The D Programming Language: A new take on C++

Matthew Soucy ()

What is D?

"Great, the last thing I need is another D in programming!" - Walter Bright

D Logo

D Logo

Six instantly-useful features

D derives much of its syntax from C

import std.stdio;

// If main is void, it's the same as if it returns 0
void main() {
    // The developers understand the awesomeness of printf
    writef("Hello, world!\n");

The "auto" keyword

Because expressions, especially those returned from some functions, can have long or complex names, the auto keyword can be used to create variables where the exact type isn't needed or important.

import std.algorithm;
import std.stdio;

void main() {
    auto x = "Hello, world!"; // string
    auto data = [1,2,3,4,5]; // int[]
    auto somefunc = (int x) {return x*x;}; // int function(int)
    // UHHHHH...don't worry about it! (It's int[])
    auto y =!somefunc().filter!q{a % 2}().array();
    writef("%s", y); // %s prints this nicely!

Variable declaration syntax

C is infamous for having an unintuitive syntax for complex function declaration.

// C style:
int *x, *y;
    // * is required for both
int* x, y;
    // x is an int*, y is an int
int *p;
    // p is a pointer to an int
int *p[13];
    // p is an array[13] of pointers to an int
int *(p[13]);
    // p is an array[13] of pointers to an int
int **p;
    // p is a pointer to a pointer to an int
int (*p)[13];
    // p is a pointer to an array[13] of int
int *f();
    // f is a function returning pointer to int
int (*f)();
    // f is a pointer to a function returning int
int (*(*f())[13])();
    // f is a function returning a ptr to an array[13]
    // of pointers to functions to int
int (*(*x[3])())[5];
    // x is an array[3] of pointers to functions
    // returning pointers to array[5] to int

Examples from K&R Sec 5.12

Variable declaration syntax (D style)

Isn't it nicer to just do it the D way?

// D style:
int* x, y;
    // D forbids declaring variables of different types in the same expression
int *x;
int y;
    // x is an int*, y is an int
int* p;
    // p is a pointer to an int
int*[13] p;
    // p is an array[13] of pointers to an int
int **p;
    // p is a pointer to a pointer to an int
int[13]* p;
    // p is a pointer to an array[13] of int
int* function() f;
    // f is a function returning pointer to int
int function() f;
    // f is a pointer to a function returning int
int function()[13]* function() f;
    // f is a function returning a ptr to an array[13]
    // of pointers to functions to int
int[5]* function()[3] x;
    // x is an array[3] of pointers to functions
    // returning pointers to array[5] to int

Examples from K&R Sec 5.12


Properties are an extension of accessors, where a "variable" lookup is converted to a function call.

@property string twice(string x) {
    return x~x;

void main() {

Smart arrays

Arrays in D can behave more like C++'s std::vector

C-style arrays are also supported:

int* p; // Same as in C
int[5] x; // Same as C's "int x[5];"
int[] y; // Smart array

y.length = 2; // Resizes
writef("Length: %s", y.length); // Gets value - property

Built-in associative arrays

D has a built-in associative array type as part of the syntax:

import std.stdio;

void main() {
    string[string] x;
    x["what"] = "Hello";
    x["who"] = "world";
    x["emotion"] = "!";
    writefln("%s %s%s",x["what"],x["who"],x["emotion"]);

Foreach loops

Foreach loops in D are kind of awesome:

import std.stdio;

void main() {
    int[] arr = [1,3,5,7,9];
    foreach(x;arr) {
        writef("%s ",x); // %s automagically deduces the type
    foreach_reverse(x;arr) {
        writef("%s ",x);
    foreach(ref x;arr) {
        x *= x; // Poor man's map()
    foreach(x;arr) {
        writef("%s ",x);


1 3 5 7 9 
9 7 5 3 1 
1 9 25 49 81

Scripting in D

D is designed so that you can even use it in scripts "without compiling"!

Example script:

#!/usr/bin/dmd -run
// If you have multiple files, you can use /usr/bin/rdmd
import std.stdio;
import std.string;
import std.random;

void main(){
    // Simple raffle name reorderer
    char[] data;
    writeln("Enter names:");
    string[] names;
    do {
        write("==> ");
        data = data.strip();
        if(data != "") names ~= data.idup;
    } while(data != "");
    writeln("--- OUTPUT ---");
    uint i=0;
    foreach(name;names.randomCover(Random(unpredictableSeed))) {
        writef("%s: `%s`\n", i, name);

Switch statements

D continues C's style of a switch statement, but with a few tricks:

string s;
// ... Do stuff, assign s
// switch now works directly on strings!
switch(s) {
case "A":
    // Do stuff
case "B","C","E":
    // Do stuff if it matches any of those
    // Same old, same old
switch(s[0]) {
case 'a' : .. case 'z': // Matches anything between a-z


Templates, combined with compile-time duck typing, allow for some powerful and simple type manipulation.

Their syntax is significantly cleaner than in C++, due to using a single binary operator ! instead of overloading <>:

template Foo(T, U) {
    class Bar { ... }
    T foo(T t, U u) { ... }
    T abc;
    typedef T* Footype; // any declarations can be templated

class Baz(T) {
    T t;

Foo!(int,char).Bar b;
Foo!(int,char).abc = 3;
Baz!int baz; // If there's only one argument it doesn't need parenthesis

Eponymous Templates

When you make an eponymous template, the template takes the value of a static variable with the same name inside it.

template factorial(int n) {
    static if (n == 1) const factorial = 1;
    else const factorial = n * factorial!(n-1);

int x = factorial!5; // Same as "int x = 120;"


The factorial example is simple, but in a lot of cases you'd prefer to be able to do a calculation both at compile and run time. For this reason, D allows compile-time function execution. Any variable what is created with enum or static is determined at compile-time.

int factorial(int n) {
    if (n == 1) return 1;
    else return n * factorial(n - 1);

int x = factorial(5); // x is initialized to 120 at runtime
static int x = factorial(5); // x is statically initialized to 120


D's standard library uses the concept of Ranges instead of Iterators like C++.

Iterators have some conceptual issues:

Iterators Ranges

A range is a single interface that allows for an alternative method of iteration.

// Taken verbatim from std.range
template isInputRange(R)
    enum bool isInputRange = is(typeof(
    (inout int _dummy=0)
        R r = void;       // can define a range object
        if (r.empty) {}   // can test for empty
        r.popFront();     // can invoke popFront()
        auto h = r.front; // can get the front of the range


More importantly, it simplifies iterating over things that may not be a "real" range.

// Reverses iteration over a bidirectional range
struct Reversed
    int[] range;

    this(int[] range) { this.range = range; }

    @property bool empty() const { return range.empty; }

    @property int front() const { return range.back; } // <- reverse

    @property int back() const { return range.front; } // <- reverse

    void popFront() { range.popBack(); } // <- reverse

    void popBack() {range.popFront(); } // <- reverse

Standard Library

The standard library (Phobos) takes a "batteries included" approach like python

Phobos contains many different modules, such as:

Much like C++'s STL and iterators, Phobos is designed almost entirely around ranges.

Because of this, any user-made range is instantly supported by the entire standard library, including a large number of algorithms for acting on those ranges (std.algorithm)

Uniform Function Call Syntax

There are often instances when one is using a library that for some reason they cannot alter, but you want to add functionality to a certain class or type.

// D allows this through the use of UFCS, which converts any call to:
// Into the actual call:
// This is used often in `std.algorithm`, which allows one to write code such as:
// As the much more legible (and easily maintainable):

Voldemort Types

Andrei discovered an interesting interaction between auto and declaring structs inside a function: Voldemort types are types that cannot be named.

import std.stdio;
import std.range;

auto generator(uint seed) { 
  struct RandomNumberGenerator {
      @property int front() {
          return ((seed / 0x10000) * seed) >> 16;
      void popFront() {
          seed = seed * 1103515245 + 12345;
      enum empty = false;
  RandomNumberGenerator g;
  g.popFront();    // get it going
  return g;

void main() {


// Taken from my ZeroMQ wrapper
mixin template SocketOption(TYPE, string NAME, int VALUE) {
    /// Setter
    @property void SocketOption(TYPE value) {
        if(zmq_setsockopt(this.socket, VALUE, &value, TYPE.sizeof)) {
            throw new ZMQError();
    /// Getter
    @property TYPE SocketOption() {
        TYPE ret;
        size_t size = TYPE.sizeof;
        if(zmq_getsockopt(this.socket, VALUE, &ret, &size)) {
            throw new ZMQError();
        } else {
            return ret;
    mixin("alias SocketOption "~NAME~";");
mixin SocketOption!(ulong, "hwm", ZMQ_HWM);

C Bindings

In addition to Phobos, D also has the external library Deimos.

Deimos is a set of D bindings to various C libraries, such as:

Much like Phobos, Deimos is supported by the open source community, and is constantly updating.

This means you can link (almost) any C library to D, and it will work normally.

extern(C) int someCrazyCFunction();

void main() {
    return someCrazyCFunction();



foreach(question;audience) presenter.answer(question);

return 0;