--- language: D filename: learnd.d contributors: - ["Nick Papanastasiou", "www.nickpapanastasiou.github.io"] lang: en --- If you're like me and spend way too much time on the internet, odds are you've heard about [D](http://dlang.org/). The D programming language is a modern, general-purpose, multi-paradigm language with fantastic support for OOP, functional programming, metaprogramming, and easy concurrency and parallelism, and runs the gamut from low-level features such as memory management, inline assembly, and pointer arithmetic, to high-level constructs such as higher-order functions and generic structures and functions via templates, all with a pleasant syntax, and blazing fast performance! D is actively developed by Walter Bright and Andrei Alexandrescu, two super smart, really cool dudes. With all that out of the way, let's look at some examples! ```d // You know what's coming... module hello; import std.stdio; // args is optional void main(string[] args) { writeln("Hello, World!"); } // Conditionals and loops work as expected. import std.stdio; void main() { for(int i = 0; i < 5; i++) { writeln(i); } auto n = 1; // use auto for type inferred variables while(n < 10_000) { n += n; } do { n -= (n / 2); } while(n > 0); // For and while are nice, but in D-land we prefer foreach // The .. creates a continuous range, excluding the end foreach(i; 1..1000000) { if(n % 2 == 0) writeln(i); } foreach_reverse(i; 1..int.max) { if(n % 2 == 1) { writeln(i); } else { writeln("No!"); } } } ``` We can define new types with `struct`, `class`, `union`, and `enum`. Structs and unions are passed to functions by value (i.e. copied) and classes are passed by reference. Futhermore, we can use templates to parameterize all of these on both types and values! ```d // Here, T is a type parameter. Think from C++/C#/Java struct LinkedList(T) { T data = null; LinkedList!(T)* next; // The ! is used to instaniate a parameterized type. Again, think } class BinTree(T) { T data = null; // If there is only one template parameter, we can omit parens BinTree!T left; BinTree!T right; } enum Day { Sunday, Monday, Tuesday, Wednesday, Thursday, Friday, Saturday, } // Use alias to create abbreviations for types alias IntList = LinkedList!int; alias NumTree = BinTree!double; // We can create function templates as well! T max(T)(T a, T b) { if(a < b) return b; return a; } // Use the ref keyword to ensure pass by referece. // That is, even if a and b are value types, they // will always be passed by reference to swap void swap(T)(ref T a, ref T b) { auto temp = a; a = b; b = temp; } // With templates, we can also parameterize on values, not just types class Matrix(uint m, uint n, T = int) { T[m] rows; T[n] columns; } auto mat = new Matrix!(3, 3); // We've defaulted T to int ``` Speaking of classes, let's talk about properties for a second. A property is roughly a function that may act like an lvalue, so we can have the syntax of POD structures (`structure.x = 7`) with the semantics of getter and setter methods (`object.setX(7)`)! ```d // Consider a class parameterized on a types T, U class MyClass(T, U) { T _data; U _other; } // And "getter" and "setter" methods like so class MyClass(T, U) { T _data; U _other; // Constructors are always named `this` this(T t, U u) { data = t; other = u; } // getters @property T data() { return _data; } @property U other() { return _other; } // setters @property void data(T t) { _data = t; } @property void other(U u) { _other = u; } } // And we use them in this manner void main() { auto mc = MyClass!(int, string); mc.data = 7; mc.other = "seven"; writeln(mc.data); writeln(mc.other); } ``` With properties, we can add any amount of validation to our getter and setter methods, and keep the clean syntax of accessing members directly! Other object-oriented goodies at our disposal include `interface`s, `abstract class`es, and `override`ing methods. We've seen D's OOP facilities, but let's switch gears. D offers functional programming with first-class functions, `pure` functions, and immutable data. In addition, all of your favorite functional algorithms (map, filter, reduce and friends) can be found in the wonderful `std.algorithm` module! ```d import std.algorithm : map, filter, reduce; import std.range : iota; // builds an end-exclusive range void main() { // We want to print the sum of a list of squares of even ints // from 1 to 100. Easy! // Just pass lambda expressions as template parameters! auto num = iota(1, 101).filter!(x => x % 2 == 0) .map!(y => y ^^ 2) .reduce!((a, b) => a + b); writeln(num); } ``` Notice how we got to build a nice Haskellian pipeline to compute num? That's thanks to a D innovation know as Uniform Function Call Syntax. With UFCS, we can choose whether to write a function call as a method or free function all. In general, if we have a function ```d f(A, B, C, ...) ``` Then we may write ```d A.f(B, C, ...) ``` and the two are equivalent! No more fiddling to remember if it's str.length or length(str)!