⚠️ Warning: This is a draft ⚠️
This means it might contain formatting issues, incorrect code, conceptual problems, or other severe issues.
If you want to help to improve and eventually enable this page, please fork RosettaGit's repository and open a merge request on GitHub.
{{task|Data Structures}}
;Task: Define the data structure for a [[Linked_List#Doubly-Linked_List|doubly-linked list]] element.
The element should include a data member to hold its value and pointers to both the next element in the list and the previous element in the list.
The pointers should be mutable.
{{Template:See also lists}}
Ada
type Link;
type Link_Access is access Link;
type Link is record
Next : Link_Access := null;
Prev : Link_Access := null;
Data : Integer;
end record;
Using generics, the specification might look like this:
generic
type Element_Type is private;
package Linked_List is
type List_Type is limited private;
...
private
type List_Element;
type List_Element_Ptr is access list_element;
type List_Element is
record
Prev : List_Element_Ptr;
Data : Element_Type;
Next : List_Element_Ptr;
end record;
type List_Type is
record
Head : List_Element_Ptr; -- Pointer to first element.
Tail : List_Element_Ptr; -- Pointer to last element.
Cursor : List_Element_Ptr; -- Pointer to cursor element.
Count : Natural := 0; -- Number of items in list.
Traversing : Boolean := False; -- True when in a traversal.
end record;
end Linked_List;
In Ada 2005 this example can be written without declaration of an access type:
type Link is limited record
Next : not null access Link := Link'Unchecked_Access;
Prev : not null access Link := Link'Unchecked_Access;
Data : Integer;
end record;
Here the list element is created already pointing to itself, so that no further initialization is required. The type of the element is marked as ''limited'' indicating that such elements have referential semantics and cannot be copied.
Ada's standard container library includes a generic doubly linked list. The structure of the link element is private.
ALGOL 68
{{works with|ALGOL 68|Revision 1.}} {{works with|ALGOL 68G|Any - tested with release [http://sourceforge.net/projects/algol68/files/algol68g/algol68g-2.7 algol68g-2.7].}} {{works with|ELLA ALGOL 68|Any (with appropriate job cards) - tested with release [http://sourceforge.net/projects/algol68/files/algol68toc/algol68toc-1.8.8d/algol68toc-1.8-8d.fc9.i386.rpm/download 1.8-8d]}} '''File: prelude/link.a68'''
# -*- coding: utf-8 -*- #
CO REQUIRES:
MODE OBJVALUE = ~ # Mode/type of actual obj to be queued #
END CO
MODE OBJLINK = STRUCT(
REF OBJLINK next,
REF OBJLINK prev,
OBJVALUE value # ... etc. required #
);
PROC obj link new = REF OBJLINK: HEAP OBJLINK;
PROC obj link free = (REF OBJLINK free)VOID:
prev OF free := next OF free := obj queue empty # give the garbage collector a big hint #
'''See also:''' [[Queue/Usage#ALGOL_68|Queue/Usage]]
ARM Assembly
{{works with|as|Raspberry Pi}}
/* ARM assembly Raspberry PI */
/* structure Node Doublylinked List*/
.struct 0
NDlist_next: @ next element
.struct NDlist_next + 4
NDlist_prev: @ previous element
.struct NDlist_prev + 4
NDlist_value: @ element value or key
.struct NDlist_value + 4
NDlist_fin:
AutoHotkey
see [[Doubly-linked list/AutoHotkey]]
Axe
Lbl LINK
r₂→{r₁}ʳ
0→{r₁+2}ʳ
0→{r₁+4}ʳ
r₁
Return
Lbl NEXT
{r₁+2}ʳ
Return
Lbl PREV
{r₁+4}ʳ
Return
Lbl VALUE
{r₁}ʳ
Return
BBC BASIC
{{works with|BBC BASIC for Windows}}
DIM node{pPrev%, pNext%, iData%}
Bracmat
link=(prev=) (next=) (data=)
C
struct link { struct link *next; struct link *prev; void *data; size_t type; };
C++
C++ has doubly linked list class template in standard library. However actual list noded are treated as implementation detail and encapsulated inside list. If we were to reimplement list, then node could look like that:
template <typename T> struct Node { Node* next; Node* prev; T data; };
C#
class Link { public int Item { get; set; } public Link Prev { get; set; } public Link Next { get; set; } //A constructor is not neccessary, but could be useful public Link(int item, Link prev = null, Link next = null) { Item = item; Prev = prev; Next = next; } }
Clojure
This sort of mutable structure is not idiomatic in Clojure. [[../Definition#Clojure]] or a finger tree implementation would be better.
(defrecord Node [prev next data]) (defn new-node [prev next data] (Node. (ref prev) (ref next) data))
Common Lisp
(defstruct dlist head tail) (defstruct dlink content prev next)
See the functions on the [[Doubly-Linked List]] page for the usage of these structures.
D
A default constructor is implicit:
struct Node(T) { T data; typeof(this)* prev, next; } void main() { alias N = Node!int; N* n = new N(10); }
Delphi
struct Node(T) { type pList = ^List ; list = record data : pointer ; prev : pList ; next : pList ; end; }
E
This does no type-checking, under the assumption that it is being used by a containing doubly-linked list object which enforces that invariant along with others such as that element.getNext().getPrev() == element. See [[Doubly-Linked List#E]] for an actual implementation (which uses slightly more elaborate nodes than this).
def makeElement(var value, var next, var prev) {
def element {
to setValue(v) { value := v }
to getValue() { return value }
to setNext(n) { next := n }
to getNext() { return next }
to setPrev(p) { prev := p }
to getPrev() { return prev }
}
return element
}
Erlang
Using the code in [[Doubly-linked_list/Definition]] the element is defined by:
new( Data ) -> erlang:spawn( fun() -> loop( Data, noprevious, nonext ) end ).
Fortran
In ISO Fortran 95 or later:
type node
real :: data
type(node), pointer :: next => null(), previous => null()
end type node
!
! . . . .
!
type( node ), target :: head
F#
type 'a DLElm = { mutable prev: 'a DLElm option data: 'a mutable next: 'a DLElm option }
Go
type dlNode struct { string next, prev *dlNode }
Or, using the [http://golang.org/pkg/container/list/#Element container/list] package:
import "container/list" var node list.Element // and using: node.Next(), node.Prev(), node.Value
Haskell
Haskell in general doesn't have mutability so the following 'mutator' functions use lazy evaluation instead.
Note that unlike naive pointer manipulation which could corrupt the doubly-linked list, updateLeft and updateRight will always yield a well-formed data structure.
data DList a = Leaf | Node (DList a) a (DList a) updateLeft _ Leaf = Leaf updateLeft Leaf (Node _ v r) = Node Leaf v r updateLeft new@(Node nl _ _) (Node _ v r) = current where current = Node prev v r prev = updateLeft nl new updateRight _ Leaf = Leaf updateRight Leaf (Node l v _) = Node l v Leaf updateRight new@(Node _ _ nr) (Node l v _) = current where current = Node l v next next = updateRight nr new
==Icon and {{header|Unicon}}==
Uses Unicon classes.
class DoubleLink (value, prev_link, next_link)
initially (value, prev_link, next_link)
self.value := value
self.prev_link := prev_link # links are 'null' if not given
self.next_link := next_link
end
J
As discussed in [[Doubly-linked_list/Definition#J]], doubly linked lists are antithetical to J's design. Defining individual elements as independent structures is even worse. Now each element of the list must contain three arrays (everything in J is an array), all so that we can implement a list.
Yo Dawg, we heard you like lists, so we put lists in your lists so you can list while you list.
Nevertheless, this is doable, though it necessarily departs from the definition specified at [[Doubly-linked_list/Definition#J]].
coclass'DoublyLinkedListElement'
create=:3 :0
this=:coname''
'predecessor successor data'=:y
successor__predecessor=: predecessor__successor=: this
)
Here, when we create a new list element, we need to specify its successor node and its predecessor node and the data to be stored in the node. To start a new list we will need a node that can be the head and the tail of the list -- this will be the successor node for the last element of the list and the predecessor node for the first element of the list:
coclass'DoublyLinkedListHead'
create=:3 :0
predecessor=:successor=:this=: coname''
)
Java
{{works with|Java|1.5+}}
{
private T element;
private Node<T> next, prev;
public Node<T>(){
next = prev = element = null;
}
public Node<T>(Node<T> n, Node<T> p, T elem){
next = n;
prev = p;
element = elem;
}
public void setNext(Node<T> n){
next = n;
}
public Node<T> getNext(){
return next;
}
public void setElem(T elem){
element = elem;
}
public T getElem(){
return element;
}
public void setNext(Node<T> n){
next = n;
}
public Node<T> setPrev(Node<T> p){
prev = p;
}
public getPrev(){
return prev;
}
}
For use with [[Java]] 1.4 and below, delete all "
JavaScript
Inherits from LinkedList (see [[Singly-Linked_List_(element)#JavaScript]])
function DoublyLinkedList(value, next, prev) { this._value = value; this._next = next; this._prev = prev; } // from LinkedList, inherit: value(), next(), traverse(), print() DoublyLinkedList.prototype = new LinkedList(); DoublyLinkedList.prototype.prev = function() { if (arguments.length == 1) this._prev = arguments[0]; else return this._prev; } function createDoublyLinkedListFromArray(ary) { var node, prev, head = new DoublyLinkedList(ary[0], null, null); prev = head; for (var i = 1; i < ary.length; i++) { node = new DoublyLinkedList(ary[i], null, prev); prev.next(node); prev = node; } return head; } var head = createDoublyLinkedListFromArray([10,20,30,40]);
Julia
{{works with|Julia|0.6}}
abstract type AbstractNode{T} end struct EmptyNode{T} <: AbstractNode{T} end mutable struct Node{T} <: AbstractNode{T} value::T pred::AbstractNode{T} succ::AbstractNode{T} end
Kotlin
// version 1.1.2 class Node<T: Number>(var data: T, var prev: Node<T>? = null, var next: Node<T>? = null) { override fun toString(): String { val sb = StringBuilder(this.data.toString()) var node = this.next while (node != null) { sb.append(" -> ", node.data.toString()) node = node.next } return sb.toString() } } fun main(args: Array<String>) { val n1 = Node(1) val n2 = Node(2, n1) n1.next = n2 val n3 = Node(3, n2) n2.next = n3 println(n1) println(n2) println(n3) }
{{out}}
1 -> 2 -> 3
2 -> 3
3
=={{header|Modula-2}}==
TYPE
Link = POINTER TO LinkRcd;
LinkRcd = RECORD
Prev, Next: Link;
Data: INTEGER
END;
Nim
type Node[T] = ref TNode[T] TNode[T] = object next, prev: Node[T] data: T
=={{header|Oberon-2}}==
MODULE Box;
TYPE
Object* = POINTER TO ObjectDesc;
ObjectDesc* = (* ABSTRACT *) RECORD
END;
(* ... *)
END Box.
MODULE Collections;
TYPE
Node* = POINTER TO NodeDesc;
NodeDesc* = (* ABSTRACT *) RECORD
prev-,next-: Node;
value-: Box.Object;
END;
(* ... *)
END Collections.
Objeck
class ListNode {
@value : Base;
@next : ListNode;
@previous: ListNode;
New(value : Base) {
@value := value;
}
method : public : Set(value : Base) ~ Nil {
@value := value;
}
method : public : Get() ~ Base {
return @value;
}
method : public : SetNext(next : Collection.ListNode) ~ Nil {
@next := next;
}
method : public : GetNext() ~ ListNode {
return @next;
}
method : public : SetPrevious(previous : Collection.ListNode) ~ Nil {
@previous := previous;
}
method : public : GetPrevious() ~ ListNode {
return @previous;
}
}
OCaml
Imperative
type 'a dlink = { mutable data: 'a; mutable next: 'a dlink option; mutable prev: 'a dlink option; } let dlink_of_list li = let f prev_dlink x = let dlink = { data = x; prev = None; next = prev_dlink } in begin match prev_dlink with | None -> () | Some prev_dlink -> prev_dlink.prev <- Some dlink end; Some dlink in List.fold_left f None (List.rev li) ;; let list_of_dlink = let rec aux acc = function | None -> List.rev acc | Some{ data = d; prev = _; next = next } -> aux (d::acc) next in aux [] ;; let iter_forward_dlink f = let rec aux = function | None -> () | Some{ data = d; prev = _; next = next } -> f d; aux next in aux ;;
# let dl = dlink_of_list [1;2;3;4;5] in iter_forward_dlink (Printf.printf "%d\n") dl ;; 1 2 3 4 5 - : unit = ()
Functional
The previous implementation is the strict equivalent of the other examples of this page and its task, but in regular OCaml these kind of imperative structures can be advantageously replaced by a functional equivalent, that can be use in the same area, which is to have a list of elements and be able to point to one of these. We can use this type:
type 'a nav_list = 'a list * 'a * 'a list
The middle element is the pointed item, and the two lists are the previous and the following items. Here are the associated functions:
let nav_list_of_list = function | hd::tl -> [], hd, tl | [] -> invalid_arg "empty list" let current = function | _, item, _ -> item let next = function | prev, item, next::next_tl -> item::prev, next, next_tl | _ -> failwith "end of nav_list reached" let prev = function | prev::prev_tl, item, next -> prev_tl, prev, item::next | _ -> failwith "begin of nav_list reached"
# let nl = nav_list_of_list [1;2;3;4;5] ;; val nl : 'a list * int * int list = ([], 1, [2; 3; 4; 5]) # let nl = next nl ;; val nl : int list * int * int list = ([1], 2, [3; 4; 5]) # let nl = next nl ;; val nl : int list * int * int list = ([2; 1], 3, [4; 5]) # current nl ;; - : int = 3
Oforth
Complete definition is here : [[../Definition#Oforth]]
Object Class new: DNode(value, mutable prev, mutable next)
Oz
We show how to create a new node as a record value.
fun {CreateNewNode Value}
node(prev:{NewCell _}
next:{NewCell _}
value:Value)
end
Note: this is for illustrative purposes only. In a real Oz program, you would use one of the existing data types.
Pascal
type link_ptr = ^link; data_ptr = ^data; (* presumes that type 'data' is defined above *) link = record prev: link_ptr; next: link_ptr; data: data_ptr; end;
Perl
my %node = ( data => 'say what', next => \%foo_node, prev => \%bar_node, ); $node{next} = \%quux_node; # mutable
Perl 6
role DLElem[::T] {
has DLElem[T] $.prev is rw;
has DLElem[T] $.next is rw;
has T $.payload = T;
method pre-insert(T $payload) {
die "Can't insert before beginning" unless $!prev;
my $elem = ::?CLASS.new(:$payload);
$!prev.next = $elem;
$elem.prev = $!prev;
$elem.next = self;
$!prev = $elem;
$elem;
}
method post-insert(T $payload) {
die "Can't insert after end" unless $!next;
my $elem = ::?CLASS.new(:$payload);
$!next.prev = $elem;
$elem.next = $!next;
$elem.prev = self;
$!next = $elem;
$elem;
}
method delete {
die "Can't delete a sentinel" unless $!prev and $!next;
$!next.prev = $!prev;
$!prev.next = $!next; # conveniently returns next element
}
}
Phix
In Phix, types are used for validation and debugging rather than specification purposes. For extensive run-time checking you could use
enum NEXT,PREV,DATA
type slnode(object x)
return (sequence(x) and length(x)=DATA and <i>udt</i>(x[DATA]) and integer(x[NEXT] and integer(x[PREV]))
end type
But more often you would just use the builtin sequences. See also [[Singly-linked_list/Element_definition#Phix|Singly-linked_list/Element_definition]].
Memory is automatically reclaimed the moment items are no longer needed.
PicoLisp
We use (in addition to the header structure described in [[Doubly-linked list/Definition#PicoLisp]]) two cells per doubly-linked list element:
+-----+-----+ +-----+-----+ | Val | ---+---> | | | ---+---> next +-----+-----+ +--+--+-----+ | prev <---+
With that, 'cddr' can be used to access the next, and 'cadr' to access the previous element.
(de 2tail (X DLst)
(let L (cdr DLst)
(con DLst (cons X L NIL))
(if L
(con (cdr L) (cdr DLst))
(set DLst (cdr DLst)) ) ) )
(de 2head (X DLst)
(let L (car DLst) # Get current data list
(set DLst (cons X NIL L)) # Prepend two new cons pairs
(if L # Unless DLst was empty
(set (cdr L) (car DLst)) # set new 'prev' link
(con DLst (car DLst)) ) ) ) # otherwise set 'end' link
# We prepend 'not' to the list in the previous example
(2head 'not *DLst)
For output of the example data, see [[Doubly-linked list/Traversal#PicoLisp]].
PL/I
define structure
1 Node,
2 value fixed decimal,
2 back_pointer handle(Node),
2 fwd_pointer handle(Node);
P = NEW(: Node :); /* Creates a node, and lets P point at it. */
get (P => value); /* Reads in a value to the node we just created. */
/* Assuming that back_pointer and fwd_pointer point at other nodes, */
/* we can say ... */
P = P => fwd_pointer; /* P now points at the next node. */
...
P = P => back_pointer; /* P now points at the previous node. */
Pop11
uses objectclass;
define :class Link;
slot next = [];
slot prev = [];
slot data = [];
enddefine;
PureBasic
Structure node
*prev.node
*next.node
value.i
EndStructure
Python
class Node(object): def __init__(self, data = None, prev = None, next = None): self.prev = prev self.next = next self.data = data def __str__(self): return str(self.data) def __repr__(self): return repr(self.data) def iter_forward(self): c = self while c != None: yield c c = c.next def iter_backward(self): c = self while c != None: yield c c = c.prev
Racket
(define-struct dlist (head tail) #:mutable)
(define-struct dlink (content prev next) #:mutable)
See the functions on the [[Doubly-Linked List]] page for the usage of these structures.
REXX
REXX doesn't have linked lists, as there are no pointers (or handles).
However, linked lists can be simulated with lists in REXX.
╔═════════════════════════════════════════════════════════════════════════╗ ║ ☼☼☼☼☼☼☼☼☼☼☼ Functions of the List Manager ☼☼☼☼☼☼☼☼☼☼☼ ║ ║ @init ─── initializes the List. ║ ║ ║ ║ @size ─── returns the size of the List [could be a 0 (zero)]. ║ ║ ║ ║ @show ─── shows (displays) the complete List. ║ ║ @show k,1 ─── shows (displays) the Kth item. ║ ║ @show k,m ─── shows (displays) M items, starting with Kth item. ║ ║ @show ,,─1 ─── shows (displays) the complete List backwards. ║ ║ ║ ║ @get k ─── returns the Kth item. ║ ║ @get k,m ─── returns the M items starting with the Kth item. ║ ║ ║ ║ @put x ─── adds the X items to the end (tail) of the List. ║ ║ @put x,0 ─── adds the X items to the start (head) of the List. ║ ║ @put x,k ─── adds the X items to before of the Kth item. ║ ║ ║ ║ @del k ─── deletes the item K. ║ ║ @del k,m ─── deletes the M items starting with item K. ║ ╚═════════════════════════════════════════════════════════════════════════╝
/*REXX program implements various List Manager functions (see the documentation above).*/
call sy 'initializing the list.' ; call @init
call sy 'building list: Was it a cat I saw' ; call @put "Was it a cat I saw"
call sy 'displaying list size.' ; say "list size="@size()
call sy 'forward list' ; call @show
call sy 'backward list' ; call @show ,,-1
call sy 'showing 4th item' ; call @show 4,1
call sy 'showing 5th & 6th items' ; call @show 5,2
call sy 'adding item before item 4: black' ; call @put "black",4
call sy 'showing list' ; call @show
call sy 'adding to tail: there, in the ...' ; call @put "there, in the shadows, stalking its prey (and next meal)."
call sy 'showing list' ; call @show
call sy 'adding to head: Oy!' ; call @put "Oy!",0
call sy 'showing list' ; call @show
exit /*stick a fork in it, we're all done. */
/*──────────────────────────────────────────────────────────────────────────────────────*/
p: return word(arg(1), 1) /*pick the first word out of many items*/
sy: say; say left('', 30) "───" arg(1) '───'; return
@init: $.@=; @adjust: $.@=space($.@); $.#=words($.@); return
@hasopt: arg o; return pos(o, opt)\==0
@size: return $.#
/*──────────────────────────────────────────────────────────────────────────────────────*/
@del: procedure expose $.; arg k,m; call @parms 'km'
_=subword($.@, k, k-1) subword($.@, k+m)
$.@=_; call @adjust; return
/*──────────────────────────────────────────────────────────────────────────────────────*/
@get: procedure expose $.; arg k,m,dir,_
call @parms 'kmd'
do j=k for m by dir while j>0 & j<=$.#
_=_ subword($.@, j, 1)
end /*j*/
return strip(_)
/*──────────────────────────────────────────────────────────────────────────────────────*/
@parms: arg opt /*define a variable based on an option.*/
if @hasopt('k') then k=min($.#+1, max(1, p(k 1)))
if @hasopt('m') then m=p(m 1)
if @hasopt('d') then dir=p(dir 1); return
/*──────────────────────────────────────────────────────────────────────────────────────*/
@put: procedure expose $.; parse arg x,k; k=p(k $.#+1); call @parms 'k'
$.@=subword($.@, 1, max(0, k-1)) x subword($.@, k); call @adjust
return
/*──────────────────────────────────────────────────────────────────────────────────────*/
@show: procedure expose $.; parse arg k,m,dir; if dir==-1 & k=='' then k=$.#
m=p(m $.#); call @parms 'kmd'; say @get(k,m, dir); return
'''output'''
─── initializing the list. ───
─── building list: Was it a cat I saw ───
─── displaying list size. ───
list size=6
─── forward list ───
Was it a cat I saw
─── backward list ───
saw I cat a it Was
─── showing 4th item ───
cat
─── showing 6th & 6th items ───
I saw
─── adding item before item 4: black ───
─── showing list ───
Was it a black cat I saw
─── adding to tail: there, in the ... ───
─── showing list ───
Was it a black cat I saw there, in the shadows, stalking its prey (and next meal).
─── adding to head: Oy! ───
─── showing list ───
Oy! Was it a black cat I saw there, in the shadows, stalking its prey (and next meal).
Ruby
Extending [[Singly-Linked List (element)#Ruby]]
class DListNode < ListNode attr_accessor :prev # accessors :succ and :value are inherited def initialize(value, prev=nil, succ=nil) @value = value @prev = prev @prev.succ = self if prev @succ = succ @succ.prev = self if succ end def self.from_values(*ary) ary << (f = ary.pop) ary.map! {|i| new i } ary.inject(f) {|p, c| p.succ = c; c.prev = p; c } end end list = DListNode.from_values 1,2,3,4
Rust
Simply using the standard library
use std::collections::LinkedList; fn main() { // Doubly linked list containing 32-bit integers let list = LinkedList::<i32>::new(); }
=== The behind-the-scenes implementation ===
Doubly linked lists present a problem in Rust due to its ownership model. There cannot be two mutable references to the same object, so what are we to do? Below are the relevant lines (with added comments) from the std implementation ([https://doc.rust-lang.org/std/collections/struct.LinkedList.html Documentation] [https://github.com/rust-lang/rust/blob/master/src/libcollections/linked_list.rs Source]).
The standard library uses the (currently) unstable Shared<T> type which indicates that the ownership of its contained type has shared ownership. It is guaranteed not to be null, is variant over T (meaning that an &Shared<&'static T> may be used where a &Shared<&'a T> is expected, indicates to the compiler that it may own a T) and may be dereferenced to a mutable pointer (*mut T). All of the above may be accomplished in standard stable Rust, except for the non-null guarantee which allows the compiler to make a few extra optimizations.
{
head: Option<Shared<Node<T>>>,
tail: Option<Shared<Node<T>>>,
len: usize,
marker: PhantomData<Box<Node<T>>>, // Indicates that we logically own a boxed (owned pointer) Node<T>>
}
struct Node<T> {
next: Option<Shared<Node<T>>>,
prev: Option<Shared<Node<T>>>,
element: T,
}
Sidef
var node = Hash.new( data => 'say what', next => foo_node, prev => bar_node, ); node{:next} = quux_node; # mutable
Tcl
{{eff note|Tcl|list}} {{works with|Tcl|8.6}} or {{libheader|TclOO}}
oo::class create List { variable content next prev constructor {value {list ""}} { set content $value set next $list set prev "" if {$next ne ""} { $next previous [self] } } method value args { set content {*}$args } method next args { set next {*}$args } method previous args { set prev {*}$args } }
Visual Basic .NET
Public Class Node(Of T)
Public Value As T
Public [Next] As Node(Of T)
Public Previous As Node(Of T)
End Class
zkl
class Node{
fcn init(_value,_prev=Void,_next=Void)
{ var value=_value, prev=_prev, next=_next; }
fcn toString{ value.toString() }
}
a,b:=Node(1),Node("three");
a.next=b; b.prev=a;
println(a.next," ",b.prev);
{{out}}
three 1
{{omit from|ACL2}} {{omit from|TI-83 BASIC}} {{omit from|TI-89 BASIC}}