Compound data type

Compound data type
You are encouraged to solve this task according to the task description, using any language you may know.

Data Structure
This illustrates a data structure, a means of storing data within a program.

You may see other such structures in the Data Structures category.

Task

Create a compound data type:

  Point(x,y) 

A compound data type is one that holds multiple independent values.

Related task

•   Enumeration

See also

• Array
• Associative array: Creation, Iteration
• Collections
• Compound data type
• Doubly-linked list: Definition, Element definition, Element insertion, List Traversal, Element Removal
• Linked list
• Queue: Definition, Usage
• Set
• Singly-linked list: Element definition, Element insertion, List Traversal, Element Removal
• Stack

11l

T Point
   Int x, y

   F (x, y)
      .x = x
      .y = y

ACL2

(defstructure point
   (x (:assert (rationalp x)))
   (y (:assert (rationalp y))))

(assign p1 (make-point :x 1 :y 2))
(point-x (@ p1)) ; Access the x value of the point
(assign p1 (update-point (@ p1) :x 3)) ; Update the x value
(point-x (@ p1))
(point-p (@ p1)) ; Recognizer for points

Output:

((:X . 1) (:Y . 2))
1
((:X . 3) (:Y . 2))
3
T

6502 Assembly

The method below is a bit unusual compared to C, where each member of a struct is stored consecutively. The addressing modes of 6502 make it much more efficient to store each member of many different structs consecutively. In other words, the index used to offset into point_x represents which instance of the data type the CPU is accessing. Of course, this assumes that you're willing to declare in advance how many active instances of that data type you'll ever have at once, a very frequent practice in the 8-bit assembly world but is absolutely ludicrous in high-level languages.

NESASM3 syntax:

MAX_POINT_OBJECTS = 64               ; define a constant

    .rsset $0400                     ; reserve memory storage starting at address $0400
point_x .rs MAX_POINT_OBJECTS        ; reserve 64 bytes for x-coordinates
point_y .rs MAX_POINT_OBJECTS        ; reserve 64 bytes for y-coordinates

VASM syntax:

MAX_POINT_OBJECTS equ 64

point_ram equ $0400                       
point_x   equ point_ram
point_y   equ point_ram+MAX_POINT_OBJECTS

So, for example, let's say we want to load our third (zero-indexed) point variable and copy it to zero page RAM addresses $00 and $01. We would do the following:

MAX_POINT_OBJECTS equ 64

point_ram equ $0400                       
point_x   equ point_ram
point_y   equ point_ram+MAX_POINT_OBJECTS

LDX #3
LDA point_x,x
STA $00           
LDA point_y,x
STA $01

Action!

Library: Action! Tool Kit

INCLUDE "D2:REAL.ACT" ;from the Action! Tool Kit

DEFINE REALPTR="CARD"
TYPE PointI=[INT x,y]
TYPE PointR=[REALPTR rx,ry]

PROC Main()
  PointI p1
  PointR p2
  REAL realx,realy

  Put(125) PutE() ;clear screen

  p1.x=123
  p1.y=4567

  ValR("12.34",realx)
  ValR("5.6789",realy)
  p2.rx=realx
  p2.ry=realy

  PrintF("Integer point p1=(%I,%I)%E",p1.x,p1.y)

  Print("Real point p2=(")
  PrintR(p2.rx) Print(",")
  PrintR(p2.ry) Print(")")
RETURN

Output:

Screenshot from Atari 8-bit computer

Integer point p1=(123,4567)
Real point p2=(12.34,5.6789)

ActionScript

package
{
    public class Point
    {
        public var x:Number;
        public var y:Number;
        
        public function Point(x:Number, y:Number)
        {
            this.x = x;
            this.y = y;
        }
    }
}

Ada

Tagged Type

Ada tagged types are extensible through inheritance. The reserved word tagged causes the compiler to create a tag for the type. The tag identifies the position of the type in an inheritance hierarchy.

type Point is tagged record
   X : Integer := 0;
   Y : Integer := 0;
end record;

Record Type

Ada record types are not extensible through inheritance. Without the reserved word tagged the record does not belong to an inheritance hierarchy.

type Point is record
   X : Integer := 0;
   Y : Integer := 0;
end record;

Parameterized Types

An Ada record type can contain a discriminant. The discriminant is used to choose between internal structural representations. Parameterized types were introduced to Ada before tagged types. Inheritance is generally a cleaner solution to multiple representations than is a parameterized type.

type Person (Gender : Gender_Type) is record
   Name   : Name_String;
   Age    : Natural;
   Weight : Float;
   Case Gender is
      when Male =>
         Beard_Length : Float;
      when Female =>
         null;
   end case;
end record;

In this case every person will have the attributes of gender, name, age, and weight. A person with a male gender will also have a beard length.

ALGOL 68

Tagged Type

ALGOL 68 has only tagged-union/discriminants. And the tagging was strictly done by the type (MODE) of the members.

MODE UNIONX = UNION(
   STRUCT(REAL r, INT i),
   INT,
   REAL,
   STRUCT(INT ii),
   STRUCT(REAL rr),
   STRUCT([]REAL r)
);

To extract the apropriate member of a UNION a conformity-clause has to be used.

UNIONX data := 6.6;
CASE data IN
   (INT i): printf(($"r: "gl$,i)),
   (REAL r): printf(($"r: "gl$,r)),
   (STRUCT(REAL r, INT i) s): printf(($"r&i: "2(g)l$,s)),
   (STRUCT([]REAL r) s): printf(($"r: "n(UPB r OF s)(g)l$,s))
OUT
  printf($"Other cases"l$)
ESAC;

The conformity-clause does mean that ALGOL 68 avoids the need for duck typing, but it also makes the tagged-union kinda tough to use, except maybe in certain special cases.

Record Type

ALGOL 68 record types are not extensible through inheritance but they may be part of a larger STRUCT composition.

MODE POINT = STRUCT(
   INT x,
   INT y
);

Parameterized Types

An ALGOL 68 record type can contain a tagged-union/discriminant. The tagged-union/discriminant is used to choose between internal structural representations.

MODE PERSON = STRUCT(
   STRING name,
   REAL age,
   REAL weight,
   UNION (
      STRUCT (REAL beard length),
      VOID
   ) gender details
);

In this case every PERSON will have the attributes of gender details, name, age, and weight. A PERSON may or may not have a beard. The sex is implied by the tagging.

ALGOL W

begin
    % create the compound data type %
    record Point( real x, y );
    % declare a Point variable %
    reference(Point) p;
    % assign a value to p %
    p := Point( 1, 0.5 );
    % access the fields of p - note Algol W uses x(p) where many languages would use p.x %
    write( x(p), y(p) )
end.

AmigaE

OBJECT point
  x, y
ENDOBJECT

PROC main()
  DEF pt:PTR TO point,

  NEW pt
  -> Floats are also stored as integer types making 
  -> the float conversion operator necessary.
  pt.x := !10.4
  pt.y := !3.14
  END pt
ENDPROC

ARM Assembly

Works with: as version Raspberry Pi

/* ARM assembly Raspberry PI  */
/*  program structure.s   */
 
/************************************/
/* Constantes                       */
/************************************/
.equ STDOUT, 1     @ Linux output console
.equ EXIT,   1     @ Linux syscall
.equ WRITE,  4     @ Linux syscall

/*******************************************/
/* Structures                             */
/********************************************/
    .struct  0
point_x:                                        @ x coordinate
    .struct  point_x + 4 
point_y:                                        @ y coordinate
    .struct  point_y + 4
point_end:                                      @ end structure point
/*********************************/
/* Initialized data              */
/*********************************/
.data
sMessResult:        .ascii "value x : "
sMessValeur:        .fill 11, 1, ' '            @ size => 11
szCarriageReturn:   .asciz "\n"


/*********************************/
/* UnInitialized data            */
/*********************************/
.bss  
stPoint:           .skip point_end               @ reservation place in memory
/*********************************/
/*  code section                 */
/*********************************/
.text
.global main 
main:                                             @ entry of program 
    ldr r1,iAdrstPoint
    mov r0,#5                                     @ x value 
    str r0,[r1,#point_x]
    mov r0,#10                                    @ y value
    str r0,[r1,#point_y]
                                                  @ display value
    ldr r2,iAdrstPoint
    ldr r0,[r2,#point_x]
    ldr r1,iAdrsMessValeur
    bl conversion10                               @ call conversion decimal
    ldr r0,iAdrsMessResult
    bl affichageMess                              @ display message


100:                                              @ standard end of the program 
    mov r0, #0                                    @ return code
    mov r7, #EXIT                                 @ request to exit program
    svc #0                                        @ perform the system call
 
iAdrsMessValeur:          .int sMessValeur
iAdrszCarriageReturn:     .int szCarriageReturn
iAdrsMessResult:          .int sMessResult
iAdrstPoint:              .int stPoint

/******************************************************************/
/*     display text with size calculation                         */ 
/******************************************************************/
/* r0 contains the address of the message */
affichageMess:
    push {r0,r1,r2,r7,lr}                          @ save  registres
    mov r2,#0                                      @ counter length 
1:                                                 @ loop length calculation 
    ldrb r1,[r0,r2]                                @ read octet start position + index 
    cmp r1,#0                                      @ if 0 its over 
    addne r2,r2,#1                                 @ else add 1 in the length 
    bne 1b                                         @ and loop 
                                                   @ so here r2 contains the length of the message 
    mov r1,r0                                      @ address message in r1 
    mov r0,#STDOUT                                 @ code to write to the standard output Linux 
    mov r7, #WRITE                                 @ code call system "write" 
    svc #0                                         @ call systeme 
    pop {r0,r1,r2,r7,lr}                           @ restaur des  2 registres */ 
    bx lr                                          @ return  
/******************************************************************/
/*     Converting a register to a decimal unsigned                */ 
/******************************************************************/
/* r0 contains value and r1 address area   */
/* r0 return size of result (no zero final in area) */
/* area size => 11 bytes          */
.equ LGZONECAL,   10
conversion10:
    push {r1-r4,lr}                                 @ save registers 
    mov r3,r1
    mov r2,#LGZONECAL
 
1:                                                  @ start loop
    bl divisionpar10U                               @ unsigned  r0 <- dividende. quotient ->r0 reste -> r1
    add r1,#48                                      @ digit
    strb r1,[r3,r2]                                 @ store digit on area
    cmp r0,#0                                       @ stop if quotient = 0 
    subne r2,#1                                     @ else previous position
    bne 1b                                          @ and loop
                                                    @ and move digit from left of area
    mov r4,#0
2:
    ldrb r1,[r3,r2]
    strb r1,[r3,r4]
    add r2,#1
    add r4,#1
    cmp r2,#LGZONECAL
    ble 2b
                                                      @ and move spaces in end on area
    mov r0,r4                                         @ result length 
    mov r1,#' '                                       @ space
3:
    strb r1,[r3,r4]                                   @ store space in area
    add r4,#1                                         @ next position
    cmp r4,#LGZONECAL
    ble 3b                                            @ loop if r4 <= area size
 
100:
    pop {r1-r4,lr}                                    @ restaur registres 
    bx lr                                             @return
 
/***************************************************/
/*   division par 10   unsigned                    */
/***************************************************/
/* r0 dividende   */
/* r0 quotient    */
/* r1 remainder   */
divisionpar10U:
    push {r2,r3,r4, lr}
    mov r4,r0                                          @ save value
    ldr r3,iMagicNumber                                @ r3 <- magic_number    raspberry 1 2
    umull r1, r2, r3, r0                               @ r1<- Lower32Bits(r1*r0) r2<- Upper32Bits(r1*r0) 
    mov r0, r2, LSR #3                                 @ r2 <- r2 >> shift 3
    add r2,r0,r0, lsl #2                               @ r2 <- r0 * 5 
    sub r1,r4,r2, lsl #1                               @ r1 <- r4 - (r2 * 2)  = r4 - (r0 * 10)
    pop {r2,r3,r4,lr}
    bx lr                                              @ leave function 
iMagicNumber:  	.int 0xCCCCCCCD

Arturo

point: #[
	x: 10
	y: 20
]

print point

Output:

[x:10 y:20]

ATS

There are numerous ways to do this. The simplest is to use an "unboxed" tuple type:

typedef point (t : t@ype+) = @(t, t)
val p : point double = (1.0, 3.0)

If one insists both that the type be unique (as opposed to an alias for a tuple) and that the notation to create a point be Point (x, y), then the following works:

datatype point (t : t@ype+) =
| Point of (t, t)
val p : point double = Point (1.0, 3.0)

AutoHotkey

Works with: AutoHotkey_L

monkeypatched example.

point := Object()
point.x := 1  
point.y := 0

AWK

As usual, arrays are the only data type more complex than a number or a string.
Use quotes around constant strings as element selectors:

BEGIN {
	p["x"]=10
	p["y"]=42

	z = "ZZ"
	p[ z ]=999

	p[ 4 ]=5
	
	for (i in p) print( i, ":", p[i] )
}

Output:

4 : 5
x : 10
y : 42
ZZ : 999

Axe

Axe does not have language support for custom data structures. However, they can be implemented from scratch using memory directly.

Lbl POINT
r₂→{r₁}ʳ
r₃→{r₁+2}ʳ
r₁
Return

To initialize a POINT at memory address L₁ with (x, y) = (5, 10):

POINT(L₁,5,10)

The caller must ensure the buffer has enough free space to contain the object (in this case, 4 bytes).

BASIC

Works with: QBasic

Works with: PowerBASIC

TYPE Point
   x AS INTEGER
   y AS INTEGER
 END TYPE

BBC BASIC

Works with: BBC BASIC for Windows

      DIM Point{x%, y%}

Bracmat

Normally, values are compounded by putting them in a tree structure. For examples, the values 3 and 4 can be put in a small tree (3.4). But since the task requires the values to be independent, the values must be changeable, which they are not in (3.4). So we go object oriented and create a 'type' Point. We show that x and y are independent by changing the value of x and checking that y didn't change. Bracmat does not have other typing systems than duck typing. The variable Point is not a class, but an object in its own right. The new$ function creates a copy of Point.

( ( Point
  =   (x=)
      (y=)
      (new=.!arg:(?(its.x).?(its.y)))
  )
& new$(Point,(3.4)):?pt
& out$(!(pt..x) !(pt..y))
  { Show independcy by changing x, but not y }
& 7:?(pt..x)
& out$(!(pt..x) !(pt..y))
);

Output:

3 4
7 4

Brlcad

In brlcad, the datatypes are geometric primitives or combinations. Here we create a lamp using a combination of previously created components:

c lamp base stem bulb shade chord plug

C

typedef struct Point
{
  int x;
  int y;
} Point;

C#

struct Point
{
  public int x, y;
  public Point(int x, int y) {
    this.x = x;
    this.y = y;
  }
}

C++

struct Point
{
  int x;
  int y;
};

It is also possible to add a constructor (this allows the use of Point(x, y) in expressions):

struct Point
{
  int x;
  int y;
  Point(int ax, int ay): x(ax), y(ax) {}
};

Point can also be parametrized on the coordinate type:

template<typename Coordinate> struct point
{
  Coordinate x, y;
};

// A point with integer coordinates
Point<int> point1 = { 3, 5 };

// a point with floating point coordinates
Point<float> point2 = { 1.7, 3.6 };

Of course, a constructor can be added in this case as well.

Clean

Record type

:: Point = { x :: Int, y :: Int }

Parameterized Algebraic type

:: Point a = Point a a  // usage: (Point Int)

Synonym type

:: Point :== (Int, Int)

Clojure

(defrecord Point [x y])

This defines a datatype with constructor Point. and accessors :x and :y :

(def p (Point. 0 1))
(assert (= 0 (:x p)))
(assert (= 1 (:y p)))

CLU

CLU has two types of compound datatypes: structs, which are immutable, and records, which are mutable. Aside from this, they work the same way.

% Definitions
point = struct[x, y: int]
mutable_point = record[x, y: int]

% Initialization
p: point := point${x: 10, y: 20}
mp: mutable_point := mutable_point${x: 10, y: 20}

The fields can be accessed using the . syntax:

foo := p.x
bar := p.y

Records, but not structs, allow updating the fields in the same way.

mp.x := 30
mp.y := 40

It should be noted that the special forms p.x and mp.x := value are really only syntactic sugar, they are equivalent to the following method calls:

foo := point$get_x(p)
bar := point$get_y(p)

mutable_point$set_x(mp, 30)
mutable_point$set_y(mp, 40)

COBOL

A compound data item description is possible in COBOL as a subdivided record:

       DATA DIVISION.
       WORKING-STORAGE SECTION.
       01  Point.
           05 x USAGE IS BINARY-SHORT.
           05 y USAGE IS BINARY-SHORT.

Here the record Point has the subdivisions x and y, both of which are signed 16-bit binary integers.

CoffeeScript

# Lightweight JS objects (with CS sugar).
point =
  x: 5
  y: 3

console.log point.x, point.y # 5 3

# Heavier OO style
class Point
  constructor: (@x, @y) ->
  distance_from: (p2) ->
    dx = p2.x - @x
    dy = p2.y - @y
    Math.sqrt dx*dx + dy*dy
    
p1 = new Point(1, 6)
p2 = new Point(6, 18)
console.log p1 # { x: 1, y: 6 }
console.log p1.distance_from # [Function]
console.log p1.distance_from p2 # 13

Common Lisp

CL-USER> (defstruct point (x 0) (y 0))  ;If not provided, x or y default to 0
POINT

In addition to defining the point data type, the defstruct macro also created constructor and accessor functions:

CL-USER> (setf a (make-point))          ;The default constructor using the default values for x and y
#S(POINT :X 0 :Y 0)
CL-USER> (setf b (make-point :x 5.5 :y #C(0 1)))  ;Dynamic datatypes are the default
#S(POINT :X 5.5 :Y #C(0 1))                       ;y has been set to the imaginary number i (using the Common Lisp complex number data type)
CL-USER> (point-x b)                    ;The default name for the accessor functions is structname-slotname
5.5
CL-USER> (point-y b)  
#C(0 1)
CL-USER> (setf (point-y b) 3)           ;The accessor is setfable
3
CL-USER> (point-y b)                    
3

Crystal

Crystal's structs work very similarly to objects, but are allocated on the stack instead of the heap, and passed by value instead of by reference. More potential caveats are noted in the language reference.

struct Point(T)
  getter x : T
  getter y : T
  def initialize(@x, @y)
  end 
end

puts Point(Int32).new 13, 12  #=> Point(Int32)(@x=13, @y=12)

D

void main() {
    // A normal POD struct
    // (if it's nested and it's not static then it has a hidden
    // field that points to the enclosing function):
    static struct Point {
        int x, y;
    }

    auto p1 = Point(10, 20);

    // It can also be parametrized on the coordinate type:
    static struct Pair(T) {
        T x, y;
    }

    // A pair with integer coordinates:
    auto p2 = Pair!int(3, 5);

    // A pair with floating point coordinates:
    auto p3 = Pair!double(3, 5);

    // Classes (static inner):
    static class PointClass {
        int x, y;
        this(int x_, int y_) {
            this.x = x_;
            this.y = y_;
        }
    }

    auto p4 = new PointClass(1, 2);

    // There are also library-defined tuples:
    import std.typecons;

    alias Tuple!(int,"x", int,"y") PointXY;

    auto p5 = PointXY(3, 5);

    // And even built-in "type tuples":
    import std.typetuple;

    alias TypeTuple!(int, 5) p6;

    static assert(is(p6[0] == int));
    static assert(p6[1] == 5);
}

Delphi

As defined in Types.pas:

  TPoint = record
    X: Longint;
    Y: Longint;
  end;

Diego

use_namespace(rosettacode)_me();

add_struct(point)_arg(x,y);

with_point(point1)_arg(4,3);

// Since no datatype is specified for the args any datatype can be passed
with_point(point2)_arg(0.033,👣);  

reset_namespace[];

E

def makePoint(x, y) {
    def point {
        to getX() { return x }
        to getY() { return y }
    }
    return point
}

EchoLisp

(lib 'struct)
    (struct Point (x y))
    (Point 3 4)
  → #<Point> (3 4)

;; run-time type checking is possible
(lib 'types)
    (struct Point (x y))
    (struct-type Point Number Number)
    (Point 3 4)
    (Point 3 'albert)
❌ error: #number? : type-check failure : albert → 'Point:y'

Ela

Ela supports algebraic types:

type Maybe = None | Some a

Except of regular algebraic types, Ela also provides a support for open algebraic types - which can be extended any time with new constructors:

opentype Several = One | Two | Three

//Add new constructor to an existing type
data Several = Four

Elena

ELENA 6.x:

struct Point
{
    int X : prop;
 
    int Y : prop;
 
    constructor new(int x, int y)
    {
        X := x;
        Y := y
    }
}

Elixir

iex(1)> defmodule Point do
...(1)>   defstruct x: 0, y: 0
...(1)> end
{:module, Point, <<70, 79, 82, ...>>, %Point{x: 0, y: 0}}
iex(2)> origin = %Point{}
%Point{x: 0, y: 0}
iex(3)> pa = %Point{x: 10, y: 20}
%Point{x: 10, y: 20}
iex(4)> pa.x
10
iex(5)> %Point{pa | y: 30}
%Point{x: 10, y: 30}
iex(6)> %Point{x: px, y: py} = pa        # pattern matching
%Point{x: 10, y: 20}
iex(7)> px
10
iex(8)> py
20

Elm

--Compound Data type can hold multiple independent values
--In Elm data can be compounded using List, Tuple, Record
--In a List
point = [2,5]
--This creates a list having x and y which are independent and can be accessed by List functions
--Note that x and y must be of same data type

--Tuple is another useful data type that stores different independent values
point = (3,4)
--Here we can have multiple data types
point1 = ("x","y")
point2 = (3,4.5)
--The order of addressing matters
--Using a Record is the best option 
point = {x=3,y=4}
--To access
point.x
point.y
--Or Use it as a function
.x point
.y point
--Also to alter the value
{point | x=7}
{point | y=2}
{point | x=3,y=4}
--Each time a new record is generated
--END

Erlang

-module(records_test).
-compile(export_all).

-record(point,{x,y}).

test() ->
    P1 = #point{x=1.0,y=2.0}, % creates a new point record
    io:fwrite("X: ~f, Y: ~f~n",[P1#point.x,P1#point.y]),
    P2 = P1#point{x=3.0}, % creates a new point record with x set to 3.0, y is copied from P1
    io:fwrite("X: ~f, Y: ~f~n",[P2#point.x,P2#point.y]).

Euphoria

Works with: OpenEuphoria

enum x, y

sequence point = {0,0}

printf(1,"x = %d, y = %3.3f\n",point)

point[x] = 'A'
point[y] = 53.42

printf(1,"x = %d, y = %3.3f\n",point)
printf(1,"x = %s, y = %3.3f\n",point)

Output:

x = 0, y = 0.000
x = 65, y = 53.420
x = A, y = 53.420

F#

See the OCaml section as well. Here we create a list of points and print them out.

type Point = { x : int; y : int }

let points = [
    {x = 1; y = 1};
    {x = 5; y = 5} ]
    
Seq.iter (fun p -> printfn "%d,%d" p.x p.y) points

Factor

TUPLE: point x y ;

Fantom

// define a class to contain the two fields
// accessors to get/set the field values are automatically generated
class Point
{
  Int x
  Int y
}

class Main
{
  public static Void main ()
  {
    // empty constructor, so x,y set to 0
    point1 := Point()
    // constructor uses with-block, to initialise values
    point2 := Point { x = 1; y = 2}
    echo ("Point 1 = (" + point1.x + ", " + point1.y + ")")
    echo ("Point 2 = (" + point2.x + ", " + point2.y + ")")
  }
}

Output:

Point 1 = (0, 0)
Point 2 = (1, 2)

Forth

There is no standard structure syntax in Forth, but it is easy to define words for creating and accessing data structures.

: pt>x ( point -- x ) ;
: pt>y ( point -- y ) CELL+ ;
: .pt ( point -- ) dup pt>x @ . pt>y @ . ;    \ or for this simple structure, 2@ . .

create point 6 , 0 ,
7 point pt>y !
.pt    \ 6 7

Works with: GNU Forth version 0.6.2

Some Forths have mechanisms for declaring complex structures. For example, GNU Forth uses this syntax:

struct
  cell% field pt>x
  cell% field pt>y
end-struct point%

Fortran

In ISO Fortran 90 or later, use a TYPE declaration, "constructor" syntax, and field delimiter syntax:

program typedemo
    type rational                                           ! Type declaration
        integer :: numerator
        integer :: denominator
    end type rational
    
    type( rational ), parameter :: zero = rational( 0, 1 )  ! Variables initialized 
    type( rational ), parameter :: one  = rational( 1, 1 )  ! by constructor syntax
    type( rational ), parameter :: half = rational( 1, 2 )
    integer :: n, halfd, halfn
    type( rational ) :: &
        one_over_n(20) = (/ (rational( 1, n ), n = 1, 20) /) ! Array initialized with 
                                                             ! constructor inside
                                                             ! implied-do array initializer 
    integer :: oon_denoms(20)                                
    
    halfd = half%denominator                       ! field access with "%" delimiter
    halfn = half%numerator
    
    oon_denoms = one_over_n%denominator            ! Access denominator field in every 
                                                   ! rational array element & store 
end program typedemo                               ! as integer array

FreeBASIC

' FB 1.05.0 Win64

Type Point
  As Integer x, y
End Type

Dim p As Point = (1, 2)
Dim p2 As Point = (3, 4)
Print p.x, p.y
Print p2.x, p2.y
Sleep

Output:

 1             2
 3             4

FutureBasic

CGRect r = {0, 0, 250, 100}
printf @"x = %.f : y = %.f : width = %.f : height = %.f", r.origin.x, r.origin.y, r.size.width, r.size.height

HandleEvents

Output:

x = 0 : y = 0 : width = 250 : height = 100

Go

type point struct {
    x, y float64
}

Groovy

Declaration

class Point {
    int x
    int y
    
    // Default values make this a 0-, 1-, and 2-argument constructor
    Point(int x = 0, int y = 0) { this.x = x; this.y = y }
    String toString() { "{x:${x}, y:${y}}" } 
}

Instantiation

Direct

// Default Construction with explicit property setting:
def p0 = new Point()
assert 0 == p0.x
assert 0 == p0.y
p0.x = 36
p0.y = -2
assert 36 == p0.x
assert -2 == p0.y

// Direct Construction:
def p1 = new Point(36, -2)
assert 36 == p1.x
assert -2 == p1.y

def p2 = new Point(36)
assert 36 == p2.x
assert 0 == p2.y

List-to-argument Substitution

There are several ways that a List can be substituted for constructor arguments via "type coercion" (casting).

// Explicit coersion from list with "as" keyword
def p4 = [36, -2] as Point
assert 36 == p4.x
assert -2 == p4.y

// Explicit coersion from list with Java/C-style casting
p4 = (Point) [36, -2]
println p4
assert 36 == p4.x
assert -2 == p4.y

// Implicit coercion from list (by type of variable)
Point p6 = [36, -2]
assert 36 == p6.x
assert -2 == p6.y

Point p8 = [36]
assert 36 == p8.x
assert 0 == p8.y

Map-to-property Substitution

There are several ways to construct an object using a map (or a comma-separated list of map entries) that substitutes entries for class properties. The process is properly (A) instantiation, followed by (B) property mapping. Because the instantiation is not tied to the mapping, it requires the existence of a no-argument constructor.

// Direct map-based construction
def p3 = new Point([x: 36, y: -2])
assert 36 == p3.x
assert -2 == p3.y

// Direct map-entry-based construction
p3 = new Point(x: 36, y: -2)
assert 36 == p3.x
assert -2 == p3.y

p3 = new Point(x: 36)
assert 36 == p3.x
assert 0 == p3.y

p3 = new Point(y: -2)
assert 0 == p3.x
assert -2 == p3.y

// Explicit coercion from map with "as" keyword
def p5 = [x: 36, y: -2] as Point
assert 36 == p5.x
assert -2 == p5.y

// Implicit coercion from map (by type of variable)
Point p7 = [x: 36, y: -2]
assert 36 == p7.x
assert -2 == p7.y

Point p9 = [y:-2]
assert 0 == p9.x
assert -2 == p9.y

Haskell

Algebraic Data Type

See algebraic data type. The different options ("Empty", "Leaf", "Node") are called constructors, and is associated with 0 or more arguments with the declared types.

 data Tree = Empty
           | Leaf Int
           | Node Tree Tree
 deriving (Eq, Show)
 
 t1 = Node (Leaf 1) (Node (Leaf 2) (Leaf 3))

Tagged Type

This is a special case of the algebraic data type above with only one constructor.

 data Point = Point Integer Integer
 instance Show Point where
     show (Point x y) = "("++(show x)++","++(show y)++")"
 p = Point 6 7

Record Type

Entries in an algebraic data type constructor can be given field names.

data Point = Point { x :: Integer, y :: Integer } 
deriving (Eq, Show)

The deriving clause here provides default instances for equality and conversion to string.

Different equivalent ways of constructing a point:

p  = Point 2 3
p' = Point { x=4, y=5 }

The field name is also a function that extracts the field value out of the record

x p' -- evaluates to 4

Tuple Type

You can make a tuple literal by using a comma-delimited list surrounded by parentheses, without needing to declare the type first:

p = (2,3)

The type of p is (Int, Int), using the same comma-delimited list syntax as the literal.

Discriminated Type

Just an algebraic data type with multiple constructors being records

data Person =
    Male   { name :: String, age :: Integer, weight :: Double, 
             beard_length :: Double }
  | Female { name :: String, age :: Integer, weight :: Double }
  deriving (Eq, Show)

Note that the field names may be identical in alternatives.

Icon and Unicon

record Point(x,y)

IDL

point = {x: 6 , y: 0 } 
point.y = 7
print, point
;=> {       6       7}

J

In a "real" J application, points would be represented by arrays of 2 (or N) numbers. None the less, sometimes objects (in the OO sense) are a better representation than arrays, so J supports them:

   NB.  Create a "Point" class
   coclass'Point'

   NB. Define its constructor
   create =: 3 : 0
     'X Y' =: y
   )

   NB.  Instantiate an instance (i.e. an object)
   cocurrent 'base'
   P =: 10 20 conew 'Point'

   NB.  Interrogate its members
   X__P
10
   Y__P
20

Jakt

struct Point {
    x: i64
    y: i64
}

fn main() {
    println("{}", Point(x: 3, y: 4))
}

Java

Starting with Java 14 you can use a record

record Point(int x, int y) { }

Usage

Point point = new Point(1, 2);
int x = point.x;
int y = point.y;

Alternately We use a class:

public class Point
{
  public int x, y;
  public Point() { this(0); }
  public Point(int x0) { this(x0,0); }
  public Point(int x0, int y0) { x = x0; y = y0; }

  public static void main(String args[])
  {
    Point point = new Point(1,2);
    System.out.println("x = " + point.x );
    System.out.println("y = " + point.y );
  }
}

JavaScript

//using object literal syntax
var point = {x : 1, y : 2};

//using constructor
var Point = function (x, y) {
  this.x = x;
  this.y = y;
};
point = new Point(1, 2);

//using ES6 class syntax
class Point {
  constructor(x, y) {
    this.x = x;
    this.y = y;
  }
}
point = new Point(1, 2);

jq

{"x":1, "y":2}

If the emphasis in the task description is on "type", then an alternative approach would be to include a "type" key, e.g.

{"x":1, "y":2, type: "Point"}

Using this approach, one can distinguish between objects of type "Point" and those that happen to have keys named "x" and "y".

JSON

{"x":1,"y":2}

Julia

Define the type:

struct Point{T<:Real}
    x::T
    y::T
end

The components of Point can be any sort of real number, though they do have to be of the same type.

Define a few simple operations for Point:

Base.:(==)(u::Point, v::Point) = u.x == v.x && u.y == v.y
Base.:-(u::Point) = Point(-u.x, -u.y)
Base.:+(u::Point, v::Point) = Point(u.x + v.x, u.y + v.y)
Base.:-(u::Point, v::Point) = u + (-v)

Have fun:

a, b, c = Point(1, 2), Point(3, 7), Point(2, 4)
@show a b c
@show a + b
@show -a + b
@show a - b
@show a + b + c
@show a == b
@show a + a == c

Output:

a = Point{Int64}(1, 2)
b = Point{Int64}(3, 7)
c = Point{Int64}(2, 4)
a + b = Point{Int64}(4, 9)
-a + b = Point{Int64}(2, 5)
a - b = Point{Int64}(-2, -5)
a + b + c = Point{Int64}(6, 13)
a == b = false
a + a == c = true

KonsolScript

Var:Create(
  Point,
    Number x,
    Number y
)

Instanciate it with...

function main() {
  Var:Point point;
}

Kotlin

data class Point(var x: Int, var y: Int)

fun main(args: Array<String>) {
    val p = Point(1, 2)
    println(p)
    p.x = 3
    p.y = 4
    println(p)
}

Output:

Point(x=1, y=2)
Point(x=3, y=4)

Lambdatalk

1) a pair
{def P {P.new 1 2}}
-> P
{P.left {P}}
-> 1
{P.right {P}}
-> 2

2) its Lispsish variant
{def Q {cons 1 2}}
-> Q
{car {Q}}
-> 1
{cdr {Q}}
-> 2

3) as an array 
{def R {A.new 1 2}}
-> R
{A.first {R}}
-> 1
{A.last {R}}
-> 2

Lang

&Point = {
	$x
	$y
}

Lasso

In Lasso, a point could just be stored in the pair type. However, assuming we want to be able to access the points using the member methods [Point->x] and [Point->y], let's just create a type that inherits from the pair type:

define Point => type {
    parent pair

    public onCreate(x,y) => {
        ..onCreate(#x=#y)
    }

    public x => .first
    public y => .second
}

local(point) = Point(33, 42)
#point->x
#point->y

Output:

33
43

LFE

Simply define a record in the LFE REPL (can also be used in include files, modules, etc.):

(defrecord point
  x
  y)

Creating points:

> (make-point x 0 y 0)
#(point 0 0)
> (set p (make-point x 1.1 y -4.2))
#(point 1.1 -4.2)

Accessing:

> (point-x p)
1.1
> (point-y p)
-4.2

Updates (note that since LFE has no mutable data, persisted updates would need to rebind the new value to the old variable name):

> (set-point-x p 3.1)
#(point 3.1 -4.2)
> (set-point-y p 4.2)
#(point 1.1 4.2)

Metadata, etc.:

> (fields-point)
(x y)
> (is-point #(x y))
false
> (is-point p)
true

Lingo

Point and Vector types are built-in. A custom "MyPoint" type can be implemented like this:

-- parent script "MyPoint"
property x
property y
on new (me, px, py)
  me.x = px
  me.y = py
  return me
end

p = script("MyPoint").new(23, 42)
put p.x, p.y
-- 23 42

Construction could also be simplified by using a global wrapper function:

-- in some movie script
on MyPoint (x, y)
  return script("MyPoint").new(x, y)
end

p = MyPoint(23, 42)
put p.x, p.y
-- 23 42

Logo

In Logo, a point is represented by a list of two numbers. For example, this will draw a triangle:

setpos [100 100] setpos [100 0] setpos [0 0]
show pos  ; [0 0]

Access is via normal list operations like FIRST and BUTFIRST (BF). X is FIRST point, Y is LAST point. For example, a simple drawing program which exits if mouse X is negative:

until [(first mousepos) < 0] [ifelse button? [pendown] [penup]  setpos mousepos]

Lua

Simple Table

Lua could use a simple table to store a compound data type Point(x, y):

a = {x = 1; y = 2}
b = {x = 3; y = 4}
c = {
    x = a.x + b.x;
    y = a.y + b.y
}
print(a.x, a.y)  --> 1 2
print(c.x, c.y)  --> 4 6

Prototype Object

Furthermore, Lua could create a prototype object (OOP class emulation) to represent a compound data type Point(x, y) as the following:

cPoint = {}                                           -- metatable (behaviour table)
function newPoint(x, y)                               -- constructor
    local pointPrototype = {}                         -- prototype declaration
    function pointPrototype:getX() return x end       -- public method
    function pointPrototype:getY() return y end       -- public method
    function pointPrototype:getXY() return x, y end   -- public method
    function pointPrototype:type() return "point" end -- public method
    return setmetatable(pointPrototype, cPoint)       -- set behaviour and return the pointPrototype
end--newPoint

In the above example, the methods are declared inside the constructor so that they could access the closured values x and y (see usage example). The pointPrototype:type method could be used to extend the original type function available in Lua:

local oldtype = type;                   -- store original type function
function type(v)
    local vType = oldtype(v)
    if (vType=="table" and v.type) then
        return v:type()                 -- bypass original type function if possible
    else
        return vType
    end--if vType=="table"
end--type

The usage of metatable cPoint which stores the behavior of the pointPrototype enables additional behaviour to be added to the data type, such as:

function cPoint.__add(op1, op2)  -- add the x and y components
    if type(op1)=="point" and type(op2)=="point" then
        return newPoint(
            op1:getX()+op2:getX(),
            op1:getY()+op2:getY())
    end--if type(op1)
end--cPoint.__add
function cPoint.__sub(op1, op2)  -- subtract the x and y components
    if (type(op1)=="point" and type(op2)=="point") then
        return newPoint(
            op1:getX()-op2:getX(),
            op1:getY()-op2:getY())
    end--if type(op1)
end--cPoint.__sub

Usage example:

a = newPoint(1, 2)
b = newPoint(3, 4)
c = a + b             -- using __add behaviour
print(a:getXY())      --> 1 2
print(type(a))        --> point
print(c:getXY())      --> 4 6
print((a-b):getXY())  --> -2 -2  -- using __sub behaviour

Maple

Point:= Record(x = 2,y = 4):

Point:-x;
Point:-y;

Output:

2
4

Mathematica / Wolfram Language

Expressions like point[x, y] can be used without defining.

In[1]:= a = point[2, 3]

Out[1]= point[2, 3]

In[2]:= a[[2]]

Out[2]= 3

In[3]:= a[[2]] = 4; a

Out[3]= point[2, 4]

Or you can just define a function.

p[x] = 2; p[y] = 3;

Data will be stored as down values of the symbol p.

MATLAB / Octave

 point.x=3;
 point.y=4;

Alternatively, coordinates can be also stored as vectors

 point = [3,4];

Maxima

defstruct(point(x, y))$

p: new(point)$

q: point(1, 2)$

p@x: 5$

MAXScript

Point is a built-in object type in MAX, so...

struct myPoint (x, y)
newPoint = myPoint x:3 y:4

In practice however, you'd use MAX's built in Point2 type

newPoint = Point2 3 4

MiniScript

Point = {}
Point.x = 0
Point.y = 0

Modula-2

TYPE Point = RECORD
  x, y     : INTEGER
END;

Usage:

VAR    point   : Point;
...
point.x := 12;
point.y := 7;

Modula-3

TYPE Point = RECORD
  x, y: INTEGER;
END;

Usage:

VAR point: Point;
...
point := Point{3, 4};

or

point := Point{x := 3, y := 4};

NetRexx

Like Java, NetRexx uses the class instruction to create compound types. Unlike Java; NetRexx provides keywords to automatically generate getters and setters for class properties and will automatically generate intermediate methods based on defaults provided in method prototypes.

/* NetRexx */
options replace format comments java crossref symbols nobinary

class RCompoundDataType
  method main(args = String[]) public static
    pp = Point(2, 4)
    say pp
    return

class RCompoundDataType.Point -- inner class "Point"
  properties indirect -- have NetRexx create getters & setters
    x = Integer
    y = Integer

  method Point(x_ = 0, y_ = 0) public -- providing default values for x_ & y_ lets NetRexx generate intermediate constructors Point() & Point(x_)
    this.x = Integer(x_)
    this.y = Integer(y_)
    return

  method toString() public returns String
    res = 'X='getX()',Y='getY()
    return res

Output:

X=2,Y=4

Nim

type Point = tuple[x, y: int]

var p: Point = (12, 13)
var p2: Point = (x: 100, y: 200)

Oberon-2

MODULE Point;
TYPE
	Object* = POINTER TO ObjectDesc;
	ObjectDesc* = RECORD
		x-,y-: INTEGER;
	END;

	PROCEDURE (p: Object) Init(x,y: INTEGER);
	BEGIN
		p.x := x; p.y := y
	END Init;

	PROCEDURE New*(x,y: INTEGER): Object;
	VAR
		p: Object;
	BEGIN
		NEW(p);p.Init(x,y);RETURN p;
	END New;

END Point.

Objeck

Classes are used for compound data types.

class Point {
  @x : Int;
  @y : Int;

  New() { 
    @x := 0;
    @y := 0;
  }

  New(x : Int, y : Int) { 
    @x := x;
    @y := y;
  }

  New(p : Point) { 
    @x := p->GetX();
    @y := p->GetY();
  }

  method : public : GetX() ~ Int { 
    return @x; 
  }

  method : public : GetY() ~ Int { 
    return @y; 
  }

  method : public : SetX(x : Int) ~ Nil { 
    @x := x; 
  }

  method : public : SetY(y : Int) ~ Nil { 
    @y := y; 
  }
}

OCaml

Algebraic Data Type

See algebraic data type. The different options ("Empty", "Leaf", "Node") are called constructors, and is associated with 0 or more arguments with the declared types; multiple arguments are declared with a syntax that looks like a tuple type, but it is not really a tuple.

type tree = Empty
          | Leaf of int
          | Node of tree * tree

let t1 = Node (Leaf 1, Node (Leaf 2, Leaf 3))

Record Type

type point = { x : int; y : int }

How to construct a point:

let p = { x = 4; y = 5 }

You can use the dot (".") to access fields.

p.x (* evaluates to 4 *)

Fields can be optionally declared to be mutable:

type mutable_point = { mutable x2 : int; mutable y2 : int }

Then they can be assigned using the assignment operator "<-"

let p2 = { x2 = 4; y2 = 5 } in
  p2.x2 <- 6;
  p2 (* evaluates to { x2 = 6; y2 = 5 } *)

Tuple Type

You can make a tuple literal by using a comma-delimited list, optionally surrounded by parentheses, without needing to declare the type first:

let p = (2,3)

The type of p is a product (indicated by *) of the types of the components:

# let p = (2,3);;
val p : int * int = (2, 3)

Oforth

Using a class :

Object Class new: Point(x, y)

ooRexx

ooRexx uses class for compound data types.

p = .point~new(3,4)
say "x =" p~x
say "y =" p~y

::class point
::method init
  expose x y
  use strict arg x = 0, y = 0   -- defaults to 0 for any non-specified coordinates

::attribute x
::attribute y

OpenEdge/Progress

The temp-table is a in memory database table. So you can query sort and iterate it, but is the data structure that comes closest.

def temp-table point
  field x as int
  field y as int
  .

Another option would be a simple class.

OxygenBasic

'SHORT FORM
type point float x,y
 
'FULL FORM
type point
  float x
  float y
end type

point p

'WITH DEFAULT VALUES
type point
  float x = 1.0
  float y = 1.0
end type

point p = {} 'assigns the set of default values


print p.x " " p.y

Oz

A point can be represented by using a record value:

P = point(x:1 y:2)

Now we can access the components by name: P.x and P.y Often such values are deconstructed by pattern matching:

case P of point(x:X y:Y) then
   {Show X}
   {Show Y}
end

PARI/GP

point.x=1;
point.y=2;

Pascal

type point = record
              x, y: integer;
             end;

Perl

Array

my @point = (3, 8);

Hash

my %point = (
   x => 3,
   y => 8
);

Class instance

package Point;

use strict;
use base 'Class::Struct'
    x => '$',
    y => '$',
;

my $point = Point->new(x => 3, y => 8);

Phix

traditional user defined type

The sequence is a natural compound data type. The following would be the same without the type point and declaring p as a sequence, apart from the run-time error. There would be no difficulty defining point to have a string and two atoms.

with javascript_semantics 
enum x,y
type point(object p)
    return sequence(p) and length(p)=y and atom(p[x]) and atom(p[y])
end type
 
point p = {175,3.375}
p[x] -= p[y]*20
puts(1,"point p is ")
?p
printf(1,"p[x]:%g, p[y]:%g\n",{p[x],p[y]})
p[x] = 0            -- fine
p[y] = "string"     -- run-time error (not pwa/p2js)

Output:

point p is {107.5,3.375}
p[x]:107.5, p[y]:3.375

C:\Program Files (x86)\Phix\test.exw:12
type check failure, p is {0,"string"}

--> see C:\Program Files (x86)\Phix\ex.err
Press Enter...

classes

Library: Phix/Class

You could also use a class (not pwa/p2js)

class point
    public atom x,y
end class
 
point p = new({175,3.375})
p.x -= p.y*20
printf(1,"p.x:%g, p.y:%g\n",{p.x,p.y})
p.x = 0         -- fine
p.y = "string"    -- run-time error

Output:

p.x:107.5, p.y:3.375

C:\Program Files (x86)\Phix\test.exw:9
type error assigning "string" to point.y

--> see C:\Program Files (x86)\Phix\ex.err
Press Enter...

PHP

# Using pack/unpack 
$point = pack("ii", 1, 2);

$u = unpack("ix/iy", $point);
echo $x;
echo $y;

list($x,$y) = unpack("ii", $point);
echo $x;
echo $y;

# Using array
$point = array('x' => 1, 'y' => 2);

list($x, $y) = $point;
echo $x, ' ', $y, "\n";

# or simply:
echo $point['x'], ' ', $point['y'], "\n";

# Using class
class Point { 
  function __construct($x, $y) { $this->x = $x; $this->y = $y; }
  function __tostring() { return $this->x . ' ' . $this->y . "\n"; }
}
$point = new Point(1, 2);
echo $point; # will call __tostring() in later releases of PHP 5.2; before that, it won't work so good.

PicoLisp

(class +Point)

(dm T (X Y)
   (=: x X)
   (=: y Y) )

(setq P (new '(+Point) 3 4))

(show P)

Output:

$52717735311266 (+Point)
   y 4
   x 3

Pike

class Point {
    int x, y;
    void create(int _x, int _y)
    {
        x = _x;
        y = _y;
    }
}

void main()
{
    object point = Point(10, 20);
    write("%d %d\n", point->x, point->y);
}

Output:

10 20

PL/I

define structure
  1 point,
     2 x float,
     2 y float;

Plain English

A cartesian point is a record with an x coord and a y coord.

Pop11

uses objectclass;
define :class Point;
   slot x = 0;
   slot y = 0;
enddefine;

PowerShell

Works with: PowerShell version 5

class Point {
  [Int]$a
  [Int]$b
  Point() {
      $this.a = 0
      $this.b = 0
  }
  Point([Int]$a, [Int]$b) {
      $this.a = $a
      $this.b = $b
  }
  [Int]add() {return $this.a + $this.b}
  [Int]mul() {return $this.a * $this.b}
}
$p1  = [Point]::new()
$p2 = [Point]::new(3,2)
$p1.add()
$p2.mul()

Output:

0
6

Prolog

Prolog terms ARE compound data types, there is no need to specifically define a type. for the purpose of this exercise you could define a rule like so:

point(10, 20).

This will create static point that can be called:

?- point(X,Y).
X = 10,
Y = 20.

terms can be passed around as values and can have a complex nested structure of any size, eg:

person_location(person(name(N), age(A)), point(X, Y)).

PureBasic

A basic structure is implemented as;

Structure MyPoint
  x.i
  y.i
EndStructure

Python

The simplest way it to use a tuple, or a list if it should be mutable:

X, Y = 0, 1
p = (3, 4)
p = [3, 4]

print p[X]

If needed, you can use class:

class Point:
    def __init__(self, x=0, y=0):
        self.x = x
        self.y = y

p = Point()
print p.x

One could also simply instantiate a generic object and "monkeypatch" it:

class MyObject(object): pass
point = MyObject()
point.x, point.y = 0, 1
# objects directly instantiated from "object()"  cannot be "monkey patched"
# however this can generally be done to it's subclasses

Dictionary

Mutable. Can add keys (attributes)

pseudo_object = {'x': 1, 'y': 2}

Named Tuples

As of Python 2.6 one can use the collections.namedtuple factory to create classes which associate field names with elements of a tuple. This allows one to perform all normal operations on the contained tuples (access by indices or slices, packing and unpacking) while also allowing elements to be accessed by name.

>>> from collections import namedtuple
>>> help(namedtuple)
Help on function namedtuple in module collections:

namedtuple(typename, field_names, verbose=False)
    Returns a new subclass of tuple with named fields.
    
    >>> Point = namedtuple('Point', 'x y')
    >>> Point.__doc__                   # docstring for the new class
    'Point(x, y)'
    >>> p = Point(11, y=22)             # instantiate with positional args or keywords
    >>> p[0] + p[1]                     # indexable like a plain tuple
    33
    >>> x, y = p                        # unpack like a regular tuple
    >>> x, y
    (11, 22)
    >>> p.x + p.y                       # fields also accessable by name
    33
    >>> d = p._asdict()                 # convert to a dictionary
    >>> d['x']
    11
    >>> Point(**d)                      # convert from a dictionary
    Point(x=11, y=22)
    >>> p._replace(x=100)               # _replace() is like str.replace() but targets named fields
    Point(x=100, y=22)

>>>

QB64

Type Point
    x As Double
    y As Double
End Type

Dim p As Point
p.x = 15.42
p.y = 2.412

Print p.x; p.y

Output:

 15.42  2.412

Quackery

The single ubiquitous compound data type in Quackery is the nest (a mostly immutable dynamic array), a sequence of items wrapped in square brackets. (Mostly immutable; i.e. immutable except under limited circumstances beyond the scope of this discussion. When we refer to changing the contents of a nest here, this is casual speech; a shorthand for saying "creating a new instance of the nest, identical the previous instance except where it differs".)

Presented here are two solutions to the task, the "quick and dirty" solution; sufficient to the task described here, and the "overkill" solution; extending the Quackery compiler to facilitate complex compound data structures akin to structs in C etc.

Quick and Dirty

The word point creates an instance of a nest with two elements, both initialised to zero. The word x specifies the location of the zeroth element within the nest, and the word y specifies the location of the first element within the nest. peek returns the value stored in a specified location, and poke changes the value stored in a specified location, returning the modified nest.

  [ ' [ 0 0 ] ] is point ( --> [ )

  [ 0 ]         is x     ( --> n )

  [ 1 ]         is y     ( --> n )

  point
  dup x peek     echo cr
  99 swap y poke 
  y peek         echo cr

Output:

0
99

Overkill

The "overkill" solution automates the process of creating new structures with the word struct{, which extends the Quackery compiler to allow the definition of complex compound data structures as follows.

  struct{
    item.0
    { item.1.0
      item.1.1
      { item.1.2.0
        item.1.2.1
        item.1.2.2
        item.1.2.3
      } item.1.2
      item.1.3
    } item.1
    item.2
  }struct mystruct

Once defined, the word mystruct will place a new instance of the described structure, with each item initialised to null, on the stack. (The behaviour of null is to place a reference to itself on the stack, as a convenience for debugging, and to allow code to identify elements within the structure that have not had a value assigned to them.)

The various names defined within the struct (e.g. item.1.2.1) return a path - a means of locating a specific item within the struct, for use by {peek} and {poke}, which have the same behaviours as peek and poke, except the they take a path to an item within a struct as an argument, rather than a number specifying an item within a nest.

Names following a } within the definition of a struct (e.g. } item.1.2) return a path to the compound data structure preceding it within the structure. In the example, item.1.2 returns the path to { item.1.2.0 item.1.2.1 item.1.2.2 item.1.2.3 }

  mystruct                 ( create new instance of a mystruct )
  dup echo cr              ( this is what it looks like        )
  789 swap item.1.3 {poke} ( change one of the items           ) 
  dup echo cr              ( this is what it looks like now    )
  item.1.3 {peek} echo cr  ( retrieve the specified item       )

Output:

[ null [ null null [ null null null null ] null ] null ]
[ null [ null null [ null null null null ] 789 ] null ]
789

The words {peek}, {poke}, null, and the building word (i.e. compiler extension) struct{ defined:

  [ witheach peek ]         is {peek}        (   { p --> x     )

  [ dip dup
      witheach [ peek dup ]
    drop ]                  is depack        (   { p --> *     )

  [ reverse
    witheach
      [ dip swap poke ] ]   is repack        (   * p --> {     )

  [ dup dip
      [ rot dip
        [ depack drop ] ]
    repack ]                is {poke}        ( x { p --> {     )

  [ this ]                  is null          (       --> [     )

  [ stack ]                 is {}.path       (       --> s     )
  protect {}.path

  [ stack ]                 is {}.struct     (       --> s     )
  protect {}.struct

  [ nextword dup
    $ "" = if
      [ $ "Unexpected end of struct."
        message put
        bail ] ]            is {}.checknext  (   [ $ --> [ $ $ )

  [ dup  $ "{" =
    over $ "}" = or
    swap $ "}struct" = or if
      [ $ "Name missing after }."
        message put 
        bail ] ]            is {}.checkname  ( [ $ $ --> [ $   )

  [ nested
    namenest take
    join
    namenest put
    ' [ ' ]
    {}.path share nested join
    actiontable take
    1 stuff
    actiontable put ]       is {}.addpath    ( [ $ $ --> [ $   )

  [ nested
    namenest take
    join
    namenest put
    ' [ ' ]
    {}.struct share nested join
    actiontable take
    1 stuff
    actiontable put ]       is  {}.addstruct ( [ $ $ --> [ $   )

  [ {}.path take
    dup -1 peek
    1+
    swap -1 poke
    -1 join
    {}.path put
    [] {}.struct put ]      is {}.{          (   [ $ --> [ $   ) 

  [ {}.struct size 3 < if
      [ $ "Badly formed struct."
        message put bail ]
    trim {}.checknext
    dup {}.checkname
    {}.path take
    -1 split drop
    {}.path put
    {}.addpath
    {}.struct take
    {}.struct take
    swap nested join
    {}.struct put ]         is {}.}          (   [ $ --> [ $   )

  [ {}.path take
    dup -1 peek
    1+
    swap -1 poke
    {}.path put
    {}.addpath
    {}.struct take
    ' [ null ] join
    {}.struct put ]         is {}.name       (   [ $ --> [ $   )

  [ trim {}.checknext
    {}.struct size
    2 != if
      [ $ "Badly formed struct."
        message put
        bail ]
     {}.addstruct ]         is {}.}struct    (   [ $ --> [ $   )

  [ ' [ -1 ] {}.path put
    [] {}.struct put
    [ trim {}.checknext
      dup $ "{" = iff
        [ drop {}.{ ] again
      dup $ "}" = iff
        [ drop {}.} ] again
      dup $ "}struct" = iff
        [ drop {}.}struct ] done
      {}.name again ]
    {}.struct release
    {}.path release ]   builds struct{       (   [ $ --> [ $   )

Finally we use struct{ etc. to fulfil the requirements go the task.

  struct{ x y }struct point
  point
  dup x {peek}     echo cr
  99 swap y {poke} 
  y {peek}         echo cr

Output:

null
99

R

R uses the list data type for compound data.

mypoint <- list(x=3.4, y=6.7)
# $x
# [1] 3.4
# $y
# [1] 6.7
mypoint$x    # 3.4

list(a=1:10, b="abc", c=runif(10), d=list(e=1L, f=TRUE))
# $a
# [1]  1  2  3  4  5  6  7  8  9 10
# $b
# [1] "abc"
# $c
#  [1] 0.64862897 0.73669435 0.11138945 0.10408015 0.46843836 0.32351247
#  [7] 0.20528914 0.78512472 0.06139691 0.76937113
# $d
# $d$e
# [1] 1
# $d$f
# [1] TRUE

Racket

The most common method uses structures (similar to records):

#lang racket
(struct point (x y))

Alternatively, you can define a class:

#lang racket
(define point% ; classes are suffixed with % by convention
  (class object%
    (super-new)
    (init-field x y)))

Raku

(formerly Perl 6)

Works with: Rakudo version #24 "Seoul"

Array

my @point = 3, 8;

my Int @point = 3, 8; # or constrain to integer elements

Hash

my %point = x => 3, y => 8;

my Int %point = x => 3, y => 8; # or constrain the hash to have integer values

Class instance

class Point { has Real ($.x, $.y) is rw; }
my Point $point .= new: x => 3, y => 8;

Set

my $s1 = set <a b c d>; # order is not preserved
my $s2 = set <c d e f>;
say $s1 (&) $s2; # OUTPUT«set(c, e)»
say $s1 ∩ $s2; # we also do Unicode

REXX

x= -4.9
y=  1.7

point=x y

---or---

x= -4.1
y=  1/4e21

point=x y

bpoint=point

gpoint=5.6  7.3e-12

Ring

see new point {x=10 y=20} class point x y

Output

x: 10.000000
y: 20.000000

Ruby

Point = Struct.new(:x,:y)
pt = Point.new(6,7)
puts pt.x        #=> 6
pt.y = 3
puts pt          #=> #<struct Point x=6, y=3>

# The other way of accessing
pt = Point[2,3]
puts pt[:x]      #=> 2
pt['y'] = 5
puts pt          #=> #<struct Point x=2, y=5>

pt.each_pair{|member, value| puts "#{member} : #{value}"}
                 #=> x : 2
                 #=> y : 5

Rust

Structs

There are three kinds of structs in Rust, two of which would be suitable to represent a point.

C-like struct

 // Defines a generic struct where x and y can be of any type T
struct Point<T> {
    x: T,
    y: T,
}
fn main() {
    let p = Point { x: 1.0, y: 2.5 }; // p is of type Point<f64>
    println!("{}, {}", p.x, p.y);
}

Tuple struct

These are basically just named tuples.

struct Point<T>(T, T);
fn main() {
    let p = Point(1.0, 2.5);
    println!("{},{}", p.0, p.1);
}

Tuples

 fn main() {
    let p = (0.0, 2.4);
    println!("{},{}", p.0, p.1);
}

Scala

case class Point(x: Int = 0, y: Int = 0)

val p = Point(1, 2)
println(p.y)   //=> 2

Scheme

Using SRFI 9:

(define-record-type point
    (make-point x y)
    point?
    (x point-x)
    (y point-y))

Seed7

const type: Point is new struct
    var integer: x is 0;
    var integer: y is 0;
  end struct;

Shen

(datatype point
  X : number; Y : number;
  ====================
  [point X Y] : point;)

Pairs (distinct from cons cells) are also supported, in which case a point would be denoted by (number * number):

(2+) (@p 1 2)
(@p 1 2) : (number * number)

Sidef

struct Point {x, y};
var point = Point(1, 2);
say point.y;                #=> 2

SIMPOL

The point type is pre-defined in [SIMPOL], so we will call this mypoint.

type mypoint
  embed
  integer x
  integer y
end type

The embed keyword is used here as a toggle to indicate that all following properties are embedded in the type. The other toggle is reference, which only places a reference to an object in the type, but the reference assigned before the property can be used. These keywords can also be placed on the same line, but then they only apply to that line of the type definition.

A type in [SIMPOL] can be just a container of values and other structures, but it can also include methods. These are implemented outside the type definition, but must be part of the same compiled unit.

type mypoint
  embed
  integer x
  integer y
end type

function mypoint.new(mypoint me, integer x, integer y)
  me.x = x
  me.y = y
end function me

SNOBOL4

	data('point(x,y)')
	p1 = point(10,20)
	p2 = point(10,40)
	output = "Point 1 (" x(p1) "," y(p1) ")"
	output = "Point 2 (" x(p2) "," y(p2) ")"
end

Standard ML

Algebraic Data Type

See algebraic data type. The different options ("Empty", "Leaf", "Node") are called constructors, and is associated with 0 or 1 arguments with the declared types; multiple arguments are handled with tuples.

datatype tree = Empty
              | Leaf of int
              | Node of tree * tree

val t1 = Node (Leaf 1, Node (Leaf 2, Leaf 3))

Tuple Type

You can make a tuple literal by using a comma-delimited list surrounded by parentheses, without needing to declare the type first:

val p = (2,3)

The type of p is a product (indicated by *) of the types of the components:

- val p = (2,3);
val p = (2,3) : int * int

You can extract elements of the tuple using the #N syntax:

- #2 p;
val it = 3 : int

The #2 above extracts the second field of its argument.

Record Type

Records are like tuples but with field names.

You can make a record literal by using a comma-delimited list of key = value pairs surrounded by curly braces, without needing to declare the type first:

val p = { x = 4, y = 5 }

The type of p is a comma-delimited list of key:type pairs of the types of the fields:

- val p = { x = 4, y = 5 };
val p = {x=4,y=5} : {x:int, y:int}

You can extract elements of the tuple using the #name syntax:

- #y p;
val it = 5 : int

The #y above extracts the field named "y" of its argument.

Stata

See struct in Stata help.

mata
struct Point {
	real scalar x, y
}

// dumb example
function test() {
	struct Point scalar a
	a.x = 10
	a.y = 20
	printf("%f\n",a.x+a.y)
}

test()
30
end

Swift

// Structure
struct Point {
    var x:Int
    var y:Int
}

// Tuple
typealias PointTuple = (Int, Int)

// Class
class PointClass {
    var x:Int!
    var y:Int!
    
    init(x:Int, y:Int) {
        self.x = x
        self.y = y
    }
}

Tcl

This can be done using an associative array:

array set point {x 4 y 5}
set point(y) 7
puts "Point is {$point(x),$point(y)}"
# => Point is {4,7}

Or a dictionary:

Works with: Tcl version 8.5

set point [dict create x 4 y 5]
dict set point y 7
puts "Point is {[dict get $point x],[dict get $point y]}"

Or an object:

Works with: Tcl version 8.6

oo::class create Point {
    variable x y
    constructor {X Y} {set x $X;set y $Y}
    method x {args} {set x {*}$args}
    method y {args} {set y {*}$args}
    method show {} {return "{$x,$y}"}
}
Point create point 4 5
point y 7
puts "Point is [point show]"

TI-89 BASIC

TI-89 BASIC does not have user-defined data structures. The specific example of a point is best handled by using the built-in vectors or complex numbers.

Transd

#lang transd

// If the Point type needs encapsulation and/or methods, it should be
// implemented as class. Otherwise, the named tuple will do.

class Point: {
    x: Double(), y: Double(),
    @init: (λ _x Double() _y Double() (= x _x) (= y _y)),
    @to-String: (λ ss StringStream() (textout to: ss 
        "Point( x: " x "; y: " y " )"))
    // ... other methods can be defined here ...
}

MainModule: {
    Point2: typealias(Tuple<Double Double>()),
    _start: (λ
        (with pt Point(2.5 3.7)
            (lout "Class: " pt)
        )
        (with pt Point2(2.5 3.7)
            (lout "\nNamed tuple: " pt)
        )
    )
}

Output:

Class: Point( x: 2.5; y: 3.7 )

Named tuple: [[2.5, 3.7]]

TXR

In TXR Lisp, a structure type can be created:

(defstruct point nil (x 0) (y 0))

If it is okay for the coordinates to be initialized to nil, it can be condensed to:

(defstruct point nil x y)

The nil denotes that a point has no supertype: it doesn't inherit from anything.

This structure type can then be instantiated using the new macro (not the only way):

(new point)         ;; -> #S(point x 0 y 0)
(new point x 1)     ;; -> #S(point x 1 y 0)
(new point x 1 y 1) ;; -> #S(point x 1 y 1)

A structure can support optional by-order-of-arguments ("boa") construction by providing a "boa constructor". The defstruct syntactic sugar does this if a function-like syntax is used in place of the structure name:

(defstruct (point x y) nil (x 0) (y 0))

The existing construction methods continue to work, but in addition, this is now possible:

(new (point 3 4)) -> #S(point x 3 y 4)

Slot access syntax is supported. If variable p holds a point, then p.x designates the x slot, as a syntactic place which can be accessed and stored:

(defun displace-point-destructively (p delta)
  (inc p.x delta.x)
  (inc p.y delta.y))

UNIX Shell

Works with: ksh93

ksh93 allows you to define new compound types with the typeset -T command.

typeset -T Point=(
  typeset x
  typeset y
)
Point p
p.x=1
p.y=2
echo $p
echo ${p.x} ${p.y}
Point q=(x=3 y=4)
echo ${q.x} ${q.y}

Output:

( x=1 y=2 )
1 2
3 4

You can also declare compound variables "on the fly" without using a defined type:

point=()
point.x=5
point.y=6
echo $point
echo ${point.x} ${point.y}

Output:

( x=5 y=6 )
5 6

Ursala

A record type with two untyped fields named x and y can be declared like this.

point :: x y

A constant instance of the record can be declared like this.

p = point[x: 'foo',y: 'bar']

A function returning a value of this type can be defined like this,

f = point$[x: g,y: h]

where g and h are functions. Then f(p) would evaluate to point[x: g(p),y: h(p)] for a given argument p. Accessing the fields of a record can be done like this.

t = ~x p
u = ~y p

where p is any expression of the defined type. A real application wouldn't be written this way because pairs of values (x,y) are a common idiom.

Vala

struct Point {
  int x;
  int y;
}

VBA

Type point
    x As Integer
    y As Integer
End Type

Vim Script

One cannot create new data types in Vim Script. A point could be represented by a dictionary:

function MakePoint(x, y)    " 'Constructor'
    return {"x": a:x, "y": a:y}
endfunction

let p1 = MakePoint(3, 2)
let p2 = MakePoint(-1, -4)

echon "Point 1: x = " p1.x ", y = " p1.y "\n"
echon "Point 2: x = " p2.x ", y = " p2.y "\n"

Output:

Point 1: x = 3, y = 2                                                           
Point 2: x = -1, y = -4

Visual Basic .NET

Structures

A simple structure with two public, mutable fields:

Structure Point
   Public X, Y As Integer
End Structure

Immutable Structures

It is generally recommended in .NET that mutable structures only be used in niche cases where they provide needed performance, e.g. when the creation of massive numbers of class instances would cause excessive garbage collection pressure, as high-performance code dealing with structs generally is of a paradigm considered "impure" from an object-oriented perspective that relies on passing by reference and directly exposing fields.

The semantics of value types in .NET mean that a new copy of a structure is created whenever one is passed by value to or from a method or property. This is particularly vexing when properties are involved, as it is not possible to mutate a structure that is returned by a property, due to the returned structure actually being an independent copy of whatever the property originally returned. The only workaround would be to store the value of the property in a temporary variable, mutate that variable, and assign the mutated variable back to the property, which involves another copy operation. When a structure is large, this copying can significantly affect performance.

On another note, algorithms relying on immutable data structures are often more easily parallelized, as they eliminate the race conditions caused by concurrent reading and writing.

Below is the same Point as above, except with an immutable API.

Structure ImmutablePoint
   ReadOnly Property X As Integer
   ReadOnly Property Y As Integer

   Public Sub New(x As Integer, y As Integer)
       Me.X = x
       Me.Y = y
   End Sub
End Structure

V (Vlang)

Vlang also supports embedding structs into other structs and assigning methods to structs.

struct Point {
    x int
    y int
}

// main() declaration can be skipped in one file programs
// we can define whether immutable or mutable by using the "mut" keyword

mut p := Point{
    x: 10
    y: 20
}

// struct fields are accessed using a dot

println("Value of p.x is: $p.x") 
println("Show the struct:\n $p")

// alternative literal syntax can be used for structs with 3 fields or fewer

p = Point{30, 40}
assert p.x == 30
println("Show the struct again after change:\n $p")

Output:

Value of p.x is: 10
Show the struct:
 Point{
    x: 10
    y: 20
}
Show the struct again after change:
 Point{
    x: 30
    y: 40
}

Wren

class Point {
    construct new(x, y) {
        _x = x
        _y = y
    }
    x { _x }
    y { _y }

    // for illustration allow Points to be mutated
    x=(value) { _x = value }
    y=(value) { _y = value }

    toString { "(%(_x), %(_y))" }
}

var p = Point.new(1, 2)
System.print(p.toString)

// mutate Point object
p.x = 2
p.y = 3
// print without using the toString method
System.printAll(["(", p.x, ", ", p.y, ")"])

Output:

(1, 2)
(2, 3)

XSLT

Data types in XSLT are expressed as XML nodes. Members of a node can be either attributes or child nodes. Access to data is via XPath expressions.

Attributes

Attributes are often used for simple values. This is how a point might be represented in SVG, for example.

<point x="20" y="30"/>

<!-- context is a point node. The '@' prefix selects named attributes of the current node. -->
<fo:block>Point = <xsl:value-of select="@x"/>, <xsl:value-of select="@y"/></fo:block>

Children

More complex, multivariate, and nested data structures can be represented using child nodes.

<circle>
  <point>
    <x>20</x>
    <y>30</y>
  </point>
  <radius>10</radius>
</circle>

<!-- context is a circle node. Children are accessed using a path-like notation (hence the name "XPath"). -->

<fo:block>Circle center = <xsl:value-of select="point/x"/>, <xsl:value-of select="point/y"/></fo:block>

Z80 Assembly

We'll declare the following C struct:

struct Point{
char x;
char y;
}

and then execute the following C code as Z80 Assembly below.

struct Point myPoint;
myPoint.x = 3;
myPoint.y = 5;

;I'm arbitrarily choosing &1100 as the memory location of our Point variable.
ld hl,&1100
ld (hl),3
inc hl
ld (hl),5
ret

zkl

The OO solution:

class Point{ var x,y; 
   fcn init(x,y){self.x=x.toFloat(); self.y=y.toFloat(); }
   fcn toString{ "P(%f,%f)".fmt(x,y) }
   fcn __opADD(P){} //+: add Point, constant or whatever
   //... __opEQ == etc
}
Point(1,2).println() //-->P(1.000000,2.000000)

which can be pretty heavy weight. [read only] lists can work just as well:

point:=T(1,2); points:=T( T(1,2), L(3,4) )

zonnon

{ref,public} (* class *)
	Point = object(ord,abs: integer)
	var
		(* instance variables *)
		{public,immutable} x,y: integer;

	(* method *)
	procedure {public} Ord():integer;
	begin
		return y
	end Ord;

	(* method *)
	procedure {public} Abs():integer;
	begin
		return x
	end Abs;

	(* constructor *)
	begin
		self.x := ord;
		self.y := abs;
	end Point;