

Parameters:

Object ownee	object to change
Object owner	source object

Information about an interface. When an object is accessed via an interface, an Interface* appears as the first entry in its vtbl.

TypeInfo_Class classinfo;

Class info returned by typeid for this interface (not for containing class)

offset to Interface 'this' from Object 'this'

Array of pairs giving the offset and type information for each member in an aggregate.

Offset of member from start of object

TypeInfo for this member

Runtime type information about a type. Can be retrieved for any type using aTypeidExpression.

nothrow @trusted size_t getHash(scope const void* p) const;

Computes a hash of the instance of a type.

Parameters:

void* p	pointer to start of instance of the type

bool equals(in void* p1, in void* p2) const;

Compares two instances for equality.

int compare(in void* p1, in void* p2) const;

Compares two instances for <, ==, or >.

pure nothrow @nogc @property @safe size_t tsize() const;

Returns size of the type.

void swap(void* p1, void* p2) const;

Swaps two instances of the type.

pure nothrow @nogc @property inout(TypeInfo) next() inout;

Get TypeInfo for 'next' type, as defined by what kind of type this is, null if none.

abstract pure nothrow @nogc @safe const(void)[] initializer() const;

Return default initializer. If the type should be initialized to all zeros, an array with a null ptr and a length equal to the type size will be returned. For static arrays, this returns the default initializer for a single element of the array, use tsize to get the correct size.

pure nothrow @nogc @property @safe uint flags() const;

Get flags for type: 1 means GC should scan for pointers, 2 means arg of this type is passed in SIMD register(s) if available

const(OffsetTypeInfo)[] offTi() const;

Get type information on the contents of the type; null if not available

void destroy(void* p) const;

Run the destructor on the object and all its sub-objects

void postblit(void* p) const;

Run the postblit on the object and all its sub-objects

pure nothrow @nogc @property @safe size_t talign() const;

Return alignment of type

nothrow @safe int argTypes(out TypeInfo arg1, out TypeInfo arg2);

Return internal info on arguments fitting into 8byte. See X86-64 ABI 3.2.3

pure nothrow @nogc @property @trusted immutable(void)* rtInfo() const;

Return info used by the garbage collector to do precise collection.

class TypeInfo_Class: object.TypeInfo;

Runtime type information about a class. Can be retrieved from an object instance by using thetypeid expression.

class static initializer (init.length gives size in bytes of class)

class name

virtual function pointer table

interfaces this class implements

base class

inheritance distance from Object

unique signature for name

static const(TypeInfo_Class) find(scope const char[] classname);

Search all modules for TypeInfo_Class corresponding to classname.

Returns:

null if not found

Create instance of Object represented by 'this'.

final pure nothrow @nogc @trusted bool isBaseOf(scope const TypeInfo_Class child) const;

Returns true if the class described by child derives from or is the class described by this TypeInfo_Class. Always returns false if the argument is null.

Parameters:

TypeInfo_Class child	TypeInfo for some class

Returns:

true if the class described by child derives from or is the class described by this TypeInfo_Class.

An instance of ModuleInfo is generated into the object file for each compiled module.

It provides access to various aspects of the module. It is not generated for betterC.

pure nothrow @nogc @property void function() tlsctor() const;

Returns:

module constructor for thread locals, null if there isn't one

pure nothrow @nogc @property void function() tlsdtor() const;

Returns:

module destructor for thread locals, null if there isn't one

pure nothrow @nogc @property void* xgetMembers() const;

Returns:

address of a module's const(MemberInfo)[] getMembers(string) function, null if there isn't one

pure nothrow @nogc @property void function() ctor() const;

Returns:

module constructor, null if there isn't one

pure nothrow @nogc @property void function() dtor() const;

Returns:

module destructor, null if there isn't one

pure nothrow @nogc @property void function() ictor() const;

Returns:

module order independent constructor, null if there isn't one

pure nothrow @nogc @property void function() unitTest() const;

Returns:

address of function that runs the module's unittests, null if there isn't one

pure nothrow @nogc @property immutable(ModuleInfo*)[] importedModules() const return;

Returns:

array of pointers to the ModuleInfo's of modules imported by this one

pure nothrow @nogc @property TypeInfo_Class[] localClasses() const return;

Returns:

array of TypeInfo_Class references for classes defined in this module

pure nothrow @nogc @property string name() const return;

Returns:

name of module, null if no name

The base class of all thrown objects.

All thrown objects must inherit from Throwable. Class Exception, which derives from this class, represents the category of thrown objects that are safe to catch and handle. In principle, one should not catch Throwable objects that are not derived from Exception, as they represent unrecoverable runtime errors. Certain runtime guarantees may fail to hold when these errors are thrown, making it unsafe to continue execution after catching them.

A message describing the error.

The file name of the D source code corresponding with where the error was thrown from.

The line number of the D source code corresponding with where the error was thrown from.

The stack trace of where the error happened. This is an opaque object that can either be converted to string, or iterated over with foreach to extract the items in the stack trace (as strings).

TraceDeallocator infoDeallocator;

If set, this is used to deallocate the TraceInfo on destruction.

pure nothrow @nogc @property @safe inout(Throwable) next() inout return scope;

Returns:

A reference to the next error in the list. This is used when a newThrowable is thrown from inside a catch block. The originally caught Exception will be chained to the new Throwable via this field.

pure nothrow @nogc @property @safe void next(Throwable tail) scope;

Replace next in chain with tail. Use chainTogether instead if at all possible.

final pure nothrow @nogc ref @system uint refcount() scope return;

Returns:

mutable reference to the reference count, which is 0 - allocated by the GC, 1 - allocated by d_newThrowable(), and >=2 which is the reference count + 1

NoteMarked as @system to discourage casual use of it.

int opApply(scope int delegate(Throwable) dg);

Loop over the chain of Throwables.

static pure nothrow @nogc @system Throwable chainTogether(return scope Throwable e1, return scope Throwable e2);

Append e2 to chain of exceptions that starts with e1.

Parameters:

Throwable e1	start of chain (can be null)
Throwable e2	second part of chain (can be null)

Returns:

Throwable that is at the start of the chain; null if both e1 and e2 are null

Overrides Object.toString and returns the error message. Internally this forwards to the toString overload that takes a sink delegate.

void toString(scope void delegate(in char[]) sink) const;

The Throwable hierarchy uses a toString overload that takes asink delegate to avoid GC allocations, which cannot be performed in certain error situations. Override this toString method to customize the error message.

nothrow @safe const(char)[] message() const;

Get the message describing the error.

This getter is an alternative way to access the Exception's message, with the added advantage of being override-able in subclasses. Subclasses are hence free to do their own memory managements without being tied to the requirement of providing a string in a field.

The default behavior is to return the Throwable.msg field.

Returns:

A message representing the cause of the Throwable

class Exception: object.Throwable;

The base class of all errors that are safe to catch and handle.

In principle, only thrown objects derived from this class are safe to catch inside a catch block. Thrown objects not derived from Exception represent runtime errors that should not be caught, as certain runtime guarantees may not hold, making it unsafe to continue program execution.

Examples:

bool gotCaught; try { throw new Exception("msg"); } catch (Exception e) { gotCaught = true; assert(e.msg == "msg"); } assert(gotCaught);

pure nothrow @nogc @safe this(string msg, string file = __FILE__, size_t line = __LINE__, Throwable nextInChain = null);

Creates a new instance of Exception. The nextInChain parameter is used internally and should always be null when passed by user code. This constructor does not automatically throw the newly-created Exception; the throw expression should be used for that purpose.

class Error: object.Throwable;

The base class of all unrecoverable runtime errors.

This represents the category of Throwable objects that are not safe to catch and handle. In principle, one should not catch Error objects, as they represent unrecoverable runtime errors. Certain runtime guarantees may fail to hold when these errors are thrown, making it unsafe to continue execution after catching them.

Examples:

bool gotCaught; try { throw new Error("msg"); } catch (Error e) { gotCaught = true; assert(e.msg == "msg"); } assert(gotCaught);

pure nothrow @nogc @safe this(string msg, Throwable nextInChain = null);

Creates a new instance of Error. The nextInChain parameter is used internally and should always be null when passed by user code. This constructor does not automatically throw the newly-created Error; the throw statement should be used for that purpose.

Throwable bypassedException;

The first Exception which was bypassed when this Error was thrown, or null if no Exceptions were pending.

@trusted void clear(Value, Key)(Value[Key] aa);

@trusted void clear(Value, Key)(Value[Key]* aa);

Removes all remaining keys and values from an associative array.

Parameters:

Value[Key] aa	The associative array.

Examples:

auto aa = ["k1": 2]; aa.clear; assert("k1" !in aa);

T rehash(T : Value[Key], Value, Key)(T aa);

T rehash(T : Value[Key], Value, Key)(T* aa);

T rehash(T : shared(Value[Key]), Value, Key)(T aa);

T rehash(T : shared(Value[Key]), Value, Key)(T* aa);

Reorganizes the associative array in place so that lookups are more efficient.

Parameters:

T aa	The associative array.

Returns:

The rehashed associative array.

V[K] dup(T : V[K], K, V)(T aa);

V[K] dup(T : V[K], K, V)(T* aa);

Creates a new associative array of the same size and copies the contents of the associative array into it.

Parameters:

T aa	The associative array.

Examples:

auto aa = ["k1": 2]; auto a2 = aa.dup; aa["k2"] = 3; assert("k2" !in a2);

pure nothrow @nogc @safe auto byKey(T : V[K], K, V)(T aa);

pure nothrow @nogc auto byKey(T : V[K], K, V)(T* aa);

Returns a forward range which will iterate over the keys of the associative array. The keys are returned by reference.

If structural changes are made to the array (removing or adding keys), all ranges previously obtained through this function are invalidated. The following example program will dereference a null pointer:

import std.stdio : writeln;

auto dict = ["k1": 1, "k2": 2]; auto keyRange = dict.byKey; dict.clear; writeln(keyRange.front);

Parameters:

T aa	The associative array.

Returns:

A forward range referencing the keys of the associative array.

Examples:

auto dict = [1: "v1", 2: "v2"]; int sum; foreach (v; dict.byKey) sum += v;

assert(sum == 3);

pure nothrow @nogc @safe auto byValue(T : V[K], K, V)(T aa);

pure nothrow @nogc auto byValue(T : V[K], K, V)(T* aa);

Returns a forward range which will iterate over the values of the associative array. The values are returned by reference.

import std.stdio : writeln;

auto dict = ["k1": 1, "k2": 2]; auto valueRange = dict.byValue; dict.clear; writeln(valueRange.front);

Parameters:

T aa	The associative array.

Returns:

A forward range referencing the values of the associative array.

Examples:

auto dict = ["k1": 1, "k2": 2]; int sum; foreach (v; dict.byValue) sum += v;

assert(sum == 3);

foreach (ref v; dict.byValue) v++; assert(dict == ["k1": 2, "k2": 3]);

pure nothrow @nogc @safe auto byKeyValue(T : V[K], K, V)(T aa);

pure nothrow @nogc auto byKeyValue(T : V[K], K, V)(T* aa);

Returns a forward range which will iterate over the key-value pairs of the associative array. The returned pairs are represented by an opaque type with .key and .value properties for accessing references to the key and value of the pair, respectively.

import std.stdio : writeln;

auto dict = ["k1": 1, "k2": 2]; auto kvRange = dict.byKeyValue; dict.clear; writeln(kvRange.front.key, ": ", kvRange.front.value);

Note that this is a low-level interface to iterating over the associative array and is not compatible with theTuple type in Phobos. For compatibility with Tuple, usestd.array.byPair instead.

Parameters:

T aa	The associative array.

Returns:

A forward range referencing the pairs of the associative array.

Examples:

auto dict = ["k1": 1, "k2": 2]; int sum; foreach (e; dict.byKeyValue) { assert(e.key[1] == e.value + '0'); sum += e.value; } assert(sum == 3);

foreach (e; dict.byKeyValue) e.value++; assert(dict == ["k1": 2, "k2": 3]);

@property Key[] keys(T : Value[Key], Value, Key)(T aa);

@property Key[] keys(T : Value[Key], Value, Key)(T* aa);

Returns a newly allocated dynamic array containing a copy of the keys from the associative array.

Parameters:

T aa	The associative array.

Returns:

A dynamic array containing a copy of the keys.

Examples:

auto aa = [1: "v1", 2: "v2"]; int sum; foreach (k; aa.keys) sum += k;

assert(sum == 3);

@property Value[] values(T : Value[Key], Value, Key)(T aa);

@property Value[] values(T : Value[Key], Value, Key)(T* aa);

Returns a newly allocated dynamic array containing a copy of the values from the associative array.

Parameters:

T aa	The associative array.

Returns:

A dynamic array containing a copy of the values.

Examples:

auto aa = ["k1": 1, "k2": 2]; int sum; foreach (e; aa.values) sum += e;

assert(sum == 3);

inout(V) get(K, V)(inout(V[K]) aa, K key, lazy inout(V) defaultValue);

inout(V) get(K, V)(inout(V[K])* aa, K key, lazy inout(V) defaultValue);

If key is in aa, returns corresponding value; otherwise it evaluates and returns defaultValue.

Parameters:

inout(V[K]) aa	The associative array.
K key	The key.
inout(V) defaultValue	The default value.

Examples:

auto aa = ["k1": 1]; assert(aa.get("k1", 0) == 1); assert(aa.get("k2", 0) == 0);

ref V require(K, V)(ref V[K] aa, K key, lazy V value = V.init);

If key is in aa, returns corresponding value; otherwise it evaluatesvalue, adds it to the associative array and returns it.

Parameters:

V[K] aa	The associative array.
K key	The key.
V value	The required value.

Examples:

auto aa = ["k1": 1]; assert(aa.require("k1", 0) == 1); assert(aa.require("k2", 0) == 0); assert(aa["k2"] == 0);

void update(K, V, C, U)(ref V[K] aa, K key, scope C create, scope U update)
if (is(typeof(create()) : V) && (is(typeof(update(aa[K.init])) : V) || is(typeof(update(aa[K.init])) == void)));

Calls create if key doesn't exist in the associative array, otherwise calls update.create returns a corresponding value for key.update accepts a key parameter. If it returns a value, the value is set for key.

Parameters:

V[K] aa	The associative array.
K key	The key.
C create	The callable to create a value for key. Must return V.
U update	The callable to call if key exists. Takes a K argument, returns a V or void.

Examples:

int[string] aa;

aa.update("key", () => 1, (int) {} ); assert(aa["key"] == 1);

aa.update("key", () => 0, (ref int v) { v += 1; }); assert(aa["key"] == 2);

aa.update("key", () => 0, (int v) => v * 2 ); assert(aa["key"] == 4);

aa.update("key", () => 0, (int) { }); assert(aa["key"] == 4);

size_t hashOf(T)(auto ref T arg, size_t seed);

size_t hashOf(T)(auto ref T arg);

Calculates the hash value of arg with an optional seed initial value. The result might not be equal to typeid(T).getHash(&arg).

Parameters:

T arg	argument to calculate the hash value of
size_t seed	optional seed value (may be used for hash chaining)

Returncalculated hash value of arg

Examples:

class MyObject { size_t myMegaHash() const @safe pure nothrow { return 42; } } struct Test { int a; string b; MyObject c; size_t toHash() const pure nothrow { size_t hash = a.hashOf(); hash = b.hashOf(hash); size_t h1 = c.myMegaHash(); hash = h1.hashOf(hash); return hash; } }

immutable size_t[pointerBitmap.length] RTInfoImpl(size_t[] pointerBitmap);

Create RTInfo for type T

enum immutable(void)* rtinfoNoPointers;

shortcuts for the precise GC, also generated by the compiler used instead of the actual pointer bitmap

@property auto dup(T)(T[] a)
if (!is(const(T) : T));

@property T[] dup(T)(const(T)[] a)
if (is(const(T) : T));

Provide the .dup array property.

Examples:

auto arr = [1, 2]; auto arr2 = arr.dup; arr[0] = 0; assert(arr == [0, 2]); assert(arr2 == [1, 2]);

@property immutable(T)[] idup(T)(T[] a);

@property immutable(T)[] idup(T : void)(const(T)[] a);

Provide the .idup array property.

Examples:

char[] arr = ['a', 'b', 'c']; string s = arr.idup; arr[0] = '.'; assert(s == "abc");

pure nothrow @property @trusted size_t capacity(T)(T[] arr);

(Property) Gets the current capacity of a slice. The capacity is the size that the slice can grow to before the underlying array must be reallocated or extended.

If an append must reallocate a slice with no possibility of extension, then0 is returned. This happens when the slice references a static array, or if another slice references elements past the end of the current slice.

NoteThe capacity of a slice may be impacted by operations on other slices.

Examples:

int[4] sarray = [1, 2, 3, 4]; int[] slice = sarray[]; assert(sarray.capacity == 0); slice ~= 5; assert(slice.capacity >= 5);

int[] a = [1, 2, 3, 4]; int[] b = a[1 .. $]; int[] c = a[1 .. $ - 1]; debug(SENTINEL) {} else { assert(a.capacity != 0); assert(a.capacity == b.capacity + 1); } assert(c.capacity == 0);

pure nothrow @trusted size_t reserve(T)(ref T[] arr, size_t newcapacity);

Reserves capacity for a slice. The capacity is the size that the slice can grow to before the underlying array must be reallocated or extended.

Returns:

The new capacity of the array (which may be larger than the requested capacity).

Examples:

int[4] sarray = [1, 2, 3, 4]; int[] slice = sarray[]; auto u = slice.reserve(8); assert(u >= 8); assert(&sarray[0] !is &slice[0]); assert(slice.capacity == u);

int[] a = [1, 2, 3, 4]; a.reserve(8); auto p = &a[0]; u = a.capacity; a ~= [5, 6, 7, 8]; assert(p == &a[0]); assert(u == a.capacity);

nothrow ref @system inout(T[]) assumeSafeAppend(T)(auto ref inout(T[]) arr);

Assume that it is safe to append to this array. Appends made to this array after calling this function may append in place, even if the array was a slice of a larger array to begin with.

Use this only when it is certain there are no elements in use beyond the array in the memory block. If there are, those elements will be overwritten by appending to this array.

WarningCalling this function, and then using references to data located after the given array results in undefined behavior.

Returns:

The input is returned.

Examples:

int[] a = [1, 2, 3, 4];

int[] b = a [0 .. 3]; b ~= 5; assert(a.ptr != b.ptr);

debug(SENTINEL) {} else { int[] c = a [0 .. 3]; c.assumeSafeAppend() ~= 5; assert(a.ptr == c.ptr); }

void destroy(bool initialize = true, T)(ref T obj)
if (is(T == struct));

void destroy(bool initialize = true, T)(T obj)
if (is(T == class));

void destroy(bool initialize = true, T)(T obj)
if (is(T == interface));

void destroy(bool initialize = true, T)(ref T obj)
if (__traits(isStaticArray, T));

void destroy(bool initialize = true, T)(ref T obj)
if (!is(T == struct) && !is(T == interface) && !is(T == class) && !__traits(isStaticArray, T));

Destroys the given object and optionally resets to initial state. It's used to destroy an object, calling its destructor or finalizer so it no longer references any other objects. It does not initiate a GC cycle or free any GC memory. If initialize is supplied false, the object is considered invalid after destruction, and should not be referenced.

Examples:

Reference type demonstration

class C { struct Agg { static int dtorCount;

    int x = 10;
    ~this() { dtorCount++; }
}

static int dtorCount;

string s = "S";
Agg a;
~this() { dtorCount++; }

}

C c = new C(); assert(c.dtorCount == 0); assert(c.s == "S"); assert(c.a.dtorCount == 0); assert(c.a.x == 10); c.s = "T"; c.a.x = 30; assert(c.s == "T"); destroy(c); assert(c.dtorCount == 1); assert(c.s == "S"); assert(c.a.dtorCount == 1); assert(c.a.x == 10);

Examples:

C++ classes work too

extern (C++) class CPP { struct Agg { __gshared int dtorCount;

    int x = 10;
    ~this() { dtorCount++; }
}

__gshared int dtorCount;

string s = "S";
Agg a;
~this() { dtorCount++; }

}

CPP cpp = new CPP(); assert(cpp.dtorCount == 0); assert(cpp.s == "S"); assert(cpp.a.dtorCount == 0); assert(cpp.a.x == 10); cpp.s = "T"; cpp.a.x = 30; assert(cpp.s == "T"); destroy!false(cpp); assert(cpp.dtorCount == 1); assert(cpp.s == "T"); assert(cpp.a.dtorCount == 1); assert(cpp.a.x == 30); destroy(cpp); assert(cpp.dtorCount == 2); assert(cpp.s == "S"); assert(cpp.a.dtorCount == 2); assert(cpp.a.x == 10);

Examples:

Value type demonstration

int i; assert(i == 0); i = 1; assert(i == 1); destroy!false(i); assert(i == 1); destroy(i); assert(i == 0);

Examples:

Nested struct type

int dtorCount; struct A { int i; ~this() { dtorCount++; } } A a = A(5); destroy!false(a); assert(dtorCount == 1); assert(a.i == 5);

destroy(a); assert(dtorCount == 2); assert(a.i == 0);

import core.lifetime : emplace; emplace(&a, A(0));

template imported(string moduleName)

Provides an "inline import", i.e. an import that is only available for a limited lookup. For example:

void fun(imported!"std.stdio".File input) { ... use File from std.stdio normally ... }

There is no need to import std.stdio at top level, so fun carries its own dependencies. The same approach can be used for template constraints:

void fun(T)(imported!"std.stdio".File input, T value) if (imported!"std.traits".isIntegral!T) { ... }

An inline import may be used in conjunction with the with statement as well. Inside the scope controlled by with, all symbols in the imported module are made available:

void fun() { with (imported!"std.datetime") with (imported!"std.stdio") { Clock.currTime.writeln; } }

The advantages of inline imports over top-level uses of the import declaration are the following:

The imported template specifies dependencies at declaration level, not at module level. This allows reasoning about the dependency cost of declarations in separation instead of aggregated at module level.
Declarations using imported are easier to move around because they don't require top-level context, making for simpler and quicker refactorings.
Declarations using imported scale better with templates. This is because templates that are not instantiated do not have their parameters and constraints instantiated, so additional modules are not imported without necessity. This makes the cost of unused templates negligible. Dependencies are pulled on a need basis depending on the declarations used by client code.

The use of imported also has drawbacks:

If most declarations in a module need the same imports, then factoring them at top level, outside the declarations, is simpler than repeating them.
Traditional dependency-tracking tools such as make and other build systems assume file-level dependencies and need special tooling (such as rdmd) in order to work efficiently.
Dependencies at the top of a module are easier to inspect quickly than dependencies spread throughout the module.