Polymorphism: Difference between revisions

 

=={{header|ActionScript}}==

{

public class Point

}

}

public class Circle extends Point

{

}

}

 

=={{header|Ada}}==

This example is constructed using a parent package and a child package. The parent package defines the Point type. The child package defines the Circle type.

type Point is tagged private;

procedure Print(Item : in Point);

Y : Integer := 0;

end record;

 

package body Shapes is

end Create;

 

The following is the child package defining the Circle type.

type Circle is new Point with private;

procedure Print(Item : Circle);

R : Integer := 0;

end record;

 

package body Shapes.Circles is

end Create;

 

The following procedure is an entry point for a program, serving the same purpose as the main function in C.

use Shapes;

 

P.Print;

C.Print;

 

=={{header|Aikido}}==

class Point (protected x=0.0, protected y=0.0) {

public function print {

c.print()

 

 

=={{header|ALGOL 68}}==

{{works with|ALGOL 68G|Any - tested with release 2.8.win32}}

 

# define the CIRCLE and POINT modes #

PRINT c1; newline( stand out )

 

{{out}}

<pre>

Circle(+4.10000000000000e +0 @ +0.00000000000000e +0,+0.00000000000000e +0)

</pre>

 

=={{header|AutoHotkey}}==

{{works with|AutoHotkey 1.1}}

AutoHotkey does not support private or protected properties and thus does not need assignment and accessor methods. Assignment and accessor methods, as well as direct assignment and access, are shown. For more information see [http://ahkscript.org/docs/Objects.htm Objects].

MyPoint.Print()

MyCircle := new Circle(4, 7, 9)

this.r := aValue

}

 

=={{header|BASIC}}==

=={{header|BBC BASIC}}==

{{works with|BBC BASIC for Windows}}

REM Create parent class with void 'doprint' method:

PROC(mycircle.doprint)

PROC_discard(mycircle{})

{{out}}

<pre>

* See [[Polymorphism/C]]

 

{

int x, y;

Point(int x0 = 0, int y0 = 0) : x(x0), y(y0) {}

virtual ~Point() {}

}

int getX() { return x; }

int getY() { return y; }

virtual void print() { printf("Point\n"); }

};

 

int r;

Circle(Point p, int r0 = 0) : Point(p), r(r0) {}

Circle(int x0 = 0, int y0 = 0, int r0 = 0) : Point(x0, y0), r(r0) {}

virtual ~Circle() {}

}

int getR() { return r; }

virtual void print() { printf("Circle\n"); }

};

 

 

'''Pattern:''' [[CRTP|Curiously Recurring Template Pattern]]

 

template <typename Derived>

class PointShape

 

// compile-time virtual function

};

 

return *this;

}

};

 

public:

Circle(int x0 = 0, int y0 = 0, int r0 = 0) : PointShape(x0, y0), r(r0) { }

~Circle() {}

const Circle& operator=(const Circle& c)

}

int getR() { return r; }

};

 

int main()

{

p->print();

c->print();

return 0;

 

}

 

 

=={{header|Clojure}}==

Clojure 1.2.

 

(print-it [this] "Prints out the Printable."))

(defn create-circle

"Redundant consturctor function."

 

=={{header|Common Lisp}}==

((x :initarg :x :initform 0 :accessor x)

(y :initarg :y :initform 0 :accessor y)))

(c (make-instance 'circle :radius 5)))

(print-shape p)

 

=={{header|D}}==

 

class Point {

writeln(p);

writeln(c);

 

=={{header|Delphi}}==

{ TPoint }

 

begin

ShowMessage('MyPoint');

 

MyPoint: TMyPoint;

Circle: TCircle;

FreeAndNil(MyPoint);

end;

 

=={{header|E}}==

def point implements pbc {

to __printOn(out) { out.print(`<point $x,$y>`) }

}

return circle

 

(It is unidiomatic to have mutation operations on an object of this sort in E, so this example has variation operations instead. __optUncall is used for serialization, and is the closest analogue to a copy constructor. E does not have destructors, but only post-mortem finalizers (which are registered after the object is created). The "extends" is only implementation inheritance; it is not necessary to enable polymorphism.)

 

def c := makeCircle(1, 1, 2)

println(p)

 

=={{header|EchoLisp}}==

(struct Point ((real:x 0) (real:y 0)))

(struct Circle ((real:x 0) (real:y 0) (real:r 1)))

(print (Circle 0 0 10))

→ ⭕️ center:[0 0] radius:10 diameter:62.83185307179586

 

=={{header|Eiffel}}==

 

POINT

inherit

Result := "Point: x = " + x.out + " y = " + y.out

end

 

 

CIRCLE

 

non_negative_radius: r >= 0

 

 

 

APPLICATION

 

end

 

 

{{out}}

Solution of this problem in Ela is similar to Haskell, as soon as Ela shares with Haskell the same features - namely, classes (typeclasses) and algebraic types.

 

instance Show Point where

circleZ = Circle 0 0

 

Class Show is defined in prelude and is effectively an abstraction for all "printable" entities.

Normally, algebraic types are analyzed using pattern matching, however, it is possible to provide a support for an "accessor style" approach by providing an instance for class Name (which is also defined in prelude). With this instance it is possible to write code like so:

 

 

=={{header|Factor}}==

 

GENERIC: print ( shape -- )

C: <circle> circle

 

 

=={{header|Forth}}==

{{works with|4tH|3.62.0}}

There are numerous, mutually incompatible object oriented frameworks for Forth. This one works with the FOOS preprocessor extension of [[4tH]]. Variadic functions in Forth are usually implemented by passing the number of parameters. Since it is highly unlikely that objects are allocated that low in memory it works. Note that X, Y and Z are passed in reverse order, which is quite common for any Forth program.

include 4pp/lib/foos.4pp

 

Circle2 => print

Circle1 new Circle Circle3

 

 

Works with any ANS Forth

 

 

 

 

;class

 

 

=={{header|Fortran}}==

Fortran provides OO features with the type mechanism. This example works with the Intel 11.1.069 compiler.

module geom

 

 

end program inh

 

 

 

 

 

 

=={{header|Go}}==

 

import "fmt"

}

 

 

=={{header|Groovy}}==

@TupleConstructor(force = true)

@ToString(includeNames = true)

void print() { println toString() }

Number r

Test Code:

def c = new Circle(x: 4, y: 3, r: 5)

 

v2.print()

assert v1 == v2

{{out}}

<pre>Verifying Point(x:3, y:4) == Point(x:3, y:4)

Polymorphism is achieved through the type class Show

 

instance Show Point where

show (Point x y) = "Point at "++(show x)++","++(show y)

 

--Constructor that sets x and r to 0

 

== Icon and {{header|Unicon}} ==

There is no destructor, as Unicon manages object destruction itself. The copy constructor is emulated by a method. Notice the 'initially' clauses. These act like constructors, in that they accept input parameters during instance construction. These parameters are null if not used, and so the initial field values are set to 0 if the entered value is null (tested using the '/' symbol).

 

# make a new copy of this instance

method copy ()

c4.print ()

end

 

=={{header|Inform 7}}==

Accessors are not needed since property values are public. Constructors and destructors are not needed since objects are statically allocated and initialized.

 

 

A point is a kind of thing.

print the origin;

print the circle of power;

 

=={{header|J}}==

 

create=: monad define

'X Y'=:2{.y

smoutput 'Point ',":X,Y

)

 

coinsert 'Point'

create=: monad define

print=: monad define

smoutput 'Circle ',":X,Y,R

 

=={{header|Java}}==

 

protected int x, y;

public Point() { this(0); }

c.print();

}

 

=={{header|JavaScript}}==

* var p = new Point(x,y);

* var p = new Point(a_point);

var out = "Circle(" + this.x + "," + this.y + "," + this.r + ")";

print(out);

 

=={{header|Julia}}==

 

It would not be idiomatic to define setters and getters for a type like this in Julia. One would just access the fields directly. There is no need to explicitly define a constructor since that is automatically provided. You only roll your own if you need more elaborate initialization or you need to set default values.

 

x::Float64

y::Float64

end

 

 

 

setx(p::Point, x) = (p.x = x)

 

x::Float64

y::Float64

end

 

 

setx(c::Circle, x) = (c.x = x)

 

 

=={{header|Kotlin}}==

 

In the JVM version of Kotlin, it is possible to declare a destructor in the guise of a 'finalize' method though there is no guarantee that this will actually be called by the garbage collector (or, if it is called, when this will be) and consequently many programmers feel it is more trouble than its worth.

 

open class Point(var x: Int, var y: Int) {

 

constructor(x: Int) : this(x, 0, 0)

constructor(x: Int, r: Int) : this(x, 0, r)

 

 

// for simplicity not calling super.finalize() below though this would normally be done in practice

 

override fun toString() = "Circle at center ($x, $y), radius $r"

for (circle in circles) circle.print()

println()

 

fun main(args: Array<String>) {

p.print()

p.y = 7 // change y coordinate

val c = Circle(5, 6, 7)

c.print()

c.r = 8

 

{{out}}

Circle at center (5, 6), radius 8

</pre>

 

=={{header|Lua}}==

Lua does not have a standard definition of objects or classes, so a basic and typical protoctype-based OOP model will be used. In Lua all objects are tables, and through the use of metatables, polymorphism can be achieved in many ways, this is only one of them.

local Point = {x = 0, y = 0}

 

function Circle:copy()

return Circle:new{x = self.x, y = self.y, r = self.r}

 

=={{header|NetRexx}}==

'''Note:''' Based on default values in method prototypes, NetRexx will automatically generate intermediate constructors and methods, thus ensuring that none are omitted.

 

options replace format comments java crossref savelog symbols binary

Rexx(getR()).format(null, 3)')'

return str

{{out}}

<pre>

 

Similar to the Python solution:

Point = object

x, y: float

 

c1.center.x = 12

{{out}}

<pre>p1: (x: 0.0, y: 0.0)

 

=={{header|Objeck}}==

bundle Default {

class Point {

}

}

 

=={{header|Objective-C}}==

 

@interface RCPoint : NSObject {

}

return 0;

 

=={{header|OCaml}}==

object (self)

val mutable x = x

c#set_x 10.0;

print c;

 

=={{header|Oforth}}==

 

 

Point method: initialize(x, y) x := x y := y ;

Point method: _x @x ;

Circle method: << "(" << @x << ", " << @y << ", " << @r << ")" << ;

 

 

Usage :

| p c |

Point new(3, 4) ->p

c println

System.Out "Attributes of this circle are : " << c _x << ", " << c _y << " and " << c _r << cr

 

{{out}}

ooRexx supports traditional class-based polymorphism. The polymorphic methods can be part of the main class sequence or brought in using mixins for multiple inheritance situations. Here is a simple example using point and circle classes in a hierarchy.

 

p = .point~new(3,2)

c = .circle~new(,2,6)

say "A circle of radius" radius "centered at location ("||self~x","self~y")"

 

{{out}}

<pre>

Method binding in ooRexx is late and dynamic. In many situations, polymorphism can be achieved merely by providing an expected method. It is not necessary for an object to be of a particular class hierarchy. In the example below, both point and circle implement a print method, but there is no class relationship between these classes other than what they inherit from the object class.

 

p = .point~new(3,2)

c = .circle~new(,2,6)

say "A circle of radius" radius "centered at location ("||x","y")"

 

{{out}}

<pre>

A circle of radius 6 centered at location (0,2)

</pre>

 

=={{header|OxygenBasic}}==

 

A compact format for the methods is used to improve layout.

type tpoint float xx,yy

type tcircle float xx,yy,rr

 

del ca : del cb : del cc

 

=={{header|Oz}}==

No accessors because we use immutable public attributes ("features").

 

feat

x

")"}

end

 

=={{header|Pascal}}==

=={{header|Perl}}==

What polymorphic function means in the context of Perl is as clear as mud. subs already can take anything as parameter by default. Destructors are automatic, so I dropped them.

package Point;

use Class::Spiffy -base;

my $c = Circle->new(r => 4);

print $c->r, "\n"; # accessor autogenerated

 

=={{header|Phix}}==

Destructors are not required, though you can use delete_routine if needed.<br>

Copy constructors are also not required, a plain '=' will do just fine (it uses copy on write semantics).<br>

You could embed routine_ids in the structures to emulate virtual functions.<br>

There are no private members here; for that I would write something that returns integer ids to the outside world.

 

 

 

 

{{out}}

<pre>

</pre>

 

Point class definition.

 

class Point

{

}

}

 

Circle class definition.

 

class Circle extends Point

{

}

}

 

Usage:

 

$point = new Point( 1, 5 );

$circle = new Circle( 1, 5, 6 );

// or

echo $circle;

 

Will result in:

 

=={{header|PicoLisp}}==

# x y

 

 

(dm print> ()

P (new '(+Point) 3 4)

C (new '(+Circle) 10 10 5) )

 

(print> P)

{{out}}

<pre>Point 3,4

So it is enough to define classes and the print method.

 

define :class Point;

slot x = 0;

define :method print(p : Circle);

printf('Circle(' >< x(p) >< ', ' >< y(p) >< ', ' >< r(p) >< ')\n');

 

To test we can use the following code:

 

lvars instance1 = newPoint();

lvars instance2 = newCircle();

;;; Use print method

print(instance1);

 

=={{header|PureBasic}}==

Using the open-source precompiler [http://www.development-lounge.de/viewtopic.php?t=5915 SimlpeOOP].

BeginProtect

EndMethod

Testcode

*circle.Circle = NewObject.Circle

 

*circle\Print()

CloseConsole()

 

=={{header|Python}}==

For the print function, use the standard __repr__ methods, used when printing an object. Destructors are not needed of course.

 

def __init__(self, x=0.0, y=0.0):

self.x = x

def __repr__(self):

return '<Circle 0x%x x: %f y: %f radius: %f>' % (

 

Usage example:

 

Or, using inheritance like some of the other solutions:

def __init__(self, x=0.0, y=0.0):

self.x = x

def __repr__(self):

return '<Circle 0x%x x: %f y: %f radius: %f>' % (

 

Usage example:

The task calls for the creation of mutable types i.e. that you are allowed to change the values of x, y, or r of a Point or Circle after they have been created.

If this is not needed, then the Python namedtuple is a good way to create immutable classes with named fields such as these.

>>> class Point(namedtuple('Point', 'x y')):

def __new__( _cls, x=0, y=0 ):

p.x = 10.81

AttributeError: can't set attribute

 

And if you don't need default arguments, this becomes:

>>> Circle = namedtuple('Circle', 'x y r')

>>> Point(3, 4)

>>> Circle(x=1, y=2, r=3)

Circle(x=1, y=2, r=3)

 

=={{header|R}}==

Copy constructors are not needed, since objects are copied by value.

Neither are destructors needed (just use the rm function).

representation(

x="numeric",

cat("This is a circle, with radius", x@r, "and centre (", x@centre@x, ",", x@centre@y, ").\n")

})

 

=={{header|Racket}}==

"Fields" come with accessors and mutators for free.

 

#lang racket

(define point%

(define/override (custom-write out) (write (show) out))

(define/override (custom-display out) (display (show) out))))

{{out}}

<pre>

"<circle% <point% 3 0> 5>"

</pre>

 

=={{header|Ruby}}==

The <tt>Kernel#dup</tt> method can be used as a copy constructor.

 

attr_accessor :x,:y

def initialize(x=0, y=0)

"Circle at #{x},#{y} with radius #{r}"

end

Example:

puts Point.new # => Point at 0,0

p = Point.new(1, 2)

d.r = 7.5

puts c # => Circle at 4,5 with radius 6

 

=={{header|Seed7}}==

[http://seed7.sourceforge.net/manual/objects.htm Seed7 object orientation] works via interfaces.

Note that ''Circle'' inherits ''x'' and ''y'' from ''Point''.

Functions which return a ''Point'' respectively ''Circle'' are used as constructors.

Note that a Seed7 constructor does not need to have the name of the type (a new Point could be created with a function called abc).

Seed7 defines copy constructor, assignment and destructor automatically.

 

const type: GraphicObj is new interface;

graph := circ;

print(graph);

 

{{out}}

 

=={{header|Self}}==

 

We create four named objects, two of which we put in the traits namespace and two in a general prototype namespace. This would normally be done in the UI.

parent* = traits clonable.

printString = ('Point(', x asString, ':', y asString, ')').

center <- point copy.

r <- 0

 

=={{header|Sidef}}==

 

}

func pp(Circle obj) {

say "Circle at #{obj.x},#{obj.y} with radius #{obj.r}";

 

Example:

var p = Point(1, 2); # create a point

pp(p); # => Point at 1,2

d.r = 7.5; # and change the radius to 7.5

pp(c); # => Circle at 4,5 with radius 6

 

=={{header|SIMPOL}}==

The <code>embed</code> keyword in the type definition states that the type itself can be embedded in another type (any type can be placed as a reference in another type).

 

embed

integer x

 

result = p.print() + "{d}{a}" + c.print() + "{d}{a}"

 

[[SIMPOL]] does not currently have access to stdin, stdout, and stderr,

=={{header|Smalltalk}}==

Like Python and Ruby, these objects do not need to be related in order to have polymorphic methods.

instanceVariableNames: 'x y'

classVariableNames: ''

!Circle methodsFor: 'polymorphism test'!

print

 

''TODO: more idiomatic mechanism for presenting objects as strings.''

 

=={{header|Swift}}==

var x: Int

var y: Int

println(p.x) // 1

p.x = 8

 

=={{header|Tcl}}==

We only do so here for convenience.

In addition, Tcl's arguments to commands, procedures and methods are all fully polymorphic by default.

oo::class create Point {

variable X Y

foreach o $objects {

$o print

 

=={{header|Wollok}}==

class Point {

var x

c.print()

}

 

=={{header|zkl}}==

Written for brevity

fcn init(xyOrPoint=0,_=0){

if(Point.isInstanceOf(xyOrPoint)) set(xyOrPoint);

c:=Circle(1,2,3);

c.println(); c.center.println();

{{out}}

<pre>